Repetition Dissociates Pointer and Content-based Representations in Visual Working Memory: Contrasting the CDA with Multivariate Shape Decoding

Read the full article See related articles

Discuss this preprint

Start a discussion What are Sciety discussions?

Listed in

This article is not in any list yet, why not save it to one of your lists.
Log in to save this article

Abstract

Memorizing a new phone number or address is hard at first, but becomes easier with repetition, as information shifts from working memory to long-term memory. Here we investigated how repetition affects the storage and transition of different aspects of mnemonic information by comparing univariate neural markers of active object storage with multivariate decoding of memory content. Thirty participants encoded lateralized stimuli from a continuous shape space into memory. Memory items were repeated six times in a row to induce learning. In line with earlier work, EEG recordings revealed that repetition led to a reduction in contralateral delay activity (CDA), a measure of active storage that has been taken to reflect a pointer-like representation of the individual object or its original source. In contrast, shape decoding during the retention and also after an impulse perturbation remained constant across repetitions. These results suggest that learning over repetitions reflects the abolishment of active and individuated object memory representations while passive, source-independent memory representations are retained.

Highlights

  • We recorded EEG as participants maintained a shape item, drawn form a continuous shape space, in working memory. These memory items repeated six times in a row before they changed to a new item.

  • We replicated the well-known CDA decrement effect from repetition, a process thought to index participants’ shift from relying on working memory to long term memory.

  • Multivariate decoding of shape identity was stable across repetitions, neither decreasing nor reliably increasing with repetition. This was true both after encoding and also following a task irrelevant ‘ping’.

  • These results dissociate the popular CDA ERP, a measure of memory load, from actual working memory content decodable via multivariate approaches.

Article activity feed