Childhood neglect is associated with alterations in neural prediction error signaling and the response to novelty

Psychol Med. 2024 Oct 24;54(14):1-9. doi: 10.1017/S0033291724002411. Online ahead of print.

Abstract

Background: One in eight children experience early life stress (ELS), which increases risk for psychopathology. ELS, particularly neglect, has been associated with reduced responsivity to reward. However, little work has investigated the computational specifics of this disrupted reward response - particularly with respect to the neural response to Reward Prediction Errors (RPE) - a critical signal for successful instrumental learning - and the extent to which they are augmented to novel stimuli. The goal of the current study was to investigate the associations of abuse and neglect, and neural representation of RPE to novel and non-novel stimuli.

Methods: One hundred and seventy-eight participants (aged 10-18, M = 14.9, s.d. = 2.38) engaged in the Novelty task while undergoing functional magnetic resonance imaging. In this task, participants learn to choose novel or non-novel stimuli to win monetary rewards varying from $0 to $0.30 per trial. Levels of abuse and neglect were measured using the Childhood Trauma Questionnaire.

Results: Adolescents exposed to high levels of neglect showed reduced RPE-modulated blood oxygenation level dependent response within medial and lateral frontal cortices particularly when exploring novel stimuli (p < 0.05, corrected for multiple comparisons) relative to adolescents exposed to lower levels of neglect.

Conclusions: These data expand on previous work by indicating that neglect, but not abuse, is associated with impairments in neural RPE representation within medial and lateral frontal cortices. However, there was no association between neglect and behavioral impairments on the Novelty task, suggesting that these neural differences do not necessarily translate into behavioral differences within the context of the Novelty task.

Keywords: early life stress; fMRI; reinforcement learning.