Can mice learn rules for categorization? Is prefrontal cortex involved?
Our new study, open access in @nature, characterizes category representations in mouse PFC using chronic 2p imaging. (1/5)
@mpi_neuro @xpieter @t_bonhoeffer @MHuebener https://doi.org/10.1038/s41586-021-03452-z

We trained mice in a head-fixed visual category learning paradigm, where they learned a spatial frequency rule or an orientation rule to categorize stimuli. We found that mice were able to generalize either rule to novel stimuli and even learned a rule-switch. (2/5)

Chronic 2p imaging throughout category learning revealed a set of neurons in mouse mPFC showing category-selective responses to the visual stimuli after learning. This category selectivity was only present for the rule that the mice used for categorization. (3/6)

By following category-selective neurons through learning and investigating other task-relevant parameters, we found that mPFC neurons showed varying degrees of mixed selectivity for category, choice and reward and that category selectivity emerged gradually with learning. (4/6)

To disentangle category-, choice- and reward influence on the neurons, we trained mice in a Go/NoGo task and tested them in a left/right task with changed motor and reward mapping.
Despite the change, a set of neurons remained category-selective, generalizing across tasks. (5/6)

In conclusion, mice can learn a rule to categorize visual stimuli, and while they learn, a representation of the categories gradually emerges in their medial prefrontal cortex. (6/6)