Fig. 6: UBC population generates a continuum of multi-scale temporal representations.
a All normalized responses displayed as a sorted heat map in linear time (n = 84). b The same sorted heat map displayed in logarithmic time (t = 0 at stimulus offset). c Peak-time vs. cell #, linear fit on the log10 data (black). d Half-width of response vs. cell #, linear fit on the log10 data (black). e Half-width to peak-time ratio vs. cell #, linear fit (black). f Width of Gaussian fits of responses on a logarithmic time scale vs. cell #, linear fit (black). g Pause time vs. cell #, linear fit on the log10 transformed data (black). h Amplitude of response vs. cell #, rectified linear fit on the log10 data (black). i Sample instantaneous firing rate in three UBCs (gray dots) and log-Gaussian fits (black). j Same as in i but in a log-log plot. k Amplitude of population response over time for different numbers of stimuli (turquoise dots, 10x, 20x, and 40×100 Hz) and shifted power-law fits (black lines). l Same as in k in a log-log plot to better illustrate the power-law decay in the tail, for 10, 20, and 40 stimuli. The critical exponent α = −1.47 with 95% CI of (−1.51, −1.44). m Representative response of a fast UBC to a 10 × 0.5 Hz sequence of 10 × 100 Hz MF bursts. n Response of an intermediate UBC to the same sequence of MF bursts o. Response of a slow UBC to the same sequence of MF bursts.
