People v. objects: a reply to Rakison and Cicchino

In our Brief Article (KBW, this issue), we present data showing that 5-month-old infants interpret the movements of inanimate objects according to the constraint of spatiotemporal continuity, but do not interpret peoples’ motion according to this same constraint. We suggest from this finding that infants may possess distinct systems for learning, interpreting, and reasoning about the actions and interactions of inanimate material objects, on the one hand, and intentional agents (of which the canonical exemplar is people), on the other. Rakison and Cicchino (R&C) raise several issues in their valuable commentary, both methodological and conceptual. Methodological Issues. R&C worry about the “small sample of infants” in each of our conditions. In fact, the number of subjects in our experiments is similar to those in infant experiments on related topics, which frequently range from 8–12 per condition (e.g. Luo & Baillargeon, in press; Oaks, 1994; Quinn & Eimas, 1998; Sodian, Schoeppner, & Metz, 2004; Spelke, Breinlinger, Macomber, & Jacobson, 1992). It is also similar to that of the original study we replicated in Experiment 1 (Spelke, Kestenbaum, Simons, & Wein, 1995). Moreover, statistical analyses are for precisely the purpose of determining the chance likelihood of a set of results, given the sample size. Both parametrically and nonparametrically, the chance likelihood of our results in Experiment 1 (Boxes) is well below the “0.05” standard (with P-values of #0.02), and in Experiment 2 (Humans) is above the standard (P-values .0.4). And the effect sizes in the two experiments differ by a full order of magnitude (0.41 versus 0.04). These analyses support our claim that infants are responding substantially differently to our human versus inanimate stimuli.