Cognition, 9 (1981) 197-236 @Elsevie Sequoia S.A., Lausanne - Printed in The Netherlands
tical description versusconfigurational airrangemeint in languageacquisition: The casei! of ciausesinlJa KENJI HAKUTA** Yale University
Abstract The description of sentence complexity in terms of grammatical relations or in terms of configurational properties of their surface structure is a basic issue in developmental psycholinguistics. The problem has been investigated in English, but these studies provide little insight because of’ the peculiar properties of word order in the language. A series of experiments with Japanese children is reported in which the configurational demands of the sentence are shown to be the critical factor.
Sentences can be described with respect to the grammatical :roles of their constituents, or in terms of the linear arrangement of the words. The extent to which the parameters specified by these altezrative concepti:Dns constrain children’s language is important in determinin’g the nature of the preparedness with which hum: ns begin the task of language acquisition. .Although the question arises in all :lspects of investigation in Fyntactic development, it has perhaps been most explicitly asked in the study of relative clauses. In this paper, I will first provide a general characterization of relative clause structures. This will be followed by a brief review of studies on relative clause comprehension with children in English, .vvhere it will be shown tha.t the evidence is inconclusive on both empirical rrnd theoretical grounds. The argument between views emphasizing grammatical description versus linear *This researchwas supported by the Peter B. Livingston Fellowship Fund of the HarvardMedical School, NSF Grants BNS7349150 and GSOC7349150 to Dr. Roger Brown, and a &sertation fund from the Departmentof Psychology and Social Relations, Harvard University. I would like to thank Roger Brown, Jill and Peter de Villiers, Haj Ross, Helen Tager-Flusberg, Steven Pinker,ar*dtwo armnymous reviewersfor helpful comments and discussion. Even more, I would like to thank my mother, Emiko Hakuta, for assistance in collecting the data and for general wisdom. FinaBy;t hanks ax:: due to Mr. Yokoyama, Ms. Asami, Ms. Toi, Ms. Abe, and other staff of the day care center wh.erethis research was conducted. **Requests for reprints should be sent to Kenji Hakuta, Department of Psychology, Yale University, BOX 11A Yale Station, New Haven, CT 06520, U.S.A.
198
Kenji Hakuta
arrangement is best resolved in languages where word order is flexible, one such language being Japanese. Thus, a series of experiments with Japanese children will be reported in which the two variables are independently manipulated. Across all experiments, in both comprehension and production, the surface configurational properties of sentences are shown to be the critical parameter. Loglc& possibilities of sentences with relative clauses Relative clauses consist of a head noun and a relativized sentence. The head noun ;&B a co-referent within the relativized sentence from which,. linguistically speaking, it has been extracted. The head noun, placed external to the sentence, can come in two positions, either before or after the relativized sentence, sketched as follows:
(a) .(bj
HEAD NOUN [ RELATIVIZED SENTENCE] 1RELATIVIZED SENTENCE 1 HEAD NOUN
S$nce these complex noun phrases are embedded within sentences, a description of how they mteract with the matrix sentence is also called for. Greenberg (1963) has noted that an overwhelming majority of the world’s langrlages place the subject before the object in the basic underlying order. This is a strictly statistical generalization, with exceptions such as Tagalog (Schachter, 19763, but for present purposes, we will assume it be a universal fact, This leaves gs with three logically possible basic orders: Subject-VerbObject (SVO), Subject-Object-Verb (SOV). and Verb-Subject-Object (VSO). These are in fact very common basic orders among the world’s languages. Since the matrix role of the complex noun phrase can be either Subject or Object (Subject Matrix and Object Matrix), the interaction of the language type @WO/SOV/VSCB)with the matrix role (Subject Matrix/Object Matrix) aud head noun position (HNIIPS] /[RS] HN) creates a variety of different sciztence configurations, appearing in Table 1. The sentence configurations irt which the [RS] element is flanked on both sides by other constituents are the center+mtedded sentences, while those with [RS] at the beginning or the end of the string are referred to as left-branching or right-branching. As it turns out, the vvorld’s languages are not evenly distributed in the grid. The ma@ity of SOV languages have the IRS] HN head noun position, while VSO languages have the HN[RS] pa&ion (Greenberg, 1963 ; Lehmann, 1973). ILuno (1974) gives an account of this universal, derived from the fact
199
Rektive clause,s
Table 1.
Logical possibilities of sentence configurations as a fuwtion of the basic order of the knguage, the matrix role of the complex noun phrase, ,and the position of the head noun Basic order
HePd noun position 3 IRS1
WIN
-
svo
Subject Matrix Object Matrix
hJIRS]-V-N N--.V-N[RS]
[RS]N-V-24 N-V-(RSJ I!
sov
Subject Matrix Object Matrix
IV[RS] -N-V N-N[RS]-V
N-[RS]N-yf
Subject Matrix Object Matrix
V-4$ RS] -N V-N-N[ RS]
V-IRS] N--N V-N-[RS] N
VSO
Table 2.
Matrix role
[RS] &N--V
Log&al possibilitiesof complex noun phrase conj?guration,.z as a jiwtion of the basic order of the ikngwge and the position qf the head (iroun Basic order
Head noun position
Focus
3 IRS1
[RSI N ~VOls_ ;sv] 0
svo
Subject Focus Object Focus
fy w gw
sov
Subject Focus Object Focus
gov1
vw
yw
Subject Focus Object Focus
y w q w
WI 0 [Ws_ [WO
VSO
s_
that centerembedding &aces the comprehensibility of sentences (e.g., Miller, 1962; N. Chomsky, 1961) Thus, the tendency for languages is to seek the head nou.n position that would result in a lower prob;:tbility of ending up with center-embedded structures. Antinucci ef al. (1979) also provide a perceptually-based explanation for facts about the diaqhroiltic change in relative clause structure. In addition to a consideration of the relative clause position in the matrix sentence, however, a thorough investigation must explore the implication of the string of elements within the relativized clause and its intt;?raction with matrix role. Assuming that the basic ord.er of the sentence is prelserved in the relative clause, SVO, SOV and VSO will yield different configurations depending on (1) whether the subject or the object noun phrase is relativized, and (2) whether the head noun comes before or after the relativized clause. The logical possibilities resulting from this combination are outlined in Table
200
Kenji Hakuta
Table 3.
Clmfgumtions of nouns and verbsasa fin&ion of the baste order of the lang.uge, the position of the head noun, the matrixrole and head noun focus Head noun position
Basic order
Sentence type
sv3
9s so 06 00
Ij[ WV] -V-N l$NY] -V-N N-V+ VN] N -V-I’$V I’]
[ EV] N-V-N [NV] 1-V-N N-V- 1VI+/]i N-V-[NV] hJ
SOV
SS sell OS 00
I$VV] --N-V N[NV] .-N-V N-N[NV] -V N-N[NY] -V
[NV]Ij-N’ [NV] 5-N-V N-(NY] N-V N-[NV] Ir_I-V
VSO
ss so OS 00
V-N[ VA’] -N V-I’J VW -N y N-N[F’A’] V-N+ Vlv]
V-[ UV] N-N V-i VA’] N-N V-N- [ VA’] I’j V-N- [ VA’] t
WI rl
2. There are two n0teworth.y interactions in Table 2. First, the order of the noun and verb within the relztive clause depends on the interaction between the basic order of the language and the focus. Thus, for SVO languages, the order for subject focus is [ VN 1, but it is [NV] for object focus. On the other hand, for SOV and VSO languages, the order remains constant: [NV1 for SOV languages and [ VN] for VSO languages. Second, whether the order of subject preceding the object is preserved or not depends on the interaction between focus and the position of the head noun. For all three basic language orders, the subject-beforeobject order is preserved in subject focus when the head noun is to the left of the relative clause. The order is preserved in object focus when the head noun is to the right of the relative clause. Now, putting together the two logical possibilities for embeddedness as a function of the matrix ro1.e and focus, we end up with Table 3. The point of the table is that a variety of sentence configurations are possible for any given sentence type depending on the basic order of the language and the position of the head noun. It serves to place in context the present line of research to be described, which was intended to help clarify issues raised by research conducted on the four sentence configurations in the uppermost left comer of the table, representing English. The following section describes the logic underlying the studies of relative clause comprehension in Englishspeaking children, and the results obtained.
Relative&uses
201
Relative clauses in English-speaking children English is an SVO language, with the head noun of relative clauses on the left. The following are examples of the four iogically possible sentence types in English (where the first letter refers to the grammatical role of the complex noun within the matrix sentence, and the second letter refers to the role of the head noun within the relative clause): (SS) (SO) (OS) (00)
The The The “nhe
duck [that licked the frog] bit the pig. duck [that the frog licked] bit the pig. duck licked the frog [that bit ,the pig]. duck licked the frog [that the pig bit] .
These four sentence types have been the center of some controversy recently over what the variables are that determine their psychological complexity for children. The controversy exists at both the theoretical and empirical levels. Theoretically, the explanations that have been proposed can be divided into two distinct classes depending on their assumptions. One class bases itself on the assumption that the grammatical description of the sentences is the important variable determining their relative complexity. This view takes the grammatical role of the variables involved in these sentences, namely the matrix role and the focus, to be the determining factor. The surface representation of these sentences is seen as not relevant. The second class of explanations homes in exactly on the factors considered irrelevant by the grammatical description viewpoint, namely the configurational properties of constituents in the surface structure of sentences. Under this view, the grammatical description of a sentence is seen as a useful summary of the ingredients that go into yielding configurational features of the sentence. While theories based on grammatical descriptions would make predictions for cross-linguistic validation that are universal for the four sentence types, theories based on the configurational properties of sentences would make predictions that are based on an interaction between these grammatical descriptions and the typological characteristics of particular languages as outlined in Table 3, since this interaction determines the configurational properties of the sentences. On the empirical level as well, there has been an alarming discrepancy in the results obtained across the several studies in English. I will describe below the nature. of both the theoretical and the empirical controversy and show how the issue can be resolved, naturally, through an investigation of Japanese. Sheldon (1974), who is the major proponent of the view emphasizing grammatical description as the variable underlying complexity, has argued
202
Kenji HakWa
that the primary determinant of complexity for these sentences is whether the grammatical functions of the two variables that define them are the same or different. Under She&ion’s Parallel Function Hypothesis, sentences are easier to comprehend when the grammatical function of the head noun withti the relative clause and the grammatical function of the complex noun phrase within the matrix sentence are the same than when they are different. Thus, SS and CO sentences should be easier than OS and SO. Apparently, the theoretical motivation underly,ing Parallel Function is that it constitutes a significant linguistic generalization not just for relative clauses, but also for pronominal co-reference and coordination reduction (Sheldon, 1974). Another reasonable prediction that assumes thke importance of the grammatical description of the sentences is based on Keenan and Comrie’s (1972) accessibility hierarcYly of noun phrases from relative clauses. After surveying over 40 languages, the authors report a hierarch.y in which languages allow relativization of noun phrases. The hierarchy is: Subject, Direct Object, Indirect Object, Object of Preposition, Possessive Noun Phrase, and Object of Comparative Particle. Reading this hierarchy from left to right, any language that allows accessibility of a given noun phrase in the hierarchy will also allow accessibility of ah noun phrases to the left of it. Thus, if a language allows an Object of Preposition to be relativized, then it will also allow Subject, Direct Object, and Indirect Object Some languages, such as Tagalog, only allow the Subject. This putative universal assumes an important role for developnrent;lll psycholinguistrcs if, as some have speculated, universal rules should appear earliest in children’s language (e.g., Ross, 1973; McNeill, 1966). Keenan (1975) has expressed his own interest in the relevance of his hierarchy for adult langzeage performance by investigating the relative frequencies of the vtious members of the Accessibility Hierarchy in written English. It wouid be a ratural extension of Keenan’s investigation to see if children wiri fmd reiativ:: clauses more difticult as they mlove down the hierarchy. With reference to the four sentence types mentioned earlier, then, the AccesGbihty Hypothesis would predict the Subject Focus sentences. (SS, OS) to be easier than theeObject Focus sentences (SO, 00). There are several competing explanations for the relative coq.Zexity of these sentences that take into account their configurational properties to varying extents. One account (Slobin, 1973), consonant with data from spontaneous speech production and imitation of relative clauses by children, considers sentences with interruption of the main clause to be more difficult than those without interruption. Thus, the center-embedded SS and SO should be more dlifficult than 6’5; and 00, which are right-branching. The difficulty with center-embedded sentences is presumably predicted owing to the heavy load placed on memory by the interruption of the main clause. As
Relativeclause3
203
Sheldon (1974) points out, however, most studies on center-embedded sentences with adults concerned multiple center-embedded structures, which are considerably more difficult than single center-embeddings. However, it is possible that for children, even a single center-embedding would caus,c difficulty. This will be called the Embeddedness Hypothesis. Smith (1974) proposes an explanation that is yet closer to the configurational properties of each sentence. His hypothesis is based on whether the sentences correspond to Bever’s (1970) Noun-Verb-Noun (NVN) strategy, in conjunction with the Minimal Distance Principle (MDP) (C. Chomsky, 1969). The MDP was originally proposed by Chomsky to account for childrn’s comprehension of sentences with complementizers, but can be extended to relative &uses where constituents are “missing”. The MDP’claims that when children find a missing noun phrase, they will assign the most recent noun phrase to its location Smith’s NVN/MDP Hypothesis predicts that OS should be easiest since NVN applies to the initial segment, and the MDP assigns the second noun of that sequence as subject of the remaining VN, yielding the correct interpretation. On the other hand, NVN correctly interprets the initial segment of SS, but MDP incorrectly assigns the second noun as the subject of the remaining VN. The NVN strategy correctlly interprets the initial segment of 00 sentences as well, but MDP does no:! apply. Since MDP mis-applies for SS and fails COapply to 00, it is not possible to predict differences between the two, but both should be more difficult than OS. And finaliy, the most difficult should be SO, since neithler NVN nor MDP can apply. Tavakolian (1978) proposes what I see as an even more local explana”lion based on the configurational properties of the sentences. She predicts that children will interpret these sentences as if they were conjoined. Her predictions work well for SS and OS, both of which have the configuration NVNVN. These configurations’ could be interpreted by the child as having the conjoined st_ructure NVN-and-VN. For SS, the sentence The igzmna that bit the tortoise kissed the pigeon, would be perceived by the child as The iguana bit the tortoise and kissed the pigeon. For SS, the sentence X”zesquirret hit the lizard that kissed the snail would be perceived as The squirrel Ilit the lizard and kissed the snail. A conjoined interpretation would yield the correct response for SS but would cause an erroneous interpretatiuln of OS. Tavakolian does not make explicit the implications of conjoined clause analysis for the other two configurations SO and 00, her only claim b’eing that the conjoined clause analysis would not apply well to these sentences. The data relevant to the predictions outlined ablove can be found 2r studies by H.D. Brown (197 I), Sheldon (1974), Tavakolian (1978) and de Villiers et &. (1979). With the exception of Brown’s study, the studies used an
204
Kenji Hakuta
act-out comprehension procedure using toy animals. Brown used a picturecued comprehension paradigm, where the child was asked to choose one of two possible pictures depicting the appropriate action. The results of each study are displayed in Fig. 1, with Subject and Object Focus sentences plotted separately, and the Matrix Role on the abscissa. The dependent measure is the percent of total score. For the act-out comprehension studies, since a breakdown of the response types was available in the published tables, the percentage of correct acting out of individual clauses out of ,the total number of clauses (two per sentence) was calculated for each sentence type. For H. D. Brown’s (1971) study, the measure is what he reports, namely the percentage of correct picture choice, for which chance level is 50%. Figure 1.
I-
Sumnzuy of comprehension data from four English studies on SS, SO, OS and 00. Closed circles represent subject focus, open circles represent object focus. Sentence type is determined in conjunction with matrix role represented on abscissa.Measure for H_ Ag,Brown’s study is per cent of correct choice of pxture, where chance is 3’”/o. For other studies, measure is per cent of in&vidual clauses acted out correctly. Adapted from H. D. Brown (1971), Sheldon (19741, Tavakolian(1978)and de Villers et al (1979). Brown
Sheldon
L
1ocl
80
‘Y
II
IUATRIX
ROLE
de VilDiers Jg!
Relative chuses
205
Altbough at first glance the results across the studies appear consistent, a close examination reveals that the only common finding is that SO is poorly comprehended. This result can be construed as evidence for all the hypo-. theses outlined above: it is not parallel function, it is object rather than subject focus, it is centerembedded, it is not amenable to NVN and MDP, and it cannot be interpreted as a conjoined structure ! Leaving SO aside, the following observations can be made: 1.
2. 3.
SS is easier than OS in Brown (1972), Sheldon ( 1974) and Tavakolian (1”978), but they are equivalent in de Villiers et al. (1979). 00 is easier than OS in Sheldon, OS is easier than 00 in Brown, and they are about equivalent in Tavakolian and in de Villiers et al. SS is easier than 00 in Brown and in Tavakolian, but they are equivalent in de Villiers et al. and in Sheldon.
The fact that such i.nconsistencies exist across these studies is quite disturbing, to say the least. Sheldon, Tavakolian and de Villiers et al. all used identical methodology and highly similar lexical items, in fact all used animals. The subject popuiations appeared comparable. The Parallel Function Hypothesis finds strong support in Sheldon’s own data, since she finds equally good comprehension of SS and 00, and poor comprehension of OS and SO. The data of Tavakolian and cf de Villiers et al., however, are not particularly supportive. In both these studies, there is roughly equal performance on OS as on 00. In addition, in Tavakolian’s data, the discrepancy between SS and 00 is very large, whic’“nParallel Function cannot explain. H. D. Brown’s data are quite contradictory to Parallel Function’s predictions, where OS is better comprehended than 00. The Embeddedness Hypothesis is embarassed by the fact that in all but the de Villiers et ul. data, SS is better than OS. This also embarasscs the NVN/MDP Hypothesis, which makes the same prediction. However, the latter’s redeeming feature is that it explicitly predicts poor performance on SO. This prediction is supported by all studies. The Conjoined Clause Hypothesis predicts successfully that SS would be better comprehended than OS, supported in Bro\un, Sheldon and Tavakolian. It does not make any predictions about the relative difficulties of SO and 00, however. Given such a conflicting array of data, surely no single hypothesis will account for all the data. However, a reasonable account might be formulated from the Conjoined Clause Hypothesis, which successfully accounts for SS being superior to OS, and the NVN/MDP Hypothesis, which accounts for SO being the most poorly comprehended. The Conjoined Clause Hypothesis was
206
Kenji Hakuta
originally formulated by : !. D. Brown (1971). He wrote, “the fact thalt [SSI was apparently easier than [OS] may be explained by the fact that i.n [SSI the fiisi subject can also correctly act as the subject of the second1verb, as in a sentence with a single subject and a compound verb, making it easier for the child to perceive the relationship expressed” (pp. 193 l-2). Tavakolian (1978) saw the implications of this in accounting not just for good performance on SS but also for the errors in interpreting OS. The Ccnjoined Clause Hypothesis accounts well for the superiority of SS over OS. It might be described in more general terms as the child automatical$ assimilating a given input sentence into a structure that s/he already knoiJs. It so happens that the conjoined clause is a good interpretation for a string with the structure NVNVU, :or English, at least. When it comes to 00 and SO, however, the conjoined clause is not a good match with the sentence configurations, and there is no reason to expect the child to force such an interpretation on these canfigurations. In the case of 00, the sequence NVNNV can be interpreted as an initial NVN segment with an additional NV remaining. We would expect the child to perform quite well on this sentence because the initial NVN sequence is straightforward. For the remaining NV sequence, given that the child correlctly interprets the fragment as agent-action, the chance of finding the correct patient is 50%. This is the same argument as would be made for SS and OS, except that in those cases, the missing noun phrase is the subject of a sequence VN. If there were no conjoined clause interpretation, 00 should be on an equal footing with SS and OS. iiowever, the conjoined clause interpretation pushes SS upwards and OS downwards on the scale of probability for correct responses, leaving 00 in between. And fmally, for SO, no existent interpretation is readily available to the child, and so we do not expect good performance. This explanation, which we wil! call the Configuration Hypothesis, is a more general hypothesis of which Tavak&an’s Conjoined Clause Hypothesis and Smith’s NVN/MDP Hypothesis might be considered special cases. The Configuration Hypothesis essentially predicts the same results as Parallel Function, except that it distinguishes between SS and 00 and between OS and SO, while Parallel Function does not. But even if the two hypotheses were not confounded in English, it would not be terribly meaningful to pursue the validity of these hypotheses with further studies until some account of why the data across the various studies are so inconsistent with each other. It is entirely possible that, had a double-blind procedure been used, we would not be worrying over the problem presently. The Parallel Functiorr faQs under the general category of explanations that looks towards the grammatical description of the sentences as the variables
Relative clauses
207
predicting sentence complexity. The Configuration Hypothesis, on the other hand, seeks to explain sentence complexity in terms of the configurational properties of the sentence, which is a function of the parameters of language described at the beginning of this section. Referring back to Table 3, it is possible ‘to see how, by looking at other languagzs, the two hypotheses can be unconfounded. I am proposing to determine the correct account for English by testing the assumptions underlying the hypotheses in other languages The study of both SOV and VSO languages is Iimportant. Taking a VSO language with the head noun to the left of the relative clause (bottommost left corner of the table), OS and 00 are right-branching while SS and SO are center-embedded. Now, taking an SOV language with head noun on the right of the relative clause, SS and SO are now left-branching while OS and 00 are center-embedded. Notice that the sentence type that ends up with the center-embedded configuration shifts depending on the basic order of the language. For VSO, SS and SO are center-embedded, while for SOV, OS and 00 are center-embedded. Thus, if grammatical description of sentences were important in predicting sentence complexity, then crosslinguistically, sentence type should have an effect regardless of the particlllar configuration that the given type takes in a given language. On the otner hand, if configuration were important, one expects variation of complexity for a given sentence type across languages depending on the configurat:.on that it takes, such as center-embeddedness. It turns out that English is a particularly bad language from the viewpcint of teasing apart the appropriateness of grammatical description and sentence configuration. There is an inherent confounding of thj=se two descriptions because of the rigid word order required by English to signal the grammatical functions of constituents. Given a grammatical description, the configuration of the sentence: is determined. The present study unconfounds these explanations by looking at an SOV language, Japanese.
Relative clauses in Japanese: experiment one Unlike English, where word order determines the grammatical role of nouns in a sentence, Japanese signals grammatical role through postpositional particles. Thus, although the predominant, canonical order of a sentence is Subject-Object-Verb (SOV), that order is free in the sense that word order is not essential for assigning grammatical role. The majolr constraint on word order in Japanese is that the main verb must be sentence!-final (Kuno, 1973), but even this constraint can be violated through dislocation of a constituent to the right of the verb.
208
Kenji Hakuta
Table4.
Bmplex Japanese sentences with relative &uses in both SOVand OSY orders Order
sentence
We SS
so OS 00
-_
SOV
osv
(No V] fi-gaN-o V (N-ga V] E-ga N-o V Nga [A’+-~ N-o V N-ga [IV-gu VI If-0 V
N-o[N-o VJb&a V N-o [N-gu V] fi-gaV [N-o v] 8-oN-&i V [IV-gu V] Eu-0N-gaV
Linguists working in Japanese have not explicitly worked out the full implications for meaning of the word order change, ributin general, it is considered an optional rule analogous to extraposition in English. N. McCawley (I 976) formulates the word order change in terms of Ross’s (1967) “scrambling rule” for Win, which simply interchanges the position of noun phrases under the condition that it is postcyclic. One important function of the scrambling rule in Japanese, as N. McCawley (1976) points out, is to make multiple centerembedded sentences comprehensible. Since Japanese is a language with tlexible word order, it allows for more than one configuration for each sentence with a given grammatical description. Thus, tak!&g the middle row of the right column in Table 3 representing Japanese, each of the four sentence types which are represented in the SOV order there can take the OSV crder, which yields a different configuration. Thus, within a language like Japanese, it is possible to unconfound grammatical description from sentence configuration. It is ideal when these factors can be separated out within a language rather than across languages, since languages can differ in so rnanJr ways other than the critical variable one is attempting to isolate. The possible sentences in Japanese are listed in Table 4. The configurations on the .A column are the same as .those representing SOV/[RS ] N in Table 3, with particles added. On the right column are the configurations that result when the sentences on the left column are i.1 the OSV order. It should be pointed out that the only syntactic cue indica!:ing relativization in Japanese is the constituent order, since there are no relatii; _ Japanese. There is generally a slight pause following the head noun,’ IJ@I --@?Gz L in I am not aware of my systematic attempts to outline such prosodic cues in Japanese. The grammatical functions of the different nouns in a sentence containing a compIex noun phrase are signalled, as in simple sentences, through particles. The particle on the head noun of the relative clause signals the role of the complex noun phrase within the matrix sentence. Thus, the
Relative clauses
209
role of the head noun within the relative clause is signallcd solely by the particle on the noun phrase within the relative clause. If that noun is marked by -ga, then the head noun is the object. If it is marked by -9, then the head noun is the subject. The sentences in Table 4 have interesting properties that help unconfound tnc variables of grammatical description and sentence configuration. If grammatical description were important, one would not expect there to be any variation in performance as a function of the different configurations that each sentence type takes. On the other hand, if configurational factors were important, then we would expect differential performance for each sentence type w;th respect to the configuration. Specifically, OS/SOV, OO/SOV, SS/ OSV and SO/OSV are center-embedded, and they all have a sentence-initial NNV sequence which mimics the simple sentence and could lead to an erroneous interpretation of the sentence. The NNV sequence for the respective sentences take the following forms: OS/SOY
oo/sov SSlOSV
SO/OSY
N-ga N-o V... N-ga N-ga N-Q/V-•
V.. . V...
N-o N-ga V . . .
As can be readily seen, VS/SOV mimics the SOV simple active sentence, while SO/OSV mimics the OSV simple active sentence, Previous experiments with simple sentences (Hakuta, 1977) showed that children found the SOV order easier to comprehend than OSV. Thus, the prediction was that the tendency to choose the fast noun as the agent would be strongest for OS/ SW and weakest for-SO/OSV. In addition, for all the sentences, it was predicted that the first two nouns would be erroneously interpreted as involved in a relationship, and that the trend would be towards interpreting the first noun as the agent in all &hesentences. The left-branching sentences, SS/SOV, SO/SOV, OS/OSV and OO/OSV do not present possibilities for such erroneous interpretations. Thus, a prediction based on considerations of configurational properties’of the sentences predicts superior performance on the left-branching over the center-embedded sentences. Sheldon’s Parallel Function Hypothesis would predict bti:iter performance on SS and 00 over OS and SO, independent of whether they Em=m&K)V or OSV order. The hypothesis could tolerate a main effect TLr iord order, but it would be embarassing to the hypothesis if there were an interaction between sentence type and the word order of the sentence. More generaliy, such an interaction would be problematic for any theory of complexity based on the grammatical description of the sentences.
210
Ktnji Hakruta
In addition to the above predictions, it was hypothesized that there would still be a main effect :forword order. SOV should overall be better comprehended than OSV, as was found for the simple sentences. Final@, the Accessibility Hierarchy Hypothesis can be tested by assessing whether Subject Focus sentences are better comprehended than Object Focus sentences. It provides a particularly critical test of the hypothesis since in Japanese, Object Focus takes the [ SV] 0 order while the Subject Focus takes the [OV] S order. Subjects Subjects were 12 children between the ages 5;3 and 6;2. In all experiments reported in this paper, they were in a public day care center in Tokyo, Japan. Materials and procedwe
Sixteen sentences were constructed from factorial c::ombinations of two levels of Matrix Role (Subject/Object), two levels of Focus (Subject/Object j, two levels of Word Order (SOV/OSV), and two replications. The nouns were all animals (e.g., alligator, guriha, camel, panda) randomly assigned to sentence frames. Verbs were ones which required distinct actions (e.g., kicked [ketta], hit [butta], licked [nameta]). For eLch sentence, the child was asked to (act out the actions on a stage ,331 which t!-re animals were placed. The experimenter read the sentences at a sJow, natural speed with short pauses following both the subject and the object of the matrix sentence. The experimenter, my mather, wars blind to the hypotheses of the study, to the extent that mothers can be expected to remain blind to the devious schemies of their children.
Overall ana@& was conducted by a 4-way ANOVA with Subjects crossed with the repeated measures factor, Sentence, nested within a cross of Matrix Holie (Subject/Clbject) by Focus (Subject/Object) by Word Order (SOV/ OSV)‘. None of the main effects proved significant (for Word Order, F’ < atrix Role, F’[l, 21 = 1.96O,p> O.lO;forFocus, F’[l, 81 = 1.744, ‘Following C&&s (1973) suggestion, both subjects a& sentences are treated as random effects, using procedtnresoutlined iz Wines (1971). Qua&Ratios (P’) are reported for ANOVAS,and individual ~~~~par%ons are made thrc~~gh 2’. Whenevermeans are rqorted, there are two starxiarddeviations. S, refixs to 4& MtJabilityacwa subjects, while S, refers to vkabili~y across sentences.
Relative clauses
Figure 2.
211
Si’ifican~ Matrix Rule by Word Order inreraction.
subJect m&+x
osv
5011 ORDER
p > 0.10). There was a highly significant Matrix Role by Word Order interaction (F’[ 1, 51 = 3 1.492, p < 0.005). The 3-way interaction of Matrix Role
by Focus by Word Order (F’ [ 1,2 ] = 3.109) approached the alpha level of 0.10, for which the critical value is 3.78. The significant Matrix Role by Word Order interaction is shown in Figure 2. Since sentences with Matrix Role Subject are left-branching when in the SOV order but center-embedded when in the OSV order, and the reverse is the case for Matrix Role Object sentences, the interaction indicates that Japanese children find center-embedded sentences more difficuit to process than left-branching sentences. This result is a rather embarassing fact for any theory which subscribes soleiy to the grammatical description of sentences .as predictors of sentence complexity. Thus, the Parallel Function Hypothesis, without major modification in which sentence configuration is incorporated, is untenable. AnalyGs of individual responses revealed that poor performance on the center-embedded sentences was mostly due to the erroneous interpretation of the initial NNV sequence as a simple sentence, as was predicted. Table 5 shows the breakdown of the frequency of responses involving the fast two nouns for each of the four sentence forms. When the response involved the fiit noun being the agent of the action, it was coded as “l-2”‘, and appears as such in the Table. When the second noun was the agent, it was coded as “2-l”. Since there were two responses for each sentence form per child, each cell has a possible total of 24. Looking at the column totals, it is evident that most responses across the sentence forms involved interpreting the NNV
212
Kmji .Yakuta
Table 5.
4ving the first two nouns in the center-embedded Frequencies of responm in,,., sentences Response
N-ga N-,0 ‘V . . . os/sov .-+
W.ga iv-ga V . . . oo/sov
N-o N-o V... SSlOSV
l-2 2-l
23 1
14 8
15 7
14 6
Tos?l
24
22
22
20
“l-2” refers tr* responses where first noun is the agent, “2-l” is the agent.
Table 6.
N-o N-ga V... so/osv
to winere second noun
Distribution of responses on thesubordinateclause in left-branching sentences Object focus
Subject focus
Response
~~O/OSV
ss/sov
os/osv
so/sov
l-2 2-l
11 11*
11 11*
16* 4
18* 2
Total
22
22
20
20
k-sponse type “l-2” refers to the fust noun being the agent, “2-1” noun is the cogent. Correct responses are marked by an asterisk (*).
where the second
seo,uence as a sentence. Within each sentence form, children appear to overlvhehhinrgly use the l-2 response in OS/SOV, since the sequence exactly mimics a simple SUV sentence. There was no tendency for children to prefer t&e 2-l response in SO/OSV which mimics a simple:!OSV serrience, but this is not surprising given that children were shown in the previous experiment to have difficulty in comprehending the OSV order. Thus, the.:e errors indicate that chimren have a tendency to interpret the first noun a:!,the agent, and this 3s enhanced for the sequence which exactly mimics the canonical SOV simple sentence. S:‘nceJapanese children m&segment the center-embedded sentences, these sentei-ia:esdo not constitute an appropriate test of the effect of Focus and of WOE Order, since the relatk*re clause k not properly parsed in these senternary -One can make specific tests of these effects in the left-branching sentr,,lrCP!!.The question of whether Subject Focus or Object Focus relative &~.!s are easier to comprehend can be answered by inspection of the numbei of ;oriect responses in the sttbordinate clause for SQ’SOV and OS/OSV,
Aelative chases
Table 7.
213
Frequencies of responses on the main clause of left-branching sentences Response
Subject-Object-Verb
Object-Subject-Verb
ss/sov
so/sov
os/osv
oo/osv
2-3 3-2
15* 6
17* 2
7 10*
8 11.*
Total
21
19
17
19
.-.-
“2-3” refers to responses where the second noun of the sernte,nceis the agent, “3-2” where the third noun is the agent. Asterisk (*) indicates correct respmse.
which are Subject Focus, and comparing this with SO/SOV and OO/OSV, which are Object Focus. Response type “2- 1” is correct for Subject Focus, and “1 -2” is correct for Object Focus. The Accessibility Hierarchy Hypothesis predicts Subject Focus to be better comprehended than Object Focus. Based on our earlier experiments with comprehension of simple sentences, however, an explanation based on the order of the subject and object predicts Object Focus to be easier, since it takes the SVO order. A breakdown of the number of the response types with respect to the: sentence forms appears in Table 3. As can be readily seen, there were more correct responses on the Object Focus than on Subject Focus. This suggests that. the order of constituents is the major factor. The effect of Word Order can best be assessed through comparison of SS,/SOV and SO/SOV with OS/OSV and OO/OSV on the number of correct main clauses acted out, since these sentences are all left-branching. Table 7 displays the breakdown of the number of correct and reversed interpretations of the main clause for each sentence form. As can be seen, there were more correct responses on the SOV order than on the OSV order, suggesting the operation of the SOV Constraint. The effects of Word Order and of Focus will be tested explicitly in the next experiment. A study by Harada et al. (1976) confirms the main findings of this experiment. The researchers tested act-out comprehension of SS, SO, OS and 00 in the SOV order in one group of children, and in OSV in a different group of children. Thus, Word Order as a main effect was a between-subjects variable, while the other variables of Matrix Role and Focus were repeated measures. The researchers do not report statistical tests performed on their data, but their subjects, ranging in age from 3 to 10, clearly found centerembedded sentences to be more difficult than left-branching sentences within the SOV and within the OSV group. It is of interest to note that even at age
IQ, chiIdre3 have difficulty with center-embedded sentences, with less than 50% of the responses being correct.
This frxperiment tested children between 5;s and 6;2 in their comprehension of SS, SO, OS and 00 ser;tences in the SOV and OSV orders. The results shc;v an interaction between the Matrix Role of the complex noun phrase snc the Word Order of the sentence. Left-branching sentences were better comprehended than center+mbedded sentences. Thus, the grammatical destiption of the sentences did not account for the differential performance of AiIdren on the various sentence types as a function of whether they were in the SOV or OSV order. This result is a major embarassment to theories, such as ParaIlei Function, that predict the psychological complexity of these ntences in terms of thek grammatical description. It appears that the configumtionaI properties of the sentences constitute the variables to be considered in accounting for comprehension of these sentences. There was a marginal effect for Word Order within the left-branching sentences, where the SOV sentences were easier than the OSV sentences. In addition, rolative clauses whose head noun role was the object tended to be easier than those with subject head noun. However, even for the older age group, these sentences were inordinately difficult, and younger children were not tested. Eqhnmt
two
The mevious experiment demonstrated that 5 and 6 year old children ;“md left-branching sentences easier to comprehend than center-embedded sentences. Thus, SS and SO types were easier when they were in the SOV order, but OS and CD were easier when they were in the OSV order. However, even for these chifdren, the sentences proved quite difficult and the sentences required modification if younger children were to be tested. The moditication chosen for this experiment was to change one of the clauses into an intransitive action, such as c_xy.When the intransitive action .is the subordinate clause, only the SS and OS types are possible, and since the main clause is transitive, both these types can be either in the SOV or OSV orders. lhs, for these sentences, it is posssible to test the effect of Sentence Type (SS or OS) md the effect of Word Order (SOV or QSV). The four sentence conf~uations are: SS/SOV [CRZED] AGENT-ga PATZEN KZCKELI F23jdSV PATIENT-0 [CRZED] AGENT-g4 KZCXED.
Relative clauses
215
OS/SOV AGENT-ga [CRIED] PATIENT-o KICKED. OS/OSV ECRIED] PATIENT-o Al GENT-ga KICKED. As in the previous experiment, a main effect for Sentence Type would indicate the importance of the grammatical description, while an interaction between Sentence Type and Word Order would indicate the importance of sentence configuration A main effect for Word Order, with SOV better than OSV, would indicate an extension of the SOV Constraint operating on simple active sentences. When the main clause is the iintransitive action, it yields two sentence type?: SS and SO. The effect of Word Order and its interaction with Sentence Type cannot be tested with these sentences since word order change can apply only to transitive actions, and the main verb in these sentences is intransitive. However, with these sentences, we can test the effect of Focus, since that parameter now varies. Thus, these sentences present a critical test of thle Accessibility Hypothesis. The two sentences are: ss so
KICKED] AGENT-ga CRIED. [ AGENT-ga KICKED 1 PA TIENT-ga CRIED. [PATIENT-o
If Keenan and Comrie’s accessibility hierarchy ha$ any implications for processing difficulty of sentences in children, we woultl expect to find the effect here, with better performance on SS than on SO. On the other hand, if camprehension of reltiive clauses were dependent on the linear order of elements, then we would expect superior performance on SO over SS, since SO follows the Subject-Verb-Object order, while SS has the Object-Verb-Subject order. Subjects Subjects were 36 children divided evenly into three age groups: Group II: 3;3-4;2, Group III: 4;3-5;2, and Group IV: 5;3-6;2. Materials and procedure The six structures described above were each replicated twice, creating a total of 12 sentences for presentation to each child. The pool of words from which the nouns and transitive verbs were selected was the same as in the previous experiments. The intransitive verbs were: cried (naita), yawned (akubi-shita), coughed (kushami-shita), and laughed (waratta). A second set of sentences was created with the same intransitive verb but with the nouns interchanged. The procedure was identical to that of previous experiment;.
216
Kenji Hakuta
scori?lg
A two-point scoring system was used, with one point given for correct performance on the transitive action and one on the intransitive action. Thus, for each sentence, a score of 0, 1, or 2 was possible.
The data were aralyzed separately for the four structures in which the subordL-ate clause was titransitive and the two structures in which the main clause was intransitive. The two analyses will be reported sequentially. For the structures in which the subordinate clause was intransitive, a 3way ANOVA was performed, with Subjects nested within Age and the repeated measures factor, Sentence, nested within a cross of Type @S/OS) by order (SOV/OSV). Age did not turn out to be significant (F’ [2,9] = 1.763), nor did it interact significantly with any of the repeated measures factors. The only significant repeated measures factor was the interaction of T;rpe by Word Order (F’[ 1,2j = 22 1.457, p < 0.005). The means involved in tie 2-way interaction appear in Table 8. Inspection of the table reveals that the interaction was due to good performance on SS/SOV and OS/OSV, and poor performance on SS/OSV and OS/S@V. This result strorngly sup ports the view that predicts sentence complexity on the basis of sentence configuration. Japanese children fmd center-embedded sentences difficult to comprehend. That there was no main effect for Type (respective means for SS and OS were 1.334 and 1.389) shows that the grammatical descrip tion of the sentences was not a good predictor of complexity. In addition, the fact that there was no main effect for Word Order (xpective means for Table 8.
iM&ns mrd s&n&& dev&iims for i!)pe by Word order h@mctiim Ordex
TYpe
soy
OSV
s
f= 1.639 SA, ‘i: OAW sl., = 0.039
t= 1.028 SA., = 0.358 s.d., = 0.039
OS
z= 1.069 rd., = 0.381 rd., = 0.059
3 = 1.70’8 !#A., = OM3 SA., = 0.059
Relative clauses
Table 9.
Breakdown of frequency of response types for SSjSOV, SS/OSV, OSjSOV and OS/OSV
SS/SQV: [Vi ] NW N-CD V -. Transitive
SS/OSV: N-O [Vi] N-p V -
Intransitive +
Total
Transitive
-
-+ -
4-P 13
6 0
53* 13
Total
60+
6
66
Intransitive
Total
+
-
+ -
13* 10
36 10
49* 20
Total
23*
46
69
.-
OS/SOV: N-m [Vi] N-OV Transitive
gJS/OSV: [Vi] N+ N-9 V
Entransitive
Total
Transitive
+ -
Intransitive
Total
+
f
Total
217
12* 0
47 4
59* 4
+ _-
55* 5
6 1
61* 7
12*
51
63
1 otal
61*
4
68
The subordinate chses are intransitive. “+” indicates correct response, ‘*-‘I indicates error. Correct responses are marked by an asterisk (*).
SOV and OSV were 1.354 and 1.368) suggests that the SOV Constraint may not be operative for complex sentences. This will be discussed further after an analysis of the errors. A breakdown of the responses into how well the children comprehended the transitive and intransitive components of the sentences is relealing. This analysis excludes those responses for which only one clause was acted out, although there were very few such cases. Responses can be ciassified into four categories:
(1) (2)
(3) (4)
Transitive Transitive Transitive Transitive
correct, intransitive correct; correct, intransitive error; error, intransitive correct; error, intransitive error.
Table 9 gives the frequency of the four types of response for each of t5e structures. As can be seen from the column and row marginals, there was very little difference across the structures with regard to the transitive clause,
218
Kenji Hakuta
the major difference being with respect to the intransitive action. The leftbranching sentences, SS/SOV and OS/OSV, showed good comprehension of the intransitive clause, but there were more errors than correct responses for the center-embedded structures SS/OSV and OS/SOV The fact that the intransitive clause was center-embedded1 caused the children to erroneously assign the first noun of the sentence as the subject of the intransitive verb. This result is exactly what a left-to-rig.ht sentence processing model, such as the Augmented Transition Network (e.g., Kaplan, 1979) would predict in the form of “garden path” sentences (Wanner et al., 1974). The fact that there is no apparent preference for SOV over OSV in the comprehension of these sentences is not surprising once we realize that Word Order for these sentences refers to the order of subject and object in the matrix sentence. Consider the left-branching sentences with the configuration V,NNV, where Vi is the intransitive verb. As can be seen in Table 9, most responses for SS/SOV and OS/OSV correctly assigned the first noun to the Vi, Thus, the first noun is indeed the subject, although not of the action of the main clause. In addition, if we assume that the noun assigned to the Vi is no longer the “fnst noun” for the processing of the remainder of the sentence, there is no reason why SS/SOV should be easier than OS/OSV. In fact, one would expect the OSV order to be easier since the second nou.n of the entire sentence in effect becomes the ‘“first noun” with respect to the main verb. When the fust noun is removed, the sequence SO V ends up as 0 V while OSV ends up as SV. A different line of experiments explicitly testing this possibibty shows superior performance on OSV over SOV orders (Hakuta, I978), but that is beyond our current scope. The major point of relevance for our present purpose :; to show that in comprehending these sentences, children appear to proces- :he sentences quite loc?Ily. The SOV Constraint may be applicable only wiien it is locally relevant, as in simple sentences, at least for comprehension. In the case of the center-embedded sentences, as Table 9 reveals, childre:n alnrost unanimously chose the first noun as the agent of Vi, erroneously. Thus once again, the first noun is the subject of the subordinate action. The SOV Constraint is not readily apparent in the context of their performance on the entire sentence because children process the sentences locally. Turning now to the SS and SO sentences in which the main clause was the intransitive zetion, the results were analyzed by a 2-way ANOVA with Subjects nested within Age and crossed with the repeated measures factor, Sentence, nested within Type (SS/SO). Once again, Age was not significant (F’I2,21 = 3.0941, nor was its interaction with Type. There was a significant main effect for Type (F’ [ 1,5 ] = 9.568, p < O.OS), and-inspection of the means revealed that children found the SO structure (X = 1.722, s.d., =
Relative clauses
Table 10. SS:
219
BreaMownof ,frequencyof response types for SS and SO
[N-o V] N-skiVi
-J-rgtg&uc
SO: [N-ga V] N-g Vi
Jntransitive
TOta1
Transitive
Jntrmsitive _--I-
+
Total
.v + __
32* 29
6 3
3a* 32
+
55* 5
13 3
68* 8
TC9M -I
61*
9
70
Tota:_,
60*
16
76
The main Aause is intransitive and the subordinate clause is transitive. “+*’indicates correct response, “-” indicates error. Correct responses are marked by an asterisk (*).
0.438, s.d.* = 0.039) easier than SS (x= 1.403, s.d., = 0.,428, s.d., = 0.098). A breakdown of the responses into the four types as was done for the other structures earlier, presented in Table 10, shows that the difference was mostly due to differential performance on the transitive action, namely the subordinate clause. This result does not support the hypothesis that Keenan and Comrie’s accessibility hierarchy would have implications for children’s sentence processing. It is more consistent with the view that the local order of elements within the complex noun phrase determines its ease of comprehension. The fract that subject focus is easier than object focus in English is best explained by the fact that the head noun in isnglish is located to the left of the relative clause, and thus a subject focus sentence results in an NV01 configuration while an object focus sentence results in an O[ SV] configuration. Since the head noun is on the right off the relative clause in Japanese, subject focus results in an [OV] S order .while object focus results in an [ SV] 0 configuration. We can derive a law which will predict ease of comprehension of subject and object focus in all languages: In languages where the head noun is on the left of tke relative clause, subject focus will be easier than object focus, whereas in languages where the head noun is on the right of the relative c.!ausc, object focus will be easier, ail other things being equal. Summary
This experiment simplified the sentences tested in the previous experiment by making one of the clauses intransitive. This reduced the number of sentence types tested. SS and OS were tested in SOV and OSV orders with the
220
KenjiHakuta
subordinate clause being intransitive, and SS and SO with the main clause intransitive. Subjects were between ages 3;3 and G;cL.A comparison of SS and OS in both word orders strongly confirmed the irk‘.eraction between Matrix Role and Word Order found in -tne earlier experi.r,..cnt. Sentences were much easier to comprehend when they were left-branching than when they were center-embedded. In the center-embedded senten.ces, most children made the error of interpreti:ng the initial NVi . . _ sequence as a unit. There was no main effect for Word Order, and thus the SOV Constraint does not appear in the context of the entire sentence. It was argued that children process the sentences at the local level. A comparison of SS and SO showed that Japanese children find the relative clauses with subject focus more difficult than those with object focus, contrary to the Accessibility Hierarchy Hypothesis,, The superior performance on object focus is attributable to an extension of the SOV Constraint to the NVN sequence.
Experiment three
IJp until this point, we have been using the terms “center-embedded” and “left-branching” as convenient Libels to capture the configurational propertiies of the sentences. This experiment will pursue the question of whether it is centerembeddedness per se that causes difficulty in understanding a sentencr:. or whether the resultant configuration is the important variab?e. In formulating the critical test, we will take advantage of the right-dislocated structure frequently found in colloquial Japanese. In these sentences, a noun phrase is placed to the right of the main verb of the sentence. Thus, a sentence with the canonical form AGENT-ga PATIENT-o BIT can be rightdislocated as AGENT-ga BIT, PATIENT-o or as PATIENT-o BIT, AGENTga. The main verb is marked in these structures by a terminal contour and, in this experiment, it was decided to mark the verb with the dimunitive - no, which is frequently used in child speech and speech to children at the eri:; (of verbs. Table 11 shows how right dislocation affects SS/SOV, SS/OSV, O?l/ SOV and OS/OSV sentences. Although it is possible to dislocate the first or the second noun phrase, in order to preserve the word order of the subject and object, the second noun of each sentence was right-dislocated. To maintain consistency in notation, although the right-dislocated sentences are now in the SVO and OVS orders, ‘I will refer to them as SOV and OSV. The original sentences !n Table 11, under the left column labelled “Matrix NNV”, are the common SOV and OSV forms. Thus, SS/SOV and OS/OSV are leftbranching while SS/OSV and OS/SOV are center-embedded and there is a stacking of nouns. Under the right column labelled “Matrix NVN” appear
Relative clauses
221
Table 11. Illustration of how n$ht dislocation affects sentence configuration Type/Order
Matrix NNV
Matrix NVN
SSlSOV ss/osv OS/SOV os/osv
[N-o V] &ga N-o V N-o [N-o V] Ega V N-ga [AJ-gaI’] b&oV [N-gu I’] N-o N-ga V
[N-o V] E-ga V, N-o N-o V, [N-o V] N_g: N-ga V, [A’-guI’] y-0 [N-ga Yj N-o V, N-ga
Confwrations under “Matrix NNV” are the standard order. Configurations under “Matrix NVN” are right-dislocated structures. Terminal contour at the end of the main verb is indicated by a comma (,).
the right-dislocated sentences. Since Japanese places the relative clause to the left of the head noun, the centerembedded sentences under Matrix NNV still remain center-embedded under Matrix NVN. However, in these sentence;, the nouns are unstacked such that there is nc longer any local mimicking of a simple sentence at the beginning of the sentences. In fact, under Matrix NVN, the four sentences have the identical configuration, NVNVN. Thus, the Matrix NVN sentences allow us to test whether centerembeddedness or configuration is the critical variable. If center-embeddedness presents difficulty to the child, then there should be a similar interaction between Sentence Type: and Word Order in the Matrix NVN sentences as there would be in the Matrix NNV sentences. On the other hand, if the configuration of the sentence is important, there should be IX interaction between Type and Word Order in the Matrix NVN sentences. Subjects
Subjects were eight children between ages 5;4 and 16;3. Materialsand procedure Sentences with both the main and subordinate clauses reversible were used for this experiment. Fortyeight sentences were constructed from ? factorial combination of two levels of Sentence Type (SS/OS), two levels of Word Order (SOV/OSV), two levels of Matrix Form (Matrix NNV/Matrix NVN), and four replications. In order to accommodate ,this large number of scntences, the pool of animals was increased. Sentences were grouped into four blocks, with each block having one example of each isentence. In addition, animals were also divided up according to these blocks. Presentation of
222
KenjTHukuta
blocks as well as of sentences within each block was randomized for each child. The animals in each block were introduced to the child immediately prior to the presentation of that block. This procedure helped maintain the child’s .interest through 48 sentences, and is highly recommended to other researchers wishing to present large numbers of sentences to young children. The entire procedure lasted about 30 minutes.
Responses were scored 1 point for each clause correct, resulting in a total possible mre of 2 for each sentence. R6milis md discussion Overall analysis was conducted by a 4-way ANOVA, in which Subjects were crossed with the repeated measures factor, Sentence, nested within a cross of Type (SS/OS) by Word Order (SOV/OSV) by Matrix Form (NNV/NVN). Main effects for Type and for Word Order were not significant (F’ < 1 ), as well as the main effect for Matrix Form (F’[ I,9 ] = 1.597, p < 0.25). The 2way interaction between Type and Word Order was significant (F’ [ 1,7] = 7.859, p < O.OS), as was the 3-way interaction of Type by Word Order by Matrix Form (F’[ 1,8] = 6.704, p < 0.05). The means, plotted ia Figure 3, Figure3.
Three-way i&ractim between Matrix Order by vpe by Word Order. Matrix
NNV
Matrix
ORDER
NVN
Relative clauses
223
reveal that the 2-way interaction was due entirely to the interaction of Type by Word Order within Matrix NNV, which also accounts for the S-way interaction. In fact, when separate ANOVAs were performed by Matrix NNV and Matrix NVN, the 2-way Type by Word Order interaction was sign&cant for Matrix NNV (F’ [ 1,s ] = 13.327, p < O-OS), and the mean square for Type by Word Order in Matrix NVN was zero. It appears that the 3-way interaction was due to better comprehension of OS/SOV and SS/OSV, the cznterembedded structures, in the Matrix NVN form, than in the Matrix NNV form. Sirce the predictions concern the variable of embeddedness (centerembedded or left-branching) and its relationship with the Matrix Form, the two sentence configurations representing each level of embeddedness within each Matrix Form were pooled. Thus, within the Matrix NNV structures, SS/ SOV and OS/OSV were pooled since they are both left-branching, and SS/ OSV and OS/SOV were pooled since they are both center-embedded. The same pairs were pooled within the Matrix NVN. Pooling of these data is justified on the grounds that it was planned a priori and, in addition, t’-tests comparing each pair of means to be pooled did not produce any differences that were even marginally significant. _The pooling of the data resulted in four means: LB(left-branching)/NtjV (X = 1.203, s.d., = 0.347, s.d.2 - 0.29 1); CE(centerembedded)/NNV (X = 0.750, s.d., = 0.189, s.d., = 0.259); LB/NVN (X = 1.125, s.d., = 0.3:!0, s.d., = 0.211); CE/NVN (2 = 1.125, s.d., = 0.284, s.d., = 0.259). Comparison of these means through t’-testsshowed that although LB was easier than CE within Matrix NNV (t’ [4] = 4.23 1, p < 0.02), there was no difference between LB and CE within the Matrix NVN (t’ < 1). Thus, there is no difference between LB and CE when we control for sentence configuration. In addition, the lack of difference between LB and CE within Matrix NVN is due to CE sentences being easier in the Matrix NVN than in the Matrix NNV form, as opposed to the other possibility, that LB might be more difficult in Matrix NVN than in Matrix NNV. This is demonstrated by the fact that while LB did not differ with respect to the two levels of Matrix Form, CE. sentences were significantly easier in Matrix NVN than in Matrix NNV (t’ [ 5 I = 3.487, p < 0.02). Individual analysis of the responses showed that seven out of the eight, children had better comprehension of CE when it was in Matrix NVN than Matrix NNV. Performance was better on Matrix NVN because of an increased comprehension of the subordinate clause, with a concommitant decrease in the garden-path type errors involving the Fist two nouns of GE/Matrix NNV sentences. The reader will recall that children erroneously segment the initial NNV ... sequence as a sentence in these structures. An “improvement score”
224,
Kenji Hakuta
was calculated for the subordinate and main clauses for the two CE sentences, which consisted c-;fthe number of correct responses for a given clause in Matrix NVN minus the corresponding number in Matrix NNV. For SS/ OSV, the improvement score for the subordinate clause was +12, while for the main clause it was -1. There was a reduction by 11 of the garden-path type errors. For OS/SW, the imprrjvement score for the subordinate clause was +I 1, while for the main Claus, it was +2. The reduction in garden-path type errors was 12. The results of this experiment show that the difficulty of center-embedded sentences in the Matrix NNV form can be removed through right dislocation, which prevents the stacking of nouns that results in garden-path errors. Centerembeddedness per se is not the critical variable in comprehension in children. The center-embedded sentences in the Matrix NVN form were no more difficult to comprehend than left-branching sentences. I &aim that center-embedded sentences tend to be difficult because they frequently cause stacking sf nouns and verbs which need to be re-ordered before the sentence can be interpreted. Such is the case in the English SO form, such as The donkey that the man kicked licked the mule. Center-embedded sentences are not difficult when the nouns are stacked such that they are separated out by verbs, as in the English SS: The donkey that licked the man kicked the mule. In Japanese as well, center-embedded sentenses are difficult when the stacking causes an erroneous assignment of nouns to verbs, as in the sentences with the NNVNV configuration. They are no more difficult than leftbranching sentences, however, when the stacking is “neatly” done, as in NVNVN. It is possible that the obligatory Terminal contour at the end of the main verb in right-dislocated sentences will cause them to be perceived as two left-branching sentences rather than as a center-embedded structure. Nevertheless, since the transitive action of the main verb involves both tbe noun to its left and the complex noun phrase to its right, they are still best considered as center-embedded. Summary
This experiment tested the hypothesis that center-embcddedness increases difficulty of sentences, which we have assumed so far, against the proposal that the stacking of nouns in such a way that they are interpreted erroneously would cause difficulty in comprehension. The second matrix noun of center-embedded sentences OS/SOV and SS/OSV and left-branching qentences SS/SOV and OS/OSV were right-dislocated, yielding a uniform unstacked configuration for both the left-branching and center-embedded sentences. Children between ages S ;4 and 63 were tested on comprehension of
Relutive clauses
225
reversible sentences created in this manner. The results show that centerembedded sentences are no more difficult than left-branching sentences when the nouns are unstacked. Thus, it is not center-embeddedness per st? that causes processing difficulty, but rather its frequently concornmitant stacking of constitut;n:s.
Experiment four The results from the studies in comprehension of complex sentences showecl that Japanese children have difficulty with sentence forms that take the conf&ration NNVNV. For this finding to be of any generality, it is important to ask the question as to whether similar difficulties are manifested with the configuration in other linguistic tasks. The present study looks at children’s immediate imitation and delayed production of these sentences. In immediate imitation, the prediction is that the NNVNV configuration should be more difficult to imitate than the NVNNV configuration, since the formelr is stacked and the latter is unstacked. In delayed production, the children will produce sentences in the NVNNV configuration and avoid the NNVNV configuration. The immediate imitation and delayed production data were obtained in the same task. Sentences in the immediate imitation phase were paired with pictures depicting the action. The same pictures were used to cue the delayed production of the sentences. The pictures were constructed such that they depicted actions described by SS and OS sentence types. In the SSpicture, the agent of a transitive action is shown performing another action, while in the OS-picture, the patient of a transitive action is shown performing another action. In the immedliate imitation phase, the model sentences came in either the: SOV or OSV order for each of the pictured types. Thus, there were four model sentence forms: SS/SOV, SS/OSV, OS/SOV, and OS/OSV. The a* sumptions underlying the immediate imitation task are that sentences which. are easier for th.a chilci will be more correctly imitated, and that errors will reflect changes zowards their preferred form of the sentences. In the delayed production phase of the task, within the rough constraint of producing sentences with relative clauses, the child should change the order of the sentences to the preferred form. The prediction is that if children find the NNVNV configuration dbfficult to produce, given that they have the choice of producing any given sentence in the SOV or OSV order, for the two pictured meanings SS and OS, they should choose the order which takes the configuration NVNNV, which is
226
Kmji Hakuta
unstacked. Thus, for the %-picture, there should be a tendency to produce the SOV order, while for the OS-picture, the OSV order should be preferred. However, since SOV is dominant in Japanese, there may be a tendency for children to produce stacked structures for OS, since it preserves the basic word order. Although word order did not appear as a main effect in the comprehension results, I argued that this was due to the fact that children processed the sentences locally. For this task, to the extent that children process the matrix sentence, one might expect an effect for word order in imitation and in delayed production. Subjects
Subjects were 36 children divided into three agp: groups: Group II: 3;9-4;8, Group III: 49-5;8, and Group IV: 5;9-6;8. Materials and procedure The pictures used as cues for the sentences contained three animals, of which
two were i’dentical. Two identical animals were pictured so that there was a natural reason for using relative clauses, highlighting their restrictive function. The third animal will be referred to as the unique animal. One of the identical animals was pictured performing either an intransitive (crying or yawning) or a transitive (eating an apple or holding a balloon) action. This action corresponds to the relative cla;ae. For the SS-pictures, this animal was pictured as the agent of a transitive action (hitting, kicking or pushing) with the unique animal. For the OS-picture, the animal was the patient of the transitive action. The %-pictures and the OS-pictures were balanced to evenly represent the relative clause actions. Two replications of each of the resulting eight combinations were created, ;~sing different animals. This produced a total of 16 pictures. An additional counterbalancing procedure involved an equal placement of the transitive action on the right and left part of the picture. The pictures were individually mounted on 8” X 11” construction paper. From each p.icture type, half of the pictures were chosen to be paired with a model sentence in the SOV order, and the other half in the OSV order. Witbin each order, half contained a relative clause with the intransitive verb and the other half with the transitive verb. A second set of sentences was created reversing the order of the sentences paired with each picture. Children were assigned randomly to either set. For each child, order of presentation of the picture/sentence pairs was randomized. Each child was introduced to the pictures, which were bound together with ring binders and made to resemble a story book, and intormed that
Relative claxses
227
4 i
s/he was to help tell stories to a pippet. The child was told to repeat the story just like the experimenter, always my mother, tells it. There were four warm-up pictures of simple actions. If s/he did not repeat the sentences verbatim, the .Md was instructed through modelling by the two experimenters, an example of mother and child interaction. Children encountered no difficulty with the task, with the exception of one child who insisted on telling her own story about the pictures. She was not included in the experiment. The immediate imitation phase of the experiment immediately followed. This part of the experiment lasted about 15 minutes. Following completion of this phase, the child was told that it was now his/her turn to tell the story, ;and that the experimenter would not say anything. Children in general readily took to this task. The pictures were shown in the same order as in the immediate imitation phase. The delayed production phase also lasted about ‘15 minutes. No child failed to complete the task, and in fact many wanted to do more. Thus, the task was intrinsically interesting to the children. The entire session was recorded on a cassette tape recorder and subsequently transcribed. Scoring The response on the immediate imitation phase was coded as correct (1) or
incorrect(O), and in addition, errors were coded if the wrong particle was used, and if there was a transposition of word order. The delayed production data were recorded as being in the SOV or OSV order if they contained a relative clause. If there was an error in the use of particles, this was separately noted. In addition, a separate category was created for those responses in which the child produced a conjoined sentence involving the two actions. hesults and discussion
Since separate analyses were conducted for the immediate imitation and the delayed production phases of this experiment, the results will be reported separately. Immediate imitation
Overall analysis for the immediate imitation data was conducted by a 4-way ANOVA, with Subjects nested within Age, crossed with repeated measures factor, Sentence, nested within a cross of Type (SS/OS) by Word Order (SOV/OSV) by Verb Type (Transitive/Intransitive). Age was significant as a
228
Kznji Hakuta
Table 12. Means and stanalzrddeviations of means in ape by
Word 0rde.r interaction in immediate imitation task Order sov
osv
= 0.944 x s.d., = 0.135 s.d., = 0.051
= 0.37.5 a s.d., = 0.330 s.d., = 0.086
x = 0.660 s.d., = 0.317 s.d., = 0.105
f = 0.660 s.d., = 0.317 s.d., = 0.0’70 --
main effect (F’ [2,26 ] = 8.7 12, p < 0.01), indicating overall better performance in older children. The means for the four sentence forms summing across the age groups are shown in Table 12. Xnthe overall ANOVA, Type was not significant (F’ < l), but there were significant effects for Word Order (F’[ 1,101 = 68.269, p < 0.001) and for the Type by Word Order interaction (F’[ 1,101 = 10.145, p < 0.01). Since there was no difference between the means for OS/SOV and OS/OSV, these significant effects can be completely accounted for by the k-qe difference between SS/SOV and SS,/OSV. Jn more ordinary language, this means that while there was no difference whe:ther an OS sentence was in the SOV or OSV order, it made a large difference for SS, which was much easier when it was SOV than when it was OSV. This is not totally in accord with the comprehension results, which would predict that OS in the OSV order should be easier since it is unstacked. But the result is not surprising when one considers the possibility that there cmld have been a trade-off bctween the Stacking Constraint and the SOV Constraint, since the unstacked configuration for OS is in the OSV order. This account of the data is supported by the errors made by the children in j.mitation, and will be discussed shotily. It appears that in this imitation task,. perh.sps because of the cue to meaning provided by the pictures, children processed the main verb to the extent that the effects of the SOV Constraint were observed. There were significant interactions of the Iepeated measures factors with Age that were of interest. Age interacted with Type (Age X Type, F’ [ 2,111 = 4-X 16, p < 0.05) and with Type by W&l Order (Age X Type X Word order, F’[2,14] = 4.901, p < 0.05). Howeveq these interactions can be explained by the fat% that even the youngest children perfoAmed close to optimum on the SS/SGV sentences, while there was improvement with age on a a3 the’ other sentence types.
Relative clauses
229
.
Table 13.
Frequency of (a) particle errors and (b) order transposition errors in immediate imitation task (a) Particle errors Order
Type
Total
sov
osv
ss OS
3 6
41 38
44 44
Total -
9
79
88
(b) Order transpositionerrors Order
Type sov ss OS -Total
-’
-
Total osv
0 11
24 0
24 11
11
24
35
The remainhg main effect, for Verb, also turned out significant (F’ [ 1,3] = 22.795, p < 0.02), with better performance on the sentences with intransitive verbs then transitive verbs. This is easily accounted for by the fact that the sentences with intransitive verbs were shorter. In fact, the interaction of Age by Verb is significant at the alpha level of 0’.10 (F’ [ 2,6 ] = 4.138), and inspection of the means reveals that while sentences with the intransitive verbs are easier for Groups II and III, the difference disappears by Group IV. An analysis of the distribution of error types with respect to the four sentence forms is highly revealing as to where the children encountered difficulty in imitation. The number of errors on each sentence form appear in Table 13. Table 13(a) gives the frequency of particle errors. For both sentence types, the particle errors appear almost entirely in the OSV order. This consisted of changing the ytiicle on the fust noun, -9, to -@. This error reflects the general preference, found also in simple sentences, for the YOV order. Table 13(b) gives the frequency of transposition errors, where the order of the constituents was switched. Transposition errors occurred exclusively for sentences in which constituents were stacked, namely SS/OSV and OS/SOV. This error reflects the tendency to create unst,acked NVNNV sequences out of stacked unes.
230
Kenji Hukuta
Delayed production
The delayed production data were analyzed in three separate ANOVAs, each using a different measure since there is no “correct” performance on this phase of the experiment. ANOVA 1 used the SOV order as the measure. Each sentence in the SOV order was given a value of 1, with all other forms assigned a value of 0. Subjects were nested within Age, crossed with repeated measures factor Sentence nested in a cross of Type (SS/OS) by Model (SOV/ OSV), where model referred to the original word order in which the sentence was presented in the immediate imitation phase of this experiment. Model was not expected to have a significant effect. ANOVA 2 used the OSV order as its measure. Each sentence in the OSV order was assigned a value of 1, and alI others 0. The structure of ANOVA 2 was identical to ANOVA 1. In ANOVA 3, the measure was sentences in the stacked configuration, where a value of 1 was assigned to each sentence in the NNVNV configuration, namely, OSV for SS and SGV for OS, and all other forms were assigned a value of 0. Since the effect of Model was already tested in ANOVAs 1 and 2, this ANOVA consisted of Subjects within Age crossed with Sentence within Type (%/OS). The predictions were that there should be significant effects for Type in AYOVAs 1 and 2, but in opposite directions. In ANOVA 1, since the measure was SOV, a higher score was expected for SS than OS. For ANOVA 2, where the measure was OSV, a higher score for OS was expected. These predictions were based on the assumption that stacked sentences will tend to be avoided in delayed production. ANOVA 3 tested the prediction that when stacked configurations are produced, they tend to be produced in the SOV order, reflecting the trade-off between the preference for the SOV order zld avoidance of stacked configurations. Thus a main effect for Type was expected, with a higher score for SS since it is in the SOV order, while the OS is in the OSV order. And finally, while an overall main effect for Age was expected, this effect was considered uninteresting since it was also expected that older children would tend to produce more relative clauses in their production overall, and the main effect would simply reflect this fact. While an Age by Type interaction was not predicted, this would be a theoretically more interesting effect open to interpretations. The results of ANOVA 1 showed significant main effects for Age (F’ [2,271 = 6.612, p < 0.005) and for Type (lFT’[1,291 = 72.1’19, p < 0.001). The direction of the effect for Type was in the predicted direction, with a higher score on SS (% = 0.764, s.d.: = 0.286, s.d.IL= 0.053) than on OS (x = 0.205, ,s.d.l = 0.262, s.d., = 0.066). The Age by Type interaction was not significant (F’ < 1). The results of ANOVA 2 showed Type once again signi-
Relativeclauses
231
ficant as a main effect (F’ [ 1,261 = 9 1.704, p < 0.001 ), and Age was approaching significance (F’ [ 2,201 = 3.026, p < 0.10). The direction of the effect for Type was again in the predicted direction, with a higher score on OS (X = 0.622, s.d.t = 0.284, s.d.* = 0.070) than on SS (X = 0.056, s.d., = 0.132, s.d.* = 0.039). The Age by Type interaction was not significant (IT;’< 1). An unexpected result consistent across both ANOVAs was the marginally significant effect for Model. For ANOVA 1, F’( 1,12) = 3.533, p < 0.10, and for ANOVA 2, F’(l ,lO) = 3.392, p < 0.10. In both cases, the effects were in the dirction predicted if children retained the word order of the sentence as it was originally presented. Inspection of individual sentences and subject responses did not reveal any outliers or biase? on particular items, and thus it must be concluded that there was a marginal retention of sentence form. This does not influence the interpretation of the other results, however, since Model did not interact significantly with any of the other factors in either of the ANOVAs. The results stronay indicate that children overwhelmingly prefer to produce sentences that are unstacked. The results of ANOVA 3, however, indicate that there is also a significant trend to produce stacked sentences in the SOV order. This is indicated by a main effect for Type (F’ [ 1,271 = 13.957, p < 0.001). Thus, there was a higher vale for OS/SOV (X = 0.205, s.d., = 0.262, s.d., = 0.066) than for SS/OSV (X = 0.056, s.d., = 0.132, s.d.z = 0.039). The effects for Age (F’ ] 2,341 = 1.327, p > 0.10) and the Age by Type interaction (F’j 1,271 = 1.365, p > 0.10) were not significant. For the reader’s convenience, the relative frequencies of the four sentence forms, SS/SOV, SS/OSV, OS/SOV and OS/OSV, are displayed in Table 14, where a score of 1 is given to each occurrence of the sentence form. The results from the delayed production phase of this experiment show that there are two constraints operating on the child in the production of sentences containing relative clauses. The fxrst constraint is that they will tend to produce sentences with unstacked configurations rather than stacked configurations, the Stacking Constraint. The second constraint, the SOV Constraint, is that they will tend to produce sentences in the SOV order. Thus, stacked sentences are produced for the reason that they fulfill the tugging demands of this latter constraint. This conclusion is supported by inspection of the data from individual children. Out of the 316children in this experiment, 21 produced at least one stacked OS sentence (which is in the SOV order), with 6 of them producing four or more. But only 8 children produced any stacked SS sentences. An inspection of the distribution of errors involving particles reveals a similarity to particle errors in the immediate imitation phase of this experiment, Table 15 gives a breakdown of the error rates for each of the four
232
Kenji Hakuta
Table 14.
Relative frequencies of the four sentence forms produced in the dekbyed production task Order
Type SOV
OSV
ss
R
= 0.764 s.d., = 0.286 s.d., = 0.053
x = 0.056 s.d., = 0.132 s.d., = 0.039
OS
B = 0.205 s.d., v= 0.262 sd., = 0.066
R = 0.622 s.d., = 0.284 s.d., = 0.070
L. .
Nun~bers represent average per response, where a score of 1 is given for each instance of a given form produced.
Table 15.
Percentage of particle errors in sentence forms produced in the deibyed production task Order
Type sov
osv
ss
7% (16/220?
50% (8116)
OS
2% (1159)
25% (44/179)
sentence forms. As can be readily seen, particle errors are clustered around the OSV order. Most of these errors, as in the immediate imitation phase, involved marking the first noun of the sentence with -ga. This is yet another manifestation of the toll taken by the SOV Constraint when the sentences were produced in the OSV order, in particular for the OS/OSV, since its appearance in the OSV order is the result of the Stacking Corstraint. Coordination provides an alternative way of describing the pictured events. Fourteen children produced 58 coordinated sentences in all, of which 39 were produced by the youngest age group (Group II), and 6 and 23 by Groups III and IV respectively. Thus, younger children tended to produce coordinate structures. There were no strong trends towards producing more coordinations with respect to either picture type. There were 32 coordinations fcrr OS-pictures, as compared to 26 for SS-pictures. One consistent trend was for the children to mention the action corresponding to the sub-
Relativeclauses
233
ordinate clause first. This is not terribly surprising since the subordinate clause actions were quite distinct from the main clause actions, and the subordinate clause always preceded the main clause verb in the modelled sentences. In order to ensure that the results of this study were not a function of the pictured meanings per se, but rather resulted from the processing constraints inherent in the sentences required for their description, a control group of 14 children between 3;8 and 6;8 was sc-!tup. The procedure was identical to the e .periment just described, except that simple sentences were paired with the pictures, without making reference to the subordinate clause action. In this control experiment, children overwhelmingly preferred the SOV order for simple sentences in both the immediate and delayed imitation phase of the task. Most importantly for our present purposes, the effect of the picture Type was tested in order to determine whether the pictures meaning might have influenced the results obtained in the main experiment. Neither the main effect for Type, nor its interaction with Model, was significant (F’ < 1). Thus, the results obtained using complex sentences cannot be attributed to the possibility that Japanese children generally prefer the SOV order for SS-pictures and OSV order for OS-pictures. Summary
This experiment used an immediate imitation and delayed production paradigm to test whether children showed the SOV Constraint and the Stacking Constraint in their production. Subjects were between 3;9 and 6;8. Children found SS sentences much easier to imitate correctly when they were in the SOV order than when they were in the OSV order, while no strong preference for word order was found for the OS sentences. Overall, both the Stacking Constraint and the SOV Ccnstraint contributed to the results obtained. Errors in imitation strongly supported this conclusion. The delayed productron results also pointed to the contribution of both constraints. Children had a strong preference for producing both SS and OS in their unstacked word order, but the tendency to produce stacked sentences was stronger for OS than for SS because the former is in the SOV order. A control conditir;\n showed that the results obtained were not a function of the pictured meanings inherent in SS and OS. Summary and conclusion
The confused state of the art in the investigation of the comprehension of complex sentences containing relative clauses in English was a primary moti-
234
Kmji HakuZa
vation underlying the investigation of complex sentences in Japanese children. The hypotheses that had been advanced to account for the English data were classified into two cksses depending on their underlying asump tions. One class assumes that the grammatical description of the sentences is the appropriate basis from which to predict sentence complexity for children. The other class: of hypotheses assumes the configurational properties of the sentences to be the important predictors oi’sentence complexity. It is impossible to unconfound these two classes of explanations in English for the simple reason that it is a language where the grammatical description determines the sentence configuration, since it has a rigid word order. It is therefore necessary to look at other languages. Japanese is an ideal language because of its variable word order, thus making it possible to unconfound the relative contributions of grammatical description and sentence configuration towards predicting sentence complexity. The data from Japanese children unequivocally point to sentence configuration as the appropriate variable. Whenever stacking of constituents occurs in a sentence such that there is local mimicking Iof simple sentences, children appear to find the simple sentence interpret.ation irresistible. When such stacking does not occur, even center-embedded sentences ale not particularly difficult. These results allow us to reject the Parallel Function Hypothesis and the Accessibility Ifierarchy Hypothesis for any language unless they are re-stated in such a way that they interact with the configurational demands of particular languages. In fact, any hypothesis stated purely in terms of the grammatical description of sentences is challenged by these result::. The question arises as to the generality of these findings in Japanese to ‘other SOV languages as well as to VSO languages. The results reported in this paper suggest a somewhat secondary role for particles in the processing of com*.llex sentences, with a greater reliance on the configurational patterns of conl;tituents. There is, however, another variable that may need to be taken into account in looking at differences within SOV and within VSO languages. Within these two rough categories of languages, there are different patterns of correlations between particles and their expected position within sentences. For example, in the case of Japanese simple sentences, there is a predominant SOV order such that the subject-marking particle -g_a most frequently appears on the fust noun of the sentence. This correlational prop erty has been observed to have a significant effect on the comprehension of simple sentences in Japanese children (Hakuta, 1977). Slobin (1978) reports data on similar simple sentences from Serbo-Croatian and Turkish, both SOV languages but with different patterns of correlation between inflections and word order, Serbo-Croatian, like Japanese, is apparently a language where word order is relatively free but not quite as free lppiTurkish in which inflec-
Relative chses
235
tions are highly regular and obligatory. Slobin reports t.hat the Turkish children had little difficulty with varying word orders, relying primarily on particles for comprehem!Hig sentences. The Serbo-Croatan children, on the other hand, reportedly “require normal marking in terms of both ward order and inflection for comprehension” (p. 24). It is thus possible that the pattern of evidence for relative clauses within SOV languages may show interesting variation depending on the extent to which the inflectional system and word order interact. Future reports on relative clauses from Slnbin’s project or Serbo-Croatian a id Turkish promise to shed light on the nature of this variation. The fact that stacked sentences that would ordinaribr cause per,eptual difficulty can be simplified through ,word order variation 18uggeststhe possibility that flexible word order might itself be a device invented by languages that are pla,,ed by such structures. Antinucci r3fal. (1979) suggest that SOV languages have such difficulty, which may cause the diachronic change from SOV to SVO. It may be the case, however, that SOV languages might as a consequence evolve flexible word order, which would either delay or obviate the need fo, change into SVO. Independent linguistic analysis should reveal the distribution of the extent of word order variation allowed by langtlages across different basic underlying structures. References Antinucci,F.,
Duranti, A. and Gebert, L. (1979) Relative clause strul:ture, relative clause perception, and the change from SOV to SVO. Cog., 7,145-176. Bevy?, T. G. (1970) The cognitive basis for linguistic structures. In J. R. Hayes (ed.), Cognitionand the Development of Language. New YoA, Wiley. Brown, H. D. (1971) Children’s comprehension of relativized En&h sentences. Chi!d Devel., 42, 1923-1936. Chomsky, C. (1969) The rdcquisitionof Syntax in CWdren from Firje to Ten. Cambrblee, MA, MIT Press. in Chomsky, N (1961) On the notion “rule of grammar”. In fioceedings of the Twelfth Syr;~posiun~ Applied Mathematics,American Mathematical Society, pp. 6-‘14. Clark, H. H. (1973) The language-as-fixed effect fallacy: a critique oflanguage statistics in psychological research. J. verb. Learn. verb, Behav., 12.335-359. de Villiers, J. G., Tager-Flusberg, H., Hakum, K. and Cohen, M. (1079) Children’s comprehension of relativeclauses.J. psycholing. Res. Greenberg, J. H. (1963) Some universals of grammarwith particular I eference to the order of meaningful elements. In J. H. Greenberg (ed.), llniversrllsof Lunguuge. Cambridge,MA, MIT Press. Hakuta, K. (1977) Word order and particles in the acquisition of Japanese. Papersand Reports on t%7d LunguageDevelopment, 13,110-l 17. Hakuta, K. (1978) Constraints on children’s comprehension of complex sentences. Paper presented at the Workshop Conference on Language Acquisition, Univzsity of Massachusetts, Amherst, April 1978.
236
Ha&a,
Kenji Hakuta
S. I., Uyeno, T., Hayashibe, H. and Yamada, H. (1976) On the development of perceptual strategies in children: a case study on the Japanese child’s comprehension, of the relative clause constructions. Acnual Bulletin of the Research Institute Logopedics Pkzniatric& University of Tokyo, 20.199-224.
Kaplan, R (1975) Transient Processing Load in Sentence Comprehension. Unpublished doctoral diswrtation, Harvard University. Keenan, E. L,. (1975) Variation in universal grammar. In R. W. Fasold and R. W. Shuy (eds.), Analyzing Variation in Language. OWashington,DC, Georgetown University Press. Keenan, E. L. and Comrie, B. (1972) Noun phrase accessibility and universal gralnmar. Paper presented at the Linguistic Society of America, Winter. Kuno, S. (1973) 77te Structure of the Japanese kmguage. Cambridge, MA, MlT Press. Kuno, S. (1974) The position of relative clauses and conjunction. ting. Ing., Y, 117 - 136. Lehmann, W. P. (1973) A structural principle of language and its implications. Lung., 49,47-66. McCawley, N. (1976) Reflexivization: a transformational approach. lit M. Shibatani (ed.), Syntax and Skmantics: Jqanese Geiterative Grammar, Vol5, New York, Academic Press. McPieill, D. (1966) The creation of lang-uage by children. In J. Lyons and R. J. Wales (eds.), Psycholinguistic Papers. Edinburgh, Fdinburgh University Press. Miller, G. A. (1962) Some psychologjcal studies of grammar. Am. Psychol., Z7, 748-762. Ross, J. R. (1967) Constraints on Variables in Syntax. Unpublished doctoral dissertation, MIT. Ross, J. R. (1973) The center. Paper presented at New Ways of Analyzing Variation in English (&WAVE) Conference, Georgetown University, Washington, DC. Schachter, P. (1976) The subject in Philippine languages: topic, actor, actor-topic, or none of the above. In C. ti (ed.), Subject and Topic, New York, Academic Press. Sheldorl, A_ (1974) The role of ;>araliel function in the acquisition of relative clauses in English. J. verb. Learn. verb. Behav.. 13, 272-281.
Siolbin, D. 1. (1973) Cognitive :jrerequisites for the development of grammar. in C. A. Ferguson and D. I. Slobin (eds.), St!,3ies of cxlifd Lunguage Development. New York, Holt, Rinehart and Winston. SLdbin,I). L (1978) Unidersar and particular in the acquisition of language. Paper presented at workshopeunference on Language Acquisition: State of the Art, University of Pennsylvti, May, 19-22. Smith, M. (1974) Relative clause furmation between 29-36 months: a preliminary report. Papus and Reports on child Lunguage Development, S, 104-110.
Tavakolian, S. (1978) The conjoined-clause analysis of relative clawes and other structures. Jn H. Goodluck and I...Solan (eds.), Papers in the Structure nnd Developme? of CiWd Language, University of Massachusetts Occasional Papers in Linguistics, Vol. 4,1978. Wanner, F., Kaplan, R. and Shiner, S. (1974) Garden paths in relative clauses. Unpublished manuscript, Harvard University. Wirier, B. 1. (1971) Statistical Rinciples in Experimental Design, 2nd ed., New York, McGraw-Hill.
ti c&ii d’une description de la complexit& des phrases en termes de relations grs:nmaticales ou en t~zmes dea prop&t& confiiurationnelles de la structure de surface est un poittt fondamental des psychc&guistlques du d&veloppement. Ce probi&me a fait l’objet de recherches en anglais mais les &ades, &ant don& les prop&t& particu:liGres de l’ordre des mats dans cette langue, n’ont pas &lair6 la question. Une s&e d’exp&iences me&es avec des enfants japonais a permis de montrer que les exigences configuxationnelles de la phrase reptisentent le facteur critique.
Cognition, 9 (1981) 237-304 @Elsevier Sequoia S.A., Lausanne - Printed in The Netherlands
Ecological laws of perceivingand acting: In reply to Fsdor and Pylyshyn (1981)* M. T. TURVEY University of Connecticut, Stows, Connecticut and tiaskins Laboratories, New Haven, Connecticut R. E. SHAW Universiitj f of Connecticut, Stows, Connecticut
E. S, REED Center itir Research in Human Learning, Minneapolis, Minnesota W. M. FFACE Trinity College, ffartford, Connecticut
1. Introduction This paper is both a reply to Fodor and Pt_lyshyn (198 1) and a systematic explication of one of Gibson’s (1979) basic claims, namely, that there are ecological laws relating organisms tc the a.ffordances of the environment. Gibson’s theory of affordances holds great promise for psychology for a number of reasons: it provides a framework for the precise formulation and testing of hypotheses about behavior and perception (e.g. E. J. Gibson, in press; Johnston and Turvey, 1980; Lee, 19110;Shaw and Bransford, 1977); it suggests a way to integrate theOphenomenological and mechanistic aspects of psychology without succumbing to either one-sided point of view (Reed, 1980; Runeson, 1977;Shaw et al., in press;Tua-:rey and Shaw, 1979); and it promises to put psychology back on the track of seeking lawful relations-as Gibson (1967, p. 122) once said, in science: “You either find causal relations or you do not’“. *This paper was written while the first author was a Wlow at the Center for Advanced W.rdy in the Behavioral Scieaces. Support from NSF’Grant BNS 76 I!2943 is gratefully acknowledged. The authci’s wish to thank Claudia Carello, Peter Kugler, Jon Barwise and, especially, John Perry, for their comments on parts of the paper. These help$ul souls should not, however, be held responsible for any of our indiscretions. Reprint requests should sent to M. T. Turvey, Dertartment of Psychology, University of Connecticut, Storm, CT 06268, U.S.A.
238 M T.lii~rvq,R.E.Shaw,E.S.Reedand W.M.Mace
Fod& and filyshyn’s (1981) most important argument is that, contrary to Gibson’s theol:I, there cannot be lawful relations between organisms (as epistemic agents) and their environments. Fodor and Pylyshyn do not think any psychological theory can be grounded in laws of nature, although they admit as valid Gibson’s hypothesis that if affordances were lawfully specified in ambient light, then dGrect visual perception of affordances would be possible. Because Fodor and Pylyshyn place such a great emphasis on Gibson’s rlcims a.bout ecological laws, and because they evaluate Gibson’s claims in the terms of the philosophy of science- terms in which Gibson’s claims have neither been criticized :Ior defended previously-our focus is this aspect of the ecological approach. We will give support to the notion of ecological laws and show how the problems of intension and intentionality are addressed in their context. At bottom line, our claim is that the ecological approach is a scientificall:: tractable approach to cognition and that what passes as the Establishment’s approach, the one championed by Fodor and Pylyshyn, is not. Readers interested in other, Iess philosophical, aspects of the ecological ‘approach to psychology should refer first to Gibson (1950, 1966, 1979) and Michacls and Care110 (198 1), and then to references cited in the text.
2. Gibson’s ecological approach in overview it is not obvious that Fodor and Pylyshyn are addressing the same subject matter as Gibson and the proponents of his ecological approach. To the extient that they are not, their arguments against Gibson miss the mark. This section of the paper 5: addressed to the failure of Fodor and Pylyshyn to make contact with Gibson’s approach and, a foxtiori, their failure to present an accurate account of Grbson:‘s enterprise for those unfamiliar with it. The ecological acJro;ach to perceiving (and acting) w3s developed by James Gibson, over more than thi;-ty ye:rs, as a framework that would do justice 20 thr: practical success o,f an organisms ‘everyday’ behavior’ . The per-
’ This object&e iy NY: 70 be confused with positions that make frequency of occurrence of behavior their primary subject -nattcr. Fodor and. Pylyshyn comment that “the goal of psychological theory construction is not t-~ prdict most (or even all) of the variance (p. 21)” as a repiy to the Gibsonian stress on sucoe~sful acr%ty. They seem to think that Gibson was the Skinner of perception, a5 they intimated even more strong!y ~IFtheir Footnote 2. However, they are not Chomsky to Gibson’s Skin(cwtinued an facing page)
Ecological lawsof perceiving and acting
239
ceiving that this approach is meant to capture is that entailed by organisms’ orientations (cf. .Jander, 1975) and adjustments of activity to their environments. It is the perceiving required to support running, flying, building, grasping and SO forth. Gibson argued that the proper “objects” of perceiving are the same as those of activity. Standing still, walking, and running are all relations between an animal and its supporting surface. Though not always explicitlll recognized (Fowler and Turvey, 1978), the supporting surface is just as mclch An essential constituent of these activities as, for instance, legs; and useful perceiving involved in controlling posture and locomotion must be directed toward the same surface. Thus it would seem that a two-team relation involving the same surface or ground can exist in both cases: an animal runs on the ground and an animal sees the ground. This much should be common sense. There is no thing between the animal and the ground in the relation. This is what Gibson has alw;lys mozzt by direct perception and it is the same as what one would mean by direct action if one were discussing activity. What begins as common sense does not remain common sense in the 1igh.i of philosophic and scientific analyses of perception. Conditions of causality and meaningfulness have dictated other types of objects such as retinal images, retinal patterns, or representations. Thinking of perception as men&d events, divorced from activity, reinforced the theoretical interpretation of such objects as objects of perception, between an organism and its environment. It does not take many examples of “illusions” and geometric decompositions of patterns to convince students that they do not see the environment, but some proximal surrogate., Bringing activity back into the story creates a dilemma, however. Organisms do not stand on or fly among images and representations. Images and representations are not the kinds of things that can be objects of ‘“physical” ac:ivity. But if this is true, how might perceptual control of activity be accbmplished? Gibson worked to establish a framework that would support both scientific analysis and the direct relation between a perceiving organism and the objects of its perceiving; that is, he sought a framework within which one
ner. The point of mentioning the frequent practical succeEaof perception is to draw attention to comkflonplacefacts that are so ordinary that they are taken fool ganted and often ignored in sustained inquiries concerning perception. Just as Cla.omskyused the regtirity and eaReof natural language pcquisition as a fact to justify treating langua;;eas a special subject matter, so Gibson and his followers have argued for the importance of doing justice to natural, effective perception. Thus, if one pursues an analogy to Chomsky’s work in linguistics, the propsr counterpart to the Gibsonian emphasis on commonplace activities is the selection of natural language as a special subject matter-not frequency of occurrence of types of utterance.
240
M. i? Tkvey, R. E. Shaw, E. S. Reed md W. M. h4uce
could do justice to the facts of both acting and perceiving without “betweenthings” (Shaw and Turvey, 198 1). How one devises such a framework is not itself obvious. Gibson had to offer a new construal of most of the major subproblems impinging on perceptual theory, as noted in the five points of his list of theoretical innovations quoted by Fodor and Pylyshyn and as discussed in Mace (1977). Gibson’s ecological approach to perception can be summarized succinctly in terms of a major denial and a major assertion, both implied in the foregoing, viz‘) the dcqial of mediating objects between an organism and its environment and the assertion of the intentionality of perception. The claim that 1’orception dc:>s not involve inference is a corollary of the denial, Gibson rejet ted the idea that organisms have to infer the properties of their environments (or even the existence of their environments) from the properties of other, putatively more primary, objects. Emphasizing Gibson’s focus on the issue of perceptual objects is critical because “direct” and “indirect” when applied to perception are contrastive terms whose meanings depend heavi+ on one another, and Gibson’s style was one of careful contrast. As Austin (1962) pointed out, the dependence is not fully symmetric. It is “indirect” that “wears the trousers” and “‘directly’ takes whatever sense it has from the con@ast with its opposite” (p. 15). It is clear what Gibson meant by “indirect” and it was in opposition to thi; that he established his meaning of “direct”. Consonant with Gibson’s contrastive style we emphasize a central contrast between the Establishment’s characterization of perceiving, as given by Fodor and Pylyshyn, and Gibson’s ecological characterization. T1:p Establishment takes its topic to be the fixation of a type of belief: perceptual belief. Thus natural problems for the Establishment would be how one sees a particular ahape as a man, as the Lone Ranger, as Tonto’s best friend, etc. In the ecological approach the latter do not represent a coherent set of examples. Rather, the subject matter is causally constrained by what can be specified in the light (for vision). Fixing the subject matter itself belongs to the scientific realm of argument, subject to empirical test and theoretical progress. Further, because the ecological approach is concerned with the perceiving that goes with acting, the canonical examples are very differtint from those of the Establishment (as will be evident in the sections that follow). As one writes pages of a manuscript, for instance, the body must be held in a posture that will allow an effective orientation of the head, eyes, arm, hands, and fingers to the desk, pen, and paper to allow writing. Although the writer’s primary awareness is directed toward the words needed to express intended ideas, the coordinated posture of the writer’s body requires the nested per-
Ecohgical hws of perceiving and acting
241
ception of the environment relative to the nested bodily structure. Fodor and Pyiyshyn’s kind of perception invo!ves clear beginnings and endings. Gibson’s kind of perception occurs in nested episodes. Thus one may change some of the details of head and body orientation without disrupting the flow of writing. Ont: might stop, stand up, and walk around the room while preserving the larger orientation of being in the room. One might remain oriented vertically to tne horizontal ground through the flux of an entire day until one hoes to sleep at night. Fodor and Pylyshyn, as Establishment theorists, concentrate on how one takes the stimulus envi-onment, appealing to verbal labels of experience to lead the way in delineatinL* subject matter. When the concentration is shifted to perceptual guidance of activity, however, it is clear that most of this continuous, nested perceiving lacks words for referring to it. There are words to talk about the focus of an activity (e.g., writing) but not about the myriad details of perceiving requirt;d to control the activity successfully. In sum, Fodor and Pylyshyn’s kind of perception (in percepts) is whatever eventuates in a perceptual judgment or belief. Gibson’s kind of perception, in contrast, is that which eventuates in the “proper” adjustment of oriented (to various levels of the environment) activity. We now turn to Gibson’s major assertion, one which he took very seriousiy . Gibson never reduced perception to r on-intentional activity. “Perceiving,” Gibson argues, “is an achievement of the individual, not an appearance in the theater of his consciousness. It is a keeping-in-touch with the world, an experiencing of things, rather than a having of experiences. It involves awareness-of instead of just awareness. It m;ay be awareness of something in the environment o: something in the observer or both at once, but there is no content of awareness independent of that of which one is aware. This is close to the act psychology of the nineteenth century except that perception is not a mental act. Neither is it a bodily act. Perceiving is a psychomatic act, not of the mind or of the body, but of a living observer” (1979, pp. 23940). Fodor and Pylyshyn read Gibson’s theory as a theor:r of mere awareness when it is, in fact, a theory of awarenmes-of. The erroneous reading is precipitated, perhaps, by the fact that all previous scientific theories of perception have attempted to reduce perception as an achievement to perception as an awareyless, where awareness is not awareness-of, but is merely a relation to a mental content (representation). The father cl’ Act Psychology, Brentano, had an insight that is often expressed as: “intentionality is the distinctive characteristic of the mental”. By this, Brentano meant (among other things) that directedness towardsobjects is a feature of living things. Often this claim is taken to mean something quite obscure, but a clear and scientifically acceptable meaning to Brentano’s
242
M. i? lI.uvey,R. E. Saw, E. S. Reed and W.K Mace
insight can be given and has been expounded at length by Merleau-Ponty (1962, 1963), Searle (1980) and Gibson (1966, 1979). Pt might be termed the “ecological concept of intentionality” (for want of a better phrase) because it takes the word “object” in discussions of intentionality much more prosaically than is usually done; it means simply the things that populate an environment. The intentionality of visual perception can work only by explaining how organisms can “come into psychological contact” with objects with which they tie not in physical or, more aptly, mechanical comact. Solving this problem of perceptual “action at a distance” is the function of Gibson’s theor\* of ecological information for perception. As Gibson (1975, p. 3 10) once note in reply to a critic: When Boynton (1975, pp. 300--l) asserts that tie are not in visual contact with objects, or edges, facets, faces or textures, we are in contact only with photons’ this assertion is loaded with epistemology. It is a strictly philosophical conclusion. I disqree with it. There is a misunderstanding of the metaphor of visual contact, one that goes back to Johannes Muller, and it is one that I discussed repeatedly in 7%~Senses Consideredas PerceptualSystems(1966). It leads to the doctrine that ail we can ever see (or at least all we can ever see directly)is lit@.” The philosophical assumption underlying virtually all research on vision, and
underlying all criticisms of Gibson, is that visual contact must be reduced to a physical or mechanical contact of the sort described above. Thus the intentionality of vision is claimed to be only apparent, and is reduced by assumption to causality of an absurdly simply sort. For centuries students of visual perception have been asserting that all that organisms ever see directly is light because (they claim) only light comes into contact with the ocular apparatus of organisms. The fact that critics of Gibson (&y.g.,Ullman, 1980) repeatedly ask how it is that optimal information gets “into” the organism shows that this symplistic doctrine of physical contact is still being invoked as the material basis of psychological contact. Thomas Reid long ago, and Mzrleau-Panty m..ore recently, showed the fallacy underlying this reduction of psychological to physical contact. It is the fallacy of assuming the consequent: if it is assumed that psychological contact must be reduced to already know cues of physicaZcontact, then it can be proven that psychological contact is a kind of physical contact. Gibson rejects this philosophical prejudice against the intentionality of psychological contact. li science is to explain .how perception is an awareness of objects in the environment then, instead of assuming ,that it is based on contact between light rays al: material things and eyes as material things, science had best discover what ;:WS of material entities could actually give rise to or account for such intentiol,~lity.
EcologicaSlawsof perceiving and acting
243
Light rays do not specify the &!ects with respect to which organisms behave, SO the idea that the light is all -Aleorganisms c.ln see must be rejected. Per-
haps there is information in li>;.it, in the sense of information specifying its source in the environment. ‘i’Alisconjecture, dating from around the mid1950s (see Gibson, 1960), ie the origjn of the ecological approach to visiori. Fodor and Pylyshyn misunderstand ,Gibson’s claims about information and specScity. They have treated his new empirical hypothesis about specificity as some sort of {incorrect) logical claim or argument, based on mere prejudice. But one cannot disprove an empirical hypothesis by assuming that it is an incorrect logical claim. The only sound way of testing Gibson’s new theory of specificity is to generate hypotheses from it, and to test them. The fundamental hypothesis of the ecological approach to vision, elaborated at great length by Gibson (1966, 1979) is that optical structure specifies its environmental source and that, therefore, mobile organisms with active visual systems that can pick up this information will see their environments and suitably adjust their activity, if and when they detect that information (and only then). Gibson’s view of the organism as active perceiver is character&d by terms like isolating and differentiating, as well as by the metaphors of hunting and clarifying (see Boring, 1967). In a re;J environment, an organism can choose (within limits) how much and which1 meaningful details to clarify. There is no definite limit to the possible detali to explore. Fodor and Pylyshyn make a great deal of the fact that organisi::s (especially people) often d: ** zonclusions about a situation that appear to be more specific than the zVailable cuptical information about that situation. They ridicule the idea that more sampling of optical structure can clarify a situation on the grounds that this idea introduces an arbitrary, unconstrained move that opens the door to a trivial interpretation of Gibson. But in a natural environment the invariablypresent option to sample further ie; an absolutely essential aspect of the adaptive behavior of organisms and, therefore, of the ecological approach to perceiving and acting. Most assuredly, part of visual perceiving is the ability of an organism to change and select samples according to its current challenges (needs, desires, etc). Controlling such changes in an environment requires perceiving changes with appro,priate directionality, such as perceiving the prey coming into vi:w or the predator going out of view. It is difficult to understand why Fodor and Pylyshyn think that the opportunity to extend the sampling of optical structure is s.3 arbitrary when it is so integral a part of the daily living of mobile, seeing organisms. Admittedly, there lvould be a great deal of slop if the problem for the scientist was to predict when an organism “decided that” something was an enclosure or a falling-off place.
244
M. T. Turvey, R. E. Shaw, E. S. Reed and W.M. Mace
But if it is an rverlapping, nested set of possibilities for activity that is being oriented to in perceiving, orderings on a dimension of levels of clarity can be approached systematically and not arbitrarily. Clarity, we should note, is not tantamount to sharp edges but rather to ‘enough’ information to pursue the ongoing activities. And the pickup of information, we should note (in reiteration of a point above), does not eventuate in judgment but in the ‘proper adjustment of activity’. In sum, what Gibson’s ecologic.al approach wishes to account for is an organism’s apprehension of its environment and how it controls its acts with respect to that environment; that is to say, an organism’s apprehension of the environment taken in a way that is relevant to the resource requirements of the organism. The issues of what counts as perception, whether or not perception shoukl be construed as judgment, whether perception is direct or indirect, and the place of inference in the scheme are secondary. If it can be assumed that Fodor and Pylyshyn would grant the proponents of the ecolog ical approach that, ideally neutral, statement of a goa12, then the overarching question takes the form: “can a theory get there from here?“. Fodor and Pylyshyn argue that Gibson’s ecological approach cannot do the work that it is meant to do because it lacks the necessary resources of inference and representation. They maintain that Gibson tried to get the job done with notions of direct perception, invariant, and information that are so unconstrained that the theory is left open to trivialization. To counter, we hold that it is the Establishment position that is insufficiently constrained. The main constraint that Fodor and Pylyshyn offer is that direct access to environmental states of affairs is limited to physical outputs of transducers linked to the basic descriptors of an energy type. This gives them one clear idea, but it then leaves a burden on inference which we argue (in Sectio:no 3 and 11) inference cannot bear. G&Vstrategy, as proponents of Gibson’s ecological approach, is to argue for a conception of natural law that allows meaningful relations between organism and environment to hold. Further, we constrain our use of the term ‘perception’ (and thus, of course, ‘direct perception’) to relations governed by such laws. Establishment theorists have a much looser usage of ‘perception’ which includes perceptud beliefs and judgments (cf. Fodor and Pylyshyn’s Footnote 9). We arc perfectly willing to admit that inference exists in cognitive life, but we This is the basic god of the ecological approach because it is assumed that perception in the service ot’ activity and orientation is the evolutionarily primary kind of perception. Having reached some understanding of this kind of perception, the ecological approach would then address the other varieties of awareness (see Gibson, 1979, p. 255 ff).
Ecohgical laws of perceiving and acting
245
wish to keep it separate from perception. We take it that the evaluation of beliefs about environmental states of aff% ordinarily rests on eui&nce. But Fodor and Pylyshyn, like Establishment theorists in general, have no discussion of evidence at all. The issue is dismissed, apparently, as one involving the justification of perceptual beliefs rather than the causation of perceptual beliefs. This dismissal is very odd givr.8. that they argue that “The psychologists’ topic is the causation of perceptual beliefs...“. Our assumption, on the contrary, is that the evidence of perception most surely plays a roleprobably the major role for almost all organisms ---in the assessment of belief. The well-motivated need to have a non-inferential source of evidence for behef is one of the reasons proponents of the ecological approach (Shaw et al., in press; Turvey and Shaw, 1979) have argued that perception should be regarded as nonpropomsitional. This argument-termed “suicidal” by Fodor and Pylyshyn- receives support from other quarters (e.g., Bar-wise, in press; Dretske, 1969). It is evident, in short, that both the Establishment and the ecological approach recognize that both natural law and inference (of some sort) play a role in the “cognitive” life of at least some organisms. The two positions differ on where to draw the line. Ecological theorists seek to extend tne application of natural law as far as possible, in part because that strategy promises a method of tight constraint (see Kugler et al., 1980, in press) and in part because we d;in hope, thereby, to explain the lawful evolution of inference in a scientifically principled way. Mablishment theorists, on the other hand, apparently want to extend cognition or intelligence as far as pas sible, choosing to limit the role 01’natural law. Regularities, then, must be accounted for by mental rules an’d representation--which are themselves constrained by very little except unsystematic intuitions and by whatever data a theorist chooses to model. The following Section highlights the logical shortcomings that infirm the Establishment’s position on perception.
3. The infeasibility of the Establishment view The Establishment view is grounded in the following two assertions: Assertion 1: An animal by virtue of its physical makeup is linked nomologitally to the conventional physical properties of light, sound, molecular distributions in the air, etc. Assertion 2: An animal by virtue of its intellectual makeup is linked nonnomologically to the behaviorally significant properties of the surroundi;=g surfaces and substances.
236
M. T. Twvey, R. E. Shm, E. S. Resd and W. M. Mace
These two assertions underly the Establishment theory that perception sho&l be described as an inferential process from evidence statements (couched in the restricted vocabulary of predicates referring to putatively basic energy variables) to belief statements (couched in the indefinitely large vocabulary of predicates referring to activity-relevant properties of the environment). On the Establishment view there are two sides to the evolution of perceptual capacity (in accord with the above assertions): evolution must (a) produce living things that are selectively sensitive to the basic descriptors of one I energy form or of s :veral energy forms and (b) provide living things with the conceptual basis needed to make correct inferences from the basic energy cescriptors to the survival relevant properties of their envirmments. The conceptual basis must include conce,? that stand for environmental properties and concepts that stand for how those properties structure energy distributions in media (Turvey and Shaw, 1979). As Fodor and Pylyshyn express it, how an animal gets from properties of the light to properties of the environment is to “infer the latter from the folmer on the basis of (usually implicit) knowledge of the correlations that connect them”. They should have added “and on the basis of (usually implicit) knowledge of the ktlds of things that populate the environment”. Presumably how evolution meets the challenge of Assertion 1 ~oulri be add.ressed by physical theory. After all, a biological transducer of An energy variable is an aggregate of physical entities-molecules-rendered as a single functional unit by a constraint that, initially low in selectivity and imprecise in function, gradually shmlpens up to high specificity and narrow precise function. Obviously the requisite physical theory would have to subsume an answer to the question of how constraints arise spontaneously-how definite structures or regularities arise in physical systems that are initially homogeneous or, more generally, how it is possible for new dynamic restraints to origir,ate in a physical system where the system’s present state variables and dynamical equations completely determine the system’s future state variables (Pattee, 1971, 1972, 1973). A careful analysis of thermal processes under boundary conditions of the sort that can be met terrestrially reveals that the tendency for molecules to organize is a very general property of a certain class of physical systems and is not indigenous to living systems. Molecular organization is aot uniquely biological; rather it is a general feature of all energy flow systems (Iberall, 1977; Morowitz, 1968, 1978; Prigogine and Nicolis, I9’7I ). Molecular organization of the kind found in transducing elements is not a property of biological systems but rather is a property of the environmental matrix in which biological systems can arise and be main-
Ecdogical lowsof perceiving and acting
247
tamed. Energy flux (from a high I$otential energy source to a low potential energy sink through an intermediary collection of matter) is a self-organizing principle that is at the core of the major physical theories currently addressing the basics of biological organization as thermodynamic phenomena, namely, Womeokinetic Theory and Dissipative Structure Theory. Insights such as these about self-organizing systems coming from nonequilibrium thermodynamics vindicate the widespread applicability of physical principles throughout the evolving cosmos. Although there are laws of nature that apply only to a biolo:yical scale this does not mean that there is any fundamental dichotomy between biology and physics (Reed, in press b). The principles are common to both scales (Iberall, 1977; Morowitz, 1968; Yates, 1980 a). A quarter of a century ago. such a separation of biology from physics wa:: being promulgated authoritatively (E,:asser, 1958). In recent years, the cznturies-old duaiism of psychology and ph: sits has been revived, and two of the more vociferous proponents of this duaristic view are Fodor ( 1975, 1980:1and Pylyshyn (1980). As will become ap;.barent,Assertion 2 of the Establishment view entails a dualistic approach tic psychology. We will argue t&t these recent endorsements of dualism rc. uce, if not destroy, the scientific credibility of psychology and that the dualistic position is based, in no small part, on a mistaken conception of what cou& as a lawful relation. It is important to underscore the relation between a pLysica1 theory of evolution in non-equilibrium, non-conservative systems and the theory of evolution synthesized from Darwin and the results acctunulated in molecular biology. The latter theory presupposes, at a minimum, self-reproduction, self-maintenance, selective irritability to significant energy dimensions and directable motility (Prigogine e:’ al., 1972; Reed, in press a). Therefore the explanation of the basic properties of living thmgs must be sought in nonequilibrium phy sits (e.g., Eigen, I 97 1; Prigogine et al., 1972). Countenaricing this fact the entrenched view of perception must turn to physical processes for the genesis of selective sensitivity to the basic descriptors of energy. But to what processes must the entrenched view turn for the genesis of knowledge that enables inferences to go through from energy-referential predicates to beliefs about the environment? It cannot be themlodynamic processes or any future elaboration of them. To turn to lawful processes would be tantamount to denying Assertion 2. Hence, the answer that the Establishment view must endorse, if it takes Assertion 2 seriously, is that the concep tual knowledge originates in a process of justifying inferences against the backdrop of the synthetic theory of evolution. It is a relatively simple matter to show that this answer is fallacious.
248
M. T. Turvey, R. E. Shaw, E. S. Reed and W,M. Mace
A~I corms of non-demonstrative inference proposed b:’ inductive logicians --enu,nerative inference, eliminative inference, and abductive inference-an be expressed as a confirmatory relation between evidence and hypothesis. The conditions of adequacy for confirmation vary among the forms of inference (see Smokler, 1968) but this is immaterial to the points we wish to make, viz,, that the very notion of inference requires the ability to project relevant hypotheses, the concurrent availability of predicates in which to frame evidence statements and in which to frame hypotheses and a concep tual commensurability between the predicates in which evidence and hypotheses are couched. To clarify, the notion of a basic set of hypotheses is explicit in eliminative and abductive inference and implied in enumerative inference. For example, one version of abduction (Hanson, 1958, p. 72) goes as follows: Some surprising phenomenon P is observed. P would be explicable as a matter of course if H were true. Hence there is reason to think that H is true. If the requisite knowledge implicated in Assertion 2 were derived from inferenc? then ii would have to be supposed that appropriate hypotheses, that is, hypotheses that were generalizations about environmental states of affairs, were already at the disposal of the animal. What is their origin? Surely the answer cannot be: “inference” for that would precipitate a vilcious regress. But if the answer is not “inference” then the only option left to the Establi!.;hment view is that the origin of the hypothesis is extra-physical and extraconceptual. These are mutually ew.clusive categories unless, of course, one’s theory Lountenances a benevolent creator. “Ihe same conclusion follows from the point about the concurrent availability of predicates. The predicates in the evidence statements stand for energy variables and by argument have their origin in * tiy:sical processes. But for any form of inference there must be available, concurrently, predicates in which to couch both evidence and hypothesis, which means for the Establishment view that there must be predicates that stand for environmental prop erties (like an obstacle to locomotion). The origin of these environmentreferenti: --red%ates cannot be inferential (infinite regress) and it cannot be physical ftir the reasons already given. The general conclusion to be drawn is that in Assertion 2 the Establishment view takes out a loan on intelligence that science can never repay: The Establishment view is not a scientifically tractable view and a fortiori a view ofperception that scieiice would be ill-advised to pursue. Let us proceed to radically alter the Establishment view so as to dilute the problem of concurrently available predicates. Let us allow that animals can
Ecological Iaws of perceiving and acting
see objects (rather
249
than just the light) in the non-epistemic sense of seeing that Dretske (1969) has isolated in the ordinary language use of the verb “sd’. That is, let US allow that seeing things is an emergent property of distributed physical processes, that there is a kind of seeing of objects that is fundamental and truly a physicA :&ate of affairs. There will then be objects to which interpreted predicates can be assigned in the Establishment scheme of things. NOW what we wish to show is that even allowing for this radical modification in the Establishment view it is still scientifically intractable. The source of intractability will by: identified as the semantic theory under which the Establishment view labors. Consider the marsh periwinkle, a snail that is found where vegetation is present in the upper intertidal zone. The snail ascends plant stems just prior to the inundation by the advancing tide of the substrate on which it moves. The evidence is that the snails when contacted by the advancing tide move under visual guidance to the nearest plant stem (Hamilton, 1977). When the tide is out the snails move about in the vicinity of the p’?nts, milling around the bases of the sterns and steering their way between t’ ‘rn. There are two characteristics of plant stems to which the marsh per%+.>lkle’s behavior is referred, viz., something that can be climbed up and something that impedes forward locomotion. Let us see how the marsh periwinkle/plant stem situation is analyzed under the Establishment view. Let us say that for a thing to be a barrier it must have the properties p, q, r. That is, (p,q,r) is the intcnsion tl of “barrier”. And let us say that for a thing to be a climb-upable thing it must have the properties s, t, u, v. That is, (s,t,u,v,) is the intension c of “climbable”. The extension of b (in t&e intertidal zone) is the plant stems and other snails. The extension of c (in the intertidal zone) is the plant stems’(other snails being unwilling and too short to comply). Thus c is coextensive with b. Our liberalization of the Establishment view allows that the marsh periwinkle can see (non-inferentially) things such as plant stems and snails but this ability, in and of itself, will not help\ the marsh periwinkle behave adaptively. This is because, under the Establishment view, what is seen are the individuals that possess the property b and the property c, that is, members of the extensions, but not the pray. j ties b and c, that is, the intensions. To paraphrase Fodor and Pylyshyn (their Section 7.1), the ability to see individuals in the extension of a property does not imply the ability to see the property. How than are we to account for the specific direc;edness of the behavior of marsh periwinkles toward plant stems, viz., their seeing them 2~ things to climb on the occasioi.1 of the incoming tide? How can the marsh periwinkle’s behavior be referential. of the intension c when its seeing is restricted to the (ambiguous) extensions? The
250
M. T. Turvey,R. E. Shaw,E. S. Reed and W.M. .Mace
answer from the Establishment view goes as foilows. “Sees a climbable thing” is an illegitimate construction for the marsh periwinkle. Rather the construetion should be *‘believes that a plant stem is the thing now seen possessing the property c”. This construction requires (i) imputing to the marsh periwinkle an internal symbol system that can represent the intension c and (ii) the notion of the marsh periwinkle as being related to that representation, and to others. Neither of these requirements unduly strains Assertion 2 above; indeed, they are implicitly contained within it. :Rt us see where all of this takes us. We modified the Establishment view sg as to give it a modicum of protection against arguments that infirm the v:ew, arguments having to do with the origin of the inferential apparatus taat is the hallnlark of the Establish.ment position. ‘To reiterate, we allowed taat seeing objects was non-epistemic and simply (sic) a matter of physics. 111the Establishment view, however, this physical state of affairs cannot be i:ttensional --it. can only be extensiomal. Put another way, the property c a.scribed to plant stems is not by virtue of a physical description under which snails and plant stems fall but by virtue of a conception that the marsh periwinkle has: the intension c is a mental representation, a concept. Herein lies an atavistic notion with which the Establishment view unabashedly concurs -a category, that is, individuals subsumed under a property, is a mental imposition on an objective world. This notion is a biased Kantianism which must always be arbitrary and relative to the current state of physics. The properties studied by contemporary physics are taken by the Establishment to be real (by and large) rather than mental, whereas the environmental properties of significance to the activities of organisms which are not studied by contemporary physics are taken by the Establishment to be mental rather than real. More generally, this notion tacitly assumes that relatively simplistic physical taxono:nies (properties of o!e:J&s) and relatively simplistic biological tnx*nomies (“plant stems”) are sufficient for the analysis of perception and action, an assuxnption that is readily refuted by behavioral observation. Gonsider, for example, the South African limpets (genus fate&z) that are preyed upon by both the starfish A4uthusterias glaciulis and the gastropod Thais d&a. .rhe limpets react to their predators by either fleeing or agressing. Whether a limpet perceives a predator to be attackable or not depends on the sire of t?e limpet relative to the size of the predator. Limpets above a certain size will &tack the small gastropod and flee the larger starfish. Limpets above a certain, larger size will’perceive both predatory species as attackable (Branch, 1979). Obviously for the limpets, the property ‘attackable enemy’ is not coextensive with the biological taxonomy of ‘predators of limpet genus Patella’. Equally obviously, the property that distinguishes p-:edator to
Ecological&TWS of perceivingmd acting 25 1
be attacked from predator to be fled will not be found in any current physical taxonomy of properties (cf. Mackie, 1970). Finally, this notion is nominalistic: to dispense with properties by reducing everything to bare individuals, their names and collections of such. Under the semantic doctrine of extensionalism, the ontic correlate of a property is its extension. It follows that if a number of individuals are collected under a rule, an intension, then that rule, that intension, must be subjcctdve --of mental origin not physical origin. Proponents of the Establishment position thus hold the view t.lat only propertyless individuals are seen and that intensions are c;onceprs; therefore, they are forced to assume that intensions are inferred from collections of individuals. Inference always involves both evidence statements and hypotheses, as noted above. For a marsh periwinkle to have the intension c it must have inferred the intension. Which means that it or an ancestor must have been able to project a hypothesis of the form “plant stems have c” which means that its or an ancestor’s internal symbol system must have been able to represent the property c. We are on the slippery slide of an infinite regress and the reason for it is commonly understood (Fodor, 1975). Any system whose present competence is defined by a logic of a certain representational power cannot progress through formal logical operations to a higher degree of competence. That is to say, it cannot come to represent more states of affairs than it can currently represent although it can come to mark off those states of affairs that do in fact obtain from those that do not. In short, the representational medium must exhibit preadaptive foresight, being able to represent all relevant states of affairs be they extant or future (see Kugler et al., in press). But there is no sensible scientific story to be told about such foresight. There is something most improper about the nominalism inherent in the Establishment view of perception, given that it is meant to be a view about living things. To force intensions into mental representations is to regard organism-environment systems as non-evolving, closed systems. As just noted, the conceptual interpretation of intension assumes a fixedness of basic intensions, given a priori, to which all systemic states of affairs, present and future, are reducible. But organism-environment systemsare open to fluxes ef energy and matter and open to various forms of cempetition. In consequence, organism-environment systems graduate from states of less structure to states of more structure through successions of stabilities and instabilities. Their syste,mic properties are a posteriori facts not 4 priori presc.riptions (see Kugler et al., 1980, in press). For evolving, open systems neither nominalism nor the conventional interpretation of its counterpart, property realism, will do. Both take intensions as given at the outset. What is
252
M. T. ?imey, R. E. Shmv,f: S. Reed and W.M Mace
needed is an understandirg of the properties of organism-environment systems as a posteriori, but 41evertheless real (Ghisehn, in press; Shaw et a!., m press; Reed, in press b). To return to the main point. Even by radically liberalizing the Establishment view of perception we reach the conclusion stated above-that the Estabiishmznt view has taken a loan out on intelligence that science cannot possibly repay. Several hints as to the reasons for this excessive borrowing have been noted and two of them will be taken up in earnest below, viz., the denial that intensions are physically specifiable and, relatedly, the claim that intensional description mandates conceptual ascription. Let us conclude this section with one further criticism of the viability of the Established view of perception. There would seem to be prima facie evidence that this view is on the right track. The “seeing machines” of artificial intelligence takes a mosaic of sh:ides of grey projected from an arrangement of opaque objects and map the mosaic to a description of the arrangement with which humal: observers would be in very reasonable agreemen t. Roughly speakmg the machines work by constructitlg successive representations of the original mosaic where the predicates in these successive representations are successively more like those that capture the propert% of the physical arrangement responsible for the mosaic. Many seeing machines (e.;g., Falk, 1972) but not all (e.g., Waltz, !975) use a variant of abductive inference, usually inference to the best exptiation (see Harman, 1965, 1968). As such they are susceptible to the above origin argument, which will come as no surprise to anybody. There is a more subtle criticism of seeing machines that we wish to focus on and it is based, in part, on the requirement alluded to above that for an inference to go through, the predicates in the evidence statements and those in the hypothese-, must be conceptually close, if not identical. Af the core of machine instantiations of the Establishment view are the noti*jns of representation and matching. To tire representational power of the first-order predicate calculus is added the computational power of serial pattern matching procedures that successively test alternative interpretations of a thing by matching evidence statements to hypotheses in the form of ge teral patterr. templates. &ut any time some such matching procedure is proposed there is the possibility that no algorithmic solution exists that can atirieve the match in less time than some exponential function of the number 0-1details to be compared in the template and the primitive stimulus descriptic In as given by encoding (Lewis and Papadimitriou, 1978). Supposedly, the biological uselessness of matching as a computational process can be circumvented by strategies that reduce the complexity of the evidence statement-
Ecological lawsof perceiving and acting
253
to-hypothesis comparison. It is of singuYar importance that in machine instantiatiDns the most obvious and most prominent strategy is to increase the grain size of description; that is, to couch the evidence sutements and the hypotheses in terms of relatively few, higher-order predicates (Hayes-Roth, 1977). Generally speaking, in order for there to be a successful match (Inference to the best explanation) it is not only necessary that the evidence predicates be rich, few in number and at the same grain size as the hypothesis predicates but it is also necessary that the evidence predicates be members of precisely the same sr.:t as (or directly comparable to) the hypothesis predicates. Given this claim we should then ask both wh;* and how ;tny given thing comes to be described in just those predicates that ore consonant with the hypothesis mediating its interpretation. This question is the unsung part of the Establishment’s central problem of mapping a stimulus to its appo,site representation(s) in memory; for ?che Establishment the so-called H6ffding function is reduced to how contact is achieved between an appropriate inter,.:a1description of the stimulus and the knowledge structures relevant to its interpretation. How the appropriate description of the stimulus is arrived at is rarely at issue. And it is not difficult to see why. Establishment theorists working on a given problem typically deal with an optimal set of predicates, bypassing the question of ho..: just those predicates would be chosen on a given occasion if there were a choice (as there must be), and focusing attention on the supposedly more important issue of how the match is effected. The theorist can refer to the determination of the requisite stimulus descriptors on a given occasion as (simply) a matter of encoding and leave the problem of how it is done to another time or to other theorists. Where the strategy just noted seeks on:: to implement matching, another strategy seeks to implement .b’oth matching and encoding. Again, however, the sought after implementation is with respect to a sir;gle class of objects (for example, a set of stick figures, a set of opaque polyhedra). As with the first strategy noted, the choice of predicates and representational format will be determined by the nature of the object class. A successful implementation of the encoding stage is equated with a procedure that successfully maps the selected objects onto the chosen set of predicates. The encoding problem gets defined as follows: Given that the set of predicates S is the proprietary set, by what means can a member of the set of objects be described reliably in terms of S? This second strategy is deceptive because it iappears to resolve the issues raised by the Hiiffding function. In actuality, by the restriction to a single class of objects it avoids the thorny aspects of the Hiiffding function with which we a;c presently concerned, namely, given a stimulus to be de-
254 M. T.Turvey,R. E. Shaw, E.S.Reed and W.M.Mace
scribed prepuratory to matching, what set of predicates should be applied
and ipz what way should they be related? The Establishment view holds that inference mediates evidence statements and hypotheses and recognizes the computational impotency that is potentially incident to numerically large predicate sets. But what mediates stimuli and evidence statements? And how fares the tractability of the computational task if the proposed mediator is inference’? In sum, and as anticipated in Section 2, the Establishment talks a great deal about how to make the right inferences and talks very little about how to get the right premises. Our suspicion, once again, is that in order to explain how the right premises are arrived at, the Establishment will have to take out a very large loan on intelligence in the form of foreknowledgeone that is not repayable. How does the ecological approach to perception distinguish from the traditional? A simple answer is that it elim.inates Assertion 2 of the Establishment view and beefs up Assertion 1 to bring fn the properties that Assertion 2 was advanced to accommodate. It holds +!natan adequate theory of perception requires not more psychology but mere physics of the kind appropriate to living things and their environments. Perception is not in the province of mental states or formal languages of representation and computation but in the province of physical principles at the scale of ecology. A core claim of the ecological approach is that an organism (as ;an epistemic agent) anQ its environment (as the support for its acts) are bound together as a synergistic system by laws. Some of these laws may repeat in many organism--environment synergies, others will be unique in the kinds of properti.es that they link. These laws that epistemically bind an organism and its environment are termed ‘ecological laws’ and it is the existence of such laws that is denied by Fodor and Pylyshyn. In Sections 4 and 5 we evaluate the grounds for this denial. 4. The argument from the philosophy of science Fodor and Pylyshyn’s argument against ecological laws is constructed with the aid or four notions--natural kind, projectible predicate, natural law and counterfactual entailment. Thus, for example, we are told (in their Section 4) that among the scientific decisions which go together to converge on directly detectable properties there is the decision to determine whether a property is projectible, the decision as to whether a generalization which involves that property is lawlike, and a decision as to whether the generahzation is counterfactual-supporting. It is painfully obvious, however, that no
Ec~~logi,~d laws of perceiving and acting
255
substantive argument can be built from the notions of kinds, projec,tibles, lawsand counterfactuals given the current state of the art. In the philosoihy of science these notions are notoriously opaque and notoriously uneven in their usage and are commonly recognized as such without undue embarrassment. Our slain intent in this paper is to make a constructive argument for ecological laws, not to engage a diatribe. However, insofar as Fodolr and Pylyshyn have chosen to ground their argument in the philosophy of science, we feel it incumbent upon us to show that that foundation is porous-lest philosophical wool be pulled over the eyes of the unwary non-philosopher. The philosophy of science provides no algorithms, nor even consistently reliable rules of thumb, for making the decisions that Fodor and Pylyshyn say will converge on directly detectable properties. Natural kind terms are said by Fodor and Pylyshyn to be the terms embodied in a statement of law. And when this claim is first introduced (Section 2.2) there is a strong intimation that natural kind terms are relative to the domain to which the law refers. If one translation of “domain” is ‘*scale of magnitudes” then it is to be expected that theoretically significant terms, natural kinds, will vary with domain to be consistent with the understanding that the laws of nature must have a definite scale (Feynmann et al., 1972). Fodor and Pylyshyn give us ‘mammals’ and ‘hearts’ as examples of natural kinds (and ‘being born before 1982’ as a non-example) and a true universal conditional relating extensions is forwarded as an example of a generalization of law, viz., ‘All mammals have hearts’. There is much controversy about biological terms such as ‘mammals’. A growing sentiment is that taxon names merely name individuals and do not pick out natural kinds. If laws subsume only naturai kinds, then taxon names cannot ent:er into natura! laws (Chiselin, 1974; Hull, 1976; 1978). The sentiment can be responded to by arguing that a species, for example, can be defined intensionally, can pick out a natural kind, but need not occur in any law (Kitts and Kitts, 1979); or by arguing that the names of taxa refer to either an individual’or a kind depending on context and wi,th either status can enter into a law (e.g., Van Valen, 1976). Freeing the conception of natural kind from the requirement of inclusion in q natural law is the tack taken by Putnam (1970 a, b); a natural kind tenn merely serves to drac? attention to commonalities among things that are superficially different; i! is a scientific convenience and an intentionally temporary one at that. Bunge’s (1977) tack is different again -a natural kind is determined by a set of lawfully related properties. These various considerations indicating the uncertain usage of ‘natural kind’ could easily be expanded (see Schwartz, 1977). But a brief summarizing comment must suffice. For the notion of natural kind to serve in the pre-
256
M. T. Turvey, R. E. Shaw, E. S. Red and W.M.Maw
mises of an argument requires tR& there be a criterion for distinguishing (natural) kinds from non (natural) kinds and an analysis of what it means for’something to be a kind, paralleling Quine’s (1960) requirements for the use of the analytical/synthetic distinction. Fodor and Pylyshyn give us no hints as to how these requiremerits are to be met. But then they are not alone (cf. Wilder, 1972). The equivocal status of natural kinds leads Fodor and Pylyshyn to argue according to the fallacy of equivocation. In Section 4 of their paper nat.ural kind status is conferred on nothing but the properties identified 3y the physical sciences in contradiction to the intimation in Section 2.2 that natural kinds a;:e relative to the domai!i of inquiry. This conferral leads them in Section 4 to limit the honorific ‘law’ to relations among conventional physical magnitudes. This htter move produces even further equivocality in their argument &cause now the intimation is that a law relates intensions (properties or magnitudes) rather than extensions (domains of propertyless individuals), but it is the extensional view of laws that Fodor and Pylyshyn introduce at the outset of their argument and it is the extensional view of laws that is doing the donkey work in the Fodor and Pylyshyn criticism of the ecological approach, There is a recently initiated dlebate that promises to be hotly contested about whether laws are properly construed as embodying extensions (the orthodox view) or intensions (see Dretske, 1978, 1979; Niiniluoto, 1978). Indeed, it will prove to be the case that the resolution of this debate on the structure of law bears significantly on the outcome of the argument between the Established view of perception and the ecological view. To reiterate, what is doing a good deal of the work for Fodor and Pylyshyn in their denial of the optical specification of ecological properties is the orthodox view of law. In anticipation, under the heterodox view of law, the Fodor and Pylyyshyn argument loses much of its sting. The notions of natural kinds and projectible predicates are misleadingly equated by Fodor and Pilyshyn. This equation is tenuous and largely absent in the literature. The theory of projection (of projectible predicates) is said to hold promise for determining natural kinds by its orginator Goodman ( 5 965). Quine ( 1970), on the other hand, sees the determination of a projectibie predicate foliowing from a scientific understanding of the notion of natural kind (see Wilder, 1972). And Putnam holds the two notions distinct in that the determination of natural kinds is a matter for physics (Putnam, 1970 a) but that the determination of projectible predicates is a matter for psycholsbj (Futnam, 1970 b). The problem of projectrble predicates is generally expressed in Goodman’s ( 1965) classic argument 3volving the predicates green and grue. Grue applies
Ecological laws0~:perceiving and acting
257
to a thing if and only if it has been observed and is green or it has not been observed ancl is blue. For some fixed number of observations of green emeralds the sentences “All emeralds are green” and “All emeralds are grue” express legitimate generalizations but only the former is a generalization of law. Goodman’s query is: Why should the presen* e,ridence favor the hypethesis that all emeralds are green rather than the cc * mry hypothesis that all emeralds are grue? Why is ‘green’ projected and ‘grue’ not? Goodftnlan’sown solution is that a predicate is projectible if it is projected suf5ciently often. A kindly reading of this solution is that it seems TVrelocate the problem from the logic of enumerative induction to the logic of abduction (e.g., Fain, 1970; Moreland, 1976). Why, at the very outset, sl’rould some predicates be preferred for projection over others? A less kindly reading of Goodman’s solution is that given by Putnam (197Ob) and noted above, namely, that it ciefines the problem as psychology’s and not philosophy’s. Passing the buck to psychology is a regressive move if all the relevant psychology, of perception, action and cognition, is as the establishment view says it is: a matter cf non-demonstrative inference. In sum, Fodor and Pylyshyn give the impression (and fire their argument with the impression) that deciding on projectibies is a routine matter rather than a matter of considerable perplexity that doggedly rebuffs the philosophy of science (e.g., Moreland, 1976; Priest, 1976; Quine, 1970; Vickers, 1967). As with natural kinds, to employ the notion of projectible predicate in the premises of an argument requires that we have at our disposal criteria for distinguishing projectible from ncnprojectible predicates and an analysis of what it means for a predicate to be projectible. Fodor and Pylyshyn provide us with neither because nobody has the slightest idea how they can be provided. Finally, let us turn to counterfactuals. In the standard view of laws, generalizations of law are contrasted with generalizations of fact in that the former are said to support or sustain contrary-to-fact conditionals, conditionals that are of the form ‘if A had been an S it would have been P’. Thus, Galileo’s law is said to sustain the counterfactual ‘if this (supported) body were unsupported it would fall with uniform acceleration’. The generalization of fact that ‘All the authors of this paper are Gibsonians’ does not sustain the counterfactual ‘If Jerry Fodor were an author of this paper he would be a Gibsonian’. There has been much puzzlement over why it is possible to advance counterfactual conditionals through the combining of laws with unfulfilled suppositions ;ihat modify the extension of the law’s subject. This putative capacity of laws is attributed by some to a kind of comic necessity (Johnson, 1925) that laws express between occurrences over and above a merely factual uniformity. Laws express that which occurs of necessity
258 M. T. 7lmgy. R.
E.Shaw,E.S.
Reedand
W.M. Mace
.
(it is said) while accidental generalizations do not. Unfortunately, a closer look at this claim that laws are counterfactually entailing reveals that it is neither so clear nor so secure as to warrant the status as a distinguishing criterion (Ayer, 1970). Indeed, rather than shedding any l&h:, counterfactuals are vague and in need of clarification (Lewis, 1973 j. For example, there are cases in which it is fair to say that a counterfactual is sustained by a generalization of fact. In some of these cases the reason lies primarily in the f&n of the counterfactual, in other cases it lies in the assumptive context in which the counterfactual is employed, in yet other cases it lies in the causal backdrop for the regularities expressed in rhe generalization of fact. Exemplary of cases of the first kind is the counterfactual YlfJerry Fodor were identical with one of the authors of this paper then he wlould be a Gibsonian’ which is sustained by the generalization ‘All the authors of this paper are Gibsonians’. With regard to cases of the second kind, consider the following scenario. Suppose that there is a Weightwatcher’s convention and that one room at the ccnvention center is a meeting place only for people who used to weigh more than 200 lb and currently weigh less than 150 lb. Let us call this group, to which one applied for membership after reducing from above 200 lb to below 150 lb. ‘The supra 200 to sub 150 club’ and formulate a generalization of fact: ‘All the people in t’his room me members of the club and currently weigh less than 150 lb’. Now under the assumption that Mary is a member of the club and does not necessarily keep her weight below 150 lb then the generalization just given would support the counterfactual ‘If Mary were in this room she would we$h less than 150 lb’. It would not do so, however, if it were assumed that Mary has kept her weight below 150 lb and is not a member of the club. An example appropriated from Mackie (1973) clarifies cases of the third kind. Given a box cor%ining some stones, we assert that none of the stones are radioactive because a nearby Geiger counter has remained quiescent. By itself the unived conditional ‘All the stones in this box are non-radioactive’ does not support the counterfactual that ‘If that other stone were in this box it would not be radioactive’. If it were known to be the case, however, that this box were the left hand box of a pair linked to a device that collects and sorts stones, casting the radioactive one into the right hand box, then the generalization would sustain the counterfactual. The uncertainty touched upon, ie., which types of counterfactuals are implied in the claim that laws entail counterfactuals, can be ex,tended (after Achinstein, 1971). Are acceptable counterfactuals to be couched in a general form ‘Anything is such that if it were ,.. . ’ or a specific form ‘If such and such
Ecological lawsof perceiving and acting
259
an item were . . . ‘. And if it is the general form, which general form? With reference to Galileo’s law., for example, should the counterfactual read ‘Any body is such that if it ware unsupported it would fall with uniform acceleration’, or should it read ‘.Anything is such that if it were a body that is unsupported it would fall with uniform acceleration’? Moreover, the issue of the dependency of counterfactual entailment on assumptive context that was identified for so-called accidental generalization can be extended to generalization of law (Achinstein, 197 1). Consider Boyle’s law ‘All gases satisfy the relationship l”Y = RT’. Does it sustain the counterfactual ‘If the substance in that jar, which happens to be lead, were an ideal gas it would satisfy the relation p V = RT ‘? If it is assumed that the lead in the jar takes on the properties of an ideal gas, the answer is “yes”. If it is assumed that there has just been a shi and (b) refer to a single state of affairs. There is a physical law at the scale of ecology that nor&ally relates (a) and (b)--as was noted in Section G---and by that law hmzo sapiens (gratefully) abides. The appearance of a straight stick. bent in water and the appearance of moving backward when walking fonivard in a room that is moving with you in the same direction but faster has, nothing to do with inference, propositions, knowledge, representation, etc. States of affairs appear to organisms as they ought to appear, and it is because they do that successful acting and knowing are possible (Shaw et al., in press). It is the very fact that appearance? are taken as being what they ought t,o be and not something else that invites, sustains and gives closure to inquiry, The stick is grasped, retrieve4 from the water, held up in the air and returned to the water. Its appearance changes from bent to straight to bent, with these appearances linked by a transformation that takes the stick from one medium (water) to another (air) and back again. At this juncture it might ble advanced that if there ksany error involved, it is not because things appear to organisms as they ought not to appear; rather, it is because organisms behave with regard to things as they ought not to behave. The ecological approach, however, resists the logical decouplin,? of perceiving and acting on wh.ich such an argument is based. Perception md action, affordance and -effectivity, are bound as dual complements; acting must be as felicitous as perceiving is veridical (Michaels znd Carello, 198 1; Shaa and Turkey 3 ! 98 1; Shaw etai., in press). CLnsider the following examp1e.s. Example 1
Sharks electrically detect things to eat and things that impede locomotion (Kalmijn, 1974). An edible bving thing such 2s a flatfish differs in ionic cornposition from the surrounding water, producing a bioelectric field partially modulated in the rhythm of the living thing’s respiratory movements. A flatf& that has buried itself 51 the sand will be detectabl:? by a shark swimming just above it. Reproducing the bioelectric field of lthe flatfish artificially, by
Ecologikullaws ofperceiving and acting
277
pass& a current between twd electrodes buried in the sand, invites the same predatory behavior. The shark digs tenaciously at the source of the field departing from the site when the act fails to reveal an edible thing (Kalmijn, 1971). Now thare is no intelligible sense in which it can be claimed that the source ought to have appeared inedible if the shark’s perception were free of error and if the shark’s perception of affordances were direct. In the niche of the shark ‘an edible thing’ and ‘electric field of, say, type F’ are nomically related. To predicate of the shark (a) ‘detects electric field of type F’ and (b) ‘takes to be an edible thing’ Is not ‘COrefer to two different states of affairs, one ctiiz. (b)) that is reached from rhe other (viz. (a)) by an inference. Rather, it is to make reference in two ways to a single state of affairs of the shark,niche system. The linking of (a) and (b) is not something that goes on in the “mind” of the shark, as the Establishment would have it. The linking of (a) and (b) is in the physics of an ecological world, namely, that system given by lthe cumplementation of the sk :.rk and it: niche. But what of the shark’s actions? Should we not classify them as being in t:rror, as being wrong? After all, the source of the electric field proved not to be an edible thing. Given the nomic relation between ‘electric field of type 13’and ‘edible’ there is no intelligible sense in which it can be said that the shark’s act of investigating the source of the field was wrong. The wrong acIion for the shark, given its niche and its appetite, would be not investigating the source of the field. Example 2 Trichogramma is a parasit:ic wasp that lays its eggs in the eggs cJf other in:er:ts. An important distinctiolh for Trichogramma .is that between propertied
things in which eggs mighit be laid and propertied things in which they might n.ot. The present, though hmited, understa:nding is that the occurrent properties ofan egg-lay-inable tiling are very rou!;;hly the conjunction of the following properties: a thing ofIminima1 volume and diameter in proportion to the size of the wasp, can be w(alked on by the wasp, a minima1 degree of exposed surface (not overly burieki) and motionless (Evans, 1978). ‘Let us term this cgonjunctio.rl the occurrent property w. It is the case that, although insect e;las naturally fit the bill (that is, exhibit w), mercury globules, glass rods, lobelia set 4~, calcium carbonate crystals and sand grains can be substituted fcrr them (Evans, 1978). y@th its ovipositLDr(egg-laying tube), the wasp will try to penetrate these thstgs exhibiting w 3nd will fait to do so. Now there is no intelligible sense in $hich it can be claimed that these various things ought to have appeared as non egg-lay-inable things if the wasp’s perception
278
M. T. Turvey,R. E. Shm, E. S. Reed and W.M. Mace
were free of error and if the wasp’s perception of affordances were direct. IIn the niche of this tiny parasitic wasp, ‘an egg-lay-inable thing’ and ‘w’ are nomically related. And we can repeat the argument voiced twice above, beginning with the claim that to predicate of Trichogrumma (a) ‘detects w’ and (b) ‘takes to be an egg-lay-inable thing’ is not to refer to two distinct states of affairs linked by inference. Moreover, there is no intelligible sense in which it can be claimed that the wasp acted wrongly in trying to pierce mercury globules, lobelia seeds, etc. To the contrary, the wrong action would have been not to try to penetrate these things exhibiting w. Example 3 Monstera gigantea is an arboreal vine whose seeds germinate on the ground
subsequent to falling from the parent plant. Soon after germinating, the seedling grows in the direction of the nearest tree, contacts the tree and ascends, losing its roots in the process. The seedling is skototropic and, in fact, always grows in the direction of the darkest sector of the horizon that comprise, more than a few degrees of the horizon (Strong and Ray, 1975). The seedling can be said to perceive a climb-upable thing. In the niche of the plant. dark sectors of the horizon that are of a minimal extent relate invariantly to climb-upable things (that is, trees) an4 the darkest se!:tor of minimal extent relates invariantly to that climb-upable thing that can be reached with a minimum of horizontal growth. If a dark cul-de-sac, a box with three sides and a top, is placed on the ground in the vicinity of seedlings, they will grow toward it and inside it. Being inside the dark box impedes the photosynthesis process crucial to the vine’s maintenance. Now, there iq no intelligible sense in which it can be claimed that the box should have appeared non climbupahle if the vine’s perception were free of error and if the vine’s perception of affordances were direct. In the niche of Monstera gigantea, ‘darkest :sector of a minimal extent’ and ‘a climb-upable thing’ are nomically related Again, to predicate of Monstera giguntea (a) ‘detects darkest sector of a minimal extent’ and (b) ‘takes it to be a climb-upable thing’ is not to identify two states of affairs mediated by inference. In the physical design of the system comprised of the vine and its niche, (a) and (b) refer to a single state of affairs. Insofar as growing into the box is detrimental to photosynthesis, should we not attribute “wrong action” to the vine that grows toward and eventually into the box? As with the other examples, it makes absolutely no sense to do so. For Monstera gigantea, on the detection of a dark sector of the horizon, the wrong action would be not growing toward it.
Ecological lawsoj perceiving and acting
279
Let u.i return to the immersed stick. Any difficulties there might have been wit;1 ‘stick appears bent’ are removed by the physical theory of refraction. Clarifying the physical grounds for the appearance (rather than clarify ing the non-demonstrative inferential grounds) does the trick. We pursue this moral with regard to geometric illusions, those in which. lines that physical measurement reveals as equal in length, straight, parallel or intersecting may be seen as unequal in length (Mtiller-Lyer illusion), curved (Wundt-Herring illusion), :non-parallel (Zollner illusion) or non-intersecting (Poggendorf illusion). For exismple, the Miiller-Lyer illusion is interpreted traditionally as exemplify.ing a measurement error. The perceiver sees difference in length between two lines that are equal to some standard of measure, say, a ruler. This observation holds for humans and for flies. The Establishment is tempted to say that the perceiver, human or fly, falsely infers from the play of light at the eyes that the two lines are of different lengths when, in fact, they are of the same length. What must be assumed to give legitimacy to this claim for perceptual error? The following come quickly to mind: (1) Whatever the proper basis of measurement for describing the figure is, it is one and the same as the basis of the measurement device by which the figure is described. (2) ‘i he perceiver as r*leasurement device quantifies over the same basis as that of the measuremerit device by which the figure is described. (To not assume this is to assume something like a mismatch between the measurement of oranges in candela/ m2 by a photometer and the measurement of oranges in kilograms by a balance. Nobody would $scribe error to the photometer because its readings did not confirm those of t,he balance). And (3) of the two measuring systems, the non-biological and th!e biological, it is the former that is privileged with regard to reality status, otherwise the inclination would be to refer to the Miiller--Lyer illusion as a physical error rather than a perceptual error. Remarka.blp, these assumptions go unchalle:iged in the absence of any independently argued grouxlds of support. And they do so because of their consistency with an assumption that is much deeper and at the core of the Establishment view, ufz., that the organism and its environmeht are logically independent. Among other things, this latter assumption gives license to the selection of’ a basis of measurement that is organism-non-referential. As argued above, there is no intelligible sense in which the straight stick in water could appear other than bent given the physical grounds for this natural phenomenon. Similarly, we shall suppose (and construct an <argument accordingly) that there is no intelligible sense in which line segments of equal length in the context of attached angles of different degrees could appear
280
M. T. Turvey,R. E, Shaw,E. S. Reed and W.M. Mace
other than unequal in length once the physical grounds for this natural phenomenon are known. The proper ccnsirual of the task for science with regard to the stick in water was to explain a difference in appearance, not an error in perception. The task reduces to the question: What physical principies are responsible for the different appearances of a straight stick (compietely) in air and a straight stick (partially) in water? We assume, therefore, that the Mtiller-Lyer figure is appearing to human and to fly as it ought to appear (that is, without the benefit of any tpistemic intervention), and that the task is explaining why two lines shouid appear equal in some contexts and unequal in others. To assume that the figure is appearing as it ought to appear is to iieny the assumptions that legalize the claim of perceptual error. In reference c3 the first asst mption, the ecological approach could not commit itself uncritically to a conventional and convenient standard of measure and, relatedly, in reference to the third assumption it could not invest reality disproportionately in an objective, organism-non-referential physics (nor conversely, in a subjective, environr,!ent-non-referential psychology). The ecological approach is committed to the empirical discovery of a basis of measurement common to both environment and organism (Shaw and Cut-
ting, 1980). This commitment follows from the assumption of organismenvironment synergy or mutuality (Gibson, 1979; Michaels and Carello, 198 I ; Shaw and Turvey, 198 1; Turvey and Shaw, 1979). Measurement of extent presupposes the determination of “chords”, that is, the differences in distance between pairs of points lying on a figure. A geometry which defines figures in texms of such differential lengths is called a chord geometry, say, as opposed to a point geometry. Since chords may be ordered in terms of length, natural numbers may be used to index the various lengths. In this way a point geometry is but a special case of a geometry whose chords approach zero length at limit. Hence, the natural number-based chord geometry can approach, at limit, the precision of measurement provided by a real number-based point geometry. However, there is one important difference between the two types of geometries: whereas measurements carried out in point geometry by necessity are infinitesimally precise, those carried out in chord geometry are no more precise than the tolerance provided by their shortest chord, This feature of chord geometry is very convenient for expressing the limited resolving power of natural measuring devices such as the human and fly visual systems. Moreover, this feature also permits a principled continuity to be defined betw Gen systems which conduct measurements at micro-levels of precision, say those limited only by quantum uncertainty, and those systems that function at more macro-levels of precisi.on, such as the human and fly
Ecologicdhws of perceiving and acting
,
28i
visual S~S~CXII~, whose tolerances are set by the dioptrics of the eye and the angular sepa;i’atiouof the receptors. Interestingly and importantly, there is a simple first-order lationship between visual resolution and body height, Resolution = k/Ji degrees, where k is a constant of proportionality and H is height (Kirschfield, 1976). This points to the fact that, although the lens eyes of large animals and the compound eyes of small animals differ in absolute resolution, they do not differ in practical resolution. A human of BP proximately 1.:m height looking at a fly 5 m away resolves tlte fly into the same number of points as a fly of approximately 2 mm height looking at another fly 5 mm away. In short, witltin some distance that is a constant proportion of the scale of the animal, visual resolution is roughly equal for the large and the small species. This is consistent with the observatiorl that large animals act with respect to things at greater absolute distance Clan do small animab. And it is important in that it gives sustenance to the thesis that the Muher-Lyer figure ought to appear in the same way to human and to fly. Suppose, for the sake of argument, that endpoints of line segments, vt-rtices, and intersections are located by natural number chord-coordinates as the centers of isoextent, chord distributions, In asymmetrically dense regions of high geometric complexity these centers will often be shifted away from the locations given by real number, Euclidean point-coordinates. For instance, in the Miiller-:&yer figure, angles that open outwardly have chord distributions with centers further out, approximated where the physical vertices tie, for the same reason. It is important to note that chord density information is intC.sic in nature; it is a function of the overall structure of a pattern. A measurement (e.g., perceptual sample) carried out on one region of the structure ir dependent upon what measurements might reveal in all other regions. By Gontrast, more conventional physical measurements are extrinsic in nature; they are insensaiive to overall structure and depend only on local circumstances. Loreover, tlte standard of measurement in chord geometries is intrinsic, bei: g dictated by the organization of the pattern to be measured (its chord distribution). The standard of measurement in Euclidean-based (physical) geometries is extrinsic, being selected for convlFnience from extraneous sources (the Bureau of Standards). The importance of intrinsic metrics is well known amcng ecologists. For example, one of the most fundamental issues in evolutionary ecology is the extent and result of competition among allied species whose geographical ranges overlap. Studying such a situation requires measuring organismic characters to evaluate the amount of divergence between individuals of competing
282
M. T. T’urvey,R. E. Shaw, E. SmReed and W,M. Mace
species. In the best studied cases, that of length of body and length of beak among avian competitors for food, it is well known that extrinsic metrics produce spurious significance. Based on his own and Schoener’s ( 1965) data, Eckhardt (1979, p. 145) recently concluded that ‘“statistically significant morphological differences as based on extrinsic m.etrics do not necessarily imply ecologically significant differences”. We suspect that the data of a large body of psychological research are similarly spurious, because ecologically relevant measures are consistently eschewed by psychologists in favor of extrinsic -and therefore uninterpretable-measures. Our complaint with those who study illusions, or who use ecologically uninteresting or unrepresentative displays, is not that they cannot, in principle, help us to understand the “mechanism” of perception, as Fodor and Pylyshyn (their Section 2.5) put it. Rather, it is that only when the ecologically relevant measurement principles are developed will we ever be able to comprehend what went on in these sorts of studies in the first place. In conclusion, if he proprietary system of measuremerit is taken to be based on intrinsic measures of structure (perhaps, chord geometry) rather than extrinsic (conventional physical) measures, then in principle the two fundamental perplexities of the geometric illusions are soluable: First, measurement by a biological system can sometimes be at odds with measurement by a non-biologic21 system because the two systems of measuring do not share common bases. Second, a structu“re embedded in one context (MiillerLyer figure with angles open inwardly) may appear to be different in magnitude from the same structure embedded in anovher contexi. (Mtiller-Lyer figure with angles open outwardly). We take this conclusion, favoring intrinsic measures, to be in the spirit of Gibson’s (1966, p. 3 13) admonition that the information for length of line is not simply len.gth of line. 9. Direct perception: the one and only gambit With the help of the arguments given in Sections 5 through 8 we can now make precise the claim of Section 2 that the ecological approach places tight constraints on the use of the term ‘perception’. The overarching constraint is that the term ‘perception’ must be reserved for designating on1.vactual states of affairs of an organism-environment system that include states of affairs involving properties of the environment taken with reference to capabilities of the organism, thereby vindicating perception as the incorrigible basis for an organism knowing its environment. Consider the statement ‘Z perceives X-having-r?, wh.ere, as before, Z stands
Ecological lawsof perceiving and acting
283
.3r organism, X for a thing and a for a property of that thing (recalling that in the realist ontology of Section 5, neither properties nor things are independehtly real),, The use of the term ‘perceives’ in the context of this statement is legitimNate if and only if the statement identifies an actual state of affairs of the organism-environment system. The following is a tentative formulation of IIvhat that evaluation entails. The perceiving of X-having-a by Z petsupposes a law, L: an ambient energy property e is nomically related to u in that it is unique and specific to a in Z’s niche. Given L, ‘Z perceives X-having4 designates an actual state of affairs if: (i) X-having*!: is present, (ii) the e resulting from (i) and L is available to Z, (iii) Z detects the e defined in (ii). Several factors contribute to the tentativeness of the foregoing. For example, the Iproper ecological defmitions of the terms “present” (in (i)) and “available” (in (ii)) have yet to be given satisfactorily. The goal of such defmitions is clear, however: both terms must be tied systematically to :ne effectivities of Z and to the occasions on which the affordances of a thing are actualized with respect to Z (see Section 11; and see Shaw et al., in press, for a pass at this problem). Although the above conditions are not expressed in precisely the form demanded by the ecological approach, they are expressed, nevertheless, in a form sufficient for our current purposes. They illustrate the one and only gambit open to the ecolo@cal approach with regard to defining perception. Recognizing the caveats, we claim that the necessary and sufficient condition for legitimately usilrg the term ‘perceives’ is met only when L and conditions (i), (ii) and (iii) hold. The incorrigible basisfor an organism knowing its en:*ironment lies in the satisfaction of L and the three conditions. Moreover, the satisfaction of L and the three conditions defines ‘directly perceives’, although, strictly spezkirng, in the ecological approach ‘directly’ is redundant. There can be no other seuse of ‘perceives’. If a cunent state of affairs of an organism-environment system can be described truly as ‘Z knows X-having* (by virtue of the fact, say, that Z, directs its behavior to X in a certain way) and L does not hold.: then it is incorrect to describ,e this %te of affairs alternatively as ‘Z perceives Xhaving4. The term ‘perceives’ must enter legally into some statements, describing some states of affairs involving Z and X, that are the necessary support for the statement ‘Z knows X-having-u’. But in the absence of a law ‘a-ness + e-ness’ it cannot be said legally that ‘Z perceives X-having-a’. More properly it should be said that ‘Z infers X-having-Q’, ‘Z judges X-having-a’, etc.
204
M. T. Turvey, R. E. Shaw,E. S. Reedand
W.M.Mace
lf condition (iii) does not hold when L and the other two conditions do hold, then we may speak of a ‘lack of perceiving’ (see Gibson, 1966; 1979; Michkels and Carello, 198 1). Similarly )we can speak of a ‘lack of perceiving’ if condition (ii) (and, by implication, (iii)) does not hold when L and condition (i) do hold. However, it would be an abuse of the term ‘perceives’, as here defined, to speak of a situation of the foregoing type as an ‘error in perceiving’ or even as a ‘failure to perceive’. The statement ‘Z perceives X-havinga' identifies a property of the organism-environment system, a property that is emergent on the fulfilling of L and the three conditions. A property can be present or not present, existing or not exir:ting, but a property cannot be right or wrong. Thus terms like ‘right’ and ‘wrong’ cannot be conjuncted with ‘perceives’. If condition (i) does not hold and L and the other two conditions do hold then we have the circumstance captured by the examples of the preceding Section. For such circumstances the proper description of the organism-environment state of affairs is roughly ‘(To) 2 appears X’-having-a’. X’ is introduced because, while it is the case that X is not present (in Example 3 of Section 8 a tree was not present), some thing is present (there was a box). Again, the realist ontology given in Section 5 does not allow thingless properties. Now the important point to be made about a statement of the kind ‘(To) Z appears X’-having-a’-the point repeatedly underscored in Section 8-is that it identifies a property of the organism-environment system that emerges ZiawfuZZy from the satisfaction of just L and the conditions (ii) and (iii). Given that it is a property one can talk about it as “present” or “not present”, but not as “right” or “wrong”. And given that if. is a nomologically based property it would be superfluous and ill-advised tcjlrefer to its etiology in the terms of inference, propositions, representation, etc. In the Section that follows we analyze the strategy of building a descrip tion of one thing from the predicates used to describe another thing. The conclusions reached in the following Section dovetail with the points expressed in the present Sec,tion. IO. Intensional description and conceptual ascription In our various examples we have cad the rnarsh periwinkle and Monstera gigantea as perceiving things that are climb-upable, the shark as perceiving things that are edible, and the parasitic wasp Trichgramma as perceiving things in which it can deposit its eggs. In these roles we are assigning to the marsh periwinkle and tropical vine the description ‘can perceive climb-upable things’, to the shark the description ‘can perceive edible things and to the
Ecdogicd laws of perceivingtad acting
285
wasp the description ‘can perceive e -iay-inable things’. These descriptions involve a borrowing of properties of the environment to predicate a property of the organism. Thus ‘can perceive climb-upable thi borrows a propin the construcerty of a kind o-f thin , namely, plant sfc;i~s,and appli tion of a property of another kind of thing, namely, marsh periwinkle or tropical vine. Thus, we have the notion of a property (climb-upable) of a perceivable thing (plant stem) and the property of a perceiver thing (marsh periwinkle) that makes it suitable tf.1perceiving that property (climb-upabte). In embedding the property ‘climb-upable’ in the property ‘can perceive climbupable things’ we have concocted what is sometimes referred to as an intensional context. The Establishment attitude with regard to intensional contexts was expressed in Section 3. It is that lo give an intensionaldescription to arzorganism h to ascribe the concept of the embedded property or properties to the organism. To construct the intensional context ‘can perceive climb-upable things’ is, in the Establishment tradition, to ac&be the concept
‘climb-upable thing’ to the organism and to claim tnat ‘perceives a climbupable thing’ is a relation of the organism to this concept. The significance of this equation of integsional description and conceptual ascription should not be undereslimated. More than anything else it is the hallmark of the Establishment view and it is the target of the criticism from a consideration of origin: in order for an organism to perceive property x, it must have the concept of property x. This mistaken equation has led to the Sisyphean struggle from the nadir of nativism to the pinnacle of empiricism, and back again down thr:: slippery slopes of conceptualism. If “inputs” require concepts to be meaningful, then concepts must precede “inputs” as in nativism; but if concepts (to be at all useful in the real world9 require “input” for their content, then “‘inputs” must precede concepts, as in et~ul;ricism (either of the ontogenetic or the phylogenetic variety). Despite centuries of widespread and unjustified optimism, this dilemma (that the mistaken equation of intension and conception gives rise to) has resisted resolution and will continue to do so (Kant notwithstanding). The problem is with the doctrine of intractable non-specilicity (Turvey and Shaw, 1979); that is, in the Establishment’s terms, the doctrine of meaningless “inputs” which requires that “inputs” be associated with meaningful concepts. Gibson repeatedly urged psychologists to reject this doctrine that shackles them to a futile oscillation from innate to acquired, and baclc again. What did we actually do when we constructed the property ‘can perceive a climb-upablle thing’? Quite uncomplicatedly, we took the concept of one sort of property of one sort of thmg to build a concept of another sort of property of another sort of thing. This procedure must always lead to inten-
286
M. i? Trzmey,R. E. Shaw, E. S. Reed and W. M. Mace
sionahty. Wh.at we wish to explore is what this procedure ha.s to do with ascribing concepts. Does this procedure of creating intensional contexts mandate-as the Establishment might seem to demand-the ascribing of concepts? Consider a balance that does not tip for objects less than 4 oz but does tip for objects greater than 4 oz. We obsenre the balance and say “only objects of 4 oz or more tip the balance”. We describe the balance’s behavior through the use of the concept of weighing 4 r’z or more and we proceed to construct the property ‘sensitive to 4 oz or more’. Surely in this case no one would wish to claim that the balance possesses the concept of ‘4 oz or more’. Rather, the claim would be that there is a law of nature that subsumes some property x of the balance and the property of weight to explain the balance’s behavior given the tiitial conditions. Strictly speaking the intensional context ‘sensitive to 4 oz or more’ is just a way of indirectly referring to the property x and the iaw. And it is to be expected that, generally speaking, describing a thing in terms of the properties of other things to which it is sen-
sitive does not mandate ascribing the concept of such properties of the thing.
Take another exiample of building intensional descriptions. To construct the property ‘fish cakes are nauseating’ is to describe fish cakes with a prop erty borrowed from humans, namely, the property of making humans vomit. It is not, of course, to ascribe a concept of this human property to fish cakes. To repeat, borrowing the property of one thing to describe another thing always leads to intensional descriptions but it obviously has nothing to do directly or remotely with ascribing concepts. So we will have to dig more deeply to uncover the conditions that apparently countenance the Establishment’s equation of intensional description and conceptual ascription when the intensional description is of the kind exemplified by ‘perceives (or registers, or senses, or detects) a climb-upable thing’. Suppose we predicate of a thing ‘sensitivity to light’. The thing in question might be a piece of cloth that is bleached by bright sunlight. Or it might be a photocell in a camera. Or it might be a human. To describe these things--cloth, photocell, human -in this way, however, would not warrant the ascriptio:,,. -2f th.;--concept of light to them and the property of ‘senses light’ would nc e be interpreted as a relation of a thing to the concept ‘light’. Presumably the equation of intensional description and conceptual ascription does not hold for ‘sensitivity to light’ because (i) energy media and the variable:: ihat conventionally describe them are not the sort of stuff that concepts are about, at least not the concepts that an organism is said to be in epistemic relation to in the Eatablishment interpretation of intensional contexts; and (ii) ‘sensitivity +o light’ is in the physical nature of cloth, photocell
Ecokyikal laws of perceiving and acting
207
and human. It has something to do with then physical design relative to that of photons. And with respect to the human, the Establishment would suppose that predicating the property ‘sensitivity to light’ falls out of Assertion 1 of the Establishment position. Groups of beans absorbing water in closely adjacent vessels are mutually sensitive to each other’s rates of absorption as expressed in the negative correlation of their respective uptakes. This is so even when the vessels in which they are housed eliminate effects of she geomagnetic field and of static biomagnetic fields, revealing that the interaction between the groups of beans is through oscirlsiiing biomagnetic fields (Brown, 1979). It is legitimate, ttrerefore, to predicate ,rf the beans: (a) ‘sensitivity to the water content or absorptive state of noighboring beans’ and (b) ‘sensitivity to a magnetic cycle’. Of course, (a) and (b) do not pick out two different properties of the beans, they just pick out the same property in two different ways. ‘Water content or absorptive state of neighboring beans’ and ‘magnetic cycle’ is a lawful, not coincidental, lconcommitance of properties in the ecological world of beans. Moreover, although (a) and (b) were built up from the properties ‘water content OF absorptive state’ and ‘magnetic cycle’, we would be reluctant to ascribe the concepts of these propertics to the beans and, obviously, rLluctant to claim that ‘sensing water content or absorptive state’ or ‘sensing magnetic cycle’ are to be interpreted as relations the beans take to the conceptual representation of these properties. More simply, it is just assumea that the behavior of the beans has something to do with the mutuality of their physical design (or the physical design of ensembles of them) and magnetic fields of a certain form. But what precisely is the reason for the reluctance to ascribe these concepts to the beans? It is probably because a magnetic field is like iight in the previous example- not the sort of stuff that the Establishment is willing to let into the class of things that get conceptually transcribed in an internal medium of representation. On the other hand, one has theniggling feeling that ‘water content or absorptive state’ is close to, if not in fact, a kind of thing that is in the class of conceptually representable. At extremes of the cycle we could predicate of the beans ‘sensed that their neighbors were sattirated’, ‘sensed that their neighbors were unsaturated’. To predicate of a . *:-a next to her was full’ or &sensedthat the person ‘sense,d that the chiid &Clrrb child sitting next to her was still hungry’ would be interpreted by the Establishment as the person being related to an internal representation (that is, concept) of Ihunger q~la food content. This surely would have to be the case for the Establishment, given that sated or unsated are coextensive properties (the extension being the child).
288
M. II: Turvey, R. E. Shaw, E. S. Reed and W.M Mace
We can entertain the following question: If only (a) could be predicated of the beans, that is ‘sensitivity ta the water content or absorptive state of neighboring beans’, because of ignorance of a law that relates the specifics of the varying water content to the specifics of a varying magnetic field, would there then be justification, in the Establishment view, to ascribe to the beans the concept of ‘water content or absorptive state’? One suspects that the answer would have to be “yes” and that what prohibits this answer is the nomic relation of water content to magnetic field. That relation means that any reason to ascribe the concept of absorptive state to the beans is also reason to ascribe the concept of magnetic field, but a property like ‘magnetic field’ is not the sort of property that the Establishment willingly allows into the representational medium. The situation of the shark and edible things is the same sort of situation as that of the beans and absorptive states of neighboring beans. We can construct these two intensional descriptions of the shark: (p) ‘can perceive edible things’ ‘and (q) ‘can detect magneti.c field of type F’. Because the properties ‘edible’ and ‘magnetic field’ are lawful, not coincidental, concommitants in the ecological world of the shark, (p) and (q) pick out not two different properties of the shark but the same property; they just happen to do so in two d.ifferent ways. Moreover, although (p) and (q) were built from the properties “edible’ and ‘magnetic field of type F’, we ought to be as reluctant ix the case of .the shark as we are in the case of the beans to ascribe the concepts of these properties to the shark. And by the same token, we ought to be reluctant to claim that ‘perceiving an edible thing’ or detecting a magnetic field of type F’ are to be interpreted as a relation of the shark to the internal representation of these properties. In short, we should assume, just as we did for the beans (and for the cloth, photocell and human with respect to light) that what (p) and (q) do is indirectly refer to a mutuality of the physical design of the shark and a property of its environment. Obviously, this conclusion about the shark and the property edible is not the Establishment’s and it is roughly apparent that the Establishment reading of’ the shark predicate ‘perceives an edible thing’, viz., a relation of the shark to the internal representation of edible, follows in part from the Establishment view of law. Let us reconsider the observations that led to the nonEstablishment conclusion about the shark. Observation 1: It is common procedure to build a property of one kind of thing through the borrowing of properties of another kind of thing. Where 2 is a kind of organism and X is a kind of thing in its environment, one builds a property of Z by borrowing a property of X. Observation 2: This way of building a property of a thing Z by borrowing a property of a thing X is just that and nothing, more. Of itself,
Ecologicallawsof perceiving and acting
289
it has no implications for ascribing to 2 the concept of the borrowed property of X. C)bl;;ervation 3: The Establishment traditionally treats this way of building properties or intensional contexts as license to ascribe to 2 a concept of thr: borrowed property of X. Observation 4: Given that Observations 2 and 3 d!o not concur it must be assumed that the Establishment believes that the content of Observation 3 follows from Observation 2 under certain conditions. That is to say, under certain conditions the intensional description of Z using a borrowed property mandates the ascription of the concept of the. borrowed property to Z. Observation 5: If for the borrowed property there is a corresponding property e in the structured energy medium in which Z is immersed then there is no need to ascribe a concept oi’ the borrowed property to Z, Observation 6: In conclusion, the Establishment draws the equation of internal description and conceptual ascription on the assumption that an energy medium is not structured by the borrowed property ofX in a way that is specific to the borrowed property. That is to say. the equation rests on an extensional view of the rela”rion of environmental things to energy distributions. What ‘else might bolster the Establishment’s equating of intensional d2scripticn with conceptual ascription? Take a fairly standard approach to semantics. in which a predicate, e.g., ‘is edible’ is seen to have two interesting features: an extension rind a meaning. Borrowing predicates to build new predicates produces predicates, such as ‘can perceive edibility’, which similarly have an extension and a meaning. The exttinsion of a new predicate can be a function of either the extension or the meaning of the borrowed predicate. The extension of the new predicate ‘bites an edible thing’ would seem to be, quite straightforwardly, a function of the extension of ‘edible thing’. Thus, if we predi’cate of shark ‘bites an edible thing’, then the shark bites a flatfish or it bites a whiting or it bites . . ., etc., etc., and we are simply ascribing to the propertied thing shark a reIation to another probertied thing. On the other hand, the extension of the new predicate ‘wants an edible thing’ is not a function of the extension of ‘edible thing’ but plausiF!y it is a function of ‘the mearhzg of ‘edible thing’. To predicate of the shark ‘wants an edible thing’ is!, therefore, to ascribe to the propertied thing shark a relation to a meaning or concept and not to a propertied thing. We have eonverged once again on the Establishment thesis: an intensional description of an organism, such as “can perceive edibility’ is to be construed as a relation of the organism to a concept. And perhaps we should recognize at this juncture that intensional descriptions of the kind ‘wants an edible thing’, ‘can perceive an edible thing’, are commonly termed intei3tional in order to underscore that, for example, on the occasion on which ihese descriptions are predicated of the shark, the concept ‘edible thing need have no extension.
240
M. T. Twvqv, R. E. Shaw, El S. Reed and W.M. Mace
It thus seems that the Establishment predilection to read an intensional description of an organism as license to ascribe concepts to the organism is bo1stere.d by a prevailing semantic theory that uses only extensions and meanings. One suspects that this license would be abrogate ., by a more richly endowed semantic theory. Let us explore briefly the implications of the arguments of Section 5 and distin&.h among the meaning, interpretation (or designation), and extension of a property. The J beaning of ‘edible’ in the ecological world of the shark is given in the dual cl.lmplementation of certain properties of the prop ertied thing shark and certain properties of certain kinds of things that, in juxtaposition with shark, actualize eating, felicitous metabolizing, etc. The meaning of ‘edible:‘, therefore, is in the province of the physical anaiysis of edibility as an affordance for the species Scyliorhinus; the meaning of ‘edible’ is not in the province of the shark. But the designation of ‘edible’ can be. In the province of the shark, it is the property ‘electric field of type F’ (among other properties suited to detection by vision and olfaction). And in that same province, the extension of ‘edible’ is the various forms of marine life that exhi.bit that property. In this semantics, which takes properties seriously, the extension of a predicate built from a borrowed predicate, such as ‘can perceive an edible thing’, would be a function of the extension of the borrowed property. In short, this constructed predicate ascribes a relation between a property of one propertied thing, viz., shark, a.nd a property of another propertied thing, viz., magnetic field, that is ba:ed on the nomic relation between ‘edible thing’ and ‘magnetic field of type F’. As with the case of the balance referred to above, and “can perceive things greater than 4 oz’, the constructed predicate for the shark of ‘sensitivity to an edible thing’ is simply a way of referring indirectZy to a lawful relation between properties. But how are we to construe the lawful relations of properties that are indirectly referred to by the foregoing intensional contexts? Generally speaking, laws of nature in their technical form neither explicitly invoke the notion of cause and effect nor allow differentiation between cause and effect (Yates, 1980b). For example, those laws that involve conservation are reverpible, or time-isotropic. The intended meanings of “cause” and “effect” are given more adequate expression by technical terms such as boundary condition, initial and final state (Margenau, 1960). We underscore this feature of natural laws in order to inhibit the temptation to blindly interpret intensional contexts involving sensing, detecting, etc., as including two properties, one of X and ,one of 2, such that the property of X causes the property of 2 or that the property of X causes a change in the property of Z. It could be said that the intensi.onal context (M): ‘marsh periwinkle perceives a climb-upable
Ecological lawsof perceiving and acting
291
thing’, :picks out a lawful relatic n between ‘climb-upable’ as that property of X whic:h is sensed and a ‘climb-upable-thing detector’ as that propi?rty of Z which cioes the sensing. This reading of the intensional context is in the spirit of promoting “grandmother detectors”, a strategy to which Fodor and Pylyshyn, among others (e.g., Ullman, 1980), appeal when giving an interpretation of direct perception (see their Section 4). Under this construal the property ‘climb-upable’ causes an effect in the ‘climb-upable’ detector. But we have just given reason for not being bound to a cause-effect interpretation of the lawful connection picked out indirectly by ti;e intensional context @!I); and we ;lave given ample reason for rejecting (M) as licensing the ascripti.on of a facsimile of the thing detected to the thing doing the detecting. The “grandmother detector” reading of (M) is formally no different from the one with which we began; that is. ‘a climb-upable-thing detector belonging to the marsh periwinkle perceives a climb-upable thing’ is a kin of ‘marsh periwinkle detects a chmb-u I;Pbhething’ and heir to the same analysis. Returning to the balance and to the beans, it makes no more sense to ascribe to them, respectively, a detector for ‘thing 4 oz or more’ and a detector for ‘absorptive states of neighbors’ than to ascribe to them the concepts of these properties. The intensional contexts of above, taking balance and beans as their respective subjects, resist sensible mterpretation in terms of concepts and detectors. And there is no reason to argue the contrary for those inte+ sional c:ontexts taking marsh periwinKle, shark, parasitic wasp and tropical vine as their subjects. The moral of this Section should be repeated. Though simple its implications are far reaching: When we borrow a property of X to construct a descripticjn of Z we should not then give to Z a concept of that property. Contravening this standard is the rampant tendency to ascribe to Z neural devices such as detectors or formal devices such as structural descriptions (see Section 3) that essentially represent the very property that issensed. This tendency is the Establishment’s: For Z to see, detect, register, perceive, or whatever, property x of X, Z must have property x in some sense, neurophyslologically or conceptuaily. (This tendency to proliferate properties by unwarranted duplication has been referred to as the first-order isomorphismfallacy (3~ Surnmerfield et al., 1981, for a discussion).) To rephrase the question with which we began this particular line of argument: How should an intensional context that involves a term such as ‘perceiving” be construed? The answer given by the ecological approach eschews rules (of computation) in favor of natural laws, representations in favor of occurrent properties, and concepts in favor of affordances. Thus, the answer we propose makes perception a function of an ecosystem rather than of an
292
M. T. Tunqv, R. E. Shaw,E. S. Reed and W.M. Mace
organism (Turvey and Shaw, 1979), and looks like this: Intensional contexts, such as (M), index an emergent property of highly distri’buted physical processes of a dynamic system, precisely the system given in the dual complement of an organism (e.g., marsh periwinkle) and its niche (cf., Kugler ef al., h press; Brindle et al., 1980; Shaw and Turvey~l981; Shaw and Todd, 1980). Very roughly, whereas the Establishment has tried to give a mechanism for perception, lirat is, an account of intensional contexts like (M), on the basis of getting the b%>rrowedproperty x of X into the organism, the ecological approach tries to give a mechanism for perception on the basis of keeping the borrowed property x where it belongs, viz., with X. To do so, however, requires the ecological approach to provide a richer semantic context in which to interpret perception; one that allows natural laws, relating occurrent properties to both animal and environment dispositions, to replace cognitive rules, relating concepts and representati.ons. 11. Toward a natural basis for intentiontity
There are two major conclusions to the deliberations of S,zcii~;ln10: 1. To describe one propertied thing Z in terms of the properties of another propertied thing X to which it is sensitive is not to ascribe the concepts of these properties to Z. 2. An intensional context of the kind ‘Z can perceive property x of X’ or ‘Z perceives property x of X and acts accordingly’ is merely an indirect way of refetig to a lawful relation of properties. Taken together, these two conclusions inform us that when the expression ‘orga~rism Z perceives the afforda.nce D of thing X’ designates an actual state of affairs, it does not mean that 2 has a concept of a that mediates the perception of a and does mean that there is a nomological basis to the perception of a by Z. And that, in a nutsheil, is the thesis of direct perception defined more explicitly in Section 9. Let us bring this thesis to bear on the problem of intentionality. A gannet swoops upwards and then dives down to the water, neatly spearing a fish in its beak on its way back up to the surface. To succeed in this act of fundamental importance to its survival, the bird must anticipate when it will make contact with the water: for if it pulls its wings back a fraction of a second too late, their lightweight hollow bones will shatter with the impact. Yet the b.ird cannot pull up and slow down to ease its impact, else it will lose its prey. This example (drawn from Lee, ‘l980) illustrates the problem posed by intentional actKty for psychology. It is tempting to say that the “cause” of the bird’s w5.ngretraction is imminent collision, but how can a future event
Ecological lawsof perceiving and acting
293
(one tha’l: might not occur) “cause” a present action? How can a collision at a later time “cause” a movement at an earlier time. While it is acceptable to say that imminent collision appears to “cause” wing retraction it is not acceptable to claim that imminent collision actually “causes” that behavior. “Causes” should not come after “effects”, and causes shlould’ definitely not be merely possible states of affairs when their effects are actual facts. Clearly, whatever is the actual “cause” of the wing retraction must exist (and not merely as a possibility) prior to the retraction, Gibson’s (1979) conception of information, the one that we have defended in det;dl in this paper, is roughly the claim tha% real possibilities are specified by current states of affairs. In the flowing optic array at the eyes of the diving gannet there currently exists information specific to a future encounter, viz., contac: with the water. There is an optical property (the inverse of the rate of dilation of the optic array structured by fish-in-water) that is lawfully related to the property ‘time-to-contact’ and a felicitious entry into the water follows if wing folding is initiated when this optical property, T(t), assumes a certain margin value (Lee, 1980). There is an important promissory note attached to the Gibsonian conception of information: Cutting the Gordian knot of intentionality. Recall L and the three conditions in Section 9 whose satisfaction defines the legitimate use of the term ‘perceives’. We wish to consider the status of condition (i), viz., X-having-a is present, in an account of intentional activity, that is, activity dire&e%! toward an object (see Section 2). There are two extreme: positions that can be ta;.:en on condition (i)‘s involvement in intentional activity: to rabidly deny it or to rabidly assert it. Condition (i) might be denied outright on the nominalist ontological grounds that only bare individuals exist and, therefo:re, X-having-u is a nonexistent environmental state of affairs. Or it might be denied on the lesser grounds that the C‘objects” to which behavior, is directed often have no extension. To emphasize the denial of condition (i)‘s involvement is likely to lead to rep;esentationalism and thence-as we . note below--to solipsism. In contrast, a rabid assertion of condition (i)‘s involvement runs into problems when inrentional activity is clearly manifest (such as a shark digging tenaciously at the site of a biomagnet field of type F) in the absence of ‘X-havirrg-a’ (an edible thing). The three-fold moral of Section 8 inculcated into the defmition of perception in Section 9 is that (a) information a’bout X-having+ must be present for perceiving X-having-a; (b) extra-r&he circumstances can be created such that the ambient energy property e that is nomologically linked to (zis present in the absence of X-havinga; and (c) when (b) is the case it must appear to Z that sorre thing-having-a is present. Here then is the way to cut the Gordian knot o’ iqtentionality.
294
M. T. llmey, R. E. Shaw, E. S. Reed and W.M. Mace.
The pick
up
of
information involves two relata that must both exist--a
propertied thing Z (an organismj and a property of a propertied thing E (ambient energy)-and therefore is relational;the specificity qf information involves two relata, one that must exist and one that in extraordinary circumstances may not exist- a property of a propertied thing E and a property of a propertied thing X (a piece of the environment)-and therefore is intentional.
How would the Establishment address the intentional activity of the gannet? To begin with, the gannet’s wing retraction is not caused by any future state of affairs, nor even by the possible fact of collisfon. The apparent “action at a distance” of the water’s surface on the bird’s wings is discounted (see Section 2). The “intentional inexistence” (to use Brentano’s phrase) of the imminent collision is replaced with the actual existence of a mental representation of the collision. The retraction of the wings is not in regard to the possible future collision but rather is in respect to an actua! mental representation (in the internal language) of the counter-factual case (“If I keep falling at this rate I will crash”). The Establishment scheme of things reduces the intentionality (dtiectsdness towards a goal) of the gannet to a self description (in the internal language) involving counter-factuals, and we are on another slippery slide of the scientifically unwelcome kind that we met in Section 3: Any reduction of intentional directedness (a quasi-relation, with one of the relata allowed to be absent on some extraordinary occasions) to self-representation (a true relation, both relata necessarily existing) must lead to solipsism (see Aquila, 1977; Fodor, 1980). In the Establishment view the gannets “direct” transduction of the cot_ventional variables of light leads to one “state of mind” (see Fodor and Pylyshyn, Section 5). By inference the gannet can move from this first state of mind to a second state of mind, viz., “perceiving that the layout is suchand-so”. In concert with Assertions 1 and 2 of the Establishment position, the gannet is directly aware of the effects of light energy on its body but not of the propertied things that constitute its environment. War, according ,to the Establishment story, can the gannet be aware directly of the propertied thing that is itself, either as body or as agent. All it can be aware of are the effects of signals from proprioceptive transducers on central processes and these do not specify a body or its actions-in short, the gannet has to infer itself and what it is doing from its proprioceptor signals. The solipsism that marks this story is suitably expressed by Dennett’s (1978) image of a person controlling a robot: The person is in a cockpit, aware only of banks of lights and switches; the trick is to pull the appropriate switches in association with given patterns of lights, leading to adaptive behavior in the environment
Ecolqical lawsof perceivingand acting
295
unavailable rbr inspection). Surely, the foregoing cannot be an account of the design principles governing any species of organism and the burden of proof must be put squarely on the Establishment’s shoulders: To show how it can oid a solipsistic account of an intentional activity such as the g&:x-ret’sdiving for food, a solipsism that follows, as the night the day, the reduction of intentionality to representation, Reducing intentionality to representation falls prey to other criticisms of the kind highlig:hted in Section 3. How did the gannet come by its counterfactual representation? If birds that experience non-felicitous collisions all die, then how cfould the content of a living bird’s mental representations involve imminent collision? If the gannet only infers that it is falling and that contact is imminent then how did it come by such accurate and precise inferences as it surely has? An appeal to theories of induction, such as Goodman’s (1965) with its analysis of projectible properties, is vacuous (as shown in Section 4) because they provid.e only post hoc procedures for determining which few of a limitless number of properties are yrojectible and hence no way of insuring the viability of any given induction. The suitability of trial and error as an ex post facto analysis for philosophers does not carry over to evolving creatures like the gannet; use of this method would eliminate not edify. To reiterate the thesis of Section 3 the Establishment must assume lawful processes playing out at the ecological scale which fashion organismniche systems !io designed that the activity-relevant properties of the niche -the projectible properties- are detected directly by the organism. If not, then the Establishment must appeal to a pre-established harmony between, say, the gannet’s internal representations and the actual environmental states of affa;rs with respect to which it lives its life. We cannot be too demure, however, about the above, ecological account of the gannet’s diving. That account does give the proper ordering of initial conditions and final conditions, and it does so without recourse to the contrived and scientifically unwelcome mt asures that mark the Establishment’s interpretation. I3ut it brings new challerr&esthat must be met if the account is to be fully satisfactory. Notably, there is the question of the marginal values of r(t): How are they selected within the system defined by the corn-plementation of the gannet and its niche? In addres,ing this question we catch a. glimpsls of how the relation between intentions and ecological laws might be defined, No law functions in isolation. Rather, the interpretation and application of any law reyuires a context cf consiraints, what night be termed the law’s nomic conWit*. These constraints are the initial conditions and final conditions, comprisin,* the law’s domain and co-domain respectively, together (which is, unfortunately,
296
M. T. Turvey,R. E. Shw, E. S. Reed and W.M. Mace
with boundary conditions, symmetry conditions and scale factors. Collecdefine the law and its relationship to other laws, more precisely, to other nomic contexts of broader and narrower scope. In concert with the ecological principle of nesting (Section 5), laws will stand in superordinate and subordinate relation to other laws (Feynman, 1965). Patently, laws relate the event of placing salt into water to the event of sait dissolving in water; the event of placing an acorn into fertile soil to the event of an oak tree growing in the soil; and, perhaps, the seeing of a prey by a hungry predator to the chasing of the prey. But exactly when the salt dissolves, or when the healthy tree matures, or when the successful predator makes the capture, is not given by the respective laws alone. There is an additional requirement, viz., that the values of the initial conditions be given: How much salt; how warm the water? How fertile the soil; how much rain? How fast the predator; how near the prey? The specific values or restricted ranges of values within which the consequent event falls are the marginal values of the final condition of the law. The specific values required to fix the parameters of the antecedent event are the marginal values of the initial condition of the law. Clearly, fixing either set of marginal values fixes, within the accuracy of the law, the other set of marginal values. This fact is important. It suggests the place o? entry of tytention into the nomic context of an ecological law. Roughly, an intention can be regarded as a convention, bas4 on a criterion, by which marginal values of the final conditions of a law might be selected 50 as to constrain the marginal values that its initial conditions can assume. To illustrate, assume a law of predation, the nomic context of which is the context of the dual subordinate laws -the laws of capture and escape. If we, as scientists, wished to selectively study, say, cases of predation which focus, on the goals of predators rather than on the goals of prey, then we would re quire a convention or decision rule. The convention would involve some cr+ terion for distinguishing cases of one kind of predation from cases of the other. What form might such a criterion take? If velocity vectors, V1 and Vz, are assigned to two organisms, 0, and 02, respectively, who are engaged in prey-predator competition, then the fixing of appropriate initial conditions (such as the distance d separating the pair and the duration t of their chase) necessarily eventuates in three final conditions: (1 ‘Jwhere V1 - V, = k, VI > V, (over some suitable marginal values of ti and 0, the prey escapes the predator; (2) where V, - V2 = -k, VI < Vz (over some suitable marginal values of d and t), the predator captures the prey; (3) where Y, - V, = 0, V1 = IT2(over some suitable marginal values of d and f), the prey and predator are engaged in a ‘no-win’ chase.
tively, these constraints
Ecological lawsof perceiving and ucting
297
TO elect to study the intention of capture is to select, from among all cases of pregdation, only those where the vector sum has a negative value,--k. Here -k is the criterion, and our convention (as students of capture) is a decision rule ll:e.g., a Kronecker de!+1 function) that uses -k to select the appropriate cases. That is, those cases for which particular marginal values of J and t hold. In short, to adopt the criterion-based convention is to constrain the marginal1 values that the initial conditions of the law of predation can assume. It is ‘to be emphasized that there is nothing arbitrary about the criterion. The three final conditions of the law of predation that the criteria designate, comprise three nonlinear @hase~of predation behavior arising from the continuous linear variation of the initial conditions of the law. Returning to the gannet, the intentions ‘dive for a fish’ and ‘alight on the water’ are expressed in the nomic context of a law that relates a property of the optical flow field to the time at which a substantial surface (here, that of a body of water) will be contacted. The time-to-contact law is nested within other laws su.ch as a law of search and a law of predation. A law of search is strongly implicated. All motile organisms from bacteria to man search for resources reSctilinearly- continuously and erratically turning between straight stretches (J;ander, 1975). Rectilinear search is efficient given that the distribution of resources is patchy (MacArthur and Pianka, 1966). The intention ‘to dive.. .’ entails that the contact with the water be headfirst and vigorous, with the wings retracted st some time prior to contact. The intention ‘to alight.. .’ entails that the contact with the water be feetfirst and gentle, with the wings spread at some time prior to contact. Presumably, the times relative to contact, at which the behaviors of retracting and spreading the wings are initiated, respectively, are not identical. That is to say, the two behaviors are initiated at different marginal values of the same optical variable, r(t). The Lltentions ‘to alight,..’ and ‘to dive...’ are thus playist,g systematic roles in the nomic context of the time-to-contact law. The intention ‘to dive.. . ’ is synonymous with the selection of a criterion k ‘specific tl> the final condition of interest (head-first, vigorous contact) and thereby to ,t”rxing,within the nomic context of the time-to-contact law, just those marginal values of the initial condition (the ,variable T(t)) requisite to yield k. T’he gannet’s intention, therefore, is tantamount to a convention that relates a categorical state (vigorous contact), associated with the final conditions of the law, to values of the initial conditions of the la!-1. How is such a convention to be interpreted? The more or ‘3s~ standard answer from the Establishment is that it is a rule that the gannet possesses of the form “when in a state of hunger and diving for food, use the set y of T(t) values”. In sharp contrast, the ecological approach would treat the conven-
298
M. T. Tim-q, R. E. Shm, E. S. Reed and I+!hi. Mace
tion as a challenge for science expressed as follows: Mow does the occasion of being hmgry bring the gannet under the aegis of one nested collection of laws rather than another and how are the nomic contexts of these laws SO coordinated as to nomologically select the initial values of the subordinate time-to-contact law on the occasion of diving for food? It is on& fitting that this discussion of intentionality concludes with the marsh periwinkle. A plant stem for the marsh periwinkle can be something to be climbed up and it can be something that impedes forward locomotion. On lthe occa.Gon of contact with t.he incoming tide the marsh periwinkle perc&es a plant stem as climb-upable, The following statements cover the situationl:
plant stem is climb-upable, b = plant stem is collide-withable, and, on the occasion of the incoming tide, I = marsh periwinkle perceives c. I is a typical intentional statement. Its truth evaluation depends on the marsh periwinkle Z, the statement c and the occasion 0. In brief, I = P(Z,c,O) where P stands for perceives. In the more general formulation P could stand for other pragmatic functions such as knows, believes, etc. Recall flom Sections 3 and 7 that in the Establishment view Z can behave differently to two coextensive properties on two different occasions if and only if Z can represent these properties to itself differently. Given that c and b zne coextensive, the Establishment reads the parenthesized symbols of I as follows: 0n the occasion 0, Z represents to itself c. In the Establishrment view the occasion of contact with the incoming tide plays the role of a “cue” that selects (in the sense of retrieves) a representation. What reading does the ecological approach give to (Z,c,O)? By argument, c is specified by e and b is specified by f (see Table 2) 2nd it follows from the thesis of direct perception that the requested reading is: On the occasion CJ,,Z relates to e. In the ecological view the occasion of conta.ct with the incom;ng tide plays the role of a state of affairs that selects (in the sense of attunes) a marsh periwinkle/ niche relation. The marsh periwinkle/plant stem situation expresses a central problern for the theory of intentionality: how an organism can take the same propertied thing to afford different acts on different occasions (cf. Shaw, et al., in press). The person who takes a bottle 1:obe a throwable thing on the occasion of a bar-room brawl and a put-intoalble thing on the occasion of needing to put out a cigarette exemplifies the problem, as does the hermit crab who takes a sea anemone to be a portable protective thing on the occasion of losing the actinians on its shell and an edible thing cn the occasion of being hungry c =
Ecological lawsof perceiving and achg
294
(von Uexkull, 1957): For both the Establishment and the ecological approach the problem is one of selective constraint. For a given thing X and an organism Z, an occasion constrains Z to one of several or many acts that X makes possible for Z. Occasions individuate affordances. To conclude, the Establishment treatment of the intentionality problem under anai,y& cc=~j:,~~sup the image of an organism on the occasion of being hungry (such as the hermit crab) moving about with a concept of food in mind anId looking for something in the environment that will match this concept; or an organism on the occasion of impending danger from the approaching tide (such as the marsh periwinkle) moving about with a concept of a thing that can be climbed up in mLnd and looking for some thing in the environment that wil! match that concept. The ecological approach’s treatment of the problem conjures up a very different image, viz., of an organism,on a given occasion, moving in the context of one set of (nested) laws rather than another The latter image expresses belief in a natural basis to intontionality whereais the former image, that of the Establishment, does not.
12. Postscript The current controversy between the Establishment view as defended by Fodor and Pylyshyn and the ecological view as defended by ourselves is continuous with a Iarger issue that has been debated endlessly by philosophers and scientists alike: Are the uniformities observed in nature expressions of an underlykzg coherent framework of laws, or are such uniformities but the insidious inventions of the human mind, applied to nature by one faculty and interpreted by another f:aculty. Kant ascribed all the apparent order in nature to formulations of pure reason. “The understarlding”, he argued, “does not draw its laws from nature but prescribes them to nature”. With respect to khe uniformities of perceiving and acting, the Establishment is similarly inclined to prescribe to nature. The physicist Sommerfield expressed the sceptieal attitude of all scientists who stubbornly pursue a nomological view of natural order--that there is an intolerable arrogance in the premise of “prescribing to nature” (Guilleman, 1968) It is a sentiment with which James Gibson would concur, and it is a sentiment wi% which we concur a$ members of that “substantial minority” who believe that Gibson% ecological ;approach to the knowings of organisms is both revolutionary and correct.
300
M. T. Turvey, R. E. Shaw, E. S. ReFedand W.M. Mace i
References Achinstein, P. (1971) Law and Explanation. Ciarendon Press, Oxford. Armstrong, D.M. (1961) Perception and the PhySiCQf World. Routledge and Kegan Paiil, London. Aquila, R. (1977) Zntentionality: A Study of Mental Acts. Pennsylvania State University Press, University Park, PA. Austin, J. L. (1962) Sense and Sensibilia. Oxford University Press, London. Ayer, A. 1. (1970)What isalaw &nature? In B. A. Brody (ed.), Readings in the Philosophy of Science. Prentice Hail,, Englewood Cliffs, NJ. Barwise, J. (in press) Scenes and other New kork. Brown, F. A. (1979) Dynamic biomagnetism associates bean seeds. Experientia, 35,468-470. Bunge, M. (1977) Treatise on Basic Philosophy. Ontobg I: The Furniture of the World. I). Reidel, Baston. Dandson, D. (1970) Mental events. In L. Foster and J. W. Swanson (eds.), Experience and Theory. Duckworth, London. Dennett, D. (1978) B?Qi?ZStOfmS.Bradford Barks. Montgomery, VT. Dretske, F. I. (1969) Seeing and Knowing. Z!nversity of Chicztgo Press, Chicago. Dretske, F. I. (1977) Laws of nature. PhtIos. 2 ci., 44, 148-268. Dretske, F. I. t 1978) Reply to Niinihroto. Phi OS.Sci., 35.440-444. Easukan,J. (1978) The universality of laws. Z%ilos.Sci, 45, 173-181. Eckhardt, R. C. (1979) The adaptive syndrc.mes of two guilds of insectivorous birds in the Colorado Rocky Mountains. Ecol. Monogr., 49, 129-149. gigen, AM.(19?1) Molecular self-organization and theearly stages of evolution. Q. Rev. Biophys., 4,149212. E&isser, W. M. (1958) 77re Physical Foundatioi;rsof Biology. Pergamon Press, Oxford. Evans, H, E. (1978) Life on a Little-Known Planet. E. P. Dutton, New York. F3& H. (1970) The very thought of grue. in B, A. Brody (ed.), Readings in the Philosophy of Science. Prentice Hail, Engiewood Cliffs, NJ. Falk, G (1972) Interpretation of nnperfect line data as a three-dimensional scene. Artif ZntelZ.,3, !Ol-144. Feynna, R. (1965) 77re Gtaracw of physica! Law. MIT Press, Cambridge, MA. Feynman, it., leighton, R. 8. and Sands, M. (1972) 77re Feynman Lectures on Physics. AddisonWesley, Reading, MA Fodor, J. A. (1975? The Lankwage of Thought Thomas Y. Crowell, New York. Fodor, J. A. i1980) Methodological solipsism aansidered as a research strategy in cognitive psychology. Behav. Br. Se& 3, 63-109. Fodor, J. A. and Pflyshyn, 2. (1981) How direct is visual perception? Some reflections on Gibson’s ‘Ecolo&al Approach’. Cog., 9, 139-196. Fowier, C. A. and Turvey, M. T. (1978) Skilf acquisition: An event approach with special reference to searching for the optimum of a function of several variables. In G. Stehnach (ed.), Information Recessing in Motor Control and Learning. Academic Press, New York.
Ecological lawsof perceiving and acting
301
Frazzetta,T. H. (1975) Com191exadaptionsin evolvingpopulations. Sinauer Associates, Sunderland, MA. Friedman, M. B. (1975) Visual control of head movements during avian locomotion. Nature (London)p 255,67-68.
*
Frost, B. 5. (1978) The opta kinetic basis of head bobbing in the pigeon. J. Exp. BioZ., 74, 187- 195. Chiselin, M. T. (1974) The Economy of Nature and the Evolution of Sex. University of California Press, Berkeley, CA. Ghiselin, M. r. (1974) A radical solution to the species problem. Syst. Zool., 23, 536-544. Ghiselin, M. ‘r., (in press) Ca?egories, life and thinking. Behav. Br. Sci. Gibson, E. J.::t(in press) The concept of affordances and their deveiopment: The renaissance of functiona$m. In W. A. Collins (ed.), Minnesota Symposium on Child Psychology, Vol. 15: The Concep\tof Development. Erlbaum, HiIlsdale, NJ. ‘Gibsor, J. J. (‘!.950) The Perception of the Visual World. Houghton-Mifflin, Boston, MA. Gibson, J. J. ($960) The infarmation contained in light. Acta Psych&, 2 7, 23-30. CI Jim, 1. J.. ~1966) Z%e Senses Considered as Perceptual Systems. Houghton-Mifflin, Boston, MA. Gibsor, J. J. (1967) Autobiography. In E. G. Boring and G. Lindzey (eds.), A History ofpSychoCogy in Aur’o,biography, Vol. 5. Appleton-Century-Crc: ts, New York. Gibson, J. J. (1?68) What gives rise to the perception of motion. Psychol. Rev., 75, 335 -346. Gibson, J. J. (1!)75) Reply to Boynton. In E. C. Carterette and M. P. Friedman feds.), Handbook of Perception, Vol. III. Academic Press, New York. Gibson, J. J. (1979). The ll’cological Approach to Visual Perception. Houghtoc-Mifflin, Boston, MA. Goodman, N. (1947) The problem of counterfactual conditionals. J. Philos., XWV, 113-128. Goodman, N. (19’55). Tact, fiction and forecast. Bobbs-Merrill, Indianapolis. Goosens, W. K. (1977) Underlying trait terms. In S. P. SChwZitZ(ed.), Naming, Necessity and Natural Kinds, Cornell Univelaity Press, Ithaca, NY. Greene, P. II. (1975) Strategies for Heterarchicul Control -_,ln Essay. 1. A Style of Controlling Comp,lex Systems. Depa,:tment of Computer Science, IlIinois Institute of Technology, Chicago. Guilleman, V. (196;j) The Srory of Quuntum Mechanics. Scribner, New York. Hamilton, P. V. (19 77) Daily mavelnents and visual location of plant stems by Littorina irrorata (Mollusca: Gastropoda). Mar. Behav. Physiol., 4. 293-304. Hanso&, N. R. (1958) htterns af discovery. Cambridge Universii v Press, Cambridge, U.K. Harman, G. H. (1965) The illfelence to the best explanation. I .rfos. Rev., 124, 88-95. Harman, G. (1968) Knowlec ge, inference and explanation. )’ riles. Q., 18, 164-173. HarrC, R. (1970) Pow?rs. Br, J Philos. Sci., 22, 81-101. Hayes-R.oth, F. (1977) The role of partial and best matches in knowledge systems. In D. A. Waterman and F. Hayes-Ri,th (eds.), Pattern Directed Inference Systems. Academic Press, New York. Hempel, G. ;md Oppenheim, P. (I948) Studies in the logic of explanation. Phil. Sci., 15, 139-175. HuU, D. L (1976) Are .lpecic:sreally individuals? Syst. Zool., 25, 174-191. Hull, D. I_ (1978) A mutter of individuality. Phil. Sci. 45, 335-360. Iberall, A. S. (1977) A Field and circuit thermodynamics for integrative physiology: I. Introduction to general notions. c.!m. ,I. Physiol./Reg. Int. Comp. Physiol., 2, R171 -R180. Jander, R. (1975) Ecolo@al aspects of spatial orientation. Ann. Rev. Ecol. Syst., 171- 188. Johnson, W. E. (1925) Logil; Part III. Cambridge University Press, Cambridge, U.K. Johnston, T. and Turvey, M. T. (1980) A sketch of an ecological metatheory for theories of learning. In G. Bower (ed.), The Psycho&y of Learning and Motivation. Vol. 14. Academic Press, New York. Kalnijn, A. 1. (1971) The electric sense of sharks and rays. f. Exp. Biol., 55,371-38X K&n& A. JI. (197,4) The detection of electric fields from inanimate and animate sources other than electric 0rgan.s. In A. l%ssard (ea.), Handbook of Sensory PhysioZogy Vol. III/J. Springer-Verlag, merlin.
shttw, E.S.Realm-iW.M.Mace 302 M. T.Trc?vejp,R.E.
Kirschfield, K. (1976) The resohrtion of lens and compound eyes. In F. Zettler and R. Weiler @Is.), Neuralfiinciples in Vision. Springer-Verlag, Berlin. Kitts, D. B. and Kitts, D. J. (1979: Biological species as natural kinds. Phil. Sci., 46,613-622. Kugler, P. N.,Kelso, J. A. S. and Turvey, M. T. (1980) On the concept of coordinative structures as dissipative structures. I. Theoretical lines of convergence. In G. E. Stelmach and J. Requin (eds.), Tutorialsin motor Behavior. North-Holland, Amsterdam. Kugler, P. N.,Kelso, J. A. S. and Turvey, ti. T. (in press) On the control and coordination of naturally developing systems. In J. A, S. Kelso and J. Clark (eds,), Development of Human Motor Skill. Wiley, New York. Lee, 9. H.(1976) A theory of visual control of braking based on information about time-to-collision. Perception,5.431-459. Lee, D. (1980) Visuo-motor coordination in spacetime. ln G. Stelmach and 1. Requin (eds.), Worials in MotorBehavior. North-Holland, Amsterdam. Levi, I. and Morgenbesser, S. (1964) Belief and disposition. Am. J. Philos. Q., I, 221-232. Lewis, D. (1973) Counterfactuals.Harvard University Press, Cambridge, MA. Lewis, H. R. and Papadimitriou, C. H. (1978) The efficiency of algorithms. Sci. Amer., 238, 96-109. Lishman, J. R. and Lee, D. N. (1973) The autonomy of visual kinesthesis. Perception,2, 287-294. MacArthur, R. H. and Pianka, E. R. (1966) On optimal use of patchy environment. Am. Nat., 100, 603-609. Mace, W. M. (1977) James Gibson’s strategy for perceiving: Ask not what’s inside your head, but what your head’s inside of. In R. Shaw and J. Bransford (eds.), Perceiving,Acting and Knowing. Erlbaum, Hillsdale, NJ. Mackie, A. M. (1970) Avoidal :ce reactioir,< of marine invertebrates to either steroid glycosides of starfish or synthetic surfaci active agents. J. Exp. Mar.Bill. Ecol., 5, 63-69. Mackie, 1. L. (1973) Truth. Probabilityand Paradox. Oxford University Press, Oxford. MacLane, S. and Birkhoff, G. (1967) Algebra. Macmillan, New York. Margenau, H. (1960) Meaning and scientific status of causality. In A. Danto and S. Morgenbesser (eds.), Philosophyof Science. New American Library, New York. Me&au-Porn@, M. (1962) The Phenomenologyof Perception. Routledge, Kegan Paul, London. Merleau-Ponty, M. (1963) The Structureof Behavior. Beacon Press, Boston, MA. Michaels, C. F. and Carello, C. (1981) Direct Perception. Prentice Hall, New York. Moreland, J. (1976) On projecting grus. philos. Sci., 43, 363-377. Morowitz, H. J. (1968) Energy P?owin Biology. Academic Press, New York. Morowitz, H. J. (1978) Foundations‘of ,Bioenergetics.,icademic Press, New York. Nii,ailuoto, I. (1978) Dretske on laws of nature. Philos.Sci., 4.5,431-439. Pattee, H. H. (1971) Physical theories of biologil;al coordination. Q. Rev. Biophys., 4, 255-276. Pattee, H. H. (1972) Laws and constraints, symbols and language. In C. H. Waddington (ed.), Towards a TheoreticalBiology. Aldine, Chicago. Pattee. H. H. (1973) Physical problems of the origin of natural controls. In A. Locker (ed.), Biogenesis, Evolution, Homeostasis.Springer-Verlag, Heidelberg. Po#er, K. (1965) 77.e Logic of Scientific Discovery. Harper and Row, New York. Priest, G. (1976) Gruesome simplicity. Hzilos. Sci., 43, 432-437. Rigogine, I. and Nicolis, G. (1971) Biological order, structure and instabilities. Q. Rev. Biophys., 4, 107-148. fiigogine, I. Nicolis, G. and Babloyantz, A. (1972) Thermodynamics of evohrtion. whys. Today, 2’5, Nos. 11 and 12. Priruile, S. S., Carello, C., and Turvey, M. T. (1980) Animal-environment mutuality and direct perception. Behav. Br. Sci., 3, 395-397.
&dbgi~al lawsof perceiving and acting
303
Putnam, H. (19700). Is semantics porsibL ” ElnH. E. Kiefer and M. K. Munitz (eds,j, Language, Belief, and hfefaphysics. State Univerrity of New Yark Press, NY. PutnaIIL Y. (197Obb). Qn propelties. in N. Rescher et ~1. (eds.), Essrrys in Honor of Carl G. Hempel. D. Pvlvshyn, 2. (1980) Computation ant1 cognition: issues in the founllations of cognitive science. Behav. __ Br. Sci., 3, 111-169. Quine, W. V. (1960) )vord and Object. MIT Press, ambridge, MA. Quine, W. V. (1970) Natural kinds. In N. Resher et al. (eds.), Essays in Honor of olrl G. Hempel. D. Reidel, Dordrecht, Holland. Reed, E. S. (1979u) The ontologicai status of natural laws. Unpublished manuscript, Center for Research in Human Learning, University of Minnesota, Minn.eapolis. Reed, E. S. (197!)b) The role of symraetry in Chiselin’s ‘Radic. 1 solution to the species problem’. Syst. Zool., 28, 71-78. Reed, E. S. (;380) Information pickt p is the activity of perceiving. Behav. Br. Sci., 3, 397-398. Reed, E. S. (in press a) The demise of mental representations: Commentary on Ghiselin. Behav. Br. Sci. Reed, E. S. (in press b) The lawfulnas of natural selection. Am. Nor. Rul:eson, S. (1977) On the possibility of ‘smart’ perceptual mechanisms. Stand. J. Psychol.. 18, ; 172-179. Rrureli, B. (1903) The Rinciples of Mathematics. Allen and Unwin, London. Scholl, J. J. and Pianka, E. R. (1980) Evolution of escape behavior diversity. Am. Nat., 215, 551-566. S&lick, M. (1949) pausality in everyday life and in recent science. In H. Fergl and W. Seilars (eds.), Readingsin PhuosophicalAnrrlysis.Appleton-Century-Crofts, New York. Schoener, T. W. (1965) The evolutixr of bill size Aifferences among sympatric congeneric species of birds. Evolution, 19. 189-W. Schwartz, S. P. (ed.) (1977) Numilrg.Necessity and NaturalKinds. Corneli University Press, Ithaca, NY. Searle, J. R. (1980) The intentionality of intention and action. Chg. Sci., 4, 47-70. Sh;aw, R. and Bransford, J. (1977) Perceiving,Acting and Knowing. Erlbaum, Hi&dale, NJ. Sh.;rw, R. and Pittenger, J. (1977) ferceiving the face of change in changing faces: Implications for * theory of object perception. In R. Shaw Lnd J. Bransford (eds.), Perceiving,Acting and Knowing.Erlbaum, HMlsdale,NJ. Sh:iw, R. E. and Cutting, J. (1980) Clues from an ecological theory of event perception. in U. BeBugi and M. Studdert-Kennedy (eds.), Signed and Spoken Lunguage: BiologicalCbnstraintson LinguisticForm. Verlag Chemie, Weinhehn, Fed. Rep. Germany. Sh;:tw, R. and Todd, J. (1980) Abrtract machine theory and direct perception. Behav. Br. Sci., 3, 400-4011. Shl&w,R. and Turvey, M. T. (1981:l Coalitions as models for ecosystems: A realist perspective on perceptual organization. In M. I;ubovy and J. Pomerantz (eds.), PerceprualOrganizufion. Erlbaum, . Hfflsdale, NJ. Shaw, R., Turvey, M. T. and Mace, W. (in press) Ecological psychology: The consequence of a commitment to realism. In W. Weim IXand D. Palermo (eds.), Cognitionand the Symbolic I+ocesses,IL Erlbaum, Hillsdale, NJ. Smokier, H. E. (1968) Contlicting conceptions of confirmation. J. Phil., 225, 300-312. Strong, D. R. and Ray, T. S. (1975 ) Host tree location behavior of a tropical vine (Monsteragiguntea)l by skotatropism. Science, 19;0, 804-806. Stmxnerfield, A. Q. et al. (1981) ‘,i$e structuring of language by the requirements of motor conrrol and perception. In U. Bellugi and M. Studdert-Kennedy (eds.), Signed and spoken 1anREtage: Biologicd constraintson lingdstic form. Verla8 Chemie Weinheim, Fed. Rep. Germmy.
304
M. T. Tiuvey,R E. Shaw,E. S. Reed md W.M. Mace
Suppe, F. (1974) Some philosophical problems in biological speciation and taxonomy. In J. A. Woj-
c&how&i (ed.), ConceptualBasisof the Uassificationof Knowledge. Verlag Dokumentation, Pullach, Munich. Turvey, M. T. (1979) The thesis of efference-mediation of vjsion cannot be rationalized. B&v. Br. Sci., 2. 81-83. Turvey, M. T. and Remez, R. (1979) Visual control of iocomotion in animals: An overview. In L. Harman (ed.), Roceedings of Conference on Inter-relationsamong the Communicative Senses, Asilomar,California,Sept. 29-G&. 2,1978. National Science Foundation, Washington. Turvey, M. T. and Shaw, R. (1979) The primacy of perceiving: An ecohrgical reformulation of perception for understanding memory. In L. G. Niion (ed.), Perspectiveson memory research: Essays in Honor of Vppsala!?nive.rsity‘:F5<h Anniversary.Erlbaum, Hillsdale, NJ. Uexkull, J, von (1957). A stroll through the worlds of animals and men. In C. H. Schiller (ed.), InstinctiveBehavior. International Universities Press, New York. Ulhnan, S. (1980) _Againstdtiect perception. Rehav.Rr. Sci., 3, 373-381. Van Valen, L (1976) Lrulividualistic classes. Phflss. SC&43.539-541. Vickers, J. (1967) Ch%acteristics of projectible predicates. J. Phil., 64, 280-286. Waltz, 0. (1?75) Understauding line drawiugs of scenes with shadows. In P. H.. Winston (ed.), The Psychologyof Computer Vision.McGraw-Hill, New York. Warren, W. H. and Tadd, J. T. (in press) Terrestrial constraints on visual information for dynamic events. In R. Shaw and W. Mace (eds.), Event Perception: An EcologicalPerspective.Erlbaum, H&dale, NJ. Wheeler, II. A. (1974) The universe as a home for man. Am. Sci., 62, 683-691. Wilder, H. T. (1972) Quine on llatural kinds. Aust. J. Phil.,50, 263-270. Wilson,N. I. (1955) Proper*l:yand description. Phil. Rev., 64, 389-404. Wilson, M. (1979) Maxwell’s condition -Goodman’s prqblem. f+. J. Phil. Sci., 30, 107-l 23. Woodbridge, F. J. E. (1913) The deception of the senses. J !%il. Psych& Sci. ,%leth.,ID. S-15. Yates, F. E. (19f;Oa) Temporal organization of metabolic processes: A bbspectroscopic apprp%. in R. Bergman and C. CobeR teds.), GrrbohydrateMetabom’km: QuantitativePhysiologyand MathematicaiModeling. Wiley, New York. Yates, F. E. (19gOb) Physical causality and brain theories. .4m. .I. Physiol., 238 (Reg. Int. Camp. PltyJiol., 7) R277-R290.
Cognition, 9 (1981) 305-309 @Elsevier Sequoia S.A., Lausanne - Plinted in The Netherlands
!HOWpoCite7:A reply to Clark and &hunk .
SUSAN MEMPER and T.)AVID THISSEN” University
of Kansas
Clark and Schunk (1988) argue that the politeness of indirect requests can be computed from the direct meaning of the utterance. This computation invoives determining the costs and benefits to the addressee (A) of’ the speaker’s (S) req.uest. Clark and Schunk consider the 18 requests listed in Table 1. ‘liable 1.
Bmnpk
Reqtuts CSed by Clurk and Schunkt
a. May 1 ask you wher: Jordan Hall is? b. Might I ask you where Jordan Hall is? Could I ask you where Jordan Hall is? C. d. Would you mind telling me where Jordan Hall is? e. Wlould it be loo mui:h troubbe to tell me where Jordan Hall is? f. Can you teh mc where Jordan Hall is? g. Could you tell me where Jordan Hall is” h. Can’t you tell me where Jordan Hall is? i. Do you know where Jordan Hall is? J- Have I already askec, you where Jordan Hall is? k. Did I ask you where Jordan Hall is? I. Dave yr~u told me w%rc Jordan Hall is? n? Do I know where Jordan Hall is? n. Will you tell me where Jordan Hall is? Wou’ld you tell me w:l,:re Jordan Hall is? 0. crc Jxdan Hall is? P. Won’t you tell me WV. Do you want to tell f;:e where Jordan Hall is? II. r. Shouldn’t you tcU mi where Jordan Hall is?
w
? Adapted from Clark an
Schunk, (19801, Table 1.
These requests are classified into six categories based on implications conveyed by their literal mean.~.ng: interrogatives questioning (1) whether S has permission to make a recfuest (a, b, c), (2) whether S is imposing on A by making a request (d, e), (.3) whether A has the ability to conform to S’s request (f, g, h, i), (4) whether A remembers a prior request (j, k, 1, m), --
*Rcprirrt Rcqucsts should be sent ‘:o Susan Kcme:or, Department of Psychology, University of Kansas, Lawrence, Kansas 66045.
306
Susan Kemper and David Thissen
(5) whether A will commit himself to complying with S’s request (n, o, p, q), and (6) whether A is obligated to comply with S’s request (r). Clark and Schunk order these categories as listed above from most to l.east polite. They further allow that requests with (a) conditional mod3 auxiliary verbs (b, g, o) are more polite than those with nonconditional modals (a, f, n), (b) that negative forms (h, p) are less polite than positive forms (f, n) and (c) that requests strongly implying desire (e, q) are less polite than those weakly implying desire (d, n). Clark and Schunk base these orderings on a consideration of the costs and benefits to A of the implications of each form. All predications from this analysis are contirmed by actual ratings, on a seven-point scale, of th.e politeness of each request. The cost-benefit view is contrasted with an idiomatic one such that: each request form is associated with a conventionally determined politeness value. This idiomatic view is rejected by Clark and Schunk on the basis of a plausibility argument: “. . . the mystery is why there is such a tight fit between the benefits and costs implied by the literal meaning and the conventional politeness value. . .” (p. 120). All of the requests considered by Clark and Schunk are interrog,atives formed from an action-assertion by fronting an auxiliary verb. Contrast their set with those used by Kemper and Thissen (1980) listed in Table 2. Table 2..
Example Requests Used by Keeper and 17tissen a. b.
C.
d. e. f. gh. i. j. k. 1. m. n. n.
Rake the leaves. P&se rake the leaves. I think you should rake the leaves. Why don’t you rake the leaves? Do you think you could rake the leaves? Dcra’tyou think you could rake the leaves? Should you rake the leaves? You should rake the leaves. Did you rake the iozzsl The leaves need to be raked. I think the leaves need to be raked. Do you think the leaves need to be raked? Don’t you think tire leaves need to be raked? Should the leaves need to be raked? Do the leaves need to be raked?
As part o+’an investigation of memory for appropriate and inappropriate ~rcquests, ‘Kernper and Thissen used multidimensional scaling (MDS) tc
Howpolite? : A reply to Clarkand ,%hunk
3’07
deter&nine the politeness of these requests. They concluded that the similarity of these forms is determined by two underlying dimensions: ( 1) a dimensian of politeness with the imperative form as the least polite and the Please + imperative form as the most polite and (2) a dimension of directness with requests explicitly mentioning the requested action ccntrasted with those involving need-assertions. Kemper and Thissen assumed that the politeness of these forms was conventionally determined with respect to the imperative or need-assertion. NeeId-assertions are more polite than imperatives; modu.lations (embeddings and interrogatives) render imperatives more polite and need-assrtions less po’1.i te. Note that there is no overlyp in the sets of requests used by Clark and Schunk and Kemper z,.ndThissen. The cost-benefit analysis cannot be easily extended to the Ken per and Thissen set; their forms do not neatly divide into those involving iermission, imposition, ability, etc. nor will conditionality, polarity, or strength account for the politeness dimension. An important difference between the two sets of requests may be the presence versus the absence of reference points provided by the imperative form and the Please + imperative form. To investigate whether the politeness rankings obtained by Clark and Schunk are obtained when such reference points are conside.red$a further multidimensional scaling project was carried out. Subjects
Forty naive native speakers of English participated; credit.
all received course
Materials
Twenty-one different forms of requests were used. The 18 stems used by Clark and Schunk were supplemented with three (imperative, Please + imperative, and Why don’t you + imperative) steins from Kemper and Thissen’s set. The requests all solicited information regarding the location of a prominent campus landmark. Twenty subjects received a booklet containing a list of pairs of requests. They were asked to indicate, on a scale ranging from -9 to +9, the relative effectiveness of the two sentences labeled as A and B. Negative numbers were to represent cases where sentence B was more effective than sentence A; positive numbers were to be assigned when sentence A w:@judged to be more effective than sentence B. An additional twenty subjects received a booklet containing the 21 requests and a sevenpoint rating scale. They were instructed to rate the politeness of each request; ‘1” was to be assigned to requests judged to be very polite while ‘7’ was to be used to very impolite requests.
Results and Discus&n The ratings were collapsed into a nine-point scale of dissimilarity. An aver-
aged lower-half matrix of dissimilarities was obtained. Non-metric scaling using the KYST-2 program (Kruskal, Young, and Seery, 1973) was carried out in two and one dimensions. Stress for the 21 request set was 0.22 and 0.26 for these two solutions, respectively. The one dimensional solution was chosen for further consideration. The averaged politeness rating of each request was also determined. Tabk 3.
Scaledand RatedPolitenessof the 21 Request Items
Do I Know Have I Did I Shouldn’t You Have You Tell Me Can’t You Do You Want Why Don’t You Won’t You Could You ILight I May I Could I i-&d It Do You Know Would You Mind Would You C&i You wiil You Rkase
MDS Dimension 1
Rated Politeness
-1.614 -1.400 -1.269 -1.072 -1 .O35 -0.926 -0.827 -0.536 -0.340 -0.224 +0.109 *0.195 +0,406 to.520 +0.680 +0.796 +0.985 +1.035 +1.259 +1.400 il.857
6.65 6.20 6.10 6.40 6.10 5.65 4.50 5.50 4.80 4.15 3.40 4.15 3.40 2.80 4.05 2.30 2.75 2.05 2.40 2.00 1.30
Table 3 lists the position of each request on the MDS dimension and the average rated politeness of each stem. The dimension appears to be one of politeness. This configuration preserves the relative locations of the imperative, Please + imperative, and Why don’t you + imperative forms as found by Kemper and Thhsen. This interpretation is confiied when the MDS solution rs compared to rated politeness. Rated politeness and scaled position on this dimension correlate r (2 1) = -0.96, p C 0.0 1.
How polite? : A reply to Clarkand Schunk
309
However, when only the 18 requests common to this solution and Clark and Schunk are considered, their politeness ratings correlate r ( 18) = 0.42, p < 0.05 with the position of the requests on the MDS dimension. Further in this solution, requests with conditional modal auxiliaries are less polite than those wMh non-conditional modals in contrast to Clark and Schunk’s finding. When imperative, Please + imperative, and Why don’t you + imperative forms are added to Clark and Schunk’$set of requests, the politeness of the requests is altered. Requests involving “ask” become less polite (t (4) = 3.91, p < 0.05) while the politeness of the requests with “tell” or “know” is not systematically &anged (t (12) = 6.05, p > 0.05). Thus it appears that the politeness of different forms of requests must be determined relative to the standard politeness values of the imperative and Please + imperative forms. A cost-benefit analysis cannot, in general, account for the politeness of a wide range of requests and the relative politeness of different forms is effected by the presence versus the absence of the imperative and Please + imperative forms. References Clark, H. H., and &hunk, D. H., (1980) Polite responses to polite requests. Cog., 8, 11 l-143. Kcmpcr, S., and Thissen, D. (1980) Dimensions of requests. Manuscript submitted for publication. Kruskal, J. B., Young, F. W. and Secry, J. B., (1973) 1tow co use KYST, 4 very flexibleprogram for multidimensional scaling and unfolding. Bell Laboratories (unpublished).
Coflirion. 9 (1981) 311-315 @Elsevier Sequoia S.A., Lausanne - Printed in The Netherlands
Discussion
Politenessin requests: A rejoinder to Kernperand T HERBER’I~ H. CLARK Stanford Llnivmity DALE H. ISCHUNK Unhwsity of Houston
In our study ‘Polite responses to polite requests”, we reported four experiments. In Experiment 1, people rated the politeness of 18 types of indirect requests, such as Could you tell me where Jordan HaI1is? In Experiments 2, 3, and 4, other people rated the politeness of variou;i responses to these requests, such as Yes, d can- it’s down the street and Dower the street. From the findings, we argued two things. First, politeness is rolughiy accounted for by a cost-benefit theory of politeness. Second, understanding such request5 appears to require understanding their direct as well as their indirect meanings. In their reply, Kemper and Thissen ( 198 1) partially redid Experiment 1 and found certain apparent discrepancies. (They did not redo Experiments 2, 3, and 4, which were a major source of support for both of our conclusions.) From these discrepancies, they concluded, “A cost benefit analysis cannot, in general, account for politeness of a wide range of requests”. They did not address our seaJnd conclusion. We suggest that Kemper and Thissen’s conclusions are premature. The discrepancies they found are not replicated in six other investigations. When we tested then explanation for the discrepancies, it was decisively disconfirmed. More generally, the independent evidence for the cost-benefit theory of politeness is so extensive-quite apart r:“;ornour own experiments-that Kemper and Thissen would need more than a partial failure to overturn it.
*Reprint requeste should he sent to: Herbert H. Clark,Deparlment of Psychology, Stanford Utirtirsity, Stanford, CA !4’ OS, U.S.A. 1See Cbgnition, 8,l I1 -143.
312
Herbert H. Clark ml Dale H. Schunk
Kemper and Thiin’s
r!pparentdiscrepancies
When we first examined Kemper and Thissen’s data, their discrepant finaings didn’t seem to make :;ense. So we redid their experiment. In our original experiment, we had asked. 30 people each to rate 54 requests, three each of 18 types; tbz 54 request:; each asked for a different piece of information. Kemper and Thissen had asked 20 people to rate only one instance each of the 18 types of requests, and all 18 requests were for the same piece of information. Did this change in procedure matter’! Tn End out, we had 3cI Wnford University students each rank order 18 requ&s, one of each type, for politeness; all I8 requests asked for the location of nearby Candlestick Park. The students judged how polite the request would be if they were asked it by another student with whom they were acquainted but not close friends. The results of this experiment do not favor Kemper and Thissen. The 18 mean ranks were, first of all, highly reliable:, with a coefficient of reliability of 0.99. These mean ranks correlated 0.88 with our original ratings, but only 0.69 with Kemper and Thissen’s (which themselves correlated only 0.40 with our original ratings). The different:: betwe:en 0.88 and 0.69 is significant, t(U) = 2.19, < 0.05. Previously published data do not favor Kemper and Thissen either. Mohan (3974) asked 80 people to judge the politeness of 27 requests (all requesting the same action). Seven of Mohan’s request types were among Kemper and T&ssen’s selection, and five were among ours. Mohan’s ratings correlated 0.83 with our original ratings, but only 0.70 with Kemper and Thissen’s. The one finding that Kemper and Thissen specifically questioned from our original study was the finding that requests with conditional modals (Could you?, Would you?) were more polite than requests with indicative msdals (Gzpl you?, Will YOU?). In our new data, Could you? was rated as more polite than GZHyou? by 23 of the 30 judges, p < 0.005 by Sign Test. And Would3 Y? was rated as more polite thai7 Will you? by 25 of 30 judges, p < 0.001 by Sign Test. rhese two contrasts, which replicate our original findings, are the reverse of Kemper and Thissen’s data; however, Kemper and Thissen do not provide statistics for their differences. As in our original study, the difference in our new data between Might I ask you? and Muy 1 usk you? was not significant. Previously published data do not favor Kemper and Thissen here either. Bates (1976) asked I2 It;l.t!an adults to judge the politeness of nine requests. They judged fl~~i
