Search Papers | Poster Sessions | All Posters
Poster A32 in Poster Session A - Tuesday, August 6, 2024, 4:15 – 6:15 pm, Johnson Ice Rink
Cognitive Integration – How Task Representations Integrate Information from Multiple Sources
Stephanie Leach1,2, Hannah Morrow1,2, Jiefeng Jiang1,2, Kai Hwang1,2 (); 1Psychological and Brain Sciences Department, University of Iowa, 2Cognitive Control Collaborative, University of Iowa
To facilitate cognitive flexibility, the human brain continuously integrates perceptual and internal state information to execute and switch between tasks. We developed a Bayesian model to investigate the neural and computational processes of integration. This model integrates the uncertainty of perceptual and internal state information into a posterior distribution, which jointly encodes beliefs of task-relevant inputs. Task decisions are then predicted by this posterior joint distribution. We designed a behavioral paradigm where subjects integrate a non-observable internal state with observable color perception to select or switch between two selective attention tasks. By applying the Bayesian model, we estimated the trial-by-trial color perception, state belief, and their associated uncertainty estimates, as well as the integrated to-be-performed task belief. This model successfully predicted human subjects’ performance, outperforming several competing models. Using model derived predictors to analyze fMRI and EEG data, we found that the medial frontal, lateral middle frontal gyrus, and intraparietal sulcus, as well as theta-to-alpha band signals encoded statistical properties of this integrated posterior joint distribution. Our findings offer a computational account for cognitive integrative functions of the frontoparietal system.
Keywords: Cognitive Flexibility Cognitive Control Integration Bayesian Generative Model