Folk psychology without principles: an alternative to the belief–desire model of action interpretation

Abstract

In this paper, we take issue with the belief–desire model of second- and third-person action interpretation as it is presented by both theory theories and cognitivist versions of simulation theory. These accounts take action interpretation to consist in the (tacit) attribution of proper belief–desire pairs that mirror the structure of formally valid practical inferences. We argue that the belief–desire model rests on the unwarranted assumption that the interpreter can only reach the agent's practical context of action through inference. This assumption betrays a deep-seated bias toward disengaged, observational interpretation strategies. On our alternative picture, the interpreter can start off on the assumption of a shared practical context and proceed to reason discourse in those cases in which this assumption runs aground. Following Brandom's non-formalist account of reason discourse, we suggest that interpreting other people's actions in terms of reasons is not a matter of following the principles of formally valid practical syllogisms, but of endorsing practical material inferences that are correct in virtue of a shared practical world.

Keywords:

• folk psychology
• belief–desire model
• theory theory
• simulation theory
• Robert Brandom

Acknowledgement

We would like to thank Marc Slors, Dan Hutto and Bob Gordon for their helpful comments on earlier versions of this paper.

Notes

Here and throughout, we use the term ‘interpretation’ in a broad sense to include all kinds of sense-making activities, such as understanding, explanation and prediction. As will become clear, though, we resist focus on explanation and prediction in third-person contexts.

Cf. Botterill (1996, 115): ‘If belief-desire psychology has a central principle, it must link belief, desire and behavior. It could be formulated like this: [Action Principle] An agent will act in such a way as to satisfy, or at least to increase the likelihood of satisfaction, of his/her current strongest desire in light of his/her beliefs’.

Fodor (1992, 283), for example, suggests that ‘normal cognitive development eventuates in the child's internalization of a tacit “metacognitive” intentional psychology: specifically, in the internalization of some version of the folk psychological theory that an agent's behavior is normally caused by his beliefs and desires’. Another example can be found in Leslie, German and Polizzi (2005, 50), who state that ‘a prediction of behavior question requires an additional ascription of desire, the integration of belief and desire, and the inferring of a resulting action’.

See, for example, Horgan and Woodward (1985, 197).

Some theory theorists explicitly restrict folk psychological theory to non-tacit knowledge of folk psychological principles represented in a mixed variety of simpler or more complex abstract models that, by their very nature, do not contain ceteris paribus clauses (e.g. Maibom 2003). The context-sensitive application of the models is delegated to other cognitive mechanisms. Yet the worries concerning complexity we raise in section three also apply to the workings of these additional cognitive mechanisms.

Thus, Goldman (2006, 29), for example, thinks that ‘a decision-making mechanism normally takes genuine (nonpretend) desires and beliefs as inputs and then outputs a genuine (nonpretend) decision. In simulation exercises, the decision-making mechanism is applied to pretend desires and beliefs and outputs pretend decisions’. In a somewhat similar fashion, Currie (1995, 158) claims that in simulating another agent ‘we tend to acquire, in imagination, the beliefs and desires an agent would most likely have in that situation, and those imaginary beliefs and desires have consequences in the shape of further pretend beliefs and desires as well as pretend decisions that mimic the beliefs, desires and decisions that follow the real case’. These ideas are made explicit in Nichols' and Stich's boxological model of ST (e.g. Stich and Nichols 1992, 1995, 1997; Nichols and Stich 2003).

If, however, one believes that the simulation routine requires mastery over the concepts of the propositional attitudes, then, as Ravenscroft (2003) argued, cognitivist ST may be committed to tacit knowledge of FP principles after all. In the case of (simulated) practical reasoning, different kinds of propositional attitude concepts are involved (beliefs, desires, decisions). One needs to keep track of these concepts in order to reach a practical conclusion, that is: one needs tacit knowledge specifying the way these attitudinal ‘tags’ are related to each other and the contents. But how can we explain our ability to tag and keep track of these different propositional attitudes? Tacit knowledge of FP principles seems to fit the job description perfectly.

Wilkerson (2001) makes a point against TT and ST that is inspired by similar considerations.

It should be clear that worries about a ‘threat of collapse’ of our proposal onto principled accounts are wrong-headed. First, the tacit knowledge (if there is such) required for co-cognition does not require knowledge concerning psychological generalizations (Davies and Stone 2001). In the case of practical co-reasoning, it only does when one assumes that such co-cognition requires ‘tagging’ the relevant contents with psychological concepts (see Ravenscroft 2003, footnote 7). Our proposal, however, is to the effect that such practical co-reasoning on default takes place in a shared context, which is to say that the contents are precisely not specified in terms of propositional attitudes. Secondly, and more importantly, on our proposal, the practical co-reasoning does not mirror formally valid inferences, but consists in proposing, endorsing or rejecting materially correct ones (see section 4). On our unprincipled account of interpretation, the interpreter's ‘knowledge’ consists in knowing how to propose, endorse or reject such material patterns of inference. It explicitly does not involve (tacit) knowledge regarding the principles of reasoning.

Brandom treats the formal principles of inference as expressive tools that enable us to make explicit the proprieties of material inferences, by saying what was done in endorsing the inference. Thus, the conditional ‘whenever it rains the streets will be wet’ is considered as a means of saying what is being done in endorsing the material inference ‘it is raining, therefore the streets will be wet’.

Here and throughout, we assume that intentionally performing an action at t is typically the reliable result of an intention to perform an action of that type at or prior to t. Explaining the intention with which one performs a certain action is therefore normally sufficient to explain the action. Brandom treats actions as ‘language exit transitions’: (reliable dispositions to respond differentially to the) acknowledging of commitments by bringing about various kinds of states of affairs (see 1994, 235).

This may be confusing for those familiar with Brandom's work. In claiming that material inferences do not have ceteris paribus clauses, we aim at a common conception of such clauses on which they are supposed to secure the monotonicity of the statements to which they belong. Brandom proposes to understand ceteris paribus clauses ‘as explicitly marking the nonmonotonicity of an inference, rather than a deus ex machina that magically removes its nonmonotonicity’ (2000, 88). We obviously do not want to object to this alternative conception of ceteris paribus here: material inferences being essentially contextual is precisely what their non-monotonicity amounts to.

According to Hutto's NPH, infants become familiar with the background norms for wielding folk psychology in practice by being exposed to ‘folk-psychological narratives’. The defining feature of these narratives is that they reveal how beliefs and desires (and other propositional attitudes) interrelate and conspire to form reasons for action. For, Hutto says, ‘it is not enough to imagine it as being sponsored by a singular kind of propositional attitude; one must also ascribe other kinds of attitudes that act as relevant and necessary partners in motivational crime’ (2008a, 26).