Abstract
Attention during cognitive tasks fluctuates between periods of optimal (accurate and consistent) and suboptimal (error-prone and variable) states. Though previous studies have associated attentional control regions (e.g., dorsal attention network, DAN) with these fluctuations, how the relationship between attentional and perceptual regions gives rise to optimal states is currently unknown. The present research uses representational connectivity analysis to examine how fine-grained visual information is transmitted from visual to higher-order cortical regions during optimal (in-the-zone) and sub-optimal (out-of-the-zone) attentional states. To accomplish this, participants (N=145) performed the gradual onset Continuous Performance Task (gradCPT) during an fMRI scan. Participants viewed a series of city or mountain scenes, responding to cities (90% of trials) and withholding responses to mountains (10%). First, a visual ROI (corresponding to bilateral lateral occipital cortex) was functionally identified as voxels sensitive to a stimulus-driven difficulty timecourse based on the similarity between the city and mountain images. Previous research has shown that increased city-mountain similarity results in response errors and slower RTs. Next, two representational similarity matrices (RSMs) were derived by computing pairwise correlations between the activation patterns for each city exemplar (n=10) within the visual and DAN ROIs. Representational connectivity (RC) was quantified as the correlation between the these two RSMs. Importantly, RC was computed separately for in-the-zone (low RT-variability) and out-of-the-zone (high RT-variability) trials. We found that the RC between the visual and DAN RSMs was greater while participants were in-the-zone (r=0.72, p< 0.001) than out-of-the-zone (r=0.19; p=0.20). Similar differences were observed in the fronto-parietal control network, while other large-scale brain networks tested showed no effect. These results suggest that optimal, but not suboptimal, states are associated with the integration of fine-grained visual information into large-scale task-positive brain networks. More broadly, this work provides a novel way to conceptualize optimal brain states within an information processing framework.
Meeting abstract presented at VSS 2017