The time scales of neural coding in auditory and visual cortices of the primate

How neurons in sensory cortices represent the environment still remains a matter of investigation. With regard to individual neurons, it is of particular interest on what time scales their responses carry information, and hence what are the time scales that define the neural code. Given that the responses of sensory neurons are temporally modulated by the environment, and given that the sensory environment can change on different time scales, however, it might well be that there is no unique time scale of neural coding in a particular sensory area. Analyzing data recorded in primate auditory and visual cortices, we show that this is indeed the case. Specifically, we recorded the responses of neurons in primary auditory cortex of alert macaque monkeys listening passively to complex and natural sounds. In addition, we analyzed the activity of neurons recorded in primary visual cortex of anesthetized macaques during the presentation of naturalistic movies. For each neuron and stimulus set, we quantified the amount of stimulus information carried by firing rates at different temporal scales (bins) and systematically compared the information derived from progressively longer time bins. From this we determined the shortest time scale (i.e. precision of neural code) that still provided all stimulus information. We found that the critical time scale of neural coding depends on the temporal stimulus dynamics. In auditory cortex, we compared responses to sequences of random tones (stimulus auto-correlation scale of below 10ms) to responses to a sequence of natural sounds (auto-correlation of more than 30ms). The temporal precision required to obtain maximal information was shorter for the tone sequence (median 5ms) than for the natural sounds (median 12ms). In visual cortex, we compared responses to natural movie epochs containing fast temporal modulation to responses to epochs of slower modulation (the temporal auto-correlation of the latter being about 2-3 times longer). We found that the time scale of optimal coding was shorter for the faster movie epochs (about 12ms) than for the slower epochs (about 20ms).These results demonstrate that the time scale of neural coding can only be specified in the context of a particular sensory environment