Answers are grounded in this paper and cite their sources.
Cognitive process models and brain connectivity analyses both provide a higher resolution into what cognitive functions are engaged during cognitive tasks than traditional methods. However, the two methods are typically applied in isolation. Our goal is to combine them, and ask which impaired cognitive functions are associated with a specific connectivity signature. Participants were 1,127 participants from the Alzheimer’s Disease Neuroimaging Initiative (ADNI), of varying cognitive status at their last included time point: Cognitively Normal = 554, Mild Cognitive Impairment = 386, Alzheimer's Disease = 187. We combine Bayesian multinomial processing tree (MPT) cognitive modelling with resting-state functional connectivity analysis guided by the Common Model of Cognition. MPT modelling showed that Learned and Delayed Retrieval processes had the largest diagnostic differences (η² ≈ 0.40), while Short-term Retrieval was unaffected. MPT parameters were correlated with widespread brain connectivity changes. However, brain connectivity changes were collinear, and mostly attributable to the diagnostic group. The exception was the Delayed Retrieval process parameter, which was associated with functional connectivity between areas associated with Working Memory and areas associated with Long-Term Memory, even when corrected for all other parameters. Combining the two analyses allowed us to link a function defined at the neural level to a function defined at the behavioral level. This article describes the potential and current limits of joining the two classes of models.
No takes yet. Share an insight, caveat, or question.
Potthoff et al. (2026) studied this question.
Synapse has enriched 8 closely related papers on similar clinical questions. Consider them for comparative context: