ERP correlates of auditory peak shifts in stimulus generalization

Poster Session E - Monday, April 15, 2024, 2:30 – 4:30 pm EDT, Sheraton Hall ABC

Chelsea Joyner¹ (cnjoyner@ksu.edu), Matthew Wisniewski¹; ¹Kansas State University

The peak shift effect is a widely observed phenomenon in which discrimination learning causes generalization gradients for learned responses to peak at a novel stimulus rather than a trained one. Associative theories posit that the peak shift arises from reweighting connections between stimulus representations and decision/response outputs. Non-associative representational plasticity perspectives explain the peak shift as resulting from changes in stimulus representations themselves. We investigated behavioral and event-related potential (ERP) expressions of the peak shift, the latter being used to characterize the processing stage(s) in which peak shifts arise. Stimuli consisted of 7 different frequency modulation (FM) rates ranging from “slow” (4.14, 6.9,7.2 octaves/s) to “fast” (9.52, 9.94, 16.56 octaves/s). Subjects were trained with feedback to respond “Target” for a 8.28 octaves/s FM rate and “Non-Target” to 6.9 octaves/s. Test trials using FM rates from the entire stimulus continuum were interspersed among these training trials. Analyses of "Target" responses on test trials revealed a gradual increase in peak shift as discrimination learning progressed. This was evident in more "Target" responses to a shifted 9.52 and 9.94 octaves/s FM rate than the trained target rate. Interestingly, two components of the ERP response to test stimuli appeared to parallel these changes seen in behavior: the P2 and a Late Posterior Positivity (LPP). Data is discussed in regard to what these components may mean for the processes involved in perceptual discrimination learning, and potential contributions from multiple learning processes in generating peak shifts.

Topic Area: PERCEPTION & ACTION: Audition