Spike timing and information transmission in biological neural networks

From 2007 Scientific Report
Jump to: navigation, search

Chloé Huetz, Philippe Tarroux in collaboration with Jean-Marc Edeline (IBAIC-NAMC Orsay


General background

A major goal in computational neuroscience is to understand the neural code used to represent perceptual information and elaborate appropriate motor responses. Most of the researches about the relationship between neural activity and cognitives functions still focus on the average firing rate of one or more neurons as the only possible measure of neural activity. However, a growing number of studies show that spike timing could constitutes a rapid and efficient neural code. Our work in collaboration with the team of Jean-Marc Edeline (NAMC, UMR 8620, Orsay), is an attempt to reveal some facets of the neural code in the auditory system. We seek to develop and use several theoretical and computational tools to understand the neural basis of auditory perception in the most possible natural conditions. These following questions are central to our work:

  • (i) - how is information about dynamic stimuli encoded by single neurons ? More particularly, what are the specific features of neuronal spike trains that are correlated with the stimulus ? With wich temporal precision is this information encoded ?
  • (ii) - how can we decode the neural code ? In other words, what are the specific characteristics of natural stimuli that evoke neural responses ?

We study these issues using computational and mathematical tools such as Information Theory, Artificial Neural Networks to understand how information is encoded in spike trains and Signal Processing, Reverse Correlation methods to explore the stimulus-response relationships. We have applied our techniques to two sets of data: first, spike trains from HVC neurons, a sensorymotor nucleus involved in both the production and perception of the songs in canary; second, spikes trains of the thalamo-cortical system in anesthetized and awake guinea-pigs.

Spike-timing allows discrimination between the bird own song and its time-reverse version.

In the songbird nuclei, the responses of HVC neurons during playback of the bird’s own song (BOS) is probably one of the most striking examples of the neural selectivity for natural stimuli. We examined to what extent spike timing carries information about natural and time-reversed version of the BOS. From a population of 107 HVC neurons a standard indicator of stimulus preference based on spike-count (the d’ index) indicates that a limited proportion of cells (about 20%) can be classified as selective for the BOS. In contrast, quantifying the information conveyed by spike trains with the metric-space of Victor and Purpura (1996, 1997) indicates that a large majority of cells (62%) display significant amounts of transmitted information, among which 77% are “temporal cells”. “Temporal cells” correspond to cells transmitting significant amounts of information when spike-timing is considered whereas no information, or lower amounts of transmitted information, are obtained when only spike-count is considered. Cells classified as selective in terms of spike-counts (d’ index) had greater amounts of transmitted information, but cells classified as non-selective (d’<0.5) can also transmit significant amounts of information (Huetz et al., 2006). Thus, information theory methods demonstrate that a much larger proportion of neurons than expected based on spike-count only participate in the discrimination between the bird own song and its time-reverse version.

Spike-timing also allows discrimination between natural and artificial vocatlizations in the thalamocortical system

Using similar techniques, we analysed spike trains of neurons in the thalamocortical auditory system at presentation of four different guinea pig vocalisations and their time-reverse versions. Most of the neurons (75-95%) responded similarly to the vocalizations and their reverse versions but for a large majority of them, analyzing the information contains in the spike trains with a temporal precision of a few tens of ms allows to discriminate between the stimuli. In addition, it appears that in awake animals the proportion of cells which carry information based on the temporal organisation of the neuronal discharges is higher than in the anesthetized animals (Huetz et al., in preparation). We also assessed whether the between vocalizations discrimination rely on the firing rate or on the temporal organisation during the short duration responses observed initially at the begining of any acoustic stimulus. Using the adaptive direct method (Chechik et al., 2006), we have observed that the first spikes latencies carried more information than the total spike count, suggesting that a code for vocalization can be achieved very quickly in the auditory cortex (Huetz et al., in preparation).

How can we decode the neural code ?

We investigated the stimulus-response relationships in the auditory thalamus and cortex of guinea-pigs. Using reverse-correlation methods such as Spectro-Temporal Receptive Fields (STRF), we analyzed single units responses to species-specific vocalizations and to dynamic noise (or Dynamic Moving Ripples, DMR). More precisely we examined to what extent the STRF of auditory cortex neurons obtained from responses evoked by artificial dynamics stimuli and by species-specific vocalisations significantly differ. Preliminary results (Laudanski et al., 2006) indicated that STRFs exhibiting significant areas (excitatory or inhibitory) could be obtained for more than half of the cells both for vocalizations and DMR. More importantly, half of these cells showed different STRFs when derived from the responses to vocalizations and to DMRs (significant differences expressed in terms of BF, bandwidth and general shape of the excitatory area). These results suggest that, in addition to provide an interesting dynamic stimulus/responses representation, STRF reveals the existence of complex stimulus-dependent neuronal signals and can therefore bring insights to the understanding of the neural code.


Huetz C., Del Negro C., Lehongre K., Tarroux P. & Edeline J-M (2004) The selectivity of canary HVc neurons for the Bird’s Own song: Rate coding, temporal coding or both ? J. Physiology (Paris), 98(4-6):395-406.

Huetz C., Del Negro C., Lebas N., Tarroux P. & Edeline J-M (2006) Contribution of spike timing to the information transmitted by HVC neurons European Journal of Neuroscience, 24, 1091-1108.

Huetz C. & Edeline J-M. (2006) From the receptive field dynamics to the rate of transmitted information: Some facets of the thalamo-cortical auditory system, Neuroembryology and Aging, 3, 230-23