Cortical Discrimination of Complex Natural Stimuli: Can Single Neurons Match Behavior?

Abstract

A central finding in many cortical areas is that single neurons can match behavioral performance in the discrimination of sensory stimuli. However, whether this is true for natural behaviors involving complex natural stimuli remains unknown. Here we use the model system of songbirds to address this problem. Specifically, we investigate whether neurons in field L, the homolog of primary auditory cortex, can match behavioral performance in the discrimination of conspecific songs. We use a classification framework based on the (dis)similarity between single spike trains to quantify neural discrimination. We use this framework to investigate the discriminability of single spike trains in field L in response to conspecific songs, testing different candidate neural codes underlying discrimination. We find that performance based on spike timing is significantly higher than performance based on spike rate and interspike intervals. We then assess the impact of temporal correlations in spike trains on discrimination. In contrast to widely discussed effects of correlations in limiting the accuracy of a population code, temporal correlations appear to improve the performance of single neurons in the majority of cases. Finally, we compare neural performance with behavioral performance. We find a diverse range of performance levels in field L, with neural performance matching behavioral accuracy only for the best neurons using a spike-timing-based code.

What can spike train distances tell us about the neural code?

“…This measure has also occasionally be used in the discrimination analysis (Wang et al, 2007), although we will now show it does not not reflect different degrees of sensitivity to the temporal structure, going from a rate code to a coincidence code. At first the spike trains are also convolved with a convolution kernel resulting in the vectors s x,y when the convolution is applied for a discrete number of sampling times.…”

“…For D R and D S we take as the time scale parameter the standard deviation of the convolution kernel. An alternative criteria could be the decay time of the convolution kernel (Wang et al, 2007). For D B , the time scale parameter is the width of the bins.…”

“…Machens et al (2003) indicated that the use of different supervised and unsupervised classifiers did not affect the results. However, other studies using the percentage of correct classifications to evaluate the discrimination do not report any comparison with other classifiers (Grewe et al, 2003;Narayan et al, 2006;Wohlgemuth and Ronacher, 2007;Wang et al, 2007;Billimoria et al, 2008). Although avoiding the selection of z, the optimality of the classifier based on the random selection of templates may differ for different time scales, cells, or conditions.…”

“…For example, the Victor distance (see Victor, 2005, for a review) has been applied to study the precision in the visual system (e. g. Victor and Purpura, 1998;Mechler et al, 1998), the auditory system (e. g. Machens et al, 2001) or the olfactory system (e. g. Macleod et al, 1998). The van Rossum distance has recently been used in a series of studies on the time scale of discrimination of natural sounds in songbirds (e. g. Narayan et al, 2006;Wang et al, 2007). The correlation-based reliability measure of Schreiber et al (2003) has also been used to analyze auditory responses (Wang et al, 2007).…”

“…The van Rossum distance has recently been used in a series of studies on the time scale of discrimination of natural sounds in songbirds (e. g. Narayan et al, 2006;Wang et al, 2007). The correlation-based reliability measure of Schreiber et al (2003) has also been used to analyze auditory responses (Wang et al, 2007). A common conclusion of many of these studies is that the optimal time scale found indicates that the temporal structure of the responses and not only the total spike counts are relevant for the neural encoding of the stimuli.…”