 On behalf of the McLean Center and the Grossman Institute for Neuroscience and Quantitative Biology and Human Behavior, I'm delighted to welcome you to this iteration of our 2015-16 lecture series on neuroethics. As you know, this lecture series was organized by John Mansell, the director of the Grossman Institute, Peggy Mason, Professor of Neurobiology, and Dan Solmezzi from the McLean Center. It's my pleasure now to introduce our speaker today, Tyler Burge. Professor Burge is the Flint Professor of Philosophy at the University of California at Los Angeles. He received his doctorate from Princeton and has taught at universities around the world, including MIT, Munich, Stanford, and Harvard. Professor Burge has published many papers and books on philosophy of mind, the history of philosophy, and epistemology. Professor Burge has worked on anti-individualism with respect to mental states. His books include Truth, Thought, and Reason, Essays on Fraga, that was Oxford University Press, Foundations of Mind from Oxford, Origins of Objectivity, Oxford 2010, and his most recent book is entitled Cognition Through Understanding, Oxford 2013. Professor Burge is a member of the National Science Foundation, the Institute Internationale to Philosophy, the British Academy, and the American Philosophical Society. Additionally, he's been awarded a Guggenheim Fellowship, is a Fellow of the American Academy of Arts and Sciences, and since 1999 has been the President of the Pacific Division of the American Philosophical Association. Today, Professor Burge will speak to us on the topic that you see behind me, do apes and very young children attribute mental states. Please join me in welcoming Professor Tyler Burge to Chicago. Please to come back to Chicago. A widespread view among philosophers and psychologists is that human children well under age two and apes have been shown to have rudimentary capacities for representing and attributing mental or psychological states. I believe that this view is mistaken. It rests on over-interpreting what relevant experiments show and on insufficiently precise understanding of mentality. I believe that the available evidence supports taking the relevant subjects to be attributing agency, not mentality. The attributions of agency do not attribute mental states. I think that we do not know whether non-human animals attribute mental states at all. I think that we do not yet know with any firmness at what stage human children attribute them. I think that the best conjecture, given present evidence, is that human children attribute mental states between ages three and six. I think that even the widely accepted view that children by age four attribute representational mental states, though probably correct, is not as firmly grounded in experimental evidence as it is widely thought to be. I will focus mainly on the view that children attribute mental states under the age two, age of two. The methodological upshot of the discussion will be that philosophers, developmental psychologists, primatologists, and cognitive neuroscientists need to develop a more sophisticated conception of mental kinds and correspondingly more sophisticated experimental methods for delineating such kinds if a lasting understanding of the phylogeny and ontogeny of the mind is to be achieved. There are two primary types of distinction that need to be more carefully attended to in theorizing about the powers of apes and very young children. I note these distinctions here to provide general orientation and I will elaborate them later. First is the need to distinguish carefully between attributions of psychologically guided action from attributions of action that do not represent the action is psychologically guided. Many psychologists and primatologists recognize the distinction between action attributions that do and do not attribute psychological sources of action. However, the recognition is commonly associated with over-simple conceptions of the resources for understanding action without attributing psychological sources for the action. A lot of action in the animal world is guided and caused by non-mental cognitive states. The agent never represents a goal or target. Who thought about such action would improve theorizing about early mentalistic attribution. Of course very young children and apes respond to actions that are in fact guided by psychological states. So young children and apes and their parents, parents of apes and children all have mental states. The issue is whether they attribute psychological states that guide the action of the beings that they interact with. Second, there is a need to distinguish perceiving from non-perceptual sensing and a need to recognize that sensing is not in itself a mental or psychological state. Primitive organisms down to the level of bacteria have sensory capacities. These capacities do not imply mentality. They do not imply consciousness or representation in any distinctively psychological sense of the term representation. Consciousness and representation in a distinctively psychological sense are the basic marks of mind. Sensory states function to causally covariate with entities in the environment, but functioning causal co-variation, what I call information registration, is not by itself representation in any distinctively psychological sense. Bacteriums or worms, sensory states co-variate with light and dark or cause to co-variate by light and dark and function to do so. That is enough for information registration. Information registration is not in itself perceptual representation or any other mental or psychological phenomenon. The evidence that would be needed for taking young children or apes to attribute mental states like perception, intention, and belief needs to be better identified. Such evidence must bring out that children attribute representational perspectives, different representational takes on given objects or properties in the environment. A few researchers have tried to test for this phenomenon. These experiments show a realization that show the presence of representational psychological states. One must show that subjects have a genuinely representational perspective, but all the experiments carried out so far underestimate what is needed. All conflate children's tracking of perspectives with children's tracking of properties. Real sensory states can track properties or even objects by their properties. Nothing in the experiments rules out such sensory non-representational tracking. Experiments are needed to show that holding plausibly indicated properties constant, the child is attuned to the other individuals acting differently with respect to a property because of different proximal stimuli or different informational access to the property. Experimental designs that clearly distinguish a child's taking another to track properties possibly in sensory stimulus dependent non-representational ways and a child's taking another to have a different representational perspective on the same property will inevitably require considerable ingenuity. Such experiments have not yet been produced. One of the points of the present work is to encourage finding such designs. I want to say here at the outset that I follow what is called Morgan's Canon throughout as I think all good science in the relevant area already does at least in its reflective methodological commitments. According to this principle, child or animal behavior is not to be explained in terms of higher psychological processes or later evolved psychological processes if the evidence can be well explained in terms of processes that stand lower in the scale of psychological evolution and development. It is true that we know that infants will come to attribute mental states. We do not know I think that non-human animals do so. Some infer that this developmental difference provides ground to take very young children to attribute mental states even if behavioral evidence supports equally well taking them to be making more modest attributions. I think that no such inference will be sustained by actual long run scientific practice. Ultimately, Morgan's Canon will assert itself. It is true that we know that adults tend to hyper-intellectualize or over-attribute mentality to others attributing mental states where there is no scientific basis for doing so. Some infer from these facts that there is ground for taking very young children to be attributing mental states when perhaps the evidence does not support their doing so. This inference really I think rather blatantly begs the question. It overlooks the fact that we know that adults attribute mental states given that we know this we're in a position to show empirically that they over-attribute them. But whether very young children and other animals attribute mental states and when the children begin to attribute them is not known apart from specific evidence that they do so. One cannot over-attribute mental states unless one has representations of mental states to attribute. Okay, that's preliminary. Modern discussion of primitive attribution of mental states begins with a paper by Primac and Woodruff from 1978. They wrote, quote, an individual has a theory of mind if he imputes mental states to himself and others. A system of inferences of this kind is properly viewed as a theory because such states are not directly observable and the system can be used to make predictions about the behavior of others. Primac and Woodruff proposed attributing a theory of mind to Schimpad's ease. They cited a 14-year-old chimp's ability to anticipate behavior of a human who was confronted with such problems as acquiring an out-of-reach fruit. They showed that the apes chose from a group of photographs the one that depicted the best way to solve the problem. Primac and Woodruff took the chimpanzee's behavior to require that the behavior attribute, quote, at least two states of mind to the human actor, namely intention or purpose on the one hand and knowledge or belief on the other. End of quote. This conclusion was criticized in commentaries on the paper. Three philosophers independently claimed that the chimpanzee might be anticipating what itself would do to solve the problem. They thought that Primac and Woodruff's conclusion would be more convincing if the ape anticipated the agent's acting in a way different from the way the ape would act. This suggestion was taken up in developmental psychology for children some years later. Verma and Perner administered what widely has become known as the false belief test. I'm going to use this term, but I use it to apply to members of the family of tests inspired by Verma and Perner's test. Emphatically, I do not imply by my usage that the tests are good tests for false belief. I'll come back to this. Children observed a puppet storing a piece of chocolate and then leaving the room. Another puppet discovers the chocolate and moves it. The first puppet returns. The children are asked where that puppet will look for the chocolate. Children younger than four years old do not say, as we would, that the puppet will look for the chocolate where he left it. Most will say that he will look where the second puppet put it. By contrast, 57% of children in the four to six-year-old range say that the puppet will look where he left it. So before four years old, they answer very differently from how we would and after four, at least 57% answer more or less as we would. This result and variants have been frequently replicated. The result was initially widely regarded as showing that three-year-olds do not make genuine false belief attributions whereas four-year-olds do. But some years later, Onishi and Balorjean produced a variant of the original false belief test. 15-month-olds passed this variant. The test was ingeniously devised to circumvent the children's having to answer questions. They were allowed to operate spontaneously. Subjects were habituated to an actor's hiding an object in one of two boxes and then retrieving the object. In the crucial condition, the actor put the object in one of the boxes. Then the subjects were shown the actor behind a screen that clearly blocked the actor's vision of the boxes. The subjects then saw the object moved to the other box. Subsequently, the screen was removed and the actor looked into one of the boxes. The subjects looked, the children, looked reliably longer when the actor took a look into the box that contained the object than when the actor looked into the box in which the actor had originally placed the object. The authors maintain that this result showed that, quote, even young children appealed to others' mental states, goals, perceptions, and beliefs to make sense of their actions. And a run of work in the same vein followed. And it's still going on. A false belief test, the false belief test was, I think, misleadingly named from the beginning. The experiment is certainly not a crucial test for attribution of belief. Neither the original test carried out by Vermeer and Perner on three and four-year-olds, nor the subsequent tests conducted by Anishin Balarjian, in themselves show anything at all about belief. Belief is a propositional attitude. To be relevant to belief, the tests would have to show that children attribute a propositional attitude with propositional structure. No false belief test even bears on this issue. The fact that children can anticipate where an individual will look, even when the individual will not look, where the children themselves would look, does not in itself show anything about whether the children attribute the specific attitude belief. Other committal representational states might produce the same behavior. The literature coming out of Anishin Balarjian's experiment says nothing at all about whether the young children are attributing propositional attitudes like beliefs to others or other more primitive mental states. The situation for false belief tests themselves, however, is worse than this point suggests. The tests not only show nothing about attribution of the specific attitude belief, they do not in themselves show that the children attribute psychological states of any kind. All the experiments bear on subjects assessing whether an action in a given circumstance will fulfill its function. All hinge on what sensory-based states the agent is acting on and whether such states are geared to yield successful agency, but neither sensory access nor agency require mentality. Ticks crawl and suck, rotifers swim and eat. There's no reason to think that such animals have mental states. Their actions are goal-oriented. Their actions have targets. Their actions are guided by sensory states or retentions of sensory states, but these organisms do not set or represent the target. Nor do sensory states constitute mentality. Rotifers sense chemicals and light. Tents ticks sense heat. By sensing heat, ticks sense arms that can be worth penetrating. These facts do not support attribution of mental states. Of course, children track actors' capabilities that are richer than those of a tick. I'll come back to this. In retrospect, some scientists who made use of the false belief tests to argue that children in their second year attribute beliefs have acknowledged that the tests by themselves do not support taking very young children to attribute mental states. They appeal to the variety and accumulation of experiments to support their views, but these appeals are vague. The variety of experimental arguments that have been made for taking very young children under the age of two to attribute mental states are all variants on the false belief tests. The admissions that false belief tests do not in themselves show much about belief mask the enormous influence that the tests have had in convincing many psychologists that children at one or another age attribute mental states. Among those who have taken false belief tests to have force in showing that children attribute false beliefs or other mental states, there remains a serious divide in the scientific community. The divide is between those who hold that children show this capacity only after age three and those who hold that children show it early in their second year. This divide is often associated with a dispute as to whether the capacity is a late emerging in eight capacity, say in the two-year-olds or the 14-month-olds, or a largely learned capacity. A range of capacities that may be relevant to win children attribute mental states certainly does seem to emerge around age four. These capacities provide some reason to think that rudimentary, metalistic attributions may occur at around this age. For example, children in the range of four to six years appear to attribute mental states in their use of language. They give coherent, verbal explanations of relevant behavior that appeal to mental terms. The explanations cite perceptions, beliefs, and desires who appear to. Their explanations follow patterns that broadly resemble explanations by adults. Children's predictions of behavior are similar. Four-year-olds are good at simple explanations of the sources of their beliefs. Say they say they're from perception, or from being told, or from figuring something out. Four-year-olds seemingly distinguish between sources that guess answers correctly and sources that know answers. Children use contrastives to mark discrepancies between an individual state and reality. They use language that appears to attribute deception and change of mind. Although the exact times when these changes emerge vary, there's a broad change in performance in these respects at roughly this age period. Those who believe that the capacity emerges at around 15 months hold that the capacities events in these experiments are present at an earlier age, but are masked by the requirements of the experiments. Those who hold that the capacity to attribute false beliefs emerges after age three, cite the just-mentioned range of behavior that emerges about then. Indeed, this group has, to its credit, always emphasized from the beginning that any basis for taking children to attribute mental states would have to bring interlocking, converging evidence together. I believe, however, that even they have underestimated the capacities of alternative explanatory schemes for explaining relevant behavior. I'm not gonna take a position on the details of the relevant development between ages 15 months and four years. I'll not discuss what is innate and what is learned. I do think that the evidence that I've just outlined about three and four-year-olds has some force, but I also think that it hasn't been investigated with sufficient precision to determine whether, for example, the children mean by their word, beliefs or desires at this stage what we mean. So I think looking at this really hard is necessary to be sure that even this evidence shows that three to four-month-olds, three to four-year-olds are attributing mental states. If I had to bet, I would say they are, but I think there's been so much confidence in the community that of course they're attributing these states that people may have jumped to the conclusion even on this evidence too quickly. I want to explain, so what I'm gonna focus on is to explain why taking experiments with children in their second year to show that they engage in attribution of false belief or indeed attribution of any other mental state is not well-founded. The conceptual difficulties that I will cite are however largely shared between proponents of an early emergence of mentalistic attribution and proponents of the later emergence. All false belief tests events a capacity on the part of infants to decouple their own point of view on the world and how to act on it from an expectation of how the agent whose actions they anticipate will act on the world. This is certainly a large developmental step. Infants at roughly ages 13 to 15 months of age take it. They show a decoupling that represents others as acting differently from the way they would act and as having states that provide different information. Slightly older infants represent other agents as having consistently different action targets from their own. This capacity constitutes another sort of decoupling. For example, at 14 months, children tend to offer an experimenter food that the infant prefers, but at 18 months children offer food that the experimenter expresses preference for even when the infant does not itself prefer the food. So there's a change between 14 and 18 months on that issue. Four-year-olds make various decoupling advances again. I think that three and four-year-olds do fail to pass false belief tests because of the performance difficulties of combining decoupling with responding to linguistic demands. Four-year-olds manage to succeed while meeting these demands. The initial decouplings in the second year of life are to be contrasted with capacities already in place to anticipate other agents' actions. So at six to seven months, human infants can identify agency and targets of action. They can anticipate efficient means of reaching these targets. They can form different anticipations as to how an agent will act depending on whether that agent has sensory access to information relevant to the action. These are capacities present in apes. Both sides of the dispute about attribution of belief at 15 months tend to agree that these earlier capacities in six to seven months, month-olds, events, mentalistic attributions, not attributions of belief, but only attributions of desire, perception, knowledge, and ignorance. Even at this earlier stage, six to seven months, a child can take lack of sensory access to prevent the agent from reaching a target. The child can take the agent's lack of sensory access to lead to behavior that differs from the action that the child would engage in given that the child has sensory access. Many psychologists take this point to amount to attributing to the child representation of different states of belief or knowledge or perception in the agent. I'm sorry, it shouldn't be belief, perception or knowledge in the agent. So these scientists don't think that knowledge requires belief. This idea, however, confuses information registration with representation in a distinctively psychological sense. I've stressed this distinction before, and I'll come back to it again. Both before and during decoupling, a child can distinguish another agent's information and actional inclinations from its own. What differentiates decoupling from this capacity is anticipation of specific actions that differ from the child's. In attributing mere lack of sensory access, a child or ape can expect the agent not to do certain things. For example, if an ape sees an obstacle between food and a rival's line of sight, it can anticipate that the rival will not head for the food. After decoupling, a child can anticipate specific actions that will fail to reach their targets because of relatively specific informational states that differ from the child's. None of these capacities specifically events attribution of mental states, in my view. There has been resistance to using false belief tests to support taking children to attribute false beliefs in their second year within the science. So Joseph Periner, one of the authors of the original false belief test, has maintained that at that young age, the test can be explained by attributing to the children a battery of learned behavioral rules. He claimed that the children anticipate that the actor will go to the empty box because the actor, the object in the box were together when last seen. These are, however, there are, however, many other experiments that purport to events false belief tests to which this rule does not apply. Of course, it is possible to propose a new behavioral rule for each experiment, but unless the behavioral rules can be delimited and shown to follow some principle, like learned association, for example, this approach will lack explanatory power and fruitfulness in gendering new experiments and predictions. And this approach has been gradually, I think, let go over the last decade or so. There does seem to be a unity in the decoupling phenomena that goes beyond learned behavioral association. Children anticipate specific goal-oriented acts by agents using a representational apparatus that identifies a target and that takes account of norms of efficiency, of sensory information and access to information and so on. It's known that children have this apparatus in their first year of life at six to seven months. One would need good reason to maintain that they're not using it in responding to false belief tests. On the score of theoretical power and promise, the behavioral rule approach seems to be at a disadvantage. I'm gonna skip another type of approach which is somewhat similar. As I've indicated, it's important to distinguish representations of agency that do and do not take actions to be guided by psychological states. There's still a very wide tendency in developmental psychology and primatology and in fact in philosophy to assume that if an event is represented as being an act by an agent, the event is represented as caused by psychological states such as intentions. Intentions represent the target of an agent's action as noted a lot of animal agency is not caused by intentions or any other psychological state. One cannot assume that representation of agency involves representation of psychological states. There are those who have emphasized in the science, emphasized a distinction between teleological, sorry, attributions and attributions of psychological sources of action. Unfortunately, the most serious and influential attempt to draw and apply the distinction has helped to lead the field to misconstrue the distinction and underestimate its power. So this is another movement within the developmental psychology. So Gargi and Shibra presented an account that distinguished attributions of agency with and without attributing specifically psychological states that represent goals. They took the contrast to mark a distinction in developmental stages in very young children. So first on their account, the explanatory ground referred to in teleological explanation is the target of the action. The explanatory ground then occurs later in time than the element to be explained, the action. They point out that this feature contrasts with causal mentalistic explanations. Second, teleological attributions refer to only three conditions on their account of it. They refer to the behavior that is taken to be a gentle. First, second, to a future state of reality that is the target or goal, perhaps including a target object or theme in relation to which the action is anticipated to be efficient. And third, to those aspects of reality that form conditions for taking the behavior to be an efficient means to bring about the in-state. There's no room in the account for attributing an antecedent state of the agent. Third, Gurgi and Shibra take teleological explanations to be quote, reality-based. The explanations must represent the world as the explainer represents it. The explainer can observe and infer actions and action targets that differ from the explainer's own. The explainer can observe that an action can fail to meet its target. The failure can be represented as resulting from lack of sensory access to an aspect of the environment. These are all relations between agent and environment. Since there's no room in the explanations for representing states of the agent, there's no room for representing states of the agent that are out of sync with reality as the explainer represents it. But none of these three features is necessary to attributions of agency that do not attribute psychological sources of agency. First, such attribution of non-psychologically guided agency can be causal. Second, it can attribute states of the agent, non-psychological action sets or sensory, non-perceptual, non-representational sensory states. And third, it need not be reality-based. It is open to the explainer to represent non-psychological action sets that are out of sync with reality. The one restriction on such attributions, which I'll discuss in just a moment, is that they not represent the relevant cognitive action sets or sensory states is psychological. Gergy and Shibra postulate a developmental transformation from teleological explanation to causal mentalistic explanation. They see teleological explanation as, quote, breaking down, unquote, when someone acts on false beliefs or pretense. They mean that that form of explanation, the teleological form, cannot account for children's attributional responses in false belief tests. But non-psychological action attribution can attribute antecedent states that are out of sync with the attributors representation of reality as long as those attributed states are non-psychological, as long as the states are not strictly false or otherwise non-veretical. Passing the false belief test does not in itself show that children have shifted from non-psychological action attributions to attributions of psychological states that cause the action. Many developmental psychologists and primatologists reject Gergy and Shibra's non-causal model for action attribution by very young children. Many hold that children attribute intention and desire before they pass the false belief test. Despite these controversies, Gergy and Shibra's claim that non-metalistic action attribution cannot account for children's passing the false belief test is widely accepted. The idea is that if a child attributes an informational state whose relation to the environment differs from the child's own psychological representational states, that informational state must itself be a psychological state. Indeed, it is widely assumed that the relevant states are false beliefs on the mere ground that they're standing states not current perceptions. I've indicated that some informational states that are not representational in a distinctively psychological sense are can be out of sync with reality. What distinction between being out of sync and being literally non-vertical or false am I relying on? Well, as I've said before, the fields of developmental psychology and primatology have been handicapped by nearly universal failure to distinguish non-perceptual sensing from perceiving. Sensing is an informational registration relation between an organism and the environment. That is, it functions to vary statistically with environmental conditions that are statistically likely to cause it. Sensory states or information registrational states that commonly guide action and other behavior. Sensory states are not in themselves distinctively mental or psychological. So ticks, caterpillars, rotifers, sense features in the environment they have no mental or psychological capacities. Representing sensory informational states of agents, very young children and apes are not thereby representing anything mental or psychological. In the absence of specific evidence, the very young children and apes represent an attribute perception, perceptual memory, belief, and other representational states that are distinctively psychological. It's appropriate to credit them only with representation and attribution of non-psychological sensory states and sensory relations. A child or ape may represent an agent as being in sensory contact with food with a capacity to retain the effect of that contact and to track the food by sensory means. A child or ape may be perceptually attuned to relations between agent and sensed object. Some of these relations allow the agent to track the object, others do not. The child need not represent the agent as perceiving or perceptually representing the food. There are many ways to track an object sensorily without perceiving it. Since perception is a special case of sensing and since there's no specific evidence that children attribute the specific special features that differentiate perception from mere non-perceptual sensing, it's appropriate to take them to be attributing the generic capacity rather than the specific capacity. One might think that sensory states are reality-based and thus cannot be what are attributed when very young children pass the false belief tests. This thought is, I think, confused. Non-psychological sensory states and their retention can be out of sync with reality and action relevant ways. This point underlies quite a lot of work in the sensory systems, on the sensory systems of animals. Consider research on circadian body clocks. Very simple organisms that lack psychological states show circadian rhythms in their activities. These rhythms are kept in sync by the day-night cycle, by sensory entrainment mechanisms, for example, by exposure to a change from dark to light every morning. If the organism is kept in a constantly lit environment, it will continue to act on the rhythm of the day-night cycle for a while. Since such rhythms are never perfectly in sync with the environment they go out of phase, if they are not entrained with exposure to changes between dark and light. So if the organism is not thus entrained, it will operate on outdated sensory information that may lead it to act in ways that will not reach its targets. The explanations need not and do not postulate representational psychological states with veridicality conditions. Consider another example. Suppose that a tick senses the heat of a human arm and begins crawling toward it. Suppose that the tick's progress is arrested but not displaced and the arm is moved so that the tick cannot sense its warmth. Suppose that the tick's sensory condition is unaffected. Perhaps its sensors are artificially heated or perhaps it's so constituted that it continues to move a predetermined distance to reach its target given that it received the sensory stimulation. The tick is then allowed to crawl again. We would expect the tick to continue crawling toward the position where the arm had been rather than toward the new position of the arm. Such expectation need not attribute an inaccurate perceptual memory or a false belief or any other representational psychological state to the tick. The tick's action fails to reach its target because the tick's sensory states and their retentions are out of sync with the action-relevant target object, the arm or food source. However, the tick's cognitive and potential and sensory states are not psychological states at all. Generic attribution of sensory states which does not represent those states as psychological does not require taking those states to be in sync with reality and action-relevant ways. By understanding that an agent's sensory states are out of sync with reality, one can predict very specifically how it will fail to reach the action's target. In false belief tests, as far as current evidence goes, when 13 to 15 month old infants see an actor hiding and retrieving an object in a box, they attribute the acts of placing the object inside the box and retrieving it. They also attribute a light-based sensory state that senses the object in the box. When the actor is screened from the scene, the children attribute retention of this sensory state. The subjects see the object moved and decouple this perception from the attribution of the retained sensory state in the actor. The actor's retained sensory state tracked the object in the first box. When the actor returns and looks in a box, the child attributes the goal or added act of retrieval and given that the child anticipates that the action will be guided by the sensory state, the child anticipates that the actor will look in the first box, even though the child has a perceptual memory of its being in the second box. The child's behavior depends on attributing actions, sensory states, retentions of sensory states, and perhaps a gentle states that cause the action, but these attributions are not in themselves attributions of mentality, much less false beliefs. The child expects the actor to be misled, but the attribution apparatus is the same apparatus that we use in anticipating the gentle behavior of a tick in parallel circumstances, except that the child is unspecific about mentality, whereas most adults would deny mentality to the tick. Since I think the child hasn't got yet a representation of mentality, or we don't have any evidence that think it does, it can't deny mentality. It is already known that about six to seven months of age human children and apes distinguish action from movement and associate targets and norms of efficiency with action. The false belief tests show that in their second year human children can decouple expectations of specific action sets, actions and targets by another being from the child's own. The child enriches the simple action explanation scheme through experience of other beings' actional targets and tendencies, but no evidence has been adduced to show that at this stage the child departs from the basic action attribution scheme that we know that it already has from six to seven month old. The idea that the child is mind reading or has a theory of mind or is otherwise attributing psychological states to the actor is a gross hyper-intellectualization at least as far as the current evidence goes. Now I'm gonna skip some discussion of the quite interesting variety of tests that have been carried out just as a quick gloss. Children even at the age of 12 to 13 months can track objects not just by the objects themselves but by associated properties of the objects. They can chain connections in quite sophisticated ways. So there's quite a lot of intelligence going on even on the part of 12 to 13 month olds in tracking, but none of these experiments really break the problem, deal with the problem of distinguishing non-psychologically guided agency as attributed by the child and psychologically guided agency. So I'm going to instead of going through a catalog of these, I think I've read all the experiments that have been done on this matter and I wouldn't have presented all of them, but instead of going through a sampling of them, I want to shift gears here and talk about another matter of attribution of mentality which may have more interest for this group. There remains a type of low level attribution whose discussion in psychology does not follow the lines that I've been criticizing. I have in mind attributions of very primitive emotions and attributions of conscious sensations. So those of you not versed in philosophy of mind, there are all kinds of different mental states so perception and belief, intention, various lower level, representational cognitive states, also emotions and then all of those are broadly representational probably, emotions are a bit of a hybrid which I'll get to. But on the one side, philosophers talk about representational states. They're states that can be, who's part of their nature, they can be true or false or accurate or inaccurate so a perceptual state is representational state. It's part of its nature that can be inaccurate. A belief is a representational state, it's part of its nature that can be true or false. On the other side, so representational states are one major block of mental states. Another kind is just conscious sensings. So for example, pain itself is a mental state because it's a conscious. It's a conscious state but pain in itself at least arguably doesn't represent anything. There are philosophers that say it does and they claim all mental states are representational states. I'm on the other side of that. I think they're two big types, one centered in consciousness itself and the other centered in representation. And it seems to me these cut across each other. I think there are conscious states that are not representational and they're representational states that are not conscious. So that's in the background. I'm going to now talk, I've been talking about representational states like perception, belief, intention, things like that. Now I'm gonna shift to the other block and talk about consciousness just a bit and emotions which somewhat straddle the two types. So some emotions I think maybe just mental as far as they are got a conscious element. They've got other aspects as well but they needn't have, I think some emotions may not have any representational aspect. Most emotions of course do. Okay, so I'm gonna start with the issues about, with these issues about sensational states like pain for example and emotions. Infants visually distinguish faces from other entities and distinguish facial expressions that in fact derive from such emotions as fear and joy. And they respond very early to differences of this between such faces. Despite occasional claims to the contrary, no one, or no responsible person in my view literally take, let's see. Despite occasional claims to the contrary, I think that no one literally perceives others emotions as distinguished from expressions that are cues to the emotion. Representing an emotion caused by an expression, representing an expression caused by an emotion, let's say the emotion anger, and functionally related to appropriate responses to anger does not constitute representing as anger the cause of the facial expression. And most researchers have been clear on this fact, on this point. Can very young children parlay their perception of emotion expression into representations of such emotions as fear, anger and joy as such? Children and animals are adept at connecting emotion expression with functional, in particular, actionable consequences of such expressions. Children quickly learn to use their perceptions of anger expressions to anticipate actions to be avoided or feared. And similarly with joy and support, say on part of the mother. Of course, such capacities have very early evolutionary roots in the animal kingdom. Agitation and tensed bodily poses constitute threat displays. What would it take to represent causes of emotion expressions as specific emotions? It is plausible that to represent an emotion, one would have to represent it as causing the expressions that one perceives. There is evidence that very young children attribute causal relations among perceived entities between their own states and their own actions, and between unperceived physical events and perceived events. The question is whether children can represent such causes as emotions. Many emotions are representational, let's say the vast majority of them. Fear of dogs is commonly associated with perception and perceptual memory of dogs. As already indicated, no good evidence for capacity in very young children or apes to represent perception or representational memory has been presented. Connecting action-causing states with sensings of functional targets seems to suffice to explain such animal and early child responses to emotion expressions. Emotions commonly have sensation-like elements, so fear is often associated with visceral phenomenal sensation. And finally, emotions have functional aspects. A function of fear is to cause avoidance or protective behavior. Most emotions in human adult life have all three aspects, representational, conscious, phenomenal, and functional. But it seems plausible that certain very primitive emotions may lack a representational aspect. So an olfactory state may be functionally associated with fear, even though the olfactory state is not perceptual or representational in a distinctively psychological sense. So perhaps one can attribute a primitive emotion as such without representing it as a representational state. To do so, one must represent the state's connection to guiding sensory states into functional upshots. Although there are unconscious emotions, I doubt that anyone can represent a specific emotion type as such unless the individual can represent conscious states. The most primitive emotions that involve some phenomenally conscious feeling, as well as the most primitive emotions involve some phenomenally conscious feeling as well as relevant functional relations. So I think that the question of whether very young children can attribute non-representational emotions may hinge on whether they can attribute conscious sensory states. I believe that there's a serious possibility that they can. Let us take pain as a simple phenomenally conscious state. Again, although there have been occasional claims that one can literally perceive or feel others' pains, I feel your pain, I think that, I ignore these claims, I think they're metaphorical. The question is whether very young children can use pain expression as a basis for representing another's pain as pain. Since pain is, I think, not in itself a representational state, representing it is not representing a representational state. But one can represented by imagining it. Contagious pain involves low level enough psychological capacities to expect that very young children may be capable of having contagious pain. Contagious pain is illustrated by the vicarious feeling of wincing after observing another's behavior resulting from a sharp pain. Plausibly, such a pain, it's a type of imagination. Experimentally, very young children appear to have such vicarious feelings. Since the vicarious feelings are guided by representations, it's plausible that in having such feelings, they represent pain in others. This is plausible, I think, not experimentally shown. Thus, it seems plausible that very young children represent in other sensations and primitive emotions that have sensational components. They feel their own pains. They would have to retain some marker of that feel in its absence through memory or imagination. And they would have to represent that feel in another, sort of, that feel in another individual. Through so-called vicarious or contagious pain. There's neural and behavioral evidence that they have simulation mirroring resources that might underlie such sensory memory and sensory imagination. There is, as yet, no strong evidence that these resources are best accounted for in psychological terms. But I think we're closer to having evidence that apes in very young children attribute sensations and elementary emotions to others than them we are to having evidence that they attribute representational states like belief and perception to others. As far as apes in very young children attribute mental states, they may attribute simple sensations and conscious emotions closely associated with that feeling, such feelings. Okay, I'm just about finished. The over attribution of theory of mind is a cautionary tale, I think, for the young scientists of primatology and developmental psychology. The terms theory of mind and mind reading have often been used in a spirit of salesmanship, not scientific disinterest. The terms systematically mislead. There's currently no evidence that supports taking apes or very young children to theorize about mind. There's no evidence that they read perceived signs as expressing anything mental. There's no evidence that they attribute states with representational content and associate them with signs or behavior, a necessary feature of any natural application of the term mind reading. Metal representational capacities may emerge very early, but so far, there's no good reason to think that they do. Significantly, the earliest constructions in language do not feature representations of psychological states. Children use early use of verbs centers almost obsessively on action. Thank you. Sure. Professor Burgi's talk is open to questions or comments. Thank you for, Mr. Anthony. I understand, of course, how the capacity to enter into states with vertical conditions, conditions of verticality, representational states, which is the least, is a far more sophisticated capacity than the capacity to just register information perhaps censor the, even if those states could go out and succeed with the environment. A huge difference in sophistication. I'm less clear about the size of the difference in sophistication between possession and the capacity to ascribe states, such as beliefs with objective representational concepts, to agents, and a capacity to ascribe to agents informational registrations that differ from one's own. That could be out of sync with the environment that generates expectation that the way that the agent prides has to be the action in addition. So, I mean, for me to describe that capacity right now, of course, required more sophisticated language. Of course, you have to distinguish from a mental attribution. So one difference between your analysis of these experimental, these physical experimental data, as opposed to saying that behavior will rule the approach, one might say, is that you have to be still ascribing a fairly sophisticated representational capacity to the very young children and the agents. One can have ascribed it as less sophisticated, but nonetheless, the capacity to ascribe to states still seems to be pretty impressive. Yes. So, what is the advantage of your view that, nonetheless, we can take of that representational capacity as a less sophisticated achievement? Yeah, I think, I do think there's a distinction between representing action and representing mere behavior and the latter is more sophisticated than the former. I'm assuming with pretty much the rest of these fields that children by the age of six to seven months are attributing action, not merely behavior. So the kind of move by Perner to deal with the 13, 14 month olds by going back to behavioral rules looked from the beginning to most people in the science to be a kind of retroactive, desperate move to save his position, that it doesn't, that mentality isn't ascribed until age four. So I don't think that the behavioral rule view is really seriously in play. Perner himself has given up that idea as applied to 12 and 13, 14 month olds. I would have thought that he would have, I don't know his work enough to know whether he considered, well look, we already have massive evidence that six to seven month olds are distinguishing ordinary movement from action. In fact, I think action attribution is probably in the perceptual systems of very, of young children. So that's being worked out right now but I believe the evidence is gonna be attribution of action is already in the perceptual system. It's not a cognitive achievement. So I'm assuming a lot there and assuming yes, action is already, action attribution is already in place very early on. I'm assuming that attribution of mere behavior won't cut it even for four to five month, six month old children or apes. But is part of action attribution, at least as you understood it, part of it is the capacity to ascribe informational states to the age? Sensory states, yeah. They're of course not gonna be able to explicate an informational state the way I can but some rudimentary capacity to recognize some relation between the environment and the agent that guides the action, that's in place by six to seven months as well. The decoupling doesn't seem to occur until about 12 or 13 months. So there is a developmental change between the six to seven months and 12 months but my argument is the attribution of action with no evident need for attributing mentality is already in place at six to seven months. It's probably in place early. It's probably virtually innate. It's probably in the perceptual system. And then something new happens at 12 to 13 months old. They're able to distinguish specific actions which are misled by the information. Although they can differentiate between their own action, what they would do and what the other one will do, even by six to seven months, what they can't do until 12, 13 months is say specifically how the misled agent will act differently. So that's what comes on then. I just say, look, you have all this other stuff in place. The natural explanation is to add the minimal amount to get from there to what the 12 to 13 month olds can do. So why postulate this big jump from teleology to action explanation to psychology when there's no evidence for it. Thank you, though. That's a very useful way to press me in. I'd be glad to talk about it further. So I guess it's working out. I just wanted to ask a question about the beginning of your talk and the argument that you made in most of the paper. So you began with the objection that the experiments that are supposed to show that young children have the ability to attribute representational states. I'm just focusing on the representational state side, the belief side. Don't work, they don't support that because there's another explanation that is in some sense simpler. But I wonder whether there, can you describe what an experiment might be like that could show that children who don't yet use language and specifically the language of mental state attribution that could show that they attribute mental states? No, I can't. I'm not an experimenter. Part of the paper I cut out a little bit spends a little more time trying to coax the experimenters into taking what we philosophers call obliquity or opacity more seriously. As I said, they confuse obliquity with tracking of properties. But I haven't been able to think of an experiment that would do it pre-language. I don't think that means much. I'm not a talented experimenter. If I were to bet, I would bet, yeah, attribution and mentality, I wouldn't bet. But it seems to me I'm open to attribution and mentality occurring before doing it through language. I'm certainly not a linguacentric theorist about this. I'm not because my teachers were and I thought the arguments they gave weren't absolutely no good. So I think we should be open as to whether we could find out that they attribute mentality before language. But I do think the most persuasive evidence right now is through the linguistic use of three to four-year-olds. Carefully hedged because even that hasn't been scrutinized, I think, hard enough. But I think it's a fair question. I get it every time I give this talk and I have to say I'm trying to stimulate the real experimenters to go do their job. I'd like to be able to do it for them but I haven't been able to figure it out myself. Thank you very much for your attention.