INFORMATION ENCODING AND THE RESPONSES OF SINGLE NEURONS IN THE PRIMATE TEMPORAL VISUAL-CORTEX

1. The possibility of temporal encoding in the spike trains of single neurons recorded in the temporal lobe visual cortical areas of rhesus macaques was analyzed with the use of principal component and informat ion theory analyses of smoothed spike trains. The neurons analyzed had responses selective for faces. 2. Provided that a correction was appl ied to earlier methods of principal component analysis used for neuron al spike trains, it was shown that the first principal component provi des by a great extent the most information, with the second and third adding only small proportions (on average 18.8 and 8.4%, respectively) . 3. It was shown that the magnitude of the second and higher principa l components is even smaller if the spike train analysis is started af ter the onset of the neuronal response, instead of before the neuronal response has started. This suggests that variations in response laten cy are at least a part of what is reflected by the second and higher p rincipal components. 4. The first principal component was correlated w ith the mean firing rate of the neurons. The second and higher princip al components reflected at least partly the onset properties of the ne uronal responses, such as response latency differences between the sti muli. 5. A considerable proportion of the information available from p rincipal components 1-3 is available in the firing rate of the neuron. 6. Periods of the firing rate of as little as 50 or even 20 ms are su fficient to give a reasonable estimate of the firing rate of the neuro n. 7. Information theory analysis showed that in short epochs (e.g., 5 0 ms) the information available from the firing rate can be as high, o n average, as 84.4% of that available from the firing rate calculated over 400 ms, and 52.0% of that available from principal components 1-3 in the 400-ms period. It was also found that 44.0% of the information calculated from the first three principal components is available in the firing rates calculated over epochs as short as 20 ms. 8. More inf ormation was available near the start of the neuronal response, and th e information available from short epochs became less later in the neu ronal response. 9. Taken together, these analyses provide evidence tha t a short period of firing taken close to the start of the neuronal re sponse provides a reasonable proportion of the total information that would be available if a long period of neuronal firing (e.g., 400 ms) were utilized to extract it, even if temporal encoding were used. The implications of these and related findings are that, at least for rapi d object recognition, each cortical stage provides information to the next in a short period of 20-50 ms, does not utilize temporal encoding , and completes sufficient computation to provide an output to the nex t stage in this same 20- to 50-ms period.