Ensembles code for associative learning in the primate lateral prefrontal cortex

• Published in: Cell reports (Cambridge) Vol. 42; no. 5; p. 112449
• Main Authors: Rouzitalab, Alireza, Boulay, Chadwick B., Park, Jeongwon, Martinez-Trujillo, Julio C., Sachs, Adam J.
• Format: Journal Article
• Language: English
• Published: United States Elsevier Inc 30.05.2023; Elsevier
• Subjects: associative learning; CP: Neuroscience; LPFC; neuronal ensemble; neuronal subspace; primates; LPFC; primates; associative learning; neuronal ensemble; CP: Neuroscience; neuronal subspace
• ISSN: 2211-1247; 2211-1247
• DOI: 10.1016/j.celrep.2023.112449

The lateral prefrontal cortex (LPFC) of primates is thought to play a role in associative learning. However, it remains unclear how LPFC neuronal ensembles dynamically encode and store memories for arbitrary stimulus-response associations. We recorded the activity of neurons in LPFC of two macaques during an associative learning task using multielectrode arrays. During task trials, the color of a symbolic cue indicated the location of one of two possible targets for a saccade. During a trial block, multiple randomly chosen associations were learned by the subjects. A state-space analysis indicated that LPFC neuronal ensembles rapidly learn new stimulus-response associations mirroring the animals’ learning. Multiple associations acquired during training are stored in a neuronal subspace and can be retrieved hours after learning. Finally, knowledge of old associations facilitates learning new, similar associations. These results indicate that neuronal ensembles in the primate LPFC provide a flexible and dynamic substrate for associative learning. [Display omitted] •LPFC neurons are selective for task rules once learned•Rules are encoded in low-dimensional neural subspaces•Neural distance between rules correlates with behavioral similarity•Rules stored in LPFC activity subspaces enable flexible memory associations The study investigated how stimulus-response associations are encoded in lateral prefrontal cortex of the brain in a learning task using working and long-term memory. The recorded neuronal activities change after learning the associations, and in a low-dimensional space, similar associations occupy closer subspaces.