Reaching for an object or that object: Context and limitations in encoding object features may constrain infant goal attribution

That very young infants attribute goals to the object-directed reaches of others is well characterized, but we understand little of what infants encode about the objects being acted on. The present study examined whether the context surrounding the reach, specifically the presence of another object, prompts infants to encode feature information of the target object. To test this, 9-month-old infants in Experiment 1 observed an actor repeatedly reach for one of two objects, while in Experiment 2, infants observed repeated reaches towards one, singly presented object. In both experiments, following habituation, infants observed the actor reach for either the same object as in the past, or a previously unseen, novel object. In Experiment 1, but not Experiment 2, infants looked longer at reaches to the novel object, indicating that this action was unexpected. Considered along with research on visual attention and infant object perception, these results suggest that the initial presence of an alternative object influenced infants’ expectations for future reaches, perhaps by prompting infants to create a feature-bound representation of the target. Thus, the context surrounding a goal-directed action may flexibly influence infants’ action processing to minimize cognitive load.

Pushmi-pullyu Representations and Mindreading in Chimpanzees

Lurz and Krachun (2011) propose a new experimental protocol designed to discriminate genuine mindreading animals from mere behavior-readers and to give evidence in favor of the claim that chimpanzees are capable of attributing internal goals to others. They suggest that chimpanzees' variety of "internal goal attribution" consists in attributing to others basic intentional representations, baptized by Millikan as "pushmi-pullyu representations" (PPs). Now, Millikan (1996, 2004a, 2004b) distinguishes what I propose to call 'pure' PPs from more complex varieties of PPs, which allow their owners to respond more flexibly to their environments. But, what would happen if we tried to differentiate, analogously, between more or less sophisticated mind-readers in virtue of the sorts of PPs that they could attribute to others? What would attributing complex PPs consist in and how would such capacity increase the predictive powers of chimpanzee mind-readers? This paper offers an answer to these questions. Based on Millikan's work, I differentiate two varieties of complex PPs. Then, I examine what a basic mind-reader, only capable of attributing 'pure' PPs, would be able to do. After that, I distinguish two more sophisticated varieties of mindreading, each consisting in the attribution of one of the complex PPs previously presented, and I show how the ability to attribute complex PPs to others comes with more potent and flexible capacities to anticipate their behavior. Finally, I offer some reasons to think that attributing complex PPs is still simpler than full-blown mindreading and I briefly evaluate the prospects of extending this proposal to infant social cognition.