Learning theories hypothesize specific circuits encode essential information for performance. For simple tasks in invertebrates and mammals, the essential circuits are known, but for cognitive functions, the essential circuits remain unidentified. Here, we show that some essential information for performing a choice task is encoded in a specific circuit in a neocortical area. Rat postrhinal (POR) cortex is required for visual shape discriminations, protein kinase C (PKC) pathways mediate changes in neuronal physiology that support learning, and specific PKC genes are required for multiple learning tasks. We used direct gene transfer of a constitutively active PKC to prime a specific POR cortex circuit for learning visual shape discriminations. In the experiment, rats learned a discrimination, received gene transfer, learned new discriminations, received a small lesion that ablated ≈21% of POR cortex surrounding the gene transfer site, and were tested for performance for discriminations learned either before or after gene transfer. Lesions of the genetically targeted circuit selectively interfered with performance for discriminations learned after gene transfer. Activitydependent gene imaging confirmed increased activity in the genetically targeted circuit during learning and showed the essential information was sparse-coded in ≈500 neurons in the lesioned area. Wild-type rats contained circuits with similar increases in activity during learning, but these circuits were located at unpredictable, different positions in POR cortex. These results establish that some essential information for performing specific visual discriminations can be encoded in a small, identified, neocortical circuit and provide a foundation for characterizing the circuit and essential information.learning | vision | protein kinase C | neocortex | herpes simplex virus vector M any current theories postulate that the essential information for specific cognitive learning problems is widely dispersed in the forebrain (1-3). Supporting such theories, lesioning studies show ablation of an entire neocortical area is required to obtain a performance deficit, and activity-dependent gene imaging and fMRI studies show activity in multiple neocortical areas during learning. Nonetheless, synaptic plasticity and neural network theories hypothesize that the essential information for cognitive learning is encoded in localized circuits, by modifying synaptic strengths (4, 5). In invertebrates, specific essential information has been mapped to specific circuits (4), and, in mammals, essential information for simple conditioning has been mapped to specific brain areas (6). The existence of specific neocortical circuits that encode essential information is suggested by the partitioning of neocortex into areas and columns (7), localized stimulation-evoked experiential learning (8), microstimulationmediated effects on cognition (9), specific fMRI studies (10), and specific human cognitive deficits (11). However, it has proven difficult to map some essential info...