Habituation Reflects Optimal Exploration Over Noisy Perceptual Samples

Abstract

From birth, humans constantly make decisions about what to look at and for how long. Yet the mechanism behind such decision-making remains poorly understood. Here we present the rational action, noisy choice for habituation (RANCH) model. RANCH is a rational learning model that takes noisy perceptual samples from stimuli and makes sampling decisions based on Expected Information Gain (EIG). The model captures key patterns of looking time documented in developmental research: habituation and dishabituation. We evaluated the model with adult looking time collected from a paradigm analogous to the infant habituation paradigm. We compared RANCH with baseline models (no learning model, no perceptual noise model) and models with alternative linking hypotheses (Surprisal, KL divergence). We showed that 1) learning and perceptual noise are critical assumptions of the model, and 2) Surprisal and KL are good proxies for EIG under the current learning context.

Individual differences in processing speed and curiosity explain infant habituation and dishabituation performance

“…We did not find evidence for an association between habituation and curiosity (Cao et al, 2022). However, we found an association between curiosity and dishabituation scores, but only for infants who did not habituate.…”

“…Although habituation and dishabituation are widely used measures in the field of infant cognition, their underlying mechanisms have been unknown to date. Previous research suggested that habituation and dishabituation might be related to processes as diverse as attention (Blaga & Colombo, 2006), learning (Rose & Feldman, 1997), processing speed (Feldman and Mayes, 1999), or curiosity (Cao et al., 2022). Here, we were able to test these competing hypotheses.…”

“…Recent theories reframe habituation as an active process: Infants might not only process information, but they might also actively decide whether and when to do so (Cao et al., 2022). Although previous research looked at curiosity as a state, with information gains triggering a momentary increase of curiosity, here we look at curiosity as a trait that differs across individuals.…”

“…Hence, this model‐based approach allowed us to assess multiple cognitive factors at the same time, and investigate their unique contribution to habituation and dishabituation. We explored whether attention (Blaga & Colombo, 2006), learning (Rose & Feldman, 1997), processing speed (Feldman & Mayes, 1999) or curiosity (Cao et al., 2022) would correlate with habituation and dishabituation, and also considered it possible that multiple cognitive factors concurrently predicted habituation and dishabituation.…”

No evidence for familiarity preferences after partial exposure to visual concepts in preschoolers and infants

“…In a recent example of such a model, Cao, Raz, Saxe, & Frank (2022) proposed that habituation and novelty preferences could be explained by a rational learner that takes noisy perceptual samples to maximize information gain (RANCH). This model accurately predicted adult looking time patterns in a self-paced habituation paradigm, reproducing both habituation and novelty preferences.…”