Latency variability of responses to visual stimuli in cells of the cat's lateral geniculate nucleus.

We constructed average histograms from responses evoked by flashing stimuli and noted previously described variations in the shape of the response profile, particularly with respect to sharpness of the peak. To express this variable, we measured the half-rise latency, which is the latency from stimulus onset required to reach half the maximum response. A short half-rise latency, which is characteristic of nonlagged cells, is associated with a brisk response and sharp peak; a long half-rise latency, characteristic of lagged cells, is associated with a sluggish response and broad peak. Nonlagged cells were readily seen; we attempted to identify cells with long latencies as lagged, but we were unable to do so unambiguously due to failure to observe lagged properties other than latency. We thus refer to these latter cells as having "lagged-like" responses to indicate that we are not certain whether these are indeed lagged cells. In addition to the histograms, we analyzed the individual response trials that were summed to create each histogram, and we used spike density analysis to estimate the initial response latency to the flashing spot for each trial. We found that lagged-like responses were associated with more variability in initial response latency than were nonlagged responses. We then employed an alignment procedure to eliminate latency variation from individual trials; that is, responses during individual trials were shifted in time as needed so that each had a latency equal to the average latency of all trials. We used these "aligned" trials to create a second, "aligned" response histogram for each cell. The alignment procedure had little effect on nonlagged responses, because these were already well aligned due to consistent response latencies amongst trials. For lagged-like responses, however, the alignment made a dramatic difference. The aligned histograms looked very much like those for nonlagged responses: the responses appeared brisk, with a sharply rising peak that was fairly high in amplitude. We thus conclude that the slow build up to a relatively low peak of firing of the lagged-like response histogram is not an accurate reflection of responses on single trials. Instead, the sluggishness of lagged-like responses inferred from average response histograms results from temporal smearing due to latency variability amongst trials. We thus conclude that there is relatively little difference in briskness between nonlagged and lagged-like responses to single stimuli.