Multivariate decoding of EEG reveals key spatial and temporal distinctions in perception and valuation

Decisions often require the integration of multiple considerations or attributes. Yet despite considerable research, debate continues about the neural mechanisms underlying the calculation of individual attribute values. One key question centers on whether attribute values are computed solely within the valuation network or also within perceptual regions specialized for identifying specific attributes. While each account makes distinct predictions concerning when attribute value signals should emerge, these hypotheses remain untested. To this end, we investigated when attribute-specific valuation emerges relative to earlier perceptual and later integrated value processes. Participants first learned to associate different face and color stimuli with specific monetary payoffs. They then made choices based on these reward-associated attributes in a multi-attribute decision making task, during which we recorded participants' EEG activity. Using multivariate classification analyses, we found that perceptual representations could be decoded early and in parallel, but that attribute value decoding emerged in a more sequential manner, followed by evidence of integrated value representations. Spatially, perceptual representations were most distinct, with attribute value representations demonstrating closer alignment with integrated value than their respective perceptual components. Importantly, attribute values amplified their perceptual representations, but only after the emergence of the integrated value representation, suggesting a post hoc attentional selection mechanism. Collectively, our findings shed new light on prior work and highlight with clearer temporal precision the evolution from perceptual to evaluative to decisional computations.