There are several different questions associated with the study of folk psychology: (1) what is the nature of our commonsense concepts of mental states?, (2) how do we attribute mental states, to ourselves and to other people?, and (3) how do we acquire our concepts and skills at mental-state attribution?Three general approaches to these questions are examined and assessed: theory theory, simulation theory, and rationality theory. A preliminary problem is to define each of these approaches. Alternative definitions are explored, centering on which questions each approach tries to answer and how it answers them. For example, simulation theorists substantially agree on the answer to question (2) but not on the answer to question (1). The paper then turns to some serious problems facing both rationality theory and theory theory.Rationality theory is faulted for its inadequate treatment of question (1) and for its implausible answers to question (2). Theory theory is faulted for the problems it encounters in explaining first-person attribution, and for its treatment of attributed reasoning about change (the “frame problem”). Turning to simulation theory, the paper argues against Gordon’s “ascent routine” account of first-person attribution and in favor of an inner detection account. Finally, the paper addresses the question of the contents of our mental-state concepts. How do these concepts incorporate both behavioral features and inner features? A dual-representation hypothesis is advanced, and linked speculatively to mirror neurons.

Let folk psychology consist in the network of concepts, and associated beliefs, in terms of which we make sense of minded performance.This paper addresses the question of how we, the folk, come to understand those concepts: this, as distinct from the separate question as to how we come to apply them in the interpretation of particular minds, our own and those of others.The argument is that even though the network of concepts is akin to a set of theoretical, interdefined terms, still it is possible to explain how we, the folk, understand them without suggesting that we are proto-scientists. The understanding required can be based on a sort of know-how: that is, a practical, untheoretical, form of knowledge.

We find it natural to say that creatures with minds can (sometimes at least) be understood ‘from the inside’. The paper explores what could be meant by this attractive but, on reflection, somewhat mysterious idea. It suggests that it may find a hospitable placement, which makes its content and appeal clearer, in one version of the so-called ‘simulation theory’ approach to grasp of psychological concepts. Simulation theory suggests that ability to use imagination in rethinking others’ thoughts and in recreating their trains of reasoning is central to our grasp and use of psychological concepts.On this view to think of another’s mind is not to think of some intricate quasi-mechanical assemblage of items in the other’s head which causally explain her behavior. If this is all that the ‘inside’ of another person, i.e. her mind, were like, then there would be no question of anything being ‘from’ it. The simulation view, however, emphasizes that thoughts essentially have content and that identifying another’s thought, and working out its possible effects, involves identifying its content and oneself entertaining thoughts with the same content.So, on this approach, to think of another’s mind is to think of a complex but rationally unified set of thoughts, a set which is conceived as had by one subject but where the contents and relations can be grasped and appreciated by another. Some of these thoughts will be indexical and the whole can thus be said to constitute the subject’s point of view on the world, both literally and metaphorically. Grasping this point of view is, the paper suggests, what is meant by speaking of ‘understanding from the inside’.

According to Dennett, the facts about consciousness are wholly fixed by the effects consciousness has on other things. But if a mental state's being conscious consists in one's having a higher-order thought about that state, we will in principle have an independent way to fix those facts. Dennett also holds that our speech acts sometimes determine what our thoughts are, since speech acts often outrun in content the thoughts they express.I argue that what thoughts we have is independent of how we express them in speech, and that this is consonant with speech acts’ often seeming to have more fine grained content than the thoughts they express. This model has the advantage, compared with Dennett’s, of accommodating our folk-psychological taxonomy of intentional states and preserving the traditional idea that speech acts express antecedent intentional states. Speech acts doubtless do sometimes have richer content than the thoughts they express, though sometimes verbally expressing a thought simply makes us conscious in a more fine-grained way of what that content is.On the higher-order-thought model, as on Dennett’s, a mental state’s being conscious is, in effect, our spontaneously interpreting ourselves as being in that state. But such spontaneous self-interpretation need not be the last word on what content our thoughts have. Even though the content of speech acts sometimes outrun that of the thoughts they express, we can explain why the two seem always to be exactly the same. Even when a speech act is richer in content than the thought it expresses, the well-entrenched pragmatic equivalence between saying something and saying that one thinks that thing ensures that one will be conscious of one’s thought as having the richer content of the speech act that expresses it. We are conscious of our thoughts as having the content that our speech acts would capture.

This paper is concerned with the question of how we arrive at knowledge of the propositional attitudes of other agents. I describe a number of methods, but focus on the method that involves arriving at conclusions about the beliefs of others from information about their assertions and acts of assent. I attempt to give a reasonably full characterization of this method.Among other things, I maintain that when it is properly understood, the method is seen to be altogether independent of simulation. Thus, one conclusion of the paper is that simulation is not in any sense a universal method. At best, it is a method that we use in a highly restricted range of situations. Another conclusion is that there are features of the method of inference from assertion and assent that tend to provide support for the theory-theory – that is, for the view that our ascriptive practice involves the implicit use of a body of principles that resembles a scientific theory.

Language has often served both as a metaphor for thought. It is highly plausible that language serves as an epistemic entre into thought and that language structures adult human thought to a considerable degree. The language metaphor is, however, uncritically extended as a literal model of thought.This paper criticizes this extension, arguing that thought is not literally implemented in language and distinguishing legitimate from illegitimate uses of language as a device for understanding thought.

It is often said that the simulation vs. theory debate must be resolved empirically. But part of the needed empirical work, perhaps the decisive part, is “armchair” testing against philosophical touchstones such as intentionality, opacity, and Moore’s paradox. I will assay Sellars' myth of “our Rylean ancestors,” frequently cited as the prototypical “theory” theory.Sellars’ laudable aim was to show how “privileged access” could be preserved without making first person ascriptions of mental predicates incorrigible or dependent on “immediate experience.” He attempted this by portraying these ascriptions as essentially theoretical: We are so conditioned that, when situational and behavioral evidence indicates certain theoretical states and episodes, we make the corresponding self-ascriptions, without having to consider the evidence.Sellars’ approach fails, however, because self-reports are actually coordinated only with verbal behavior – typically, the outward-looking “expression” of a state in the non-mental ‘Rylean’ language. (Such coordination is illustrated, in the case of belief ascription, by Moore’s paradox, and more generally, by what I call “ascent routines.”) Thus our self-reports could not be a product of theory-based training – nor directly theory-based, as some psychologists suggest. Given this coordination, the only way to achieve Sellars’ aim – without sacrificing nivocality – is to suppose ascriptions to others to be “third person self-ascriptions” based on the other’s situation and behavior, as the simulation theory maintains. Sensitive to nonverbal as well as verbal behavior, they prevail over first person ascriptions. This leads to a certain conception of people: peepholes through which the world reappears, possibly transformed.

In this paper, I examine Daniel Dennett’s well-known intentional systems theory of belief and desire from the perspective of physicalism. I begin with a general discussion of physicalism. In the course of that discussion, I present familiar ways that one might attempt to reconcile physicalism with belief-desire realism. I then argue that intentional systems theory will not provide a reconciliation of physicalism and belief-desire realism.