Prediction of Individual Melodic Contour Processing in Sensory Association Cortices From Resting State Functional Connectivity

Abstract

Recent studies suggest that it is possible to predict an individual brain's spatial activation pattern in response to a paradigm from their functional connectivity at rest (rsFC). However, it is unclear whether this prediction works across the brain. We here aim to understand whether individual task activation can be best predicted in local regions that are highly specialised to the task at hand or whether there are domain-independent regions in the brain that carry most information about the individual. To answer this question, we used fMRI data from participants at rest and during an auditory oddball paradigm. We then predicted individual differences in brain responses to melodic deviants from their rsFC both across the whole brain and within the auditory cortices. Predictability was consistently higher in sensory association cortices: In the local (auditory cortex) parcellation, the best predicted area was the right superior temporal gyrus (STG), an auditory association area, while in the global parcellation, the best predicted network was the bilateral visual association cortex. Our results indicate that individual differences can be predicted in paradigm-relevant areas or general areas with high inter-individual variability. Predicting individual task activation from rsFC may be of clinical relevance in cases where patients are unable to carry out a certain task, such as, to inform surgical targets.