Chronometry on Spike-LFP Responses Reveals the Functional Neural Circuitry of Early Auditory Cortex Underlying Sound Processing and Discrimination

Experimental design, Go/No-go task. A, Monkey waits during a rest period with hands on a lever and attends to the stimuli (PT, Coo, or pink noise; presentation time, 300 ms). To obtain a water reward, the monkey must release the lever when pink noise is presented. The next trial starts 600 ms after the previous stimulus onset. B, The spectrogram (time, frequency, and power) of PT and Coo stimuli. The frequency of PTs matches the fundamental frequency of the Coo.

One representative session from each monkey, where simultaneous recordings from two areas spikes and LFP could be obtained. First row indicates spike rasters (cyan and magenta dots) and firing rates (blue and red) computed using Gaussian smoothing (10-ms window) for PT (cyan/blue) and Coo (magenta/red) stimuli. The second row depicts the trial-by-trial LFP waveforms using the same color code as for spikes. The averaged LFP responses are plotted in blue and red. The spike-LFP responses in two auditory cortical areas A1 and LB were recorded during the same session in each monkey.

A, Estimation of trial-averaged onset and selection latency using ms-by-ms t test on raw time series. For spikes, the binary time series was transformed to a SDF by convolving single trial spike trains with assymetric exponential functions having different growth and decay time constants, 1 and 20 ms, respectively, following Thompson et al. (1996). A ms-by-ms t test was performed on the distribution SDFs in a given session from different conditions (see text for details). For onset, prestimulus rest period was used to compute the SDF. Analyses were performed across both monkeys and for each monkey individually. Error bars were plotted at 95% significance levels. B, Estimation of trial-averaged onset and selection latency from AccLLR distributions using ms-by-ms t test; p = 0.01 was chosen as threshold for significance.

A, Estimation of trial-by-trial onset and selection latency using AccLLR. We follow the same pattern of presentation as in Figure 4 and report the group-level analysis and individual monkey analysis. Error bars were computed at 95% significance levels by pooling all trials and sessions in a monkey. See text for details of Methods. B, Estimation of trial-by-trial onset and selection latency using AccLLR on pooling all trial information within a stimulus category to a single trial in each session and setting accuracy to 100%. Error bars were computed at 95% significance levels.