Fig. S6

Further characterization of SLO-MO activity, dependence on complex trajectories and swimming, related to Figure 3

(A) Trajectories through dimensionally reduced network space for seven individual fish, related to Figure 3. Dimensionality reduction done with principal-component analysis (PCA).

(B) Distance in PC space (first 3 components) between backward pre-displacement and zero pre-displacement, and forward pre-displacement and zero pre-displacement activity, for 9 fish. Since distances are zero or greater, noise causes distance to always be positive even before the pre-displacement at ∼2 s. (Shaded regions: SEM.)

(C and D) SLO-MO cells showed a transient decrease in activity during swimming, after which they returned to activity levels encoding past displacements.

(E and F) Ranked correlation of cell activity to fish position for example fish of Figures 3B and 3C separated by trials with and without swimming in pre-displacement and delay periods.

(G) Neurons (in an example fish) with increasing activity following three consecutive backward or forward pre-displacements, showing integration. (Shaded regions: SEM in all panels.)

(H) Neurons (in an example fish) showing integration over backward pre-displacements. The green trials are offset by one additional displacement 5 s before the triplet displacement; this is reflected in a persistent increase in neuronal activity.

(I) Integration across triplet displacements by neurons that respond positively to forward displacements.

(J) Additional example of integration by SLO-MO neurons during forward pre-displacements of varying durations followed by delay periods.

(K) Individual neuron traces showing integration across triplet pre-displacements, fast imaging of single planes at 33 Hz. The firing rate of neuron 1 is low enough that individual spikes are visible in the calcium trace, showing persistently increasing spike rate after backward pre-displacements and persistently decreasing spike rate after forward pre-displacements. The firing rate of neuron 2 appears to be higher and, although individual spikes are not visible due to sampling rate and calcium indicator limitations, firing rate can be seen to increase following a backward pre-displacement relative to a forward pre-displacement.