The late positive component in recognition memory is linked to mnemonic evidence accumulation.

Recognising objects from memory requires an integration of sensory and mnemonic information. This process has been theorised to occur via a stochastic evidence accumulation process implemented within the parietal cortex. Recent electroencephalographic (EEG) evidence indicates that the widely studied parietal Late Positive Component (LPC) shows characteristics of such a mnemonic accumulation signal. Here, we formally investigated this hypothesis using generative computational modelling that links a trial-level memory strength variable in the Diffusion Decision Model (DDM) with LPC amplitudes prior to the time of the decision. We recorded EEG from 24 participants making recognition judgements based on either studied or novel words. Each participant completed up to three testing sessions. We replicated recent findings that the LPC ramps up and peaks around the time of the decision and corresponding motor response. LPC amplitudes also covaried with accuracy and response time, as expected of a neural correlate of memory strength. By fitting DDMs to LPC amplitudes and behavioural data using specialised neural network tools, we demonstrate that LPC amplitudes are selectively associated with the rate of evidence accumulation, signifying memory strength. This association was stronger for previously studied words compared to novel words, and strongest at the time window immediately prior to the recognition decision. Our findings, therefore, recast the LPC as a neural signature of mnemonic strength linked to the rate of evidence accumulation during recognition memory decisions.