Figure 3.

Snap-shake behavior is induced by CRF and epitomizes indecision regarding escape. (A) Mean (± SEM) number of snap-shake behaviors expressed per minute without treatment (N = 8), following icv injections of aCSF (= 0 ng CRF, N = 8), or corticotropin-releasing factor (CRF; N = 8 each for 500 or 2,000 ng). (B) Percentage of social interactions that include snap-shake behavior in Escape (light gray bars, N = 10) and Stay (dark gray bars, N = 10) phenotypes. (C) Mean (± SEM) number of snap-shake behaviors per second of interaction immediately following retreat/explore- (Esc; bars without hatching), submission- (Sub; bars with left hatching), and aggression-related (Agg; bars with right hatching [hatching not visible]) behaviors in Escape (light gray bars) and Stay (dark gray bars) phenotypes. Escape behavior precedes snap-shake significantly (^*one-way ANOVA) more often than other behaviors in both phenotypes (⁺T-test, Escape significantly greater than Stay; N = 10 each). (D) Positive linear regression (r² = 0.48, P < 0.019, N = 10) between escape speed, normalized for daily changes in the escape learning curve (see Figure 1C), and number of snap-shakes performed. The dotted line represents the demarcation between slower (above) and faster (below) than average escape, and results from the normalization formula: individual latency to escape in seconds/group mean escape latency (s) for that day.