Abstract
The attentional blink reflects a ubiquitous bottleneck with selecting and processing the second of two targets that occur in close temporal proximity. An extensive literature has examined the attention blink as a unitary phenomenon. As a result, which specific component of attention - perceptual sensitivity, choice bias, or both - is compromised during the attentional blink, and their respective neural bases, remains unknown. Here, we address this question with a multialternative task and novel signal detection model, which decouples sensitivity from bias effects. We find that the attentional blink impairs specifically one component of attention - sensitivity - while leaving the other component - bias - unaffected. Distinct neural markers of the attentional blink were mapped onto distinct subcomponents of the sensitivity deficits. Parieto-occipital N2p and P3 potential amplitudes characterized target detection deficits, whereas long-range high-beta band (20-30 Hz) coherence between frontoparietal electrodes signaled target discrimination deficits. We synthesized these results with representational geometry analysis. The analysis revealed that detection and discrimination deficits were encoded along separable neural dimensions, whose configural distances robustly correlated with the neural markers of each. Overall, these findings provide detailed insights into the subcomponents of the attentional blink and reveal dissociable neural bases underlying its detection and discrimination bottlenecks.