Altered amygdala and striatal responsivity and prediction error encoding of temporal reward dynamics in post-trauma psychopathology

TL;DR

This study finds altered brain activity in trauma-related mental health conditions, particularly in regions involved in reward prediction and timing.

Contribution

The study identifies disrupted amygdala and striatal responsivity to temporal reward prediction errors in post-trauma psychopathology.

Findings

PTP individuals showed reduced amygdala and putamen activation during unexpected reward receipt.

Blunted amygdala deactivation correlated with lower anhedonia in PTP individuals.

Model-based analysis revealed altered prediction error encoding in PTP-related brain regions.

Abstract

Diminished positive affect is a hallmark of post-trauma psychopathology (PTP), e.g. posttraumatic stress disorder and major depressive disorder, yet the circuit dysfunction underlying these PTP symptoms is poorly understood. Computational models offer a powerful framework to probe learning processes by linking such processes to brain activity in regions involved in updating reward predictions. Here, we examined fMRI neural responsivity to temporal prediction errors—deviations in timing of expected reward delivery—and applied a temporal difference (TD) learning model to characterize model-based prediction error neural encoding in individuals with PTP (N = 45; 32 females) and trauma-exposed healthy controls (N = 45; 25 females). Participants learned that a visual cue reliably predicted timing of a primary reward (oral juice bolus). To induce temporal prediction errors, we introduced occasional “catch” trials, which unexpectedly extended the cue-outcome interval. Compared to controls, the PTP group showed blunted activation in the left amygdala and putamen to reward receipt at unexpected vs. expected times and blunted deactivation to reward absence at unexpected vs. expected times (corrected p’s < 0.05). Interestingly, attenuated amygdala deactivation was associated with lower anhedonia (ρ43 = –0.50, p < 0.001), suggesting a possible mechanism for preserving hedonic capacity. Model-based fMRI revealed diminished amygdala and putamen prediction error encoding at reward delivery in PTP but exaggerated insula and putamen encoding at cue presentation (corrected p’s < 0.05). We provide new evidence of disrupted PTP amygdala-striatal reward responsivity and computational learning signals, demonstrating an unrecognized pervasiveness of dysfunction extending to the fundamental learning domain of cue-reward timing.

Figures