SPARSENESS OF THE NEURONAL REPRESENTATION OF STIMULI IN THE PRIMATE TEMPORAL VISUAL-CORTEX

1. To analyze the selectivity and the sparseness of firing to visual s timuli of single neurons in the primate temporal cortical visual area, neuronal responses were measured to a set of 68 visual stimuli in mac aques performing a visual fixation task. The population of neurons ana lyzed had responses that occurred primarily to faces. The stimuli incl uded 23 faces. and 45 nonface images of real-world scenes, so that the function of this brain legion could be analyzed when it was processin g natural scenes. 2. The neurons were selected to meet the previously used criteria of face selectivity by responding more than twice as muc h to the optimal face as to the optimal nonface stimulus in the set. A pplication of information theoretic analyses to ?he responses of these neurons confirmed that their responses contained much more informatio n about which of 20 face stimuli had been seen (on average 0.4 bits) t han about which (of 20) nonface stimuli had been seen (on average 0.07 bits). 3. The sparseness of the representation of a scene or object p rovided by each of these neurons (which can be thought of as the propo rtion of stimuli to which the neuron responds, and which was fundament al to understanding the network operation of the system) can be define d as [GRAPHICS] where r(1) is the firing rate to the ith stimulus in t he set of n stimuli. The sparseness has a maximal value of 1.0. It was found that the sparseness of the representation of the 68 stimuli by each neuron had an average across all neurons of 0.65. This indicates a rather distributed representation. 4. If the spontaneous firing rate was subtracted from the firing ate of the neuron io each stimulus, so that the changes of firing rate, i.e., the responses of the neurons, were used in the sparseness calculation, then the ''response sparsenes s' had a lower value. with a mean of 0.33 for the population of neuron s, or 0.60 if calculated over the set of Faces. 5. Multidimensional se aling to produce a stimulus space represented by this population of ne urons showed that the different faces were well separated in the space created. whereas the different nonface stimuli were grouped together in the space. 6. The information analyses and multidimensional scaling provided evidence that what was made explicit in the responses of the se neurons was information about which face had been seen. Information about which nonface stimulus had been seen was not made explicit in t hese neuronal responses. These procedures provide an objective and qua ntitative way to show what is ''represented'' by a particular populati on of neurons. 7. The response sparseness value obtained shows further that this population provides a distributed representation of informa tion about which face is being seen. This type of distributed represen tation is very efficient for fine discriminations between een the memb ers of a stimulus set: in this case, faces.