Abstract

Computational models suggest that visual attention is directed via activation profiles in neural ‘attentional priority maps’ spread across retinotopically organized brain regions. Prior work has proposed that these maps independently index the physical salience and behavioral relevance of stimuli, as determined by examining the topography of activation across regions while stimulus features (e.g., contrast) and task demands (e.g., attention) are independently manipulated (e.g., Sprague et al., 2018). Previously, these spatial maps have been quantified by measuring functional MRI while participants attended to one of two stimuli presented in the periphery. Thus, it remains unknown how more diverse modulations of attention—attention to fixation and attention to both stimuli—impact multiple stimulus representations in cortical priority maps. To probe this question, we used a selective attention task in which participants monitored random line stimuli and reported whether the stimuli cohered into a clockwise or counterclockwise spiral. When cued to attend to both stimuli, they reported the direction of the first item to cohere. If cued to fixation, participants maintained attention on a central fixation cross and responded whenever the horizontal or vertical line increased in size. This allows us to better survey the landscape of attentional control and more concretely assess how attention in different configurations—towards fixation, selective towards one of several stimuli, and distributed across multiple stimuli—changes stimulus representations. Using a spatial inverted encoding model, we were able to reconstruct images of priority maps from retinotopic regions in visual and parietal cortex which contained representations of each visual stimulus. We found that attention modulated stimulus representations in reconstructed spatial maps across human cortex dependent on condition, providing further evidence that these maps encode the relative priority of stimuli based on both their salience and relevance.