Extended Data Fig. 1: Further analysis of C. elegans olfactory navigation behavior.

a, Visualization of the two main chemotaxis strategies: weathervaning (left) and biased random walk (right). Animals weathervane by bending their direction of forward movement in a favorable direction, either toward the attractive odor or away from the aversive odor²³. Animals execute a biased random walk by increasing the likelihood of initiating reorientations (red) when they are moving in an unfavorable direction in the odor gradient (forward movement shown in blue)²⁰. b, Weathervaning behavior in olfactory gradients. To test for the presence of weathervaning, we examined the curving rate of forward movement when animals had different directions to the odor (θ). Curving rate is the change in the animal’s heading divided by their change in displacement over 1 s, which can be thought of as a measure of how much and which way animals are bending forward runs (see Methods for further details). As previously shown²³, animals bend runs toward attractive odors (butanone and diacetyl, sign of θ is the same as the sign of the curving rate). We further saw evidence that animals weathervane away from aversive odors (octanol and nonanone, sign of the curving is the opposite of the sign of θ). Data show mean ± 95% CI. c, Example animal movement path during chemotaxis showing a single, isolated reorientation (blue) and two examples of repeated reorientations that form pirouettes (green). d, Fraction of reorientations that turn the animal in the correct dorsal or ventral direction, comparing with odor to no odor. The diacetyl and nonanone with odor plates were recorded on different days, so each has its own no odor control, while 10 mg ml⁻¹ pyrazine, 1:10 propyl acetate, and 1:1000 2,4,5-trimethylthiazole were recorded on the same days and so have the same no odor control. Two-sided Wilcoxon rank-sum test with Bonferroni correction (from left to right, p = 0.0068, p < 0.0001, p < 0.0001, p < 0.0001, p < 0.0001). n = 12-18 recordings, 15 plates each of pyrazine, propyl acetate, and 2,4,5-trimethylthiazole; and 13 no odor plates. Black dot shows data mean. e, Change in direction (∆θ) executed by animals that start with a large or small angle to the odor (θ). Data presented here are the same as in Fig. 1j but additionally include animals that turned away from the odor (negative reorientation angle), demonstrating that amplitude modulation is more notable in animals that have turned in the ‘correct’ direction. n = 18 recording plates. Data show mean ± 95% CI. f, Change in direction (∆θ) executed by animals that start with a large or small angle to the odor (θ). Note that the ‘goal’ for octanol is the opposite of the goal turn for butanone—animals that begin facing toward odor (purple) are best served by executing larger angle turns to turn away from the odor, while animals that begin facing away from the odor (green) are best served by executing small angle turns. We indeed see such a behavior modulation. Two-sided Wilcoxon rank-sum test with Bonferroni correction (p = 0.0002, p = 0.0002). n = 18 recording plates. Data show mean ± 95% CI. g, Fraction of reorientations that turn the animal in the correct dorsal or ventral direction, split by the length of the preceding forward run. Runs are 0–2 s or 2–30 s. Not significant by two-sided Wilcoxon rank-sum test. n = 18 recording plates. Black dots show data mean. h, Fraction of reorientations that turn the animal in the correct dorsal or ventral direction among low-angle reorientations, split by reorientation length. Short reorientations are less than 0.5 body lengths. Not significant by two-sided Wilcoxon rank-sum test. n = 18 recording plates. Black dots show data mean. i, Fraction of reorientations that move the animal in the correct dorsal or ventral direction, split by the type of postreversal turn executed. Omega turns are >135°, mid-angle turns are 40–135°, and low-angle turns are 0–40°. Not significant by two-sided Wilcoxon rank-sum test with Bonferroni correction. n = 18 recording plates. Black dots show data mean. j, Bearing to odor at the end of isolated reorientations, the first reorientation of a pirouette, or the last reorientation of a pirouette for animals in butanone (left, pink) and octanol (right, blue) gradients. Pirouettes are defined as clusters of consecutive reorientations separated by less than 13 s. Two-sided Wilcoxon rank-sum test with Bonferroni correction (from left to right: butanone p < 0.0001, p = 0.017, p < 0.0001; octanol p < 0.0001, p = 0.0004, p < 0.0001). n = 17-18 recording plates per odor. Data are mean ± s.e.m. k, Fraction of animals that execute a reorientation in the next 13 s depending on their bearing to the odor at the end of their previous reorientation. Animals that end an individual reorientation in an unfavorable direction (away from butanone or toward octanol) are more likely to reverse again. Two-sided Wilcoxon rank-sum test with Bonferroni correction comparing slopes of the reorientations rate (p < 0.0001, p < 0.0001). n = 17–18 recording plates. Data are mean ± 95% CI. l, Schematic showing the sharp odor gradient. In the sharp gradient, 5–10 animals begin on NGM agar and are surrounded by agar containing the aversive odor octanol (see Methods for more details). By contrast, in the shallow gradient, 20–100 animals begin on minimal agar and are placed equidistant from a distal spot of an odor and a spot of ethanol (which is used to dilute the odor). m, Animals can direct reorientations dorsally or ventrally. Which choice is ‘correct’ in a sharp-boundary context depends on the animal’s initial bearing to the aversive octanol boundary. n, Example animal encountering octanol during whole-brain calcium imaging, showing that the difference between the baseline and octanol agar is identifiable by eye. The agar boundary is indicated with a black arrow. o, Latency to reverse following encounter with the agar boundary in multi-animal recordings. All animals begin on baseline agar and periodically either cross agar with the aversive odor octanol (‘octanol’) or baseline agar (‘no odor’). Each dot is the time in seconds to reverse after each boundary encounter, averaged across all encounters from one plate with 5–10 worms. Two-sided Wilcoxon rank-sum test (p < 0.0001). n = 20 recording plates per condition. Data are mean ± s.e.m. p, Recording setup for whole-brain calcium imaging. Top, an example animal from NIR imaging, which is used for behavioral data collection. Bottom, an example fluorescent head from spinning disc confocal imaging, which is used to image neuronal activity. To examine aversive olfactory responses, animals begin the recording on baseline agar (NGM) but are surrounded by a sharp octanol gradient. q, Fraction of animals that start a reversal in a 10 s interval in whole-brain calcium imaging recordings. ‘Octanol’ specifically looks at whether animals start a reversal in the 10 s following an octanol encounter. ‘Spontaneous’ is from data looking at whether animals reverse in a randomly chosen 10 s interval of spontaneous movement on baseline agar (not octanol). This fraction is calculated by looking at the fraction of animals that reverse in a fixed number of random intervals, which is chosen based on the number of intervals where the animal was on octanol (n = 132). Each dot shows one random sample of data. This process was then repeated 500 times to generate the distribution in black. Statistics compare this distribution to the actual fraction of animals reversing on octanol in a one-sided test. A one-sided test was used as we wanted to test if octanol would increase reversal probability. The octanol value was at the 99th percentile of the dataset (**p < 0.01). r, Distribution of signal variation in neuron activity in all recorded datasets compared to previously published GFP controls, recorded under identical image acquisition settings and with the same image processing software for trace extraction¹⁵. Signal variation is calculated as the s.d. of non-normalized neuron fluorescence divided by the mean of non-normalized fluorescence, as defined in ref. ¹⁵. This captures the extent to which a given trace shows fluctuations over time. Turning neurons are cells whose activity is associated with head curvature, which are further profiled in Fig. 2. Other neurons are all other recorded GCaMP+ cells in our datasets. The GFP control strain is SWF467, which expresses pan-neuronal GFP and mNeptune. s, Sensory neuron activity (ASH, ADL, and AWC) as animals encounter the aversive octanol barrier is shown in blue. Black line shows neuron activity during similar epochs of spontaneous forward movement. ASH is the sensory neuron that senses octanol, AWC senses butanone, and ADL has been shown to have a role in sensing high concentrations of octanol^31,32. Gray dashed line shows octanol encounter. Two-sided Wilcoxon rank-sum test comparing average activity on octanol and during spontaneous movement postoctanol encounter (p = 0.0001), data are mean ± 95% CI. t, ASH activity in animals that reverse away from octanol barrier is shown in blue, showing decreasing neuron activity as the animals initiate reversals and successfully leave the octanol. Black line shows ASH activity during spontaneous reverse movement. Black dashed line shows reversal start, and the red shading shows the reversal. Data are mean ± 95% CI. For all panels, significance is noted as: NS (not significant), *P < 0.05, **P < 0.01, ***P < 0.001 and **P < 0.0001. For panels with multiple comparisons, symbols denote Bonferroni-adjusted P values.