%0 Journal Article %A Christophel, Thomas %A Yan, Chang %A Stopak, Lee %A Hetzer, Stefan %A Haynes, John-Dylan %T Evidence for non-frontal control of sensory working memory %B Journal of Vision %D 2018 %R 10.1167/18.10.364 %J Journal of Vision %V 18 %N 10 %P 364-364 %@ 1534-7362 %X Items held in visual working memory can be quickly updated, replaced, removed and even manipulated in accordance with behavioral goals. Current views postulate that memorized contents are represented by mnemonic activity distributed across the full cortical hierarchy. It has however remained unclear which neural mechanism exert control over these distributed stores to select what items should be remembered and which can be forgotten. Here, we use multivariate pattern analyses (MVPA) to identify fMRI activity patterns representing executive control processes supervising these flexible stores. In the task, subjects had to memorize two stimuli - one visual grating and one auditory tone that were sequentially presented. After an extensive delay, a cue stimulus instructed the subjects whether the first or the second stimulus should be used for a successive change discrimination task (at the same time also implying which stimulus could be forgotten). Multivariate decoding indicated that memory for the cued item persisted throughout the 30 second trial and mnemonic activity selectively declined when the item was dropped from memory. To identify areas involved in control, we used MVPA to identify activity patterns that are selectively carrying information regarding the instruction conveyed by the cue ('first stimulus' or 'second stimulus'). We found that transient neural activity in inferior parietal and superior temporal cortices carried information about what items should be retained during cue presentation. These selection-specific activity patterns generalized across both numerical and alphabetical selection cues. We found no such evidence in prefrontal cortex. In summary, we our findings point to posterior, but not frontal areas as carrying information to exert control of distributed stores. Thus, our results suggest that selection of memorized items can be controlled in a distributed and decentralized fashion thus questioning the notion of a prefrontal central executive supervising all working memory function. Meeting abstract presented at VSS 2018 %[ 1/22/2021 %U https://doi.org/10.1167/18.10.364