Autism and the "Theory of Mind" Debate

Robert M. Gordon and John A. Barker

Recent research has established that by the age of 4 most developmentally normal children understand that people sometimes respond to the world not as it actually is but merely as they believe it to be. With this understanding, children are better able to anticipate the behavior of others and to attune their own behavior accordingly. In mentally retarded children with Down's syndrome, attainment of such competence is delayed, but it is generally acquired by the time they reach the mental age of 4, as measured by tests of nonverbal intelligence. Thus from a developmental perspective, attainment of the mental age of 4 appears to be of profound significance for acquisition of what we shall call psychological competence: possession of the skills and resources people routinely call on in the anticipation, explanation, and social coordination of behavior.[1]

There is one notable exception, however. Most autistic children lack much of this competence even at significantly higher mental ages, according to a number of recent experimental studies. Most do not seem to understand that people's actions and emotions are contingent on their beliefs. Rather than treating other people as subjects with "points of view," they frequently give the impression of "treating people and objects alike." Asked what the brain does, autistic children speak of it as making people move or run, whereas most children first mention thinking or feeling. Philosophers have sometimes found it useful to invent imaginary people who treat their fellows as, literally, mindless beings.[2] The exotic creatures of philosophical fiction appear congenial and well-adjusted, however, in comparison to those real people who are severely handicapped by autism. People with autism, even many of the most intelligent among them, apparently never succeed in developing a normal understanding of many of the psychological dimensions of human existence, and as a result, they fail to achieve normal interactions with other human beings.

Many autistic people are also abnormal in respects other than psychological competence. The majority are mentally retarded. But their striking failure in many psychological tasks does not appear to be accountable in terms of any broad deficit in intellect. It has therefore been argued by a group of leading researchers that autism is characterized by a specific deficit in psychological competence, or what they call possession and use of a "theory of mind" (Baron-Cohen, Leslie, & Frith 1985; Leslie 1987; Frith 1989; Baron-Cohen 1990).

Some Pivotal Experimental Results

According to many accounts, the major watershed in the development of psychological competence is the capacity to deploy the notion of belief, and in particular the capacity to attribute beliefs that are false or contrary to fact. A classic experimental study, Wimmer & Perner 1983, focuses on false belief and establishes that in normal children this capacity becomes apparent at approximately the age of 4. In the original experiment, children are presented a story, illustrated with puppets, in which the protagonist places an object in one location and subsequently, while he is out of the room and without his knowledge, someone else relocates the object. Where does Maxi go to retrieve his candy when he returns to the room? We of course would predict that he will look for it in the wrong place, and so do most children of age 4 or older. Children under about the age of 4, however, point to or otherwise indicate the actual present location of the candy. Evidently they are unable to adjust for the fact that the protagonist was not in a position to know that the object was relocated. They treat all the facts presented to them in a story as accessible to the protagonist, as if nothing were beyond his ken. It doesn't matter whether something happens in plain sight of the protagonist or whether he is epistemically handicapped. The experiment has been replicated a number of times and it has held up very well and been supported by other research.

Even more striking than the results of the original experiment were those reported in Baron-Cohen, Leslie, & Frith 1985. In this study, a test similar to the one used in Wimmer & Perner 1983 was applied to a group of clinically normal 4-year-old and 5-year-old children, a group of mentally retarded children with Down's syndrome (mean chronological age = 10, mean nonverbal mental age = 5) and a group of high-functioning children with autism (mean chronological age = 11, mean nonverbal mental age = 9). It was found, on the one hand, that the mentally-retarded Down's syndrome subjects gave the "right" response about as frequently as the normal children did. On the other hand, most of the high-functioning autistic subjects gave the wrong response. Even those who had attained the mental age of 9 typically performed at the 3-year-old level on false-belief tasks. Despite being "smarter" than the other subjects, the autistic children appeared to suffer from a specific deficit in at least this aspect of psychological competence.

Results consistent for the most part with those reported in Baron-Cohen, Leslie, & Frith 1985 have been obtained in a variety of subsequent studies.[3] In addition, a later study by Baron-Cohen, Leslie, & Frith (1986) employed a largely non-verbal test in which children were asked to put the frames of a picture story into the proper sequence. Where the sequence was one of mechanistic causality, children with autism performed at least as well as normal children and those with Down's syndrome. But where the "right" sequence (recognized immediately by normal adults) depicts a story involving false belief, the performance of autistic children was no better than chance, and far worse than that of normal and Down's syndrome subjects.

Is Psychological Competence Based on a Theory?

We shall assume in this paper that the available evidence largely favors the hypothesis of a specific deficit in psychological competence. Whether we should characterize it as a "theory of mind" deficit, following Baron-Cohen, Leslie, and Frith, depends on whether we wish to accept the implicit assumption that psychological competence consists in the possession and use of a theory. This is indeed a popular assumption. For the past quarter-century a dominant view in philosophy and the cognitive sciences has been that the resources that underlie common-sense explanations and predictions of behavior chiefly consist in a tacit body of propositional knowledge roughly comparable to a scientific theory. The alleged theory posits unobservable mental states such as beliefs, desires, intentions, and feelings, linked to each other and to observable behaviors by "law-like" principles. These principles are applied to observable situations by way of logical inferences that generate predictions and explanations of behavior. The theory is supposedly called upon by people of all cultures and virtually all levels of intelligence to explain and predict the behavior of others. To apply the theory, it is said, one neither needs nor typically possesses any conscious awareness of the principles one is applying. Careful reflection, however, can often bring them to light. And once they are provided with verbal garb, the principles typically seem obvious, commonplace, and platitudinous‑‑which is taken to be good evidence that they constitute unquestioned presuppositions of a tacit folk theory.

The view that common-sense explanations and predictions of behavior are theory-based is called the "theory" theory. Despite general acknowledgement that this view has serious deficiencies, it was until recently widely conceded that it had no plausible alternative: it was, as Jerry Fodor put it, "the only game in town." Indeed, it has been presupposed in most debates in the philosophy of mind, particularly the debate between those such as Fodor, who think our tacit common-sense theory likely to be vindicated in large part by future science and those (the "eliminativists") who believe future science will show it to be radically mistaken and misconceived.

According to some developmental psychologists who accept the theory-theory, children acquire psychological competence by a process of theory-construction and theory-change, replacing inadequate laws or principles with better ones, progressing toward mature conceptions of mental states such as belief and desire. Consider how on this view one might explain the difference between the way a 3-year-old and a 4-year-old answers the question, "Where will Maxi go to get his candy?" A 3-year-old might reason (consciously or unconsciously) as follows:

To get x, people will go to wherever x is. And Maxi's candy is in location B. Therefore, to get his candy, Maxi will go to location B.

(The same conclusion could also be reached using an immature notion of belief shaped by the principle, "If it is the case that p, then people believe that p.")

In contrast, a typical 4-year-old could deploy the notion of a belief that may or may not correspond to fact. The child might reason, consciously or unconsciously, along the following lines:

Principle: People who put an object in location l will typically believe just afterward that it is currently in location l.

So, just after Maxi put his candy in location A, he believed that his candy was currently in location A.

Principle: People who believe that something is currently in location l will typically continue to believe that it is in location l, unless they come to believe it (was) moved.

Principle: Typically people come to believe that x (was) moved only if they saw x move (being moved) or are told that x (was) moved.

But Maxi neither saw his candy being moved nor was told that it was moved.

So, when Maxi returned, he believed that his candy was currently in location A.

Principle: To get x, people will go to wherever they believe x is.

Therefore, to get his candy, Maxi will go to location A.

The assumption that psychological competence consists in the possession and use of a theory is shared by many recent investigators of the development of psychological competence: the field of investigation has even come to be called, "the child's theory of mind." Developmental psychologists inherited the term "theory of mind" from an article that asked whether the chimpanzee has a theory of mind (Premack & Woodruff 1978). The authors of that paper used the term very loosely, however. They gloss the term as follows:

In saying that an individual has a theory of mind, we mean that the individual imputes mental states to himself and others. (p. 515)

This suggests that Premack and Woodruff were introducing the term "theory of mind" merely as an abbreviation of "capacity to attribute mental states." If that were their sole intent, then of course they would be leaving open the question, "What is the basis of the capacity to attribute mental states?" But their choice of the term theory suggests that they were not leaving this question open but rather taking for granted a certain answer to the question. They were accepting the conjecture that it is possession and use of a theory that gives individuals (human beings and possibly chimpanzees) the capacity to attribute mental states. Indeed, they did attempt to justify their use of the term theory on the grounds that mental states are not observable and that people are able to make predictions on the basis of these attributions. And at the time they were writing there seemed to be no plausible alternative way of explaining how people might make predictions on the basis of attributions of states that lie "behind" behavior.

Some of the developmental psychologists who borrowed the term "theory of mind" from Premack and Woodruff also slide too easily from "capacity to attribute mental states" to "theory of mind" (as noted in de Gelder 1993). They vacillate between the innocuous assertion that by age 4 children are able to attribute mental states and the extravagant speculation that by age 4 children have a relatively mature theory on the basis of which they are able to attribute mental states. Considerable care should be given, therefore, to interpreting Baron-Cohen's claim that people with autism have a "theory of mind" deficit. If he is to be taken as saying only that normal children and those with Down's syndrome have a capacity to make a type of attribution that autistic children do not make, then he is putting forward a very interesting claim that does appear to be warranted by the experimental results he cites. But if he is to be taken as saying that normal children and those with Down's syndrome possess and use a theory (or part of a theory) which autistic children alone do not possess and use, then he is making a claim that appears to be called into question by some of the experimental results he cites in its support.[4] For children afflicted with Down's syndrome, who evidently have the psychological competence characteristic of their mental age, are notably deficient in theoretical abilities; whereas children afflicted with autism, who appear to be psychologically incompetent, are not generally deficient in theoretical abilities (again adjusting for mental age). Why would autistic children who in some cases do even better than the average normal child in mastering theories of other sorts fare so badly in this particular domain? And, even more counter-intuitive, how can it be that Down's children with IQ's as low as 50 are able to master this "theory" about as well as normals do (albeit a few years later), when they master no other theory? In short, one drawback of the theory-theory is that it is not easily squared with the findings regarding the psychological competence of children with Down's syndrome and the psychological incompetence of relatively intelligent children with autism.

We think the evidence suggests that the psychological competence that normal children develop but autistic children do not develop is based not on a theory but rather on a skill that does not essentially require the aptitudes tested in typical "IQ" tests. That is why retarded children with Down's syndrome, even those who in large degree lack these further aptitudes, prove to be psychologically competent. (Of course, one may reasonably expect the relevant skill to be refined and enhanced by these further "IQ" aptitudes. Hence it is not a consequence of our view that low-IQ Down's syndrome children are as likely as normal children to become successful novelists, playwrights, or psychologists.)

Although clearly not deficient in theoretical powers relative to their mental age, autistic children are in other respects very markedly different from other children of the same mental age, including those with Down's syndrome. Some of these differences can reasonably be attributed in large part to their impaired psychological competence. For example, autistic people are notoriously odd and limited in their capacity to communicate with others, both verbally and nonverbally. They also have a severely diminished capacity to enter into normal social relationships, especially with peers. Although these handicaps no doubt set them further behind the pack in psychological competence, the more important causal relationship seems to be the converse: without a level of psychological competence beyond that attained by most autistic people, normal communication and normal social relationships are altogether impossible (Baron-Cohen 1988; also Frith 1989). Thus whatever may prove to be the key to the psychological incompetence of autistic individuals will probably also be the key to their social and communicative impairments.

Pretending and Simulating

There are, however, other autistic abnormalities that cannot be attributed chiefly to psychological incompetence. Among these is the often-remarked failure of autistic children to engage in spontaneous pretend play, whether in conjunction with others or alone. Normal children and Down's syndrome children spontaneously initiate pretend-play and develop the ability to participate in complex, interactive forms of it before they develop the psychological competence required in false belief tasks. In stark contrast, the behavior of children with autism characteristically remains almost totally devoid of any signs of spontaneous pretend-play. The impoverishment of pretend play, particularly the absence of role play and mime play with imaginary objects, is well known. Although studies have shown that many autistic children can, with appropriate prompting, engage in some forms of pretend play (Lewis & Boucher 1988; Ungerer & Sigman 1981), the play is characterized by lack of spontaneity and by stereotypical, inflexible, and repetitive patterns. What is most conspicuous is the absence of other-regarding pretending, typified by role play and joint pretend play, in which two or more children act on a shared pretense (Harris, forthcoming).

Might a faulty capacity for pretense, especially for other-regarding pretense, severely degrade a person's capacity to ascribe mental states? It would, according to the theory of psychological competence that we favor, the mental simulation theory. This theory asserts that psychological competence fundamentally depends on the capacity to use one's own cognitive and motivational resources to "simulate" other people, a capacity that calls on no special theory of mental states.[5] For example, we often predict what another will decide to do by making a decision ourselves‑‑a "pretend" decision, of course, made only in imagination‑‑after making adjustments for relevant differences in situation and past behavior. According to our version of the simulation theory, such vicarious decision-making also underlies the capacity to explain the behavior of others in terms of mental states, and probably also the very capacity to grasp the concepts of such states as belief and desire. If pretending is the key to making and understanding mental state ascriptions, then of course a developmental pathology such as autism that severely restricts the capacity to pretend should also severely restrict a child's capacity to make and understand such ascriptions, even if in other respects the child's intelligence is normal.

To help the reader understand how mental simulation can yield predictions and explanations of behavior, it will be instructive to examine the logical structure of pretend play. Children enter informal games of make-believe by initially pretending something to be true that they do not believe to be true¾for example, that certain globs of mud are pies. By combining the initially stipulated premise with their existing store of beliefs and calling upon their reasoning capacity, they are able to obtain answers to questions not addressed in the initial premise. In the mudpie example, they would typically be able to answer the question, "How many pies are there?" And where there is more than one player, their answers would typically agree: barring a stipulation to the contrary, the answer is the same as the number of (approximately pie-shaped) mud-globs. "Which pie is biggest?" The biggest mud-glob, of course, unless otherwise stipulated. This productive feature of many games of make-believe, pointed out by Kendall Walton, closely parallels our understanding of subjunctive conditionals, as Gareth Evans suggested.[6] Wondering if the bridge would have collapsed had there not been a heavy snowfall, we pretend there hadn't been a heavy snowfall, and then ask whether, in the world thus modified in our imagination, the bridge collapsed. This, with some amendments, is the so-called Ramsey test for evaluating subjunctive conditionals.[7]

But there is a further productive feature of games of make-believe. What the child does with the mudpies depends not only on the stipulated pretend-facts along with his existing perceptions and beliefs, but also his existing desires, values, and norms. Together, these fix or at least constrain the child's answer to the question: What shall I do with these pies?" This further productive feature of games of make-believe parallels our typical understanding of conditionals concerning our own actions under hypothetical or counterfactual conditions. Gordon (1986) makes the connection with pretense explicit in describing how one might predict what actions one would take upon hearing footsteps coming from the basement:

To simulate the appropriate practical reasoning I can engage in a kind of pretend-play: pretend that the indicated conditions actually obtain, with all other conditions remaining (so far as is logically possible and physically probable) as they presently stand; then‑‑continuing the make-believe‑‑try to "make up my mind" what to do given these (modified) conditions. I imagine, for instance, a lone modification of the actual world: the sound of footsteps from the basement. Then I ask, in effect, "What shall I do now?" And I answer with a declaration of immediate intention, "I shall now...." This too is only feigned. But it is not feigned on a tabula rasa, as if at random: rather, the declaration of immediate intention appears to be formed in the way a decision is formed, constrained by the (pretended) "fact" that there is the sound of footsteps from the basement, the (unpretended) fact that such a sound would now be unlikely if there weren't an intruder in the basement, the (unpretended) awfulness of there being an intruder in the basement, and so forth.

As in pretend play, an initial premise‑‑here, the hypothetical condition‑‑is added to one's store of beliefs, desires, and other inputs to intention-formation and decision-making. In one important respect, however, this kind of simulation is unlike children's games of make-believe (and also unlike rehearsals and drills). Although the simulation may be accompanied by autonomic arousal and some expression of emotion, it stops short of overt action. One does not carry out the decision, say, to call the police, even in overt pretend-play. Our motivational, emotional, and decision-making systems are running "off-line," as it were, disengaged from their natural output systems.

The simulation theory says that in predicting, explaining and interpreting another's behavior, likewise, we run the explanation or prediction through our own motivational and emotional systems, utilizing our own capacity for practical reasoning and decision-making. In simulating another, however, it is often not sufficient to imagine being in the other's situation‑‑that is, to employ our imagination merely to ask, "What would I do, believe, want, and feel were I in Smith's situation?" For this would leave open the further question, "What about Smith: what does Smith do, believe, want, or feel in that situation?" Rather than simulate ourselves in Smith's situation, we must simulate Smith in Smith's situation (as it appears to Smith).

This would entail a further, more complex use of pretense. When one is predicting one's own actions or reactions in hypothetical conditions, the initial premise‑‑for example, that there is a sound of footsteps coming from the basement‑‑is simply stipulated. But when we explain or predict another's behavior in such a situation, one may in addition have to make adjustments of various kinds. These might be based on knowledge of the other's actual behavior in related situations in the past: one tries to become in imagination a person who might have acted as the other did in such situations. In false belief tasks like those presented in Wimmer and Perner 1983 and Baron-Cohen, Leslie, & Frith 1985, the needed adjustment is a simple one: just ignore one or more of the facts, and then carry on as before. In the Maxi example, what would need to be ignored is what happened while Maxi was outside, namely that his candy was moved. In effect, one follows Maxi as he goes outside, and thus fails to "see" what subsequently transpires inside: looking back, one sees only the outside of the house. Then, within the context of this pretense, one simply states where the candy is. One will then predict correctly that he will look for his candy, not where it actually is, but where it was before it was moved. Notice that, instead of invoking a special folk-psychological principle that people typically assume that objects (to put it crudely) tend to stay put, one simply relies on one's own assumption that objects tend to stay put. (For a more detailed account of the methodology of mental simulation, see Gordon 1992.)

Until and unless children develop the capacity and the motivation to make these imaginative adjustments, we should expect them to explain and predict as if everything they themselves count as fact were accessible to the other as a basis for action and emotion. We should also expect them to allow no false propositions into the other's data base: no false beliefs, in other words. Because people with autism are in general severely deficient in the capacity for pretense, particularly other-regarding pretense, they would be unable to make the adjustments posited by the simulation theory. Therefore, the theory predicts, correctly, that these people will generally not succeed in the false belief tasks, even where in other respects their intelligence is normal.

Pretending and the "Theory of Mind Mechanism"

We argued earlier that the theory-theory is hard put to explain why children with Down's syndrome, who are generally poor theorists, have far more psychological competence than relatively intelligent children with autism, who are generally good theorists. The simulation theory would explain this in part by the fact that children with Down's syndrome, though poor theorists, spontaneously engage in complex, interactive forms of pretend play; whereas children with autism, though often good theorists, do not spontaneously engage in such play. But unlike the simulation theory, the theory-theory does not hold psychological competence to depend on a capacity for certain kinds of pretending.

A further move, however, is available to theory-theorists. They may argue that a theory of mental states differs in fundamental ways from theories in other domains, particularly in the nature of the concepts it employs. For one thing, it is a theory of intentional states, of states that are about something. And it may be suggested that possession and use of a theory of intentional states requires a special cognitive mechanism, and perhaps further that the same special mechanism is needed for pretense. This is proposed by one theory-theorist, Alan Leslie, who posits a special computational mechanism, which he dubs the "Theory of Mind Mechanism," as requisite for overt pretend play as well as the understanding and recognition of intentional states (Leslie 1987; Leslie & Roth 1993). This proposal would parallel the simulation theorist's suggestion that both are implemented by "off-line" processing: utilizing systems that are normally dedicated to perception, cognition, motivation, emotion, and decision-making, but using them in at least partial isolation from their normal input and output systems. The simulationist's idea is that the partial independence from input systems would explain, among other things, the freedom we have, in pretend play and in our representation of the content of others' mental states, to portray the world in contrary-to-fact ways; yet, because output systems are not engaged in the normal way, this free play goes on within a "protected" context in which some of the normal consequences, especially but not only some of behavioral consequences, do not actually ensue.

Leslie maintains that it is the Theory of Mind Mechanism that makes this protected free play possible. What the mechanism does (to give a greatly simplified account of a rather complex theory) is to enable these systems to function at a higher semantic level: instead of manipulating object-level representations, as in their normal engagement with the world, they are enabled to manipulate representations of representations, or "metarepresentations." The chief matter at issue between Leslie's hypothesis and the off-line hypothesis of the simulation theorist would seem to be this: Do our systems operate at this higher representational level in pretense and in mental state attribution, or do they operate in much the same way as in real-world engagements, but off-line? We believe Leslie's answer misrepresents the nature of pretense: the recognition and understanding of pretense might arguably involve metarepresentation, but not, ordinarily, the production of pretense. We also think the off-line hypothesis explains more, requires fewer specially dedicated resources, and comports better with evidence from other domains, such as imitation and mimicry. Although mimicry is discussed briefly in the following section, detailed discussion of Leslie's hypothesis would not be appropriate in a short essay on autism. We intend to discuss the hypothesis elsewhere, because it appears to be the only way the theory-theory can be squared with the evidence that in the populations tested, psychological competence is better correlated with capacity for pretense than with capacity for theoretical understanding.

Simulation and Imitative Behavior

The capacity for simulation involves not only the deliberate procedure of "putting oneself in the other's place" but also a number of automatic, unconscious responses. For example, there is subliminal muscular mimicry of the bodily postures and especially the facial expressions of others. Where the other's face bears an expression of emotion, adoption of a similar expression tends to produce a similar emotion in oneself. Even when it does not produce a like emotional response, it at least gives the simulator the wherewithal to recognize the other's emotion. The automatic response to facial expressions is complemented by another mechanism. Like many other animals human beings have an automatic tendency to direct their eyes toward the target of a conspecific's gaze. This mechanism automatically turns one's own attention from the other's response to its environmental stimulus: to the "object" of the other's attention or emotion (what it is about), or the "object" (aim, goal) of the other's action.[8] The tendency is activated particularly when another exhibits startle, terror, or some other strong reactive emotion, or shows attentiveness and interest. In normal children, all of this emerges in the first year. If psychological competence essentially depends on a capacity to simulate others, these imitative mechanisms prove to be important, perhaps even essential, steppingstones to competence. In particular, they facilitate finding the environmental explanation of another's action and emotion.

In the case of children with autism there is strong evidence that at least the gaze-tracking response is largely absent and some evidence of deficiency in the tendency to mimic emotional expression. (Baron-Cohen & Ring 1993; Meltzoff & Gopnik 1993). Thus we should expect these children to be deficient in both the tendency to search for reasons for (or objects of) action and emotion and the capacity to locate them in the environment. These problems are not likely to show up in the artificial tasks presented in most false belief experiments, which call primarily for predictions of action rather than explanations. Furthermore, they offer the subject only a narrative rather than a live, expressive protagonist. It is no surprise, then, that in the specific behavior tested in those experiments, the autistic individual seems for the most part to resemble a normal 3-year-old. But where the task is one of explaining another's behavior in terms of an environmental stimulus or "object," and the subject is allowed to see the other's facial expression or overt behavior and track the direction of the other's gaze, the simulation theory predicts that children with autistism will perform far below the level of normal 3-year-olds.

Conclusion. We claimed that the evidence suggests that the psychological competence that normal children develop but autistic children do not develop is based not on a theory but rather on a skill. That skill, it appears, includes a capacity for egocentric recentering and a capacity to be engaged as an agent in a world imagined to deviate somewhat from the actual world. These capacities appear to be intact in children with Down's syndrome but deficient in children with autism. Also deficient are some of the ancillary imitative mechanisms that would ordinarily facilitate simulation, particularly by producing emotional responses that copy those exhibited by others and by turning the simulator's attention toward the environmental causes and "objects" of others’ emotions and actions.

It should be remarked that this is the first philosophical paper we know of on the topic of autism. We have tried to bring to bear some recent philosophical thinking about the nature and acquisition of mental concepts and the nature of pretense. But we acknowledge that this is at most a small contribution toward understanding what is still a mysterious pathology.

Acknowledgements

The authors are indebted to Simon Baron-Cohen and Paul Harris for comments that were extremely helpful, especially in filling in some of the gaps in our knowledge of recent empirical research. Baron-Cohen kindly furnished preprints of some of the papers since published in Baron-Cohen, Tager-Flusberg, and Cohen 1993. We thank the Editors for a number of suggestions that helped shape our presentation.

References

Baron-Cohen, S. 1988. Social and pragmatic deficits in autism: cognitive or affective? Journal of Autism & Developmental Disorders, 18: 379-402.

Baron-Cohen, S. 1990. Autism: a specific cognitive disorder of "mind-blindness." International Review of Psychiatry, 2, 81-90.

Baron-Cohen, S., & Cross, P. 1992. Reading the eyes: evidence for the role of perception in the development of a theory of mind. Mind and Language, 7:172-86.

Baron-Cohen, S., Leslie, A.M., & Frith, U. 1985. Does the autistic have a "theory of mind"? Cognition, 21, 37-46.

Baron-Cohen, S., Leslie, A.M., & Frith, U. 1986. Mechanical, behavioral, and intentional understanding of picture stories in autistic children. British Journal of Developmental Psychology, 4, 113-25.

Baron-Cohen, S., & Ring, H. 1993. The relationship between EDD and ToMM: Neuropsychological and neurobiological perspectives. In P. Mitchell & C. Lewis, eds., Origins of an understanding of mind. Lawrence Erlbaum Assoc.

Baron-Cohen, S., Tager-Flusberg, H., & Cohen, D.J., eds., 1993. Understanding other minds: perspectives from autism. Oxford University Press.

Davies, M. & Stone, T., eds. 1994. Mental Simulation: Philosophical and Psychological Essays. Oxford: Blackwell Publishers

de Gelder, B. 1993. Intentional ascription, autism, and troubles with content. In J. Verschueren, ed., Pragmatics at issue. Amsterdam/Philadelphia: J. Benjamins.

Evans, G. 1982. The Varieties of Reference, edited by J. McDowell. Oxford: Oxford University Press.

Fodor, J. A. 1975. The Language of Thought. New York: Thomas Y. Crowell.

Fodor, J. A. 1978. Propositional attitudes. The Monist, no. 4. Reprinted in Representations, edited by J. A. Fodor. Cambridge, MA.: MIT Press.

Fodor, J. A. 1987. Psychosemantics. Cambridge, MA: MIT Press.

Frith, U. 1989. Autism: explaining the enigma. Oxford: Blackwell.

Goldman, A. I. 1989. Interpretation Psychologized. Mind and Language, 4:161-85.

Goldman, A. I. 1992. In Defense of the Simulation Theory. Mind and Language, 7:104-19.

Gordon, R. M. 1986. Folk psychology as simulation. Mind and Language, 1, 158-71.

Gordon, R. M. 1992. The simulation theory: Objections and misconceptions. Mind and Language, 7:11-34

Harris, P. 1989. Children and Emotion. Oxford: Blackwell.

Harris, P. 1992. From Simulation to Folk Psychology: The Case for Development. Mind and Language, 7:120-44.

Harris, P. forthcoming. The work of the imagination.

Leslie, A.M. 1987. Pretense and representation: The origins of "theory of mind." Psychological Review, 94: 412-26.

Leslie, A., & Roth, D. 1993. What autism teaches us about metarepresentation. In Understanding Other Minds: Perspectives from Autism, edited by S. Baron-Cohen, H. Tager-Flusberg, & D. J. Cohen. Oxford: Oxford University Press.

Lewis, V. & Boucher, J. 1988. Spontaneous, instructed and elicited play in relatively able autistic children. British Jnl of Developmental Psychology, 6:325-39.

Meltzoff, A. & Gopnik, A. 1993. The role of imitation in understanding persons and developing a theory of mind. In S. Baron-Cohen, H. Tager-Flusberg, & D.J. Cohen, eds., Understanding Other Minds. Oxford: Oxford University Press.

Newell, A. 1980. Physical symbol systems. Cognitive Science, 4: 135-83.

Premack, D. & Woodruff, G. 1978. Does the chimpanzee have a "theory of mind"? Behavioral and Brain Sciences, 4, 515-26.

Sellars, W. 1956. The myth of the given: Three lectures on empiricism and the philosophy of mind. In Minnesota Studies in the Philosophy of Science, vol. 1. Minneapolis: University of Minnesota Press.

Stich, S. & Nichols, S. 1992. Folk psychology: Simulation or tacit theory? Mind and Language, 7, 35-71.

Ungerer, J. A. & Sigman, M. 1981. Symbolic play and language comprehension in autistic children. Journal of American Academy of Child Psychiatry, 20, 318-37.

Walton, K. L. 1973. Pictures and make-believe. The Philosophical Review, 82: 283-319.

Wimmer, H. and Perner, J. 1983. Beliefs about beliefs: representation and constraining function of wrong beliefs in young children's understanding of deception. Cognition, 13, 103-28.

NOTES