Is an infant a people person?.

5-month olds apply the constraint of continuous motion to objects but not to people. They propose, based on these data, that humans are interpreted in terms of social actions whereas inanimate objects are construed in terms of object physics. We believe that care must be taken, however, before strong conclusions can be drawn from their empirical findings. We also find their proposal fails to address an important developmental issue, namely, what mechanism might underlie infants' ability to learn about people, animates more generally, and inanimates.

Limitations of the empirical findings. The main finding of the experiments is that infants looked for different durations at the one box versus two box test events but equally long at the one person versus two person test events. There are a number of reasons why we have reservations about this finding. First, infants' looking behavior in Experiment 2 must be interpreted with discretion, but they nonetheless rely on it as the basis for their conclusions about the nature of infants' representations for people and inanimates. We are wary, however, of interpreting these null results as meaningful, particularly in light of such a small sample of infants in each condition (N=10).

Additionally, our reading of the data is that one condition alone—the continuous habituation condition with the box in Experiment 1—may be driving force behind the statistical significance of results. That is, it is not clear that infants in both conditions in Experiment 1 responded by looking longer at the anomalous event than the event consistent with that seen during habituation. Examination of the data reveals that the looking times for the discontinuous habituation condition differ only minimally, and further analyses on the visual fixation times within each condition were not presented. This concern is further borne out by the number of infants who looked longer at one test trial over the other. In the continuous habituation condition with the box this difference is considerable (9 vs. 1); in the discontinuous habituation condition there was no discernible difference (6 vs. 4).

More problematic, it is unclear whether any differences between Experiments 1 and 2 were indeed due to the involvement of people versus boxes. A crucial interpretation is that infants responded differently in the two experiments because they apply different constraints to humans than to other things in the world. This argument relies on the assumption that infants understood that the moving objects in Experiment 2 were humans, yet offer no evidence to confirm that infants interpreted the experimental stimuli this way. They do provide analyses that show that infants did not view the human displays as more interesting than the box displays; however, it remains to be seen to what aspect of the events infants in Experiment 2 were habituating and responding.

It has been shown that infants as old as 10 months of age have difficultly processing both local and global dynamic cues that are presented concurrently. Perhaps then, the complexity of the events with humans affected to what aspects of those events infants attended. One possibility is that infants in Experiment 1 processed the events in terms of continuous or discontinuous motion whereas infants in Experiment 2 processed the locally dynamic nature of the stimuli rather than their global motion as they moved across the screen. This would explain why infants responded equivalently to the two test trials in Experiment 2 but not in Experiment 1. We suggest that any number of stimuli that are more locally dynamic than boxes could have produced the same results as those reported in Experiment 2; for example, we predict that infants would respond the same way to a robot with moving parts as they did to the human stimuli.

Finally, the anomalous nature of the self-propelled boxes raises questions about how infants interpreted the events and if they apply different constraints to people and inanimates. On the one hand, we believe that infants apply constraints relating to continuous movement differently to people and inanimates; on the other, they claim that infants do not apply constraints relating to self-propulsion differently to people and inanimates. We wonder whether it is reasonable to make both claims concomitantly, and what the theoretical basis for this position might be. Indeed, there are data that infants as young as 6 and 7 months of age expect people but not inanimates to be self-propelled and goal-directed. We find it difficult to reconcile such findings with infants' interpretation of the boxes’ anomalous self-propulsion.

Mechanisms of learning about the animate–inanimate distinction. Even if one were to accept the basic findings presented in the article, we question how infants learn about the properties of animates and inanimates. The lack of a developmental approach—with only one age group tested—and evidence relating to which aspect of the event infants processed does not help in this regard. The authors claim that by the 5th month of life there are "different modes of construal for humans versus inanimate objects". We are not entirely sure how to interpret this phrase. In the subsequent text, infants possess different interpreting systems, one for social phenomena, and one for object physics. We read this to mean that the authors' proposal is that infants have a number of mechanisms or modules—perhaps innate, perhaps not—that process different kinds of information in the world.

It is unclear to us how infants might acquire such systems by the 4th month of life, if they do at all. For instance, if separate social and physical modes of construal are not innate, how are they formed through experience? Is the process of learning fundamentally different for social phenomena than physical phenomena? And are there separate modes of construal for animates and people? But at the same time, we believe that the use of terms such as "modes of construal" raises more questions than answers and does not provide any insight into key questions relating to development or mechanism.

Our view is that infants demarcate the physical and social domains through general learning mechanisms such as associative learning. However, we suggest that the result is not separate systems for learning that process different kinds of input. Rather, we posit that through experience infants associate actions that affect them emotionally and, to some extent, physically—what we think of as social actions—with people and not with other, inanimate things in the world. At the same time, infants also associate aspects of physics (e.g., solidity, gravity) with both people and inanimates because they receive only confirmatory evidence that such laws apply to both categories. This is why we consider it unlikely that infants would apply the constraint of continuous movement to objects but not to people, and we stress the need for caution in over-interpreting the limited dataset that is currently available