Resolving the Attentional Bottleneck: Cross-Modal Representational Alignment Reveals Late-Stage Working Memory Consolidation

The attentional blink reveals the severe capacity limits of conscious human processing, yet whether this bottleneck arises from early perceptual decay or late-stage consolidation failure remains highly debated. To disentangle these psychological stages, the present study employed causally silenced cross-modal representational similarity analysis (EEG-fMRI RSA) to trace the fine-grained spatiotemporal information flow of target processing. Our results revealed a striking spatiotemporal double dissociation: although the early primary visual cortex successfully extracted the physical features of the targets, a stable cross-modal psychological representation did not emerge until 414 ms post-stimulus, an effect entirely dominated by the right parieto-frontal network. This finding provides decisive spatiotemporal evidence that the attentional bottleneck stems not from a failure of perceptual extraction, but from restricted late-stage working memory consolidation. By pinpointing the critical temporal threshold and the gating network for conscious access, this study offers robust cross-modal support for the Global Workspace Theory.