Fig. S3

Examples of single, non-averaged trials, related to Figure 1

(A) Trajectories of three example fish (see Figure S4A for average trajectories of those and other fish). Trials are consecutive in groups of three, stacked vertically. For visualization, the start of the first of each triplet is shifted to zero. Little fictive “probe swims” are visible and magnified in the dashed circles; these probe swims are often present right after the start of a backward pre-displacement and at the start of swim periods and may inform the fish that it’s in an open-loop (pre-displacement) or closed loop (swim) period. We hypothesize that the probe swim in the pre-displacement may reduce subsequent swim attempts in the pre-displacement and delay periods, whereas in the swim period it may enhance further closed-loop swimming. In these fish, slow drift is present that can be minimal (left fish), backward (middle fish), or forward (right fish), i.e., positional homeostasis is not perfect. Figure S4A quantifies this drift and Figures S4B–S4F proposes an explanation in a control theory framework. The lower two rows illustrate “unstable trials,” showing that swimming behavior is variable and likely driven not only by visual flow but also other factors (e.g., fluctuating “internal states,” “noise,” etc.).

(B) Single-trial SLO-MO activity traces during 6.5 min of data for two example neurons with different tuning properties in one fish.