Infection-related images reduce cortical edge metastability: a specification-curve analysis of a block-design fMRI study

Abstract

Cues of infection evoke disgust, attentional capture, withdrawal and shifts in social orientation, a set of responses attributed to the behavioral immune system. Responses this varied have to be coordinated, as the effects of inflammation are, so they should be visible in whole-cortex coupling. We predicted that infection-related images hold cortical dynamics in a narrower repertoire, reducing cortical edge metastability, the temporal variability of mean cofluctuation across parcel pairs. In two fMRI runs, 44 adults viewed alternating 18-s blocks of infection-related and neutral images; the primary test was specified before we looked at the data. That test gave Cohen's d = -0.31 (permutation p = .042), and 263 of 480 specifications were significant, all negative (joint p < .001). With global signal regression, the synchrony-based reduction survived removal of the stimulus-locked response, detrending, and motion, white-matter and cerebrospinal-fluid regression (d = -0.47), recurred in the second run (d = -0.44), and was smaller on strictly alternating blocks (d = -0.36). Without global signal regression, the pre-specified effect did not survive white-matter and cerebrospinal-fluid regression. Individual-difference predictions were not supported, and participants' contrasts did not replicate across runs. Infection-related images reduce the variability of residual global cofluctuation at the group level; whether this is specific to pathogen cues remains open.