(A) Recording session structure: after recording from neurons during virtual navigation, the same stimuli were presented in an ‘open loop’ configuration where the flow speed of the virtual environment was decoupled from the animal’s movement on the treadmill. (B) Trial onset, landmark, and reward example neurons under these two conditions. (C, D) Response amplitudes of all task engaged RSC neurons during decoupled stimulus presentation (Kruskal-Wallis: p<0.0001, Wilcoxon signed-rank pairwise comparison with Bonferroni correction indicated in (D)). (E, F) Population activity and population vector cross-correlation during decoupled stimulus presentation for short (left) and long (right) trials. (G) Local cross-correlation at animal’s location is smaller during decoupled stimulus presentation. (H) Mean location reconstruction error. Reconstructing animal location from population vectors is significantly less accurate when the animal is not actively navigating (unpaired t-test, short trials and long trials: p<0.0001). (I) Traces of example neuron activity overlaid during virtual navigation (top) and decoupled stimulus presentation (bottom) with transients highlighted in orange. Ticks along bottom indicate peaks of transients around the neuron’s peak response. (J) Spread of transient peak location around peak mean response measured as the standard error of the mean of (standard error of mean of transient peak location – mean peak location). Solid bars: virtual navigation (VR), open bars: decoupled stimulus presentation (DC). (K) Average number of transients/trial during virtual navigation and decoupled stimulus presentation. (L) Average amplitude of transients in VR and DC conditions. Boxplots show median, 1st - 3rd quartile, and 1.5 interquartile range.