Measures of infant looking times, habituation, and preferences are core empirical devices in the study of infant cognition and its development. While this has been the case for many years, interpretations of these looking data are still rife with controversy.;This study takes a modeling approach, providing a motor account of looking with the aims of clarifying the knowledge that looking behaviors can reveal, and shedding light on the neural processes required to generate the patterns of looking behavior observed in the lab. Looking in the model is not merely an output; it actively and autonomously controls exposure to stimuli. The model instantiates dynamically plausible patterns of neural activation to control looking behavior in continuous time. Thus the approach is motivated by issues of stability and flexibility in motor control. Gaze must be stabilized for the infant to learn about a stimulus, without getting too "stuck" to attend to other stimuli.;Rather than replicating gross measures of total looking times, the model's "output" is a looking direction in continuous space at each moment. Thus, compared to past models of habituation, it can be applied to a much wider range of experimental paradigms, including presentations of multiple stimuli at once, as well as the common practice of turning displays on and off in response to infant looks. Across various experimental paradigms, the model exhibits patterns of habituation and looking preferences like those observed with infants. It also captures microscale phenomena such as distributions of fixation lengths.
展开▼
