Derivation of Idiographic Model-based Parameters from Resting State Neuroimaging Data

Abstract

To work towards a complete and holistic understanding of human cognition, an essential task is to relate idiographic, or individual, parameters representing cognitive functioning to interactions between connected brain networks identified by neuroimaging methods. Here, using the ACT-R cognitive architecture, two models of cognitive strategy under two separate conditions of a working memory N-back are implemented. Individual participants who completed the N-back task are modeled by the determination of a set of parameter values which result in accurate model prediction of a given participant’s response time and accuracy under the task. These parameter-based characterizations, in turn, demonstrate reasonable predictions of individual behavior, and are interpreted as reflecting differences between individuals in the contribution of multiple cognitive components to working memory function. A method of determining a subset of parameters that are critical for prediction of behavior is applied before the resulting parameters are related to functional connectivity measures in individual resting-state fMRI data. It is observed that ACT-R model parameters interpreted to represent goal-directed attention, memory decay, and processing speed each reflect the coupling between the task-negative default mode system and a set of task-positive brain networks; however, coupling between different regions within these networks was specific to individual parameters. This approach seeks to validate previously reported associations between behavior, cognitive constructs, and brain activity while simultaneously extending these associations to account for differences in behavior and ability between individuals.