Last week we discussed how intentions and intentionality might be detected or recovered from the stream of action actually observed. In general, we weren't much attracted to the Premack & Premack account, in which intentionality, and possibly some specific kinds of intentions (e.g., escape, responsiveness, and overcoming gravity) are directly apprehended in actions given the relevant triggering conditions (e.g., self-propelled object, goal-directedness). Part of our dissatisfaction arose from the fact that the account relies on psychological primitives which themselves seem intentionally laden. We spent some time getting clear on the details of the Newtson & Enquist approach to investigating adults' organization of everyday action, and left this research uncertain as to what it ultimately tells us about detection of intentions and intentionality. This uncertainty arises in part because Newtson and colleagues seem to devote little attention to probing the relation between so-called "breakpoints" and adults' intentional analysis, instead concentrating on the empirical facts concerning identification, perception of, and memory for "breakpoints." In later discussion we talked about the importance of distinguishing between an observer's detection of an action as intentional/non-intentional versus the observer's detection of the content of the intention being enacted, and the observer's detection of the specific goal being targeted. This in turn developed into an extended discussion concerning the need for distinguishing between intentions, goals, and desires, and relations between these notions.

Today we will focus on what kind of developmental story might be told concerning the ability to detect intentionality and to recognize the content of intentions.

A first issue concerns when in development children can safely be said to detect intentions and intentionality in others' action. Baldwin & Baird review a body of literature on this topic, and for the seminar we are focusing on the Meltzoff reenactment paradigm, in particular. Meltzoff presents evidence that 18-month-olds can be credited with both of these abilities. After viewing adults execute actions in which they accidentally failed to accomplish their goals, infants spontaneously reenacted these actions with successful outcomes. These findings suggest that infants inferred the content of adults' intentions and projected their goals, despite the fact that infants never actually observed the achievement of these goals. Likewise, the findings suggest that infants appreciated that adults' failures were accidental, and therefore did not attempt to reenact the faulty action sequences that led to such failure.

Question 1: Do we buy this interpretation?

Question 2: What do infants' abilities in the Meltzoff research and the other work reviewed by Baldwin & Baird suggest about how detection of intentions/intentionality actually is accomplished?

A second issue concerns whether there is reason to attribute skills for detecting intentions/intentionality to infants considerably younger than 18 months. Some relevant work is again reviewed by Baldwin & Baird, and Woodward, Gergely et al., and Johnson pursue some specific questions related to this issue. Woodward demonstrates that infants of 9 months selectively encode the goal of an actor's intentional reach, but do not do so when the actor appears to place her hand on an object unintentionally. 5-month-olds showed a similar, but weaker, pattern of results. These findings might indicate an early ability to a) discriminate between intentional versus unintentional actions, b) specifically encode goals for intentional actions, and c) formulate action concepts such as "reaching" versus "resting." Gergely et al. show that 12-month-olds are surprised when a self-propelled ball previously seen to act "rationally" pursues an "irrational" path. These findings might indicate that 12-month-olds take the "intentional stance" toward objects, such that they attribute intentionality to self-propelled objects perceived to act rationally. Johnson investigates the conditions under which 12-month-olds will follow head-turns, an ability that many researchers have previously linked to the emergence of intentional understanding. She finds that head-turn following in 12-month-olds is linked to whether the head has a face or has previously engaged with them in contingent interaction. She also finds that both of these factors (face and contingent interaction) play a role in adults' attributions of intentionality. These findings might indicate that early following of head turns indeed indexes emerging intentional understanding.

Question 3: Which of these accounts do we buy, and where are gaps that need filling?

Question 4: If we grant these accounts, what do the findings tell us about how detection of intentions/intentionality is actually accomplished?

A third and fourth issue concern what develops in children's ability to detect intentions/intentionality in others' actions, and how it develops. Baldwin & Baird argue that a lot develops with regard to detecting the content of others' intentions, because a) such detection is heavily influenced by world knowledge, and b) such detection is influenced by developments in theory of mind. They have little to say about developments in the ability to detect intentionality. Regarding the "how" question, they suggest that a foundational ability to analyze action plays a crucial role in making possible developments in intentional understanding, and that action analysis continues to play a crucial role in supporting the generativity of adult detection of the content of intentions.

Question 5: What's wrong about this analysis?

Question 6: What's right about this analysis?