The gradient model of brain organization in decisions involving 'empathy for pain'

Influential models of cortical organization propose a close relationship between heteromodal association areas and highly connected hubs in the default mode network. The gradient model of cortical organization proposes a close relationship between these areas and highly connected hubs in the default mode network, a set of cortical areas deactivated by demanding tasks. Here, we used a decision-making task and representational similarity analysis with classic empathy for pain stimuli to probe the relationship between high-level representations of imminent pain in others and these areas. High-level representations were co-localized with task deactivations or the transitions from activations to deactivations. These loci belonged to two groups: those that loaded on the high end of the principal cortical gradient and were associated by meta-analytic decoding with the default mode network, and those that appeared to accompany functional repurposing of somatosensory cortex in the presence of visual stimuli. In contrast to the nonspecific meta-analytic decoding of these loci, low-level representations, such as those of body parts involved in pain or of pain itself, were decoded with matching topics terms. These findings suggest that task deactivations may set out cortical areas that host high-level representations. We anticipate that an increased understanding of the cortical correlates of high-level representations may improve neurobiological models of social interactions and psychopathology.