Consistent Neural Representation of Affective States between Watching and Recall using fNIRS

People tend to recall similar feelings when remembering the same event. Yet the neural mechanisms by which emotional valence and arousal are represented during perception and later recall remain unclear. Here, we used functional near-infrared spectroscopy to record prefrontal cortex activity while participants viewed and then freely recalled short video clips designed to elicit four affective states (amusement, calmness, disgust, and sadness). Multidimensional scaling revealed that both modality-general (shared across viewing and recall) and modality-specific (unique to each phase) dimensions of valence and arousal were significant, and cross-modal classification successfully decoded valence and arousal. Our results support a hierarchical model in which a stable core affect is preserved across perception and memory, while context-dependent refinements adapt emotional representations during recall. This work highlights fNIRS as a useful tool for studying emotional memory in ecologically valid settings and lays groundwork for multimodal imaging of affective replay.