Dissertations, Theses, and Capstone Projects

Date of Degree

9-2025

Document Type

Master's Thesis

Degree Name

Master of Science

Program

Cognitive Neuroscience

Advisor

David Johnson

Committee Members

Tony Ro

David Johnson

Tatiana Aloi Emmanouil

Wanda Mercado

Subject Categories

Cognitive Neuroscience | Cognitive Psychology | Cognitive Science | Experimental Analysis of Behavior

Keywords

Prediction Errors (PE), aversive learning, fear conditioning, physiological measures, Intolerance of Uncertainty (IU), State trait anxiety (STAI-T)

Abstract

Prediction error (PE) is the discrepancy between expected and actual outcomes. It is a central mechanism in associative learning particularly within fear-conditioning paradigms. Identifying physiological markers of PE in aversive contexts has important implications for refining learning models and enhancing clinical interventions for anxiety-related disorders. This study examined skin conductance response (SCR) as a potential physiological index of PE by analyzing both anticipatory and omission-related responses across multiple learning phases. Participants completed a Pavlovian fear-conditioning paradigm in which two conditioned stimuli (CS+) signaled threat at different reinforcement probabilities: CS+100 (certain threat) and CS+50 (uncertain threat), alongside a non-reinforced safety cue (CS-).

Repeated measures ANOVAs revealed robust acquisition effects, with significantly higher SCRs to CS+ compared to CS−, and habituation across trial blocks. Extinction analyses showed successful reduction of SCRs across all cues, with no significant differences between stimulus conditions or individual difference groups. During recovery, reinstatement produced a reemergence of fear responses across all stimuli, whereas spontaneous recovery reflected only early vs late differences, without stimulus-specific effects.

To determine whether offset SCRs reflect prediction error (PE)-driven learning, a repeated-measures correlation was conducted. Results revealed a non-significant relationship between omission responses and subsequent anticipatory SCRs, suggesting offset SCRs do not serve as a direct marker of PE. Exploratory ANCOVAs indicated that individuals with higher trait anxiety showed heightened omission responses to uncertain threat (CS+50), though no effects emerged for intolerance of uncertainty (IU). To test whether offset SCRs reflect PE-driven learning, a repeated-measures correlation (rmcorr) was conducted to test whether greater omission-related responses (SCR at offset for non-reinforced CS+ trials) predicted smaller subsequent anticipatory SCRs (CS+ onset). Contrary to the PE hypothesis, results revealed a non-significant and near-zero correlation (r = 0.039, p = .782, 95% CI [-0.234, 0.306]), suggesting offset SCRs may not encode PE signals, but instead reflect generalized arousal or affective relief.

Further exploratory ANCOVAs incorporating individual differences revealed a significant interaction between trait anxiety (STAI-T) and omission-related responses to the threat cue- whereby individuals with higher trait anxiety showed elevated omission responses specifically to uncertain threat (CS+50), but not to certain threat (CS+100) or safety cues. No significant effects were found for Intolerance of Uncertainty (IU).

Overall, these findings suggest that while SCR effectively tracks fear acquisition and extinction, offset SCRs may not reflect prediction error in a learning-specific sense. Instead, they may index emotional or attentional processes related to uncertainty resolution. These results contribute to current literature in clarifying the neural and physiological mechanisms underlying fear learning and highlight the need for more precise biomarkers of PE in translational anxiety research.

Download

Share

COinS