Poster #6 - Investigating prediction error in auditory and visual domain of infants using pupillometry

• In Event: 3-206 - Poster Session 13
  In Poster Session: PS 13 Section - Attention, Learning, Memory

Integrative Statement

Prediction, or the ability to use past experiences to generate expectations about future
sensory input, influences learning. Predictions can be correct or incorrect and numerous adult studies have demonstrated how prediction errors can support cognitive changes. Recent work in developmental psychology have found evidence that infants experience prediction error in the visual domain, suggesting predictions can support cognitive development early in life. Prediction error in the auditory domain, however, was not investigated. The goal of this current study is to compare the prediction error response for auditory and visual stimuli of 6 months old infants.

To index prediction error, we used eye-tracking as it allowed us to measure pupil diameter and design the task to be gaze contingent. Pupil diameter has been shown to reflect cognitive surprise in infants, which is closely linked to prediction error. Furthermore, the gaze-contingent nature of the task allowed us to make sure that the stimuli were presented only when the infant was paying attention, which is especially important for a learning paradigm.

The task is composed of two conditions: 1) auditory predicting visual (AV) and 2) visual predicting auditory (VA). In the AV condition, infants learn that a sound predicts a visual stimulus (e.g., A1 -> V1) and in the VA condition, infants learn that a visual predicts an auditory stimulus (e.g., V2 -> A2). After familiarization, infants viewed omission trials, such that A1 was not followed by V1 in the AV condition, and V2 was not followed by A2 in the VA condition, but a blank screen and silence. Using a blank screen, instead of an unexpected stimulus, was critical because it allowed us to isolate surprise from stimulus processing as the presentation of any stimulus causes pupil diameter to change. The order of condition was counterbalanced across participants. 

Comparing present and omission trials in both conditions will allow us to see how pupil diameter (i.e. level of prediction error) change as infants’ predictions were unexpectedly being violated and if it depends on the domain it occurs in (i.e. visual versus auditory). Preliminary analysis of data from ten 6-month-old infants show that there might be a domain-specific prediction error response. Specifically, in the AV condition, the pupil size for omission trials is larger than present trials, however, in the VA condition, the pupil size for omission trials is very similar to present trials. Following data collection of this task, we will use concurrent measures of eye-tracking and fNIRS to investigate the relationship of neural and behavioral measures of prediction error. Overall, preliminary data suggest that prediction errors are experienced in both the auditory and visual domain, but that it may be attenuated in the auditory domain.

Authors

• Felicia Zhang, Princeton University
  Presenting Author

• Lauren Emberson, Princeton University
  Non-Presenting Author