Precision of sensory representations in cortex is indexed by alpha power and pupil dilation, and tracks the dynamics of correlated noise
In the authors' words
Sensory representations fluctuate in precision from moment to moment, shaping the fidelity of perception and memory. Population-coding theory proposes that these fluctuations are constrained by correlated variability within neural populations, yet linking these mechanisms to behaviour and brain state in humans has remained challenging. Here we combined EEG with an encoding-decoding framework to reconstruct orientation-selective population responses and quantify representational precision dynamically over time. Precision predicted subsequent working-memory performance and evolved in close correspondence with the dynamics of correlated variability between decoded orientation channels. Following stimulus onset, correlated variability was selectively suppressed among similarly tuned channels, the pattern predicted to maximise coding fidelity. Prestimulus alpha power and pupil size independently predicted representational precision, linking these dynamics to endogenous cortical and neuromodulatory state. Together, these findings identify candidate mechanisms and accessible physiological markers of representational precision in humans.
Appeared: Thursday, November 19. bioRxiv. Preprint, not yet peer-reviewed.