Multiplexed BOLD oscillations reveal the interplay of normalization and attention

Monkey electrophysiology has linked attention to divisive normalization, yet noninvasive evidence in humans remains limited. We use frequency-tagged fMRI to isolate visual cortical populations that simultaneously encode multiple competing inputs. We show that responses of these sites are suppressed during inattention and enhanced during attention – consistent with the normalization model, which predicts that attention selectively disinhibits competing inputs – offering a noninvasive translational bridge to study fine-grained computations underlying attentional selection.