Extended Data Fig. 9: latSC neuronal activity prior to spout contact encodes tongue position.
a, Example lick cycle correlated latSC neuron. Black bar indicates window used for cross-correlation analysis. b, Top, Mean-subtracted firing rate of the neuron in a. Middle, Mean-subtracted tongue volume. Data in a and b are aligned to L4 protrusion onset. Bottom, Normalized cross-correlation coefficient histogram. Black horizontal line, correlation coefficient from shuffled data. c, Percentage of neurons correlated with lick cycle across all combinations of trial types. Note that the circles are not proportional to the actual percentages. d, Top, Distribution of maximum cross-correlation coefficient across all lick cycle correlated neurons (n = 242). Bottom, Distribution of lags for lick cycle correlated neurons. Lag was taken at the time which maximized the cross-correlation coefficient for each neuron. e, Two example latSC neurons that are modulated prior to L4. Data are aligned to L4 protrusion onset, and sorted by condition. Within each condition, data are sorted by tip position in the ML plane at L4 protrusion offset. Black bar, analysis window used for GLMs. f, GLM schematic. We sought to test whether tongue position was encoded in pre-contact L4 latSC neuronal activity. Given the close temporal proximity (~120 ms) between past (L3) contact and L4 re-aiming, any relationship we might observe between tongue position and latSC neuronal activity may be confounded by past L3 contact location. We included both variables to test whether L4 tongue position and/or past L3 contact location was encoded in latSC neural activity. Note that, we are using GLMs to examine encoding of tongue position, not as a model of the latSC. g, Mean ± IQR McFadden’s pseudo-R2 across all latSC neurons (n = 320 neurons). pseudo-R2 is a measure of goodness-of-fit for GLMs, with higher values indicating better model fits. Note that pseudo-R2 values of 0.2-0.4 represent excellent model fits. h, Relative contribution of each predictor across all latSC neurons (n = 320 neurons). Tip and contact position are separated into their ML, AP or DV components. Error bars represent the bootstrapped IQR. i, Percentage of neurons encoding L4 tip position (left) and previous (L3) contact position. Tip and contact position are broken down into neurons encoding only ML, AP or DV components of position, two of these components, or all three.
