Abstract
A defining feature of human cognition is our ability to respond flexibly to whatwe see and hear, changing how we respond depending on our current goals. Infact, we can rapidly associate almost any input stimulus with any arbitrarybehavioural response. This remarkable ability is thought to depend on afrontoparietal "multiple demand" circuit which is engaged by manytypes of cognitive demand and widely referred to as domain general. However, itis not clear how responses to multiple input modalities are structured withinthis system. Domain generality could be achieved by holding information in anabstract form that generalises over input modality, or in a modality-taggedform, which uses similar resources but produces unique codes to represent theinformation in each modality. We used a stimulus-response task, withconceptually identical rules in two sensory modalities (visual and auditory), todistinguish between these possibilities. Multivariate decoding of functionalmagnetic resonance imaging data showed that representations of visual andauditory rules recruited overlapping neural resources but were expressed inmodality-tagged non-generalisable neural codes. Our data suggest that thisfrontoparietal system may draw on the same or similar resources to solvemultiple tasks, but does not create modality-general representations of taskrules, even when those rules are conceptually identical between domains.