The dimensionality of neural coding for cognitive control is gradually transformed within the lateral prefrontal cortex

Implementing cognitive control relies on neural representations that are inherently high-dimensional and distribute across multiple subregions in the prefrontal cortex (PFC). Traditional approaches tackle prefrontal representations by reducing them into a unidimensional measure (univariate amplitude) or using them to distinguish a limited number of alternatives (pattern classification). By contrast, representational similarity analysis (RSA) enables flexibly formulating various hypotheses about informational contents underlying the neural codes, explicitly comparing hypotheses, and examining the representational alignment between brain regions. Here, we used a multifaceted paradigm wherein the difficulty of cognitive control was manipulated separately for five cognitive tasks. We used RSA to unveil representational contents, measure the representational alignment between regions, and quantify representational generality vs. specificity. We found a graded transition in the lateral PFC: The dorsocaudal PFC was tuned to the information about behavioural effort, preferentially connected with the parietal cortex, and representationally generalisable across domains. The ventrorostral PFC was tuned to the abstract structure of tasks, preferentially connected with the temporal cortex, and representationally specific. The middle PFC (interposed between dorsocaudal and ventrorostral PFC) was tuned to individual task-sets, ranked in the middle in terms of connectivity and generalisability. Furthermore, whether a region was dimensionally rich or thin co-varied with its functional profile: Low dimensionality (only gist) in the dorsocaudal PFC dovetailed with better generality, whereas high dimensionality (gist plus details) in the ventrorostral PFC corresponded with better ability to encode subtleties. Our findings, collectively, demonstrate how cognitive control is decomposed into distinct facets that transition steadily along prefrontal subregions. SignificanceCognitive control is known to be a high-dimensional construct, implemented along the dorsocaudal-ventrorostral subregions of PFC. However, it remains unclear how prefrontal representations could be dissected in a multivariate fashion to reveal (1) what information is encoded in each subregion, (2) whether information systematically transforms across contiguous PFC subregions as a gradient, (3) how this transformation is affected by functional connectivity. Here we shed light on these issues by using RSA to decode informational composition in the PFC while using participant-specific localisers to facilitate individually-tailored precision. Our findings elucidate the functional organisation of PFC by revealing how a trade-off between dimensionality and generalisability unfolds in the PFC and highlight the strength of RSA in deciphering the coding of cognitive control.