Task relevance modulates the cortical representation of feature conjunctions in the target template

Reshanne R Reeder,Michael Hanke,Stefan Pollmann

doi:10.1038/s41598-017-04123-8

Reshanne R Reeder, Michael Hanke + Show 1 more

Open Access

https://doi.org/10.1038/s41598-017-04123-8

Copy DOI

Abstract

Little is known about the cortical regions involved in representing task-related content in preparation for visual task performance. Here we used representational similarity analysis (RSA) to investigate the BOLD response pattern similarity between task relevant and task irrelevant feature dimensions during conjunction viewing and target template maintenance prior to visual search. Subjects were cued to search for a spatial frequency (SF) or orientation of a Gabor grating and we measured BOLD signal during cue and delay periods before the onset of a search display. RSA of delay period activity revealed that widespread regions in frontal, posterior parietal, and occipitotemporal cortices showed general representational differences between task relevant and task irrelevant dimensions (e.g., orientation vs. SF). In contrast, RSA of cue period activity revealed sensory-related representational differences between cue images (regardless of task) at the occipital pole and additionally in the frontal pole. Our data show that task and sensory information are represented differently during viewing and during target template maintenance, and that task relevance modulates the representation of visual information across the cortex.

Highlights

But a recent study reconstructed working memory content activated in frontal cortex using encoding model techniques[22]
We analyzed each subject’s accuracy, reaction time (RT; only correct trials included), and Inverse Efficiency Score (IES; an accuracy-weighted RT value (RT/accuracy) that controls for speed-accuracy tradeoffs) for each of the search tasks
How does the brain represent a target in preparation for visual search? Here we investigated the representation of task relevant feature dimensions as attention templates, i.e., working memory representations of targets in an upcoming search task

Summary

Introduction

But a recent study reconstructed working memory content activated in frontal cortex using encoding model techniques[22]. We used representational similarity analysis (RSA23, 24) to investigate the similarity of activity patterns elicited by task relevant and task irrelevant feature dimensions (e.g., a target defined by its orientation vs SF) during viewing and target template maintenance prior to visual search. Our time periods of interest were the cue period (i.e., the stimulus processing period) and the delay period prior to the onset of the search display (i.e., the target template maintenance period). Based on the working memory literature, as well as on previous neuroimaging studies of dimension-specific visual search[27], we expected the task relevant feature dimension to be represented maximally differently from the task irrelevant feature dimension (see Fig. 2a) in classical attention regions extending from the frontal eye fields to posterior parietal cortex along the intraparietal sulcus. The former would signal the result of a dimension change whereas the latter would indicate that left frontopolar cortex receives information about attended as well as not attended features, a precondition for detecting changes in the unattended dimension that may require a re-weighting of dimensional attention[27]

Methods

Results

Conclusion