Fig. 1

Figure 1. Zebrafish switch from randomly directed to biased escapes upon visual detection of barriers

(A) Left: overview of experimental setup. Right: typical large-angle short-latency escape turn after tap delivery.

(B) Escape trajectories after tap induction, color coded to reflect time, are shown for 3 conditions (no barriers in visible light, barrier in visible light, barrier in the dark). Trajectories begin 20 ms after tap command and trajectories with barrier approaches to the right are reflected across the y axis. 2D histograms depict probability of the escape trajectory passing through a spatial bin for each condition. White asterisk indicates starting position of the fish. Barriers are 12 mm wide and 6 mm high, and fish are tapped when they pass within 2 mm of the barrier (N = 98, n = 967; N = 22, n = 128; N = 21, n = 102).

(C) Distribution of preference index for left versus right escapes in no barrier conditions (N = 98).

(D) Pairwise comparison of barrier avoidance index (p = 4.77e−5) and collision rate (p = 2.3e−4) in fish tested in both the light and dark (same fish as B middle and right panels). Error bars mark 95% CI of the median.

(E) Avoidance index per fish per window of approach angle to the barrier in the light (yellow) and in the dark (light gray). In the light, escapes are not biased in the frontal 25° of visual space but are biased (95% CI above 0) for all other windows. Solid lines represent mean and shaded areas 95% CI. Scale bars in trajectory plots = 2.5 mm; scale bars in 2D histograms = 1.25 mm. Area in dotted box in left (B) panel reflects area plotted in 2D histograms. See Figure S1 for latency and turn amplitude statistics and Figure S2, which shows that naturally biased fish in control barrier-free conditions invert their preference when a barrier is on the side of their bias.

See also Figures S1 and S2.