Fig. 8: Working model of RSC processing and representational stability.

a We hypothesize that current visual sensory features are matched to stored reference features (green arrow), which then triggers a motor output (magenta arrow) from a collection of possible choices during the association process (black arrow). Our model also postulates that sensory features are more stably represented in the RSC (solid green lines), which is supported by strong connectivity with primary and secondary visual areas^24,27,59, and that motor choice is less stable (dashed magenta lines), possibly favoring flexibility during the decision process. In addition, reinforcement signals (cyan arrows) help maintain stable responses of RSC neurons tuned to task features, particularly those related to trial outcome and environmental context. b Previous studies have reported higher levels of stability in sensory and motor cortices^5,6,9,60. Our results in retrosplenial cortex suggest that stability in association areas, which typically comprise neurons multiplexing different streams of information, follows a different function. First, sensory inputs are represented with lower stability than in the region they originate from. Motor coding, although stable within a session, changes considerably between days, which may lead to more flexible remapping to new contingencies. We expect that motor representations will become very stable again in more peripheral motor cortices.