Modeling Infant i's Look on Trial t: Race-Face Preference Depends on i's Looking Style

Abstract

When employing between-infant designs young infants' looking style is related to their development: Short looking (SL) infants are cognitively accelerated over their long looking (LL) peers. In fact, looking style is a within-infant variable, and depends on infant i's look distribution over trials. For the paired array setting, a model is provided which specifies the probability, πi ∈ [0, 1], that i is SL. The model is employed in a face preference study; 74 Caucasian infants were longitudinally assessed at 3, 6, and 9 months. Each i viewed same race (Caucasian) vs. other race (African) faces. Infants become SL with development, but there are huge individual differences in rate of change over age. Three month LL infants, trueπ^i<1/2, preferred other race faces. SL infants, trueπ^i>1/2, preferring same race faces at 3, and other race faces at 6 and 9 months. Looking style changes precede and may control changes in face preference. Ignoring looking style can be misleading: Without considering looking style, 3 month infants show no face preference.

Modelling processes in infant research

“…There is no other way to capture latent individual differences than to model them. To give an example regarding the visual pair preference task, in Thomas and Fassbender (2017), we defined a latent variables mixture model that explicitly recognizes individual differences in the manner with which infants look. Specifically, we defined a model parameter π i for the probability of an infant being a short looker, resulting in (1 À π i ) being the probability of an infant being a long looker.…”

“…In Thomas and Fassbender (2017), we analysed infant fixations of African and European faces presented pairwise side by side for 15 trials. The position of the stimuli was as in previous visual pair comparison tasks investigating face-race preference in infancy.…”