Classification of optic flow sensitive (OFS) descending neurons (DN) in Eristalis tenax.

a Receptive field of a representative OFS DN1 recorded from a male hoverfly. Colour coding indicates the local maximum spike frequency (red, inset, panel b), the direction of the arrows shows the local preferred direction (LPD, red arrowhead, panel c) and their length the local motion sensitivity (LMS, red, panel c). As the stimuli were not perspective corrected, the receptive field map reflects the visual monitor. b Contour line representing the 50% receptive field boundary based on local maximum spike frequency (colour coding, panel a). Inset: example response at one central location to eight directions of motion (black circles), showing local maximum spike frequency (local max, red line) above spontaneous rate (spont, black dotted line). c Local preferred direction (LPD) map of the same example neuron at locations where local motion sensitivity (LMS) exceeds 50% of the maximum. Inset: example response at one location to eight directions of motion with a sinusoidal fit (black line), illustrating LMS (red line) and LPD (red arrowhead). The data in the inset corresponds to the red arrow. d Receptive field of a representative male OFS DN2. e 50% receptive field contour and receptive field centre of the neuron shown in panel d. f Preferred direction map of the same OFS DN2. g Distribution of receptive field preferred directions across 100 male reference neurons, colour-coded by neuron classification. Dashed lines in corresponding colours indicate the thresholds used for neuron type classification. h Receptive field centres of the same 100 reference neurons, using the same colour coding. i Relationship between preferred direction and receptive field centre for the 100 neurons. Note that the scale is the same for the azimuth and elevation, and the scale bar in panels a, d, h and i correspond to both the x- and y-axis.

Sex-based comparison of direction sensitivity in OFS DNs.

a Polar plot showing the preferred direction of male (blue) and female (red) OFS DNs in response to a full-screen, full-contrast sinusoidal grating (spatial wavelength 7°, temporal frequency 5 Hz) moving in eight different directions (see also Supplementary Fig. 3). Individual data points represent the response amplitude and preferred direction of each OFS DN (red, Supplementary Fig. 3j). Larger, salient circles indicate the population median, with error bars showing the interquartile range. The dashed lines indicate the directional thresholds used to classify neuron type, as OFS DN1 (male: N = 9; female: N = 12) or OFS DN2 (male: N = 20; female: N = 14). Asterisk indicates a statistically significant difference (p < 0.05, Watson-Williams two-sample test). b Comparison of male and female OFS DN1 responses to translational optic flow at 0.5 m/s: sideslip, lift and thrust; and rotational optic flow at 50 °/s: pitch, yaw and roll (male: N = 9; female: N = 12). c Comparison of male and female OFS DN2 responses to the same optic flow stimuli (male: N = 20; female: N = 14). Data presented as median and interquartile range.

Velocity response functions in male and female OFS DNs.

a Example stimulus profile over time, with roll velocity on the y-axis. b Representative spike histogram from a single trial from a male OFS DN2, smoothed using a 100 ms square-wave filter with 0.025 ms resolution, and time-aligned to the stimulus shown in panel a. Grey shading in panels a and b highlight the analysis windows used to calculate response. c Example extracellular raw data traces extracted from the analysis windows in panel b, illustrating neuronal responses to roll velocities of –150, 10, and 200 °/s. d Average spike frequency (grey circles) was calculated for each repetition from each neuron (N = 1 neuron, n = 18 repetitions), and the median of these (black) was used for further analysis. e Velocity response functions of OFS DN1 in male (blue) and female (red) hoverflies in response to roll (N = 4 males, 5 females), sideslip (N = 5, 6), lift (N = 4, 4), and thrust (N = 3, 5). f Velocity response functions of OFS DN2 to roll (N = 10 males, 8 females), sideslip (N = 6, 7), lift (N = 6, 7), and thrust (N = 6, 7). Data in panels e and f are presented as median and interquartile range. Asterisks indicate statistically significant differences, two-way ANOVA with Šídák’s multiple comparisons test (** p < 0.01 and **** p < 0.0001), see also Table 1.

Statistical summary of neural velocity response functions.

Results from two-way ANOVA analyses evaluating the effects of stimulus velocity and sex on neural responses (OFS DN1 and DN2). Asterisks denote levels of statistical significance: *** p < 0.001, **** p < 0.0001.

Morphological reconstruction of OFS DNs.

a Schematic diagram of the hoverfly central brain and thoracic ganglia showing the projections of OFS DN1 (green) and OFS DN2 (orange). Key anatomical landmarks are labelled: CB, central brain; OL, optic lobe; CC, cervical connective; T1 LN, prothoracic leg nerve; T2 LN, mesothoracic leg nerve; T3 LN, metathoracic leg nerve; TAG, thoracic-abdominal ganglion; FN, frontal nerve; PtN, pterothoracic nerve; HN, haltere nerve; AbN, abdominal nerve. b Confocal image of a reconstructed male OFS DN2. White boxes indicate regions magnified in panels d - i. c A magnification of the thoracic ganglion in panel b. d Input dendrites of the same OFS DN2 around the sub-oesophageal ganglion. e Input dendrites of a female OFS DN2. f Input dendrites of a male OFS DN1. g Output projections of the same OFS DN2 neuron as in panels b, c and d. f Output projections within the thoracic ganglia of the neuron in panel e. h Output projections of the neuron in panel f.

WBA velocity response functions in male and female hoverflies.

a A single frame from a video labelled with trained DeepLabCut models35,36,55. Coloured dots correspond to those in panel b, used to extract the wing beat amplitude (WBA). b Pictogram illustrating a higher wing beat amplitude on the left wing (WBAL) compared to the right wing (WBAR), suggesting a turn to the right. Equations used to calculate wing beat amplitude difference (WBAD) and wing beat amplitude sum (WBAS). c Example stimulus with sideslip velocity on the y-axis. d Representative WBA for the left (pale colour) and right (salient colour) wings from a single trial, time-aligned with the stimulus shown in panel c. Grey shading in panels c and d indicate the analysis windows used to calculate WBAD and WBAS. e Magnified view of example WBA, showing behavioural responses to sideslip velocities of -2, 0.2, and 2 m/s. f Average WBAS (grey circles) calculated across repetitions (N = 1 animal, n = 19-36 repetitions). Black circles represent the median WBAS per stimulus velocity, used for further analysis. g WBAD in male (blue) and female (red) hoverflies in response to different optic flow velocities: roll (N = 5 males, 7 females), sideslip (N = 6, 6), lift (N = 7, 5), and thrust (N = 6, 5). h WBAS to the same stimuli in the same animals. Data in panels g and h are presented as median and interquartile range.

Statistical summary of behavioural velocity response functions.

Results from two-way ANOVA analyses evaluating the effects of stimulus velocity and sex on behavioural responses (WBAS, WBAD, head angle, foreleg extension, hind leg extension and hind leg difference). Asterisks denote levels of statistical significance: ** p < 0.01, *** p < 0.001, **** p < 0.0001, after performing a Bonferroni correction54 for 6 comparisons of data from the same videos.

Response onset to roll and lift stimuli in male and female hoverflies.

a Time to response onset in OFS DN1 to roll (+50 °/s) or lift (+0.5 m/s) stimuli in males (blue, N = 5) and females (red, N = 8). b Response onset in OFS DN2 to roll (+50 °/s) or lift (-0.5 m/s) measured in males (N = 20) and females (N = 14). c Onset of WBAS responses to roll (-200 °/s) and lift (+2 m/s) in males (N = 5 for roll, 9 for lift) and females (N = 7 for roll, 5 for lift). Data are presented from individual neurons (a, b) or animals (c) with lines indicating median. Asterisks indicate statistically significant differences, two-way ANOVA with uncorrected Fisher’s LSD test (** p < 0.01 and *** p < 0.001).

Velocity response functions of the head, fore- and hind legs in male and female hoverflies.

a A single frame highlighting the points used to track the head (yellow and dark green), thorax (orange), forelegs (pale and dark blue), and hind legs (magenta and pale green). b The head angle (HA) was defined as the angle between the medio-posterior head and the longitudinal thorax axis. c The foreleg extension was defined as the distance between the proximal and distal point, after calculating the average for the left and the right leg. d The hind leg extension was defined as the distance between the knee and a lateral point on the mid-abdomen, after calculating the average for the left and the right leg. We also calculated the difference between the right and the left hind leg. e The head angle in male (blue) and female (red) hoverflies in response to different optic flow velocities: roll (N = 5 males, 7 females), sideslip (N = 6, 6), lift (N = 7, 5), and thrust (N = 6, 5). f The foreleg extension to the same stimuli in the same animals. g The hind leg extension in the same animals. h The difference between the left and right hind leg extension in the same animals. The data in panels e, f, g and h are presented as median and interquartile range. Asterisks indicate statistically significant differences, two-way ANOVA with Šídák’s multiple comparisons test (* p < 0.05, ** p < 0.01), see also Table 2.

a The four clusters as identified by us (see Fig. 1g, h and i) displayed on a polar plot where the distance from origo illustrates the azimuthal position and the position along the circumference the preferred direction, colour coded as in the main paper (pale green: OFS DN1 LHS; dark green: OFS DN1 RHS; yellow: OFS DN2 LHS; orange: OFS DN2 RHS). b The Calinski-Harabasz value52 as a function of cluster number when using only the azimuthal location of the receptive field centre and the preferred direction (red), or when adding one other receptive field criterion (grey, receptive field centre elevation, receptive field height or width, maximum LMS, the number of positions with LMS values exceeding 50% of the maximum, or LPD variance, see Methods), or all receptive field criteria (blue). c The resulting z-score for azimuthal location and preferred direction, and the four clusters using k-means clustering in Matlab.

Exclusion criteria and receptive field comparisons in male and female OFS DNs.

a Distribution of maximum local motion sensitivity (LMS) in 100 male reference neurons (grey histogram, N = 100) compared to the neurons used for quantification in the rest of the paper (N = 35 males, 36 females). Six neurons with a maximum LMS below 20 spikes/s (red cross-hatched region or dashed line) were excluded from further analysis. b Number of stimulus positions with LMS greater than 50% of the neuron’s maximum LMS (black and red arrows in Fig. 1c and f). Neurons with fewer than 4 positions were excluded. c Distribution of LPD variance. Neurons with directional variance exceeding 30° were excluded. d Classification of male neurons based on receptive field centre and preferred direction. The top panel shows the preferred directions across neurons, with dashed lines indicating the directional thresholds used for neuron type classification. The middle panel shows receptive field centre locations, and the bottom panel illustrates the relationship between receptive field centre and the receptive field preferred direction for all 35 male neurons, with 2 neurons excluded from further analysis (grey). Note that the scale is the same for the azimuth and elevation, and the scale bar corresponds to both the x- and y-axis. e Same classification criteria applied to 36 female neurons, with 7 neurons excluded (grey). f The receptive field width and height at the 50% contour line for the remaining OFS DN 1 (N = 10 males, 12 females) and OFS DN2 (N = 23 males, 17 females). Individual data points from independent neurons, with black lines representing the median.

Schematic illustration showing the steps used to standardize data across recordings.

a. Top panel, ventral-side-up position of hoverfly during electrophysiology recordings. b Response of an example OFS DN2 neuron to a full-screen, full-contrast sinusoidal grating (spatial wavelength 7°; temporal frequency 5 Hz) moving in eight different directions. The red arrowhead shows the neuron’s preferred direction, and the red line indicates response amplitude. c Preferred direction of male hoverfly OFS DNs (N = 29). Neurons are colour coded by classification. d Preferred direction of female hoverfly OFS DNs (N = 26). Neurons are colour coded by classification. e We rotated the data 180° to display the dorsal visual field up. f Direction sensitivity of the same neuron as in panel b after rotating the data 180°. g Preferred direction of male hoverfly OFS DNs after rotating the data 180°. Neurons are colour coded by classification, with dashed lines indicating the receptive field thresholds. h Preferred direction of female hoverfly OFS DNs after rotating the data 180°. i To display all neural data as RHS neurons, we flipped LHS neurons along the midline (yellow to black). i Resulting direction sensitivity after flipping along the midline of the same neuron as in panels b and f. k Preferred direction of male hoverfly OFS DNs when displayed as RHS only. l Preferred direction of female hoverfly OFS DNs when displayed as RHS only.

Comparison of spontaneous activity and responses to stationary stimuli.

a Spontaneous activity (open shapes) and responses of a male OFS DN1 to the stationary starfield (filled shapes), from data extracting directional sensitivity (as in Fig. 2; circles) or velocity response functions (as in Fig. 3; triangles). b Spontaneous activity and responses of OFS DN2, displayed as in panel a. c WBAS before stimulus presentation (open squares) or when viewing the stationary starfield (filled squares), quantified during velocity response function experiments (as in Fig. 5). d Hind leg extension before stimulus presentation (open diamonds) or when viewing the stationary starfield (filled diamonds), quantified during velocity response function experiments (shown in Fig. 7). The data points show data from individual neurons (a and b) or animals (c and d), with lines representing the median. Asterisks denote statistical significance, two-way ANOVA with uncorrected Fisher’s LSD test: * p < 0.05, ** p < 0.01 and *** p < 0.001.

Morphological quantification of OFS DN2.

a Schematic diagram of the hoverfly central brain and thoracic ganglia showing locations where the cervical connective (CC) and the OFS DN2 (orange) widths were measured. b Width measurements of OFS DN2 axons along the cervical connective at four anatomical landmarks: (1) immediately posterior to the central brain, (2) upper third of the CC, (3) lower third of the CC, and (4) immediately anterior to the thoracic ganglion. Data are shown for 3 male (blue) and 3 female (red) hoverflies. c Width of the anterior end of the cervical connective in the same individuals. All data are presented as individual animals with the horizontal lines indicating median.

Impact of stimulus size on neural and behavioural velocity response functions.

a Electrophysiological recordings were performed with the hoverfly orientated ventral side up, facing the centre of a horizontally oriented visual display. b In behavioural experiments, the hoverfly was tethered at 32° angle, parallel to the ground, facing the centre of a vertically aligned visual display. c Velocity response functions of male OFS DN2 neurons when the stimulus covered either the full screen (blue) or the central square (black, N = 7). d Velocity response functions of WBAS in male hoverflies when the stimulus covered either the full screen (blue) or the central square (black, N = 5). All data are presented as median and interquartile range.

Response onset to sideslip.

Onset of head angle, WBAS, fore- and hind legs to sideslip (+2 m/s) in males (N = 6) and females (N = 6). There was no significant difference between body parts, nor males and females (two-way ANOVA, p = 0.08, p = 0.67 and p = 0.84, for body parts, sex and interaction respectively). All data are presented as individual animals with the horizontal lines indicating median.