Benjamin Claessens
Affiliation: CUNY
Category: Philosophy
Keywords: Teleological belief, Purposive inference, Action representation, Action explanation, Cognitive models, Difference-making
Date: Wednesday 3rd of September
Time: 14:30
Location: Room 154 (154)
View the full session: Mind Reading
Imagine a game of tag: one person attempts to touch someone, while everybody else tries not to be touched. When contact finally occurs, players switch roles. Somehow, we spontaneously represent such behaviors as goal-directed; we intuitively recognize that the players do not run arbitrarily, but run for the sake of distinct and changing purposes. And irrespective of the purposive facts, we human animals seem effortlessly adept at forming such purposive beliefs. But how and why do we acquire the particular purposive beliefs that we do?
Often, this question is directed at the phenomena of agency and action: how do we come to represent behaviorsâincluding our ownâas agentive, or, in some more minimal sense, as goal-directed? In turn, that question is often rephrased in terms of mental states: how do we recognize the beliefs, desires, intentions, or preferences, by which a given behavior is made intelligible as purposive? For example, perhaps the chaser runs because she desires to tag, and believes she must run to do so. So, according to this rationale, we represent the playersâ behavior as goal-directed, because we can somehow identify their mental states.
Recently, however, psychologists and philosophers have argued that this mentalizing strategy gets things backwards. Rather, we can identify mental states, because we independently represent some events as purposive. In other words, according to the teleological account, we identify purposes first, and mental states later. For example, Gergely and Csibra (2003), drawing on Dennett (1987), argue that we implicitly adopt a âteleological stanceâ. We therein assume that the players are instrumentally rational, such that they are disposed to select the most efficient behavioral means to an end. Under that assumption, and given a set of evidence regarding the playersâ behaviors and environmental constraints, we can then infer the purpose which best explains the behavior. Other teleological theorists emphasize the evidential and explanatory role of objective normative reasons (Perner et al., 2018). On this view, it is our recognition of evaluative facts that allows us to interpret behavior as goal-directed; perhaps we recognize, for example, that it would be good for the chaser to catch another player.
One of the main arguments in support of the teleological position is that it best explains data in developmental psychology. Young children, between 3 and 4 years old, clearly show the ability to understand behavior as goal-directed (Csibra, 2008). And yet, at that age, their capacity for mental state attribution has seemingly not yet fully developed. Proponents of the teleological account argue that children first develop a capacity for goal-directed representations, and only much later do they refine their predictions with a theory of mind (Gergely et al., 1995; Perner et al., 2018; Burge; 2018).
In what follows, I will assume that our cognitive models are built upon a foundation of perceptually basic belief. But crucially, we somehow inferentially extend those models, and thereby form purposive beliefs about events that far outstrip our perception. Are mental state attributions necessary to make those purposive inferential extensions? Or, can we somehow infer purpose, without yet mentalizing? Existing teleological accounts do not adequately address these questions, for they focus on explanations of phenomena, such as false-belief tasks, in which perceived content is represented as goal-directed.
In this paper, I defend a theory of purposive inference, and thereby advance the general teleological position. I will argue that we can indeed infer purpose without mentalizing. Specifically, we navigate the world with cognitive models that are populated with purposive difference-makers: events that appear to make a difference to preceding events. In this respect, our concepts of cause and purpose are temporally symmetrical; causes make a prospective difference, while purposes make a retrospective difference. And, as I will argue, so too are our causal and purposive judgments mutually exclusive; among our representations of events, causes preclude purpose, just as purposes preclude cause. Accordingly, our cognitive models are not causal per se; cause and purpose are twin inferential supports, with which we build a single representational structure.
My overall argument will be abductive. Existing theories of agency and action representation widely share a common feature. Namely, they each posit, at least, either two distinct mental states, or two distinct computations: one to determine the likely outcomes of a given behavior, and another to determine the desirability or value of those outcomes. This distinctionâbetween prospects and preferencesâlies at the core of contemporary decision theory; ostensibly, both types of ingredient are necessary to predict, to explain, and perhaps to cause, goal-directed behavior. But, once we add purposive difference-makers to our cognitive models, we can account for that same predictive, explanatory, and causal work, with one type of representational mental state, and one computation: the inference by which we form a purposive belief. So, all else being equal, we ought to believe that the simpler theory is true.
Additionally, I will argue that, given this theory, we can explain why purposive beliefs are a cognitive default (Kelemen, 1999), and why they interact, systematically, with causal beliefs (Knobe & Fraser, 2008), and judgments of moral responsibility (Foster-Hanson & Lombrozo, 2022). In doing so, we need not assume a principle of rationality, nor identify objective evaluative facts; we instead use entirely descriptive data, as input to a rational inferential process. We thereby acquire purposive beliefs.