Feedback inactivation strongly increases behavioral detection thresholds.

<p>(a) Relevant anatomy diagram showing the main brain areas considered. Behavioral responses were recorded before and after lidocaine, a sodium channel antagonist, was injected bilaterally into the nPs (top left inset), which will inactivate feedback input onto ELL PCells (red cross). (b) Top: Stimulus waveform (blue) and its envelope (red) showing contrast as a function of time. Bottom: EOD frequency from a representative example individual fish before (dark brown) and after (light brown) feedback inactivation. The behavioral detection threshold strongly increased after feedback inactivation (compare the position of the leftmost and rightmost black circles). (c) Whisker-box plots comparing population-averaged detection thresholds computed from behavior before (dark brown) and after (light brown) feedback inactivation to those obtained from PCells after feedback inactivation computed using firing rate (green, left) and phase locking (green, right) and to those computed from firing rate in EAs (blue). Overall, PCell firing rate was a much better predictor of behavior than phase locking. The similarity of detection thresholds obtained from EA and PCell firing rate to that of behavior after feedback inactivation strongly suggests that, for low contrasts (<15%), phase locking in EAs is detected by PCells but is not decoded by downstream brain areas to give rise to behavior. “*” indicates significance at the <i>p</i> = 0.05 level using a Wilcoxon sign rank or Kruskal-Wallis test (behavior). The data can be downloaded at <a href="https://figshare.com/s/93707200732db87bb80f" target="_blank">https://figshare.com/s/93707200732db87bb80f</a>. EA, electrosensory afferent; ELL, electrosensory lateral line lobe; EOD, electric organ discharge; nP, nucleus praeeminentialis; n.s., not significant; PCell, pyramidal cell; Ts, torus semicircularis; VS, vector strength.