Encoding of categories by noncategory-specific neurons in the inferior temporal cortex

In order to understand how the brain codes natural categories, e.g., trees and fish, recordings were made in the anterior part of the macaque inferior temporal (IT) cortex while the animal was performing a tree/nontree catego rization task. Most single cells responded to exemplars of more than one ca tegory while other neurons responded only to a restricted set of exemplars of a given category. Since it is still not known which type of cells contri bute and what is the nature of the code used for categorization in IT, we h ave performed an analysis on single-cell data. A Kohonen self-organizing ma p (SOM), which uses an unsupervised (competitive) learning algorithm, was u sed to study the single cell responses to tree and nontree images. Results from the Kohonen SOM indicated that the collected neuronal data consisting of spike counts was sufficient to account for a good level of categorizatio n success (approximately 83%) when categorizing a group of 200 trees and no ntrees. Contrary to intuition, the results of the investigation suggest tha t the population of category-specific neurons (neurons that respond only to trees or only to nontrees) was unimportant to the categorization. Instead, a large majority of the neurons that were most important to the categoriza tion was found to belong to a class of more broadly tuned cells, namely, ce lls that responded to both categories but that favored one category over th e other by seven or more images. A simple algebraic operation (without the Kohonen SOM) between the above-mentioned noncategory-specific neurons confi rmed the contribution of these neurons to categorization. Thus, the modelin g results suggest (1) that broadly tuned neurons are critical for categoriz ation, and (2) that only one additional layer of processing is required to extract the categories from a population of IT neurons.