Item

Abstract

As most cortical neurons are broadly tuned to various stimulus parameters, it is inevitable that individual neurons participate in the representation of more than one visual object. Vice versa, accurate representations of individual objects for example in short-term memory that can support reliable decisions require the participation of large neuronal populations. To provide evidence in favor of population codes, we have recently analyzed simultaneously recorded multi- and single-unit signals derived from arrays of single-ended microelectrodes (Waizel et al., SfN 2007). Multi-contact electrodes like tetrodes (tts) which have a real 3D-structure provide signals that allow for estimating the position of the recorded neurons by triangulation. Here we set out to study whether recording from 3D-tts would improve the quality of sorting and hence allow for the extraction of more information about the stimulus. Based on single trial firing rate values, we calculated one-way ANOVAs at 1 significance thresholds and performed subsequent posthoc comparison (Scheffé) in order to detect stimulus selectivity and determine stimulus specificity for the activity at each single site, respectively. In order to investigate the coding of distributed neuronal ensembles, we computed binary activity patterns for all active electrodes in the array and determined their stimulus selectivity and specificity. Compared to what we found previously in single microelectrode recordings, the number of object selective or even specific recording sites increased up to 3 times which provides highly specific sites in 3 out of 4 sessions (3000 trials, 13.5 million spikes). Given that our monkeys always performed the memory task with a set of twenty visual stimuli, we found highly specific sites coding for only one object which revealed up to 18 of 19 possible pairwise comparisons. According to the proposal that single neurons participate in more than one specific object representation we also found bi- or even tri-object-dependant sites (average 27 significant pairwise comparisons per session) and never non-systematic object specificity. As clusters of triangulation-reconstructed spikes tend to have inter-cluster regions with smooth transitions which could potentially reflect synchronous spikes, we wanted to know how much information could be carried by these signals. After removing spikes between clusters, we found object specificity highly decreased (in one session only 6 out of previous 38 significant pairwise comparisons remained). These results suggest that the use of tetrodes with a real 3D-structure provides more information about neuronal object representations.