Interface problems in the explanation of action

Abstract

When doing mental ontology, we must ask how to individuate distinct categories of mental states, and then, given that individuation, ask how states from distinct categories interact. One promising proposal for how to individuate cognitive from sensorimotor states is in terms of their representational form. On these views, cognitive representations are propositional in structure, while sensorimotor representations have an internal structure that maps to the perceptual and kinematic dimensions involved in an action context. This way of thinking has resulted in worries about the interface between cognition and sensorimotor systems – that is, about how representations of these distinct types might interact in performing actions. I claim that current solutions to the interface problem fail, because they have not sufficiently abandoned intuitions inspired by faculty psychology. In particular, current proposals seek to show how cognitive states can enforce prior decisions on sensorimotor systems. I argue that such “determination” views are the wrong kind of views to adopt, given the form distinction. Instead, I offer a proposal on which propositional representations can at best bias us toward certain kinds of action. This kind of view, I argue, appealingly distributes the explanation of action across distinctive contributions from cognitive and sensorimotor processing.

Keywords:

• action
• intention
• interface problem

Acknowledgements

The author would like to thank Myrto Mylopoulos, Ben Sheredos, Sebo Uithol, Jason Winning, and three anonymous reviewers for helpful comments on earlier drafts. The author would also like to thank the participants in the 2015 Summer School on Philosophy of Action at the University of Tübingen for stimulating discussion about the nature of action.

Disclosure statement

No potential conflict of interest was reported by the author.

Notes on contributor

Daniel C. Burnston is an Assistant Professor in the Philosophy Department at Tulane University, and a member of the Tulane Brain Institute. He writes on issues at the intersection of philosophy of mind and philosophy of neuroscience.

Notes

1. There are differences amongst causal theorists about whether intentions are reducible to belief-desire complexes, and whether all intentional actions must involve intentions (Davidson 1980). I will target the causal theory as expressed by theorists who think that intentions are distinct mental states, which stand in causal relations to the representations that guide and control the action – for example, Bratman (1999), Searle (1983), and Pacherie (2000).

2. Positing determination relations does not entail that there are only unidirectional causal relations between representations. Pacherie (2008), for instance, suggests that bottom-up interactions allow propositional intentions to monitor and correct downstream processing. This is compatible with the initial content-determination being top-down, and indeed, the interface problem is raised here again in the reverse direction. I thus won’t engage further with this proposal.

3. M&P occasionally talk in terms of “prototype” actions encoded in a motor schema. B&S raise the possibility that motor representations encode the “invariant” aspects of action. However, positing prototypes or invariants faces the same problem as positing highly abstract schemas. Given that we act on particulars, not prototypes, there will always be a gap between the content of the prototype/invariant and the particular representations tokened – a gap which determination views have no resources to fill.

4. As noted in the introduction, I am making the assumption here that, since natural language and propositional states have a similar structure, we can learn about the kinds of interactions undergone by the latter by studying the former. Defenders of determination views are free to attempt to define a difference between lexical and propositional representations but, given the definition of cognitive representations in terms of form, this seems a difficult strategy. I won’t say more about it here; hopefully the positive appeal of the biasing view will be sufficient to obviate this kind of move.

5. Now, some have been tempted to say that only dimensions, and not features are potentiated (e.g. Memelink and Hommel 2013), but this seems overly strong. At least in the case of color, there are specific advantages to tokening “green” as opposed to simply tokening “color” (although these advantages don’t seem to hold for orientation). But even admitting some feature-specific effects is not really much help in the overall discussion. “Green,” of course, is a determinable, and what counts as green for subjects can depend on the relevant contrast in the environment (Hardin 2008). An alternative, and as far as I can see unproblematic account, is to interpret the tokening of “green” as potentiating a dimension relatively – namely, biasing certain values slightly more than others, where the details are determined by the interaction of this relative biasing with the actual information available in the environment.

6. Indeed, when one asks subjects to perform a task that is incongruent with the affordances of an object (for instance, in the effects cited in Section 3.1), error rates generally go up, suggesting, again, that the resources of the intention we “decide on” aren’t sufficient to determine the action.

7. I believe this perspective can be scaled up if combined with recent philosophical approaches that focus on imagination (Nanay 2016) and skill (Railton 2009) in decision-making, and that this is a necessary project to undertake if we adopt the form distinction. I won’t pursue it further here, however.

8. Anscombe (2000) used a grocery list example as an analogy for describing when an act is intentional – one intends to get butter if ending up with margarine is a failure of the action, not the list. My point is different: I think the role of the list is an appropriate model for the role of intention.