Schedule of Events | Search Abstracts | Invited Symposia | Symposia | Poster Sessions | Data Blitz

Individualized Parcellation Reveals Specificity of Dorsal Attention Network Function in Sustained Attention

Poster Session D - Monday, March 9, 2026, 8:00 – 10:00 am PDT, Fairview/Kitsilano Ballrooms

Janet Z. Li¹ (jl4646@drexel.edu), Lotus Shareef-Trudeau¹, Tiara Bounyarith¹, Dave Braun¹, Michael Esterman^2,3, Aaron Kucyi¹; ¹1) Department of Psychological and Brain Sciences, Drexel University, ²2) Neuroimaging Research Center for Veterans (151), VA Boston Healthcare System, ³3) Department of Psychiatry, Boston University School of Medicine

Sustained attention enables stable, goal-directed behavior and fluctuates between optimal “in the zone” and suboptimal “out of the zone” states. Trial-wise reaction time variability (RTV) during continuous performance tasks indexes these attentional states and has been linked to fluctuations between dorsal attention network (DAN)- versus default mode network (DMN)-dominant states. However, these networks are typically treated as monolithic, despite evidence of comprising multiple distinct networks. Although such network distinctions have been demonstrated in cognitive domains including memory and semantic processing, they have not been systematically examined in relation to sustained attention. Here, we tested whether RTV fluctuations preferentially relate to specific networks within canonical DAN and DMN using individualized mapping. Twenty healthy adults completed two fMRI sessions, each including an 8-minute gradual-onset Continuous Performance Task. Individualized DAN and DMN subnetworks were identified using Multi-Session Hierarchical Bayesian Modeling, and RTV–BOLD associations were quantified using z-scores extracted from general linear model analysis. At the group level, DAN-A showed the highest mean RTV-BOLD correlation, followed by DAN-B, while DMN networks showed near-zero or negative correlations. A one-way repeated measures ANOVA revealed a significant main effect of network (F(3, 57) = 6.53, pHF = 0.007). One-sample t-tests against zero revealed a significant positive RTV–BOLD association in DAN-A (t (19) = 3.15, p = 0.005), whereas DAN-B and both DMN networks did not differ from zero (all p > 0.35). Our results suggest that RTV-indexed attentional fluctuations selectively recruit DAN-A, reflecting inefficient or maladaptive hyper-engagement of the DAN’s core sustained goal maintenance component.

Topic Area: METHODS: Neuroimaging