Neural Signatures of Automatic Auditory Regularity Detection Reflect Individual Differences in Explicit Short-Term Memory

Abstract

Perception is shaped by the statistical structure of the environment, reflecting the brains capacity to detect and exploit regularities in sensory input. This process requires the maintenance of contextual information over time, yet the nature of the memory mechanism supporting automatic structure learning remains unclear. Here, we ask whether auditory regularity processing relies on a dedicated sensory buffer or instead draws on domain-general mnemonic resources. We related neural indices of regularity processing measured with EEG during passive listening to behavioural performance on an explicit auditory short-term memory task. Human participants (N=30; both sexes) passively listened to regularly repeating (cycles of 5.5 seconds) or random tone sequences, while sustained and tone-locked neural responses were extracted as complementary markers of predictability tracking and prediction-error signalling. Individual differences in explicit memory performance, quantified using a delayed match to sample task, systematically predicted both neural measures: high performers showed enhanced sustained responses and attenuated tone-evoked responses (most pronounced in the N2 time window; 250-400 ms post-onset) to regular sequences, whereas low performers showed no reliable modulation by sequence structure. These findings demonstrate that the memory processes engaged automatically during auditory pattern analysis are not encapsulated, but instead draw on shared mnemonic resources, providing a link between predictive perception and individual variability in sensory memory capacity.