<p>Attention to relevant information is necessary to solve a categorization task. Prior studies monitoring peck location indicated that pigeons narrowly focus their attention on the relevant features of a category exemplar, whereas computational modeling suggested that pigeons’ attention can be widely distributed. To help resolve these apparently contradictory results, we deployed both strategies – peck tracking and computational modeling – in the same experiment. Pigeons’ peck-tracking scores neatly aligned with their category choices and with the results of the computational modeling. In addition, pigeons displayed distinctive learning and attentional patterns during training and testing. Half of our pigeons showed preferential, even exclusive, use of the most relevant category features, but the other half were inclined to use several of the available features. These results highlight the importance of combining an individual differences approach along with the pursuit of general cognitive principles.</p>