Abstract
Looking time habituation dominates studies of visual perception and cognition in infants and yet remains poorly understood. Inspired in part by psychophysical approaches, we analyzed individual patterns of habituation using regression techniques (Thomas & Gilmore, 2004). The goal was to determine whether individual infants show consistent patterns of habituation within a single session. One group (n=57) viewed patterns of optic flow depicting motion of the viewpoint 45 deg to the left or right of center. A second group (n=73) viewed movie clips of two different infant faces. Each infant viewed 7 trials of one display (familiar), followed by 7 trials of a different (novel) display. Linear and nonlinear functions were fit to looking time data for each infant and goodness of fit assessed by the Bayesian Information Criterion (BIC). On average, younger (3–4 month-old) infants looked longer than older (5–6 mo-old) ones, F(1,126)=8.99, p=.003, faces elicited longer looking, F(1,126)=8.52, p=.004, and the familiar display drew longer looks than the novel one, F(1,126)=9.17, p=.003. Nevertheless, infant looking time patterns differed across the two phases. Fewer than 1/5 infants habituated to both the familiar and novel displays according to the standard 50% decrement from baseline criterion. Moreover, for 57/130 infants, a flat line through the mean provided the best fit to the observed looking time data in both phases of the experiment. When looking times declined, a 3-parameter exponential function provided the best fit. The results suggest that few infants show systematic look time patterns to both familiar and novel displays within the same experimental session. While average looking times across infants may show statistically detectable trends, individual data do so inconsistently. The results call into question the generality of claims made about infants' perceptual or cognitive abilities based on group, not individual-level habituation data.
Supported by NICHD (HD041476󔼉)