Fig. 2.

Reverse-correlation analysis of thermotactic behavior. (A) Behavioral event-triggered averaging of themotactic response. (A, Top and Middle) Larvae crawling on a temperature gradient that is time-varying but spatially uniform (Top) are subjected to random white noise thermal flicker between 16 and 18 °C and up to 1 °C/s changes (Middle). Run-to-turn transition events are indicated by red dots, with a surrounding time window from a single transition (large red dot) shaded in gray for illustration. (A, Bottom) The temperature stimulus histories for all such transitions are then averaged together. n = 60 wild-type (Canton-S) larvae. (B) Reverse-correlation response filter computed by the event-triggered averaging in A. (B, Upper) Averaged temperature (T) stimulus history. (B, Lower) Averaged change in temperature (dT/dt) stimulus history. A sharp decrease in temperature immediately precedes the run-to-turn transition (defined as time t = 0). The red dot indicates the turn initiation, and the blue dot indicates the median time of the preceding turn-to-run transition. (C) Reverse-correlation filter for the turn-to-run transition, where t = 0 is defined to be the time corresponding to the maximum body bend angle during an accepted head sweep. (C, Upper) Averaged temperature (T) stimulus history. (C, Lower) Average change in temperature (dT/dt) stimulus history. A sharp increase in temperature precedes the turn-to-run transition. The blue dot indicates the accepted head sweep time, and the red dot indicates the median preceding run-to-turn transition time. (D) Predicting the run-to-turn transition probability. (D, Upper) Larvae were subjected to a temperature waveform from 17.5 °C to 16.5 °C starting at t = 0 s (n = 90 animals). The reverse-correlation filter in B convolved with the temperature waveform yields a predicted probability of turning over time (green trace), which compares with experimental measurement (black markers).