Cognitive processes are disentangled at cortex-wide scales
Costa, R. M.; Luo, J. K.; Salvino, P. S.; Ackert-Smith, L. A.; Ibarra, S. M.; Pinto, L.
Show abstract
Most decisions involve multiple cognitive processes. Recent findings suggest that these processes are distributed across the cortex1-21, but that single regions implement them via orthogonal population-activity patterns14,21-33. How are these local geometries combined across the cortex? Here, we designed a virtual-navigation task for mice that dissociates the accumulation and short-term memory of sensory evidence, and choice. Combining dimensionality reduction and decoding models with cortex-wide widefield Ca2+ imaging, we observed distributed but near-orthogonal coding subspaces for these different cognitive processes, and that this geometry breaks down during erroneous choices. Further, only the memory subspace corresponded to a spontaneous activity-timescale hierarchy, suggesting that it co-opts intrinsic circuit properties. Thus, we reconcile previous findings by showing that cortex-wide dynamics supporting distinct cognitive processes are disentangled.
Matching journals
The top 1 journal accounts for 50% of the predicted probability mass.
Similar papers in this journal
Similar papers in this journal
- A cortical circuit mechanism for coding and updating task structural knowledge in inference-based decision-making 98%
- Ventral frontostriatal circuitry mediates the computation of reinforcement from symbolic gains and losses 97%
- Universal statistics of hippocampal place fields across species and dimensionalities 97%
Similar papers in this journal
- Slowly evolving dopaminergic activity modulates the moment-to-moment probability of movement initiation. 98%
- Layer 6 ensembles can selectively regulate the behavioral impact and layer-specific representation of sensory deviants 97%
- linc-mipep and linc-wrb encode micropeptides that regulate chromatin accessibility in vertebrate-specific neural cells 97%
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.