The effects of the post-delay epochs on working memory error reduction

Accurate retrieval of the maintained information is crucial for working memory. This process primarily occurs during post-delay epochs, when subjects receive cues and generate responses. However, the computational and neural mechanisms that underlie these post-delay epochs to support robust memory remain poorly understood. To address this, we trained recurrent neural networks (RNNs) on a color delayed-response task, where certain colors (referred to as common colors) were more frequently presented for memorization. We found that the trained RNNs reduced memory errors for common colors by decoding a broader range of neural states into these colors through the post-delay epochs. This decoding process was driven by convergent neural dynamics and a non-dynamic, biased readout process during the post-delay epochs. Our findings highlight the importance of post-delay epochs in working memory and suggest that neural systems adapt to environmental statistics by using multiple mechanisms across task epochs.Author summary: In daily life, we often need to store information temporarily (during a delay epoch) and retrieve it later when required. While this ability may seem simple, it poses significant challenges at the neural level. Neural activity is inherently highly variable (“noisy”), so how can we maintain accurate memory despite this variability? Previous research has primarily focused on neural processes during the delay epoch, however, the neural processes after delay remain (post-delay epochs) poorly studied. The post-delay epochs are critical for information retrieval—for instance, they are time periods when we write down something we just heard of, or, in a laboratory setting, when an animal makes a saccade to indicate a memorized color to earn a reward. In this study, we investigated the computational role of the post-delay epochs in helping noisy neural networks reduce memory errors. Our findings revealed that post-delay processes adapt to environmental statistics, and identified key neural mechanisms that reduce memory errors, highlighting the importance of post-delay epochs in supporting robust working memory.