Neuronal detection of social actions directs collective escape behaviour

Living in groups provides animals with a variety of advantages, including protection from danger and predation^{1,2,3,4,5,6,7,8}. In fish schools and bird flocks, an approaching predator or other threatening stimulus can drive collective escape and avoidance behaviours^{5,6,20,21,22,23,24}. Individuals in these groups need not directly sense the threat to avoid danger if they act in response to the movements of their threat-informed neighbours, a benefit of collective living referred to as the ‘many eyes effect’^25,26. Despite the prevalence of collective movement and predator avoidance in nature^1,2,3,22,27, little is known about the neural mechanisms that allow individuals to obtain relevant, actionable information from their partners^3,11,12. The neural basis of social recognition has been characterized in a variety of invertebrate and vertebrate species^{15,18,28,29,30,31}, but it remains unclear how individuals recognize the specific actions of their social partners to engage in adaptive behaviour that benefits the individual and the group.

We address this question by studying the brain and behaviour of D. cerebrum, a micro glassfish that engages in collective schooling behaviour using visual perception of social partners^13,15,32 and that is amenable to large-scale in vivo neural activity imaging^15,33,34,35. This model system provides us with experimental access to the sensory cues and neural circuits that underlie social information transmission during collective responses to threats. Here we examine the perceptual and neural basis of collective escape behaviours in adult D. cerebrum and identify a neural signature of social action detection in central visual circuits that engages escape behaviour in observers to produce collective threat avoidance.

We first characterized the escape behaviour of adult D. cerebrum in groups of four by measuring the positions and postures of fish responding to a looming visual object, which mimics an approaching predator (Extended Data Fig. 1a,b and Methods). Visual looming stimuli drive robust escape and defensive behaviours in aquatic and terrestrial animals³⁶. We built a 265 mm×265 mm square arena to deliver directional looming stimuli to groups of D. cerebrum, in which two of the four walls were lined with a video screen displaying a textured floor and dim sky (Fig. 1a). Looming stimuli were rendered as dark spheres moving towards the centre of the arena from one of four randomly selected directions (3–6 s in duration, delivered every 7 min; Methods). These stimuli elicited a short-latency escape behaviour (Fig. 1b,c and Extended Data Fig. 1c). We investigated how the presence of social partners may inform escape behaviour by comparing the behaviour of single fish to fish in groups of four. Single fish escaped by increasing their speed by 6.0 ± 4.6 mm s⁻¹ in the 0.5 s after loom offset (n = 18), whereas a randomly chosen fish in a group of 4 increased their speed by 17.3 ± 2.7 mm s⁻¹ (n = 20). Compared to single fish, fish in groups showed a greater increase in speed (Fig. 1d,e and Extended Data Fig. 1d–f) and escaped by a greater distance (Fig. 1f and Extended Data Fig. 1g). We observed these escape behaviours in groups of different sizes, but not in single fish (Extended Data Fig. 1h–m). Escaping fish also rapidly increased the magnitude of their turning, but their turns were directionally unbiased³⁷ (Fig. 1g; Extended Data Fig. 1n,o). These results demonstrate that D. cerebrum collectively escape in response to visual looming stimuli.

Whereas some dense animal collectives show contractions or directional escape waves in response to external threats^{6,20,21,22,36,38}, we observed that D. cerebrum groups scattered after exposure to a looming stimulus (Fig. 1h). Groups of four rapidly increased their mean inter-animal distances to 150.3 ± 12.9% of their baseline (Fig. 1i and Extended Data Fig. 1p–r) and returned to the group approximately 10 s afterwards (Extended Data Fig. 1q). These fish also scattered in response to a mechanical tap stimulus, with a short latency³⁹ (Extended Data Fig. 2a–c and  Methods). Similar to visually evoked collective escape, groups of fish escaped more rapidly than individuals to taps (Extended Data Fig. 2d–h), increased their mean inter-animal distances to 118.9 ± 4.9% of their baseline (Extended Data Fig. 2i–l) and increased turning (Extended Data Fig. 2n–p).

Although collective escape to visual and mechanical stimuli are not identical (Extended Data Fig. 2q–s), we conclude that groups of D. cerebrum exhibit a collective escape behaviour by rapidly scattering. In response to either stimulus, groups only dispersed briefly (Extended Data Figs. 1q and 2l) and no more than a random arrangement (Extended Data Figs. 1r and 2m). This group scattering resembles flash expansions^38,40,41, a type of collective escape strategy that may improve the fitness of group members by preventing predators from following all animals in the group at once^40,42,43. However, we note that during escapes D. cerebrum are not actively avoiding each other; rather, they transiently depart from their typically coherent group structure.

D. cerebrum school using their vision of social partners^13,15,32, and thus we hypothesized that individuals modify their escape behaviours by obtaining information from conspecifics using vision. We tested this hypothesis by quantifying and manipulating vision during escape behaviour. We first tested the mechanosensory escape of groups in light and darkness. Loss of visual cues did not influence the fast tap-evoked escape behaviour of single fish or groups, with comparable changes in short-latency speed and travel distance (Extended Data Fig. 3a–j). However, groups required vision to effectively scatter; darkness reduced tap-evoked scattering and prevented group reformation (Extended Data Fig. 3k–n). Therefore, vision of social partners facilitates collective scattering in response to a mechanical startle, consistent with prior findings of the importance of vision in the collective escape of other fish species^5,6,26.

Although the mechanical threat stimulus is delivered synchronously to all group members, the visual loom stimulus is perceived asynchronously by individual fish, depending on the proximity of each fish to the approaching visual object; the size of the looming sphere in each individual’s field of view depends on their position in the arena relative to its approach (Fig. 2a and Extended Data Fig. 4a). We measured swimming speed after the loom passed different angular thresholds in the field of view for each fish, ranging from 1 to 30° (Methods). We observed that speed immediately increases upon passing a threshold of 7°, whereas smaller angles evoke delayed escape behaviour (Fig. 2b,c and Extended Data Fig. 4b). The time from passing the angular threshold to increasing speed is not significantly different amongst trials with an angular size threshold of 7° or higher (Fig. 2c). This critical angle for loom-evoked escape is similar to those reported for adult zebrafish⁴⁴, but smaller than those reported for larval zebrafish⁴⁵. We also observed a shift to increased turning amplitude at these larger visual angles (Fig. 2d).

Within groups of four or more D. cerebrum, we examined the angular and temporal distribution of the individuals’ loom angles across the duration of the loom stimulus (Extended Data Fig. 4c–f). We calculated the cumulative probability of the loom stimulus reaching the 7° critical angle across individual fish and observed a similar temporal pattern (Extended Data Fig. 4g,h), obtaining a cut-off time (80.5% through the loom) for all trials when half of fish see the loom at critical angle size or larger (Fig. 2e). We then classified individual fish as ‘early-informed’ or ‘late-informed’ depending on whether that individual’s time to critical loom angle was before or after the cut-off time, respectively (Fig. 2e and Extended Data Fig. 4i). Across stimulus durations, early-informed fish increased their speed more rapidly than late-informed fish as the looming stimulus approached (Fig. 2f,g).

To determine whether individuals benefitted from having early-informed social partners, we categorized groups as predominantly ‘early’ or predominantly ‘late’ (Fig. 2e and  Methods). Early-informed fish comprised 72.1% (153 out of 212 fish) of the individuals in early groups and 16.9% (35 out of 207 fish) of those in late groups. We found that early-informed individuals moved at similar speeds regardless of the status of their group (Fig. 2h,i), suggesting that their direct observation of the looming stimulus drives their escape behaviour. By contrast, late-informed individuals increased their speed earlier when they were members of early groups, in contrast to those in late groups (Fig. 2j,k and Extended Data Fig. 4j). Additionally, fish characterized as ‘not-informed’—those that never reach the 7° loom threshold—only increased their speed if they were members of an early-informed group (Fig. 2l,m). Furthermore, the speed of late- or not-informed observers increases with greater number of early-informed fish in their group (Fig. 2n,o and Extended Data Fig. 4k). These results indicate that individuals with limited direct perception of a looming threat stimulus can nonetheless execute a timely escape when their social partners are informed of the threat.

Our results suggest that fish can escape after receiving indirect information about a threat from the visual perception of their social partners’ escape behaviours. Testing this requires experimental conditions in which single D. cerebrum can observe conspecifics escaping but cannot directly sense a threat. To present fish with experimentally controllable social partners, we developed a visual virtual reality environment populated by three-dimensional fish models whose appearance and movement closely match those of adult D. cerebrum (Fig. 3a, Extended Data Fig. 5a,b and  Methods). We found that individual adult D. cerebrum readily engage with these virtual partners, swimming at an average distance of 71.1 ± 13.3 mm from the monitor when they are displayed, compared to 134.4 ± 17.7 mm at baseline (Fig. 3b,c).

We developed a closed-loop paradigm in which virtual fish would execute an escape action when real D. cerebrum came within 16 mm of the virtual groups (>60 s inter-stimulus interval; Fig. 3d and Extended Data Fig. 5c). Because D. cerebrum scatter during collective escape (see Fig. 1h and Extended Data Fig. 2i), we reasoned that real fish would often escape away from the screen, rather than into it. We observed that individual fish executed avoidance behaviours to these escaping virtual fish (Fig. 3e), which scaled with increasing numbers of escaping virtual partners (Fig. 3f,g). This escape behaviour was comparable in magnitude, but delayed in onset, to escape from direct perception of a threat; when we replaced the conspecific escape action with a visual looming stimulus, observers escaped at a faster rate (Extended Data Fig. 5d,e and Methods). This finding is in agreement with prior work showing that socially transmitted information decays with distance^6,46,47, a strategy to prevent continuous unabated transmission of threat information^22,46.

Therefore, we conclude that collective escape in D. cerebrum is mediated by the visual perception of the actions of one’s social partners: observing an escaping conspecific promotes one’s own escape. For individual fish to achieve this, their visual systems must be capable of detecting these escape actions.

Neurons in the visual systems of D. cerebrum and zebrafish are responsive to stimuli whose size, shape and motion patterns resemble conspecifics^15,18. These objects robustly activate populations of neurons in the optic tectum (homologue of superior colliculus in mammals), a retino-recipient midbrain structure that mediates innate visually driven orienting behaviours in vertebrates ranging from fish to birds to primates^16,17,36,48. To examine how neural populations in the D. cerebrum visual system respond to the escape behaviours of conspecifics, we recorded neural activity while fish observed virtual social motion stimuli with routine swimming movements versus those with sudden escape behaviour (Methods).

Using a D. cerebrum transgenic line expressing a nuclear-localized calcium indicator in almost all neurons, we performed large-scale two-photon calcium imaging from fish tethered within a panoramic visual display environment¹⁵ (Fig. 4a). We displayed simplified biological motion stimuli^15,18,49: fish-sized spherical objects whose motion in virtual space is derived from recordings of real D. cerebrum groups moving with their characteristic burst-and-glide kinematics (Extended Data Fig. 6a). Tethered D. cerebrum observed objects that either moved in a typical swimming pattern (biological motion group), moved normally then escaped (biological motion-escape group), or moved with smooth non-biological motion (linear motion group), in addition to a looming stimulus (Extended Data Fig. 6b and Methods). We presented each stimulus for 6 s, in a pseudorandom order with a 15 ± 1 s inter-stimulus interval, and recorded neural activity from 12 adult D. cerebrum, observing each stimulus 7–10 times.

We recorded from 3,267 ± 92 neurons in a 978 μm × 978 μm field of view in each of the 50 planes in a total of 12 fish (1–2 planes per imaging session, 1–4 sessions per fish; Extended Data Fig. 6c). After motion correction and cell extraction (Methods), we sorted neurons by their anatomical position to identify cells in the midbrain optic tectum, as well as surrounding diencephalic, forebrain, and hindbrain regions (Fig. 4b and Extended Data Fig. 6d). We identified neurons responsive to visual stimuli across all recorded brain areas (Fig. 4b and  Methods), including cells in portions of the optic tectum corresponding to the retinotopic position of stimuli presented to the contralateral eye (Extended Data Fig. 6e). We also observed visually driven cells in the dorsal thalamus and telencephalon, regions identified as responsive to biological motion in juvenile zebrafish^18,50. We classified neurons by their responses to visual stimuli, identifying cells with significant responses to individual or combinations of stimuli (Extended Data Fig. 6f–i; n = 23,452 neurons in 12 fish, P < 0.05). Many cells were activated by looming stimuli (Extended Data Fig. 6h–k), which are innately salient³⁶ and expand to occupy most of the visual space. Although many neurons responded to multiple stimuli (Extended Data Fig. 6j), a decoder trained on neural activity was able to distinguish amongst stimulus types (Extended Data Fig. 6l–q and Methods).

We focused our analysis on the smaller objects presented to similar positions in visual space: the two biological motion stimuli (routine swimming and escape) as well as the non-biological motion stimulus. Of these stimuli, the magnitudes of neural responses to biological motion stimuli were greater than those of responses to the linear motion object (P = 0.0021 and P = 2.43 × 10⁻¹⁰, respectively), and responses to the biological motion stimulus that escaped were greater than those to swimming motion (P = 0.0056; Fig. 4c). These responses were unaffected by the fish’s swimming movements during imaging (Extended Data Fig. 6r,s).

Thus, stimuli of social relevance to the observer increase the visual response of cells in the midbrain and thalamus. Similar types of neuronal gain enhancement have been identified in the tectum or colliculus of other species in response to innately salient visual stimuli^16,17. We hypothesize that conspecific escape and social actions are particularly salient to D. cerebrum, which therefore drives an increased response in visual neurons.

To determine how specific social actions are encoded in neural populations, we compared the timing of neural responses to objects with routine biological motion versus those with sudden escapes. We found a subpopulation of neurons whose activity increased upon the escape action of objects with biological motion, at the midway point of the stimulus (Fig. 4d,e and Extended Data Fig. 7a,b). These same neurons also increased their activity at the offset of the routine swimming stimulus, when the objects vanish at the end of the trial (Fig. 4d,e and Extended Data Fig. 7b). This stimulus-offset response was specific to the stimuli that move with biological motion; these cells were not driven by the offset of the linearly translating object (Fig. 4d,e and Extended Data Fig. 7b). Across all recording sessions, neurons preferentially responded to moments of stimulus escape and offset (Fig. 4f). Cells whose activity peaked after the escape action also peaked after offset of the biological motion stimulus (Extended Data Fig. 7c). Although we also observed a subset of cells that peaked in response to the offset of the linear motion object (Extended Data Fig. 7b), these responses were not clearly associated with the actions of objects with biological motion (Extended Data Fig. 7d,e).

We classified neurons as those that responded specifically to the escape or disappearance of biological motion objects (social-offset), versus those that responded to the offset of any stimulus (general-offset) (Fig. 4g and Extended Data Fig. 7f). We examined the spatial location of social-offset neurons (Extended Data Fig. 7g, n = 12 fish) and found 45.2% (2,357 out of 5,208 total) of these cells in the tectum and 13.4% (702 out of 5,208 total) in the dorsal thalamus (Fig. 4h and Extended Data Fig. 7h). Neither response type was modulated by the observer’s swimming movements (Extended Data Fig. 7i).

These data suggest that a subset of visual neurons that track objects with biological motion, such as social partners, are responsive to the disappearance of those objects. As adult D. cerebrum are capable of high-speed movements (Fig. 1 and ref. ³⁹), a social partners’ sudden disappearance from their position in the visual field can be an indication of their rapid escape.

To selectively respond to the actions of social partners, D. cerebrum must be able to recognize social partners before they escape. We propose that this is achieved by detecting their species-typical movement patterns: burst-and-glide biological motion. To test this, we recorded neural activity while presenting fish with realistic virtual D. cerebrum (Fig. 3a and Extended Data Fig. 5a), which drive robust stimulus engagement (Fig. 3b,c), unlike simple spheres (Extended Data Fig. 8a,b). We programmed three virtual fish to follow a defined trajectory in visual space, moving with either biological motion (burst-and-glide) or linear motion; after 5 s, stimuli either escaped, vanished, or stopped moving while a looming sphere appeared (Fig. 5a). These stimuli enable us to examine how the visual system distinguishes between the motion and actions of virtual conspecifics, and whether these visual neurons respond to any spatially localized visual threat.

We recorded from 3,269 ± 114 neurons in a 978 μm × 978 μm field of view in each of the 28 imaged planes (1 plane per imaging session, 1–5 sessions per fish, n = 9 fish; Extended Data Fig. 8c,d). Many neurons responded to each stimulus type, with the greatest response to the highly salient looming stimulus (Fig. 5b–d and Extended Data Fig. 8e–g). We trained linear classifiers to decode trial type from neural activity (Methods). We found that classifiers distinguished between patterns of virtual conspecific motion (biological versus linear motion), with increased classification accuracy during stimulus presentation and before the action (classification accuracy = 72.6%; Fig. 5e). However, classifiers were unable to effectively determine whether the observed action was an escape or disappearance (classification accuracy = 50.2%; Fig. 5f). By contrast, the classifier effectively decoded looming from escape or disappearance after biological motion (classification accuracy = 66.7% and 66.1%, respectively; Fig. 5g–i and Extended Data Fig. 8h), suggesting that these populations are not responding uniformly to all forms of visual danger. Motor activity did not modulate the neurons that responded to this stimulus set (Extended Data Fig. 8i–k), and these neurons were found across all brain regions imaged (Extended Data Fig. 8l,m). Cells that were responsive to the escape of virtual fish swimming with biological motion also responded to the disappearance of those fish, compared to the disappearance of virtual fish moving with linear motion (Fig. 5j).

Together, these results demonstrate that the visual system of D. cerebrum can identify objects that move with its species-typical burst-and-glide biological motion, and respond to the escape actions of those social partners. However, we find that these neurons do not distinguish between whether those social partners execute a kinematically realistic escape action or simply vanish.

To determine whether a fish’s visual perception of disappearing conspecifics affects their behaviour, we allowed individual freely moving D. cerebrum to engage with virtual fish models (as in Fig. 3a and Extended Data Fig. 5a). We programmed groups of five virtual fish to swim with either biological (burst-and-glide) or linear motion. When real D. cerebrum came within 16 mm of the groups, virtual fish either escaped, vanished, or did not alter their trajectory (Fig. 5k and  Methods).

We examined the behaviour of fish in response to the actions of virtual conspecifics swimming with biological motion, and found that either their escape or disappearance caused disengagement from the virtual group, resulting in them moving further than the control (no action, 40.4 mm; escape, 84.1 mm; disappear, 69.8 mm; Fig. 5l,m). There was no significant difference between behavioural responses to virtual fish escaping or disappearing (P = 0.546). By contrast, when virtual conspecifics moved with non-biological linear motion, real fish did not show a behavioural distinction between no action and their escape or disappearance (no action, 66.0 mm; escape, 65.6 mm; disappear, 61.8 mm; Fig. 5n,o). Notably, behavioural withdrawal from the offset of virtual conspecifics differs from the rapid escape from a direct threat (a loom; Extended Data Figs. 5d and 8n,o and Methods). These data demonstrate that D. cerebrum will rapidly disengage from conspecifics that escape or vanish, but only if those conspecifics move with biologically realistic burst-and-glide motion. This behavioural response to visual detection of escaping conspecifics can produce threat avoidance at the level of a small group, where escape amplitude decays with distance from directly threatened individuals^6,21,22.

By analysing the perceptual and neural basis of escape behaviour in D. cerebrum groups, we found that the information transmission underlying this collective behaviour is mediated by the visual detection of conspecifics that suddenly escape or vanish. Our neural recordings indicate that this is accomplished by populations of visually responsive neurons, primarily in the optic tectum and thalamus, that are activated by the rapid disappearance of mobile visual objects with biological motion. Neural encoding of a social offset can be an effective means for animals to detect the escape actions of social partners, and thus execute an escape themselves. The ability to respond to the sensory experiences of one’s social partners (for instance, their detection of a predator that causes escape) is a benefit of living in a group, and exemplifies the ‘many eyes’ effect^1,3,25,26. Our work provides evidence of how neural computation at the level of individual animals can produce emergent collective behaviours at the level of the group^3,11,12. Although we focus on collective escape behaviour here, similar visual communication mechanisms may be at work during other collective behaviours, including in the context of navigation, foraging and decision-making^1,2,3,4,5,6.

Our large-scale neural recordings demonstrate the central role of the optic tectum in D. cerebrum’s visual perception of escaping social partners. The tectum (or colliculus in mammals) enables salience-based attention and neuronal gain across vertebrates^16,17,36. Whereas most studies have investigated responses to salient stimuli without social features, socially relevant visual stimuli are known to activate tectal neurons in fish^15,18, mice⁵¹ and primates⁵². Therefore, the rapid visual detection of social actions with innate salience may be driven in part by conserved midbrain circuits across species. Here we classify social stimuli as those that move with biological motion, but biological motion alone may not always be sufficient to classify objects as being a social agent. Fish may use a combination of many sensory cues to identify conspecifics, including mechanosensory⁵³, auditory^35,39 or chemosensory⁵⁴ information. Integration of multisensory cues is likely to contribute to many collective behaviours, including escape and predator avoidance.

It remains unknown whether the specific social-offset response that we observe here emerges from the on–off organization of presynaptic retinal ganglion cells⁵⁵, or if it is generated by local circuits in the midbrain or thalamus^16,56,57. In addition to the tectum, we found neurons for social action detection in the thalamus and telencephalon, which are associated with object detection, biological motion perception and social behaviour in zebrafish^18,50,56. Interconnected tectal, thalamic and forebrain circuits^18,56 may play an important part in classifying the species-typical movements of social partners to guide collective behaviour. Although we found distinctions between escape behaviours driven by looming or social stimuli, the mechanisms that we describe here need not be exclusive to social stimuli; offset responses could generalize to the rapid disappearance of other visual items that have captured an animal’s attention.

Visual attention to social actions may be particularly important for the adaptive behaviour of schooling fish. Aquatic animals commonly experience a more restricted visual range compared with terrestrial animals¹⁹, and D. cerebrum have been found in turbid waters with reduced visibility^14,34. Therefore, individual animals may be particularly attuned to the actions of their neighbours, whose movements can inform observers of stimuli beyond their individual sensory range. As D. cerebrum are capable of high-speed escape movements³⁹, an escaping social partner could rapidly vanish from the short reach of an individual’s field of view. The scattering behaviour that we observed here may be a consequence of these environmental features, as the socially transmitted signal for danger (disappearing conspecifics) may not provide the observer with sufficient information about the direction of the threat—only its imminence. For species that show collective escape waves^{6,20,21,22,36}, social transmission of threat information may also include information about threat direction. Future studies can examine how group composition, ecological constraints and neurobiological mechanisms influence the diverse predatory avoidance strategies of animal collectives^1,11,12.

Our results provide an instructive example of how the sensory ecology of an organism impacts the behavioural significance of its neural computations and perceptual abilities. Danionella will be a useful model clade to explore these concepts^34,58, as different species occupy distinct ecological niches, such as clear versus turbid waters. Further insights into these natural behaviours will be gleaned by advancing experimental techniques for neural circuit dissection in Danionella and other emerging model systems capable of collective behaviour¹².

Fish husbandry

D. cerebrum^{13,14,15,33,35,54,60,61,62,63} were raised and maintained in conventional zebrafish housing systems (Aquaneering; system water temperature 29 ± 0.5 °C, pH 7.0, conductivity 600 µS cm⁻¹), under 14:10 h light:dark cycles, and were fed 2–3 times a day. D. cerebrum were bred in communal tanks of ~40–60 individuals, enriched with ~5 cm silicone tubes to encourage spawning. Eggs were collected during the first 1 h of daylight, and 1–2 h after morning and afternoon feeding. Embryos (0–5 days post-fertilization (dpf)) were raised in egg-water (5 mM NaCl, 169 μM KCl, 330 μM CaCl₂, 161 μM MgSO₄∙7H₂O, and 0.15% methylene blue dissolved in reverse osmosis purified water) in an incubator at 29 ± 0.5 °C.

Larvae (5–14 dpf) were housed in static tanks in the housing systems and cocultured with L-type rotifers (Brachionus plicatilis), while water level was raised ~1 cm per day. At 21 dpf, the housing tank was connected to the water circulation system and fish were fed with artemia. At 8–10 weeks post-fertilization, D. cerebrum reach adulthood and become fertile. Wild-type D. cerebrum were provided by A. Douglass and B. Judkewitz. The transgenic line Tg(elavl3:H2B-GCaMP6s)⁶⁴ was provided by B. Judkewitz.

The behavioural and imaging experiments with D. cerebrum were approved by the government authorities (US Animal Care, and Institutional Animal Care and Use Committee (IACUC), (USDA Registration Number 93-R-0437)) and carried out in accordance with the US federal and California state law to enforce the Animal Welfare Act (AWA) under the California Health and Safety Code.

Behavioural experiments setups

All behavioural experiments used D. cerebrum that were at least 8 weeks old, and included both sexes. Behavioural recordings were conducted in a room with a portable electric heater to maintain water temperature between 28–30 °C. Data acquisition was automated and timestamped in BonsaiRx^65,66 (https://bonsai-rx.org). All behavioural arenas were built with custom cut acrylic (black walls and transparent bottom with white styrene to diffuse bottom lit light sources) and encircled with black curtain to control luminance. Light and dark environments are controlled with a bottom-projected white or dark background (AnyBeam Pico Projector, HD301M1-H2).

Behaviours were illuminated with infrared light (CM-IR130-850NM, CMVision Technologies) from the bottom, and acquired at 121 Hz from a camera mounted above (FLIR Grasshopper, 33-534) with an 8 mm/F1.8 lens (Edmund Optics, 15-626) and long-pass filter (Edmund Optics, 12-767).

Behavioural tracking and processing

Social LEAP-Estimates Animal Poses (SLEAP)⁶⁷ was used to train models for tracking the location and posture of D. cerebrum. Each video was proofread after inference and identity tracking to create h5 files that contain positional information of all tracking nodes of each animal for the given behavioural session. The h5 files and a corresponding timestamped csv were processed with custom Python code (3.12) for further analysis (see Code availability). Models were trained with a nine-node skeleton covering the two eyes and evenly distributed points along the midline of the body and tail.

For each time point, animal centroid was defined as the xy coordinates of the middle point of the two lobes of the swim bladder. A nose point was defined as the centre point of the 2 eyes, and the orientation of the animal was defined as the direction of the vector pointing from the centroid to the nose point and smoothed by a 5-frame running average (~40 ms). Speed and angular speed of the animal was calculated by taking the derivatives of centroid coordinates and the heading direction and smoothed by a 10-frame running average (~80 ms), respectively. Distance and alignment of each fish to all its neighbours were identified for each time point. All x and y coordinates were then transformed from pixel to mm by measuring the arena edge, a known physical length, in Fiji. We saved csv files with the timestamps for each video frame and any stimulus presented. Python code aligned all arrays to the behaviour before further analysis.

Visual loom-evoked escape assay

In the loom-evoked escape behaviours, D. cerebrum were left to habituate for 20 min in a square arena (265 × 265 mm) where two of the walls were lined with 10.1-inch LCD monitors (HAMTYSAN, 10.1 inch rendered at 1,024 × 600 pix). Each screen was mapped as a 14 × 20 cm window into a 3D virtual environment, eye perspective at centre point of area, 2 cm above the bottom, using BonVision package⁶⁶. Each experiment consisted of 5× 2-min recordings, and at the 1-min mark a dark sphere of 4.3 cm would appear in one of the corners (x = ±30, y = ±15, z = 1.5 cm) (Extended Data Fig. 1b) and move linearly to the centre of the arena (x = 0, y = 0, z = 1.5 cm) within 3–6 s. With a total virtual displacement of 33.5 cm, the loom moved at a constant speed of 5.6–11.2 cm s⁻¹, creating a length/velocity (L/V) ratio ranging from 384 to 769 ms. Both the starting position and moving duration were randomly determined. We separated each trial by 5-min intervals and used groups of 1, 2, 4 and 8 fish.

Baseline swimming speed in the 30 s before loom onset was fitted with Gaussian mixture model with four Gaussian components (determined by Bayesian information criterion) for single fish and fish in a group of four separately (Extended Data Fig. 1e). The boundary between the second and third components was defined as the threshold between low and high swimming speed. We defined escape latency as the first time point when a fish reached high speed if that time event occurred within 1 s after the end of loom (Extended Data Fig. 1f). We calculated the change in speed from looming stimulus by normalizing each fish’s speed 0.5 s after the end the stimulus to its mean speed 30 s prior to the onset of the loom (Fig. 1d,e). Travel distance before and after stimulus was calculated with integration of speed over 1 s (Fig. 1f and Extended Data Fig. 1g,l,m) window.

Turning behaviour was calculated by taking the absolute heading direction changes in 0.05-s bins, and plot it as probability density function (PDF) or cumulative distribution function over continuous 250-ms block (Extended Data Fig. 1n,o). Same statistical curves were plotted with 100-ms blocks to visualize the median turning angles (Fig. 1g). Inter-fish distance was calculated for each pair of fish within a group. The positions of each fish in the group were shifted by a random number independently to create the shuffled data used to calculate temporally shuffled inter-fish distance (Fig. 1h), and the trajectories of all fish were shuffled post end of loom to create chance level of spatially shuffled inter-fish distance (Extended Data Fig. 1r). Change in inter-fish distance was computed by subtracting the mean over 1 s before loom onset, and the mean at 0.5 s after the stimulus are used for statistical comparison (Fig. 1i).

Eye angle tracing of the loom was performed via two-dimensional ray-casting. In each video frame during loom stimulus presentation, we calculated the coordinates of the virtual sphere, and drew angles from the fish’s midpoint (between the two eyes) to the edges of the 4.3 cm sphere. The angular size of the loom is the absolute angle difference between the vectors to the left and right side of the sphere (Fig. 2a,b, Extended Data Fig. 4a,b). Speed from 6 s before the threshold was used as baseline to calculate changes (Fig. 2b,c).

We created Boolean arrays to identify when each fish first senses the loom at the threshold angle of 7°. These arrays were resampled to align trials of different lengths (speeds of 5.6–11.2 cm s⁻¹), and construct the cumulative probability curve of reaching loom threshold (Fig. 2e and Extended Data Fig. 4b). Using trials aligned from loom start to loom end, we identified the time through the trial where 50% of group members pass the angle threshold. This time point was used to assign individual and group information status in the subsequent analysis. For individuals, they were labelled as ‘early-informed’ if loom reaches critical angle at an earlier time than this cut-off. Otherwise, individuals were labelled as ‘late-informed’, including those that never reached critical angles and did not have a valid reaching time (that is, the angle of the loom never reached 7°). The mean of the threshold-reaching time in a group was also compared to the same cut-off to assign the group as ‘early group’ or ‘late group’ (Fig. 2e). Therefore, early groups contained more early-informed individuals, and vice versa. To compare behaviour across trials, we calculated the average speed in time bins 5% of the original trial length.

Mechanosensory tap-evoked escape assay

In the tap-evoked escape behaviours, D. cerebrum were left to habituate for 20 min prior to the experiment in a 350 mm circular arena filled with 1 l of system water¹⁵. Each experiment consisted of 5 repetitions of 2-min recordings, where an actuator (Adafruit Industries, TAU0730TM-14) mounted beneath one corner of the arena was moved by a 300-ms voltage pulse that delivered a mechanical tap at the 1-min mark. Each trial was separated by 7 min, and experiments included groups of 1, 2, 4 or 8 fish. For the light–dark experiments of tap escape, groups of 1 or 4 fish from a different cohort were used. Behaviours were run with either a white or black background bottom projection.

Baseline swimming speed in the 30 s before tap onset was fitted with Gaussian mixture model with four Gaussian components (determined by Bayesian information criterion) for single fish and fish in groups of four separately (Extended Data Fig. 2g). The boundary between the second and third components defined the threshold between low and high swimming speed. Escape latency within 1 s after the end of tap was identified as the first time point when a fish reached high speed (Extended Data Fig. 2h). The mean speed in the 30 s prior to the tap was used as the baseline for each individual fish to calculate how their speed changed 0.5 s after the tap (Extended Data Fig. 2d,e). Travel distance before and after the stimulus was calculated by integrating the speed over a 1 s (Extended Data Fig. 3d,e,i,j) window.

Turning behaviour was calculated by taking the absolute heading direction changes in 0.05-s bins, and plot it as PDF or cumulative distribution function over continuous 250-ms block (Extended Data Fig. 2n,o). Same statistical curves were plotted with 100-ms blocks to visualize the median turning angles (Extended Data Fig. 2p).

Inter-fish distance was calculated for each pair of fish within a group. The positions of each fish in the group were shifted by a random number independently to create the shuffled data used to calculate temporally shuffled inter-fish distance (Extended Data Fig. 2i,k), and the trajectories of all fish were shuffled post end of tap to create chance level of spatially shuffled inter-fish distance (Extended Data Fig. 2m). Change in inter-fish distance was computed by subtracting the mean over 1 s before tap onset, and the mean at 0.5 s after the stimulus are used for statistical comparison (Extended Data Figs. 2j and 3l).

Virtual conspecifics engagement and social escape assay

We created a 3D virtual reality (VR) environment using Panda3D (https://www.panda3d.org/), and populated it with five virtual male D. cerebrum (Extended Data Fig. 5a) modelled in Blender (v3.5.1). The Danionella model was adapted from an image in Kadobianskyi et al.⁵⁹. The VR scene was captured by a virtual camera positioned in VR space such that the virtual D. cerebrum appeared on the bottom third of the monitor. These virtual conspecifics appeared on a single monitor, with the opposing monitor displaying an empty background. Each VR D. cerebrum appeared ~12 mm in length on the display and had animated tail movements that coincided with the burst phase of their burst-and-glide kinematics (Extended Data Fig. 5b). The virtual school was programmed to swim along the width of the monitor by translating their x position in VR space, with small, random deviations in z to simulate slight changes in depth. The school always remained in the field of view of the VR camera, and therefore appeared on the monitor, as each fish executed a 180° turn when they reached the boundary of environment.

The environment was rendered on the monitors of the same square arena used in loom-evoked escape assay (Fig. 3a). To establish whether the sight of our virtual school attracted real fish, we first implemented an open loop paradigm (Fig. 3b,c). Here, a single D. cerebrum was left to habituate for 5 min in a square arena before being shown 4× 5-min no stimulus followed by 5-min stimulus on, with 5 swimming virtual D. cerebrum. At the end of each stimulus-on period, all virtual fish were assigned a random escape heading spanning 150° towards the orientation of the virtual camera so they rapidly swam away from the virtual camera into the horizon of the VR environment (Fig. 3d). These actions were not contingent on the position of the real fish with respect to the virtual fish.

For the social escape assays, we implemented a closed-loop paradigm wherein the proximity of the real fish to the virtual school triggered an escape. For this, we again allowed a single D. cerebrum to habituate to the environment for 5 min. After 1 min of the 5-fish virtual school being displayed on the monitor, the test fish’s engagement with the virtual school (test fish distance to centre of virtual school <16 mm) triggered either an escape or disappearance of VR fish depending on the experiment (Figs. 3d,e and 5k and Extended Data Fig. 5c). The overhead camera served to track the centroid position of the fish in real time using the KNN background subtractor function from the openCV library. The number of escaping or disappearing virtual fish was randomized for experiments with differing numbers of escapees, with each fish experiencing a 1-, 3- and 5-fish-escape event from the group of 5 total in the school (Fig. 3f,g). Each trial was separated by 60 s wherein we displayed an empty background after the escape. The virtual fish then re-appeared on the screen and the process repeated for a total of three trials per fish. For trials examining the response to linear movement of the virtual school, we programmed the virtual fish to move at the average velocity of a burst-glide cycle and removed the tail movement animation. In the case of disappearing fish, they vanished at the onset of ‘escape’. We used a similar closed-loop trigger to initiate a black looming stimulus that emerged from the centre of the virtual school (Extended Data Fig. 8n,o). The looming stimulus expanded hyperbolically over the span of 0.75 s, averaging 54° s⁻¹, to a final size of 50 mm on the LCD monitor.

Fish immobilization and holding chamber for in vivo two-photon calcium imaging

We modified the head-tethered preparation and holding chamber design from Zada et al. 2024¹⁵. In brief, adult D. cerebrum were transiently anaesthetized with tricaine (120 mg l⁻¹ MS-222 for 1–2 min) before they were removed from water and placed ventral side up on a 24 × 30 × 1 mm coverslip (12545B, Fisher Scientific). Two pieces of pre-cured SYLGARD wedges were fixed on the coverslip with UV-cured plastic (Bondic) to support the fish body from behind the eyes. 5–10% low-melting point agarose (UltraPureTM LMP Agarose, Invitrogen) was added to cover the body from swim bladder to the proximal tail, and on the side of head in front of the SYLGARD pieces. Another horizontal strip of UV-cured plastic was added to connect the coverslip and the agarose for more stability. Agarose from the side of the eyes, gills, and the tip of the tail were removed with a scalpel. Once all components were dried, the fish was placed back into a holding dish and ventilated with oxygenated water to ensure gill movement. Fish that recovered and showed spontaneous gill ventilation were then moved into the holding chamber under the two-photon microscope.

The triangular chamber was built from two monitors (HAMTYSAN, 7-inch 800 × 480 LCD). The two screens occupied ~260 ° of visual space in azimuth (130 ° each side, with ~10 ° gap in the front, and ~90 ° gap behind the fish), and ~105 ° of visual space in elevation (52.5 ° below and above). The chamber was filled with fish housing water (28 °C, ~500 ml). The water level was maintained ~1 cm above the midline of monitor height (10 cm) by gravity flow of fresh fish housing water and a custom vacuum made from a bent borosilicate electrode (WPI) constantly removing excess water throughout the duration of the experiment.

The back wall and floors were made of transparent acrylic for IR lights and recording camera access. A 4.5 mm column of clear acrylic was used as a pedestal to position head-fixed fish at a standard positioning within this arena (intersection of each screen’s midpoint). Coverslips were suspended from the acrylic pedestal using two 3 ×1 mm magnets (Ethcool) that was embedded into the top of the pedestal and another one added when anchored.

Visual stimuli

Dot stimuli

BonsaiRx was used to record the timing of microscope frame acquisitions, and display visual stimuli. Visual stimuli were displayed on the screens using cube mapping in BonVision⁶⁶, where each screen displayed a viewport in a virtual environment. The virtual environment was composed of a 6,000 mm² plane with a smoothed white-noise pattern (‘seafloor’). The fish’s position in this VR world was in the centre (0,0 in xy coordinates), 10 mm above the seafloor.

For nonsocial stimuli, a dark sphere was programmed to appear in the front of fish in the virtual environment. After 1 s, it either moved horizontally to the periphery (linear motion stimulus) or moved toward the virtual camera (thus, expanding in size; loom stimulus) over 5 s (Extended Data Fig. 6b). For the social stimuli, behavioural recordings of swimming and escaping D. cerebrum from both top-down and lateral view were processed to obtain realistic position changes of fish in 3D space (Extended Data Fig. 6a). Behaviour tracking (121 and 200 Hz for the top-down and lateral side view) was resampled to the refresh rate of monitors (60 Hz) and applied to spheres. The diameter and distance of the spheres in the virtual environment were calibrated to match their size with live fish. The two smaller dots swam from the back to front of the head-tethered fish in the VR environment over 6 s (swim stimulus) or executed a downward and outward escape at the 3 s mark and kept swimming farther (Fig. 4d and Extended Data Fig. 6a,b). All stimuli within a session were delivered to one visual field (left or right).

Virtual D. cerebrum stimuli

A custom Python script was used to record the timing of microscope frame acquisitions and display visual stimuli. To simulate the presence of conspecifics, we constructed a 3D environment using Panda3D, positioning two virtual cameras to match the perspective of the head-tethered fish. A virtual school comprising three D. cerebrum were used, populating one visual field (left or right).

For stimulus types with biological motion, the virtual school moved with realistic burst-and-glide kinematics (rapidly accelerating and then exponentially decelerating at ~1 Hz, Extended Data Fig. 5b) and with animated tail movements that corresponded with the burst phase of motion. For stimulus types with linear motion, the virtual school moved with a constant velocity and without animated tail movements. The start and end position of each fish, as well as the average speed, were matched to stimuli swimming with biological motion. Movement in each trial lasted for 5 s, which corresponded to 4 burst-glide cycles, and was followed by either an escape or disappearance of the virtual fish, or the emergence of a looming stimulus. Escape trajectories spanned 45–135° away from the heading of the head-tethered fish. That is, the virtual fish veered to the left when they were displayed on the left screen, and they veered to the right when displayed on the right screen. To avoid contaminating our imaging path with green/blue light from the visual display, we lined the LCD monitors with red gel light filters (LILIYA, 8.5 × 11 inch cut to size) and red-shifted the virtual environment rendering with a graphic shader (Background colour is set as Light Coral (hex code f08080); ambient light RGB set to 1/0.5/0.5 to bias the light reflection from the seafloor and objects toward longer wavelength).

For both experiments, each recording session started with a 1–2 min baseline with only the seafloor background scene visible. We then used a trial structure to present each of the stimulus types to head-tethered fish for 10 trials each in a pseudorandomized order, with randomized 14–16 s inter-trial intervals. All stimuli were displayed underneath the water surface of the recording arena and on the same side of the monitor within the recording session (15–20 min). Concurrently, we recorded the tails of these fish using a FLIR Grasshopper camera (33-534) affixed with a long-pass filter (Edmund Optics, 12-767) at 30–120 Hz. Each video frame was also timestamped by the Bonsai or Python script for post hoc data synchronization.

Two-photon calcium imaging

Two-photon microscopy was performed with a ThorLabs Bergamo II multiphoton microscope controlled by ThorImageLS 4.3 and illuminated by an fs-pulsed 80-MHz Ti:S laser (MaiTai DeepSee; SpectraPhysics). GCaMP6s fluorescence was imaged with a 16×/0.8 W Nikon objective at an excitation wavelength of 930 nm and ≤10 mW excitation power. Frames of 1,024 × 1,024 pixels covered a field-of-view of 978 × 978 μm.

We imaged fish at a diagonal, to include as much of the brain in our field of view as possible^15,18. Depth of the imaging plane was determined by maximizing the number of cells in the optic tectum within the field of view. Dual-plane or single-plane imaging sessions were conducted for each fish if it maintained healthy spontaneous movement for the entirety of the experiments, where the stimuli came from each side of the visual field twice each. For two-plane imaging, we used fast z mode in ThorImageLS 4.3 to move across 50–90 μm at 5 Hz. For the single-plane imaging, a 2–3 frame average was used to produce a final effective imaging rate at 5–7.6 Hz.

Imaging analysis

We only proceeded with experiments where fish remained healthy at the end of recording and did not have substantial z-movements or drift during recording. For sessions with detected shifts in z, we trimmed the z-shifted frames if the remaining uncorrupted experimental length was more than half of the original time series, and analysed neural activity in this epoch.

Imaging data were motion corrected in Suite2p⁶⁸, followed by cell extraction. Time series were inspected using Rastermap⁶⁹ to ensure there was no z-motion contamination or slow drift, and we only included neurons with continuous measurements and which were classified as a cell by Suite2p (“iscell”=1). A cellpose⁷⁰ classifier was trained for cell extraction. The anatomical region where each neuron is located was further classified with manually defined boundaries; cells outside these boundaries were not included for region-specific analyses. We excluded the first 5% of each imaging trial to avoid the influence of sound-evoked activity upon the initiation of scanning. Stimulus timing and calcium traces were aligned and resampled to a common 10 Hz sampling rate for further analysis. Then the fluorescence traces were detrended, smoothed with a 1-s rolling mean, and z-scored. The z-scored fluorescence and the median xy coordinates for each cell within these boundaries were used for further analysis.

We identified stimulus-responsive cells by conducting a t-test between the mean z-scored fluorescence during the 6-s stimulus and the mean z-scored fluorescence in the 5-s baseline pre-stimulus. Neurons were identified as responsive to a certain stimulus type if the significance is P < 0.05. Offset cells are defined by peak response time (1) at or later than 6 s to biological motion (BM) stimulus onset and (2) between 3 and 6 s to BM-escape onset. Among them, those peaked at or later than 6 s to the linear stimulus onset are categorized as general-offset cells, and those peaked before 6 s are categorized as social-offset cells. For the support vector machine (SVM) models, we only used experiments where there were at least 8 trials for all four stimulus types. The decoders were trained on the principal components (those that explain 90% of variance) of neural activity from all responsive cells in each recording session. For longer sessions, 8 trials were randomly selected for each stimulus type and split with a stratified K-fold cross-validator (eightfold). Twenty-five iterations of training and testing were performed.

Tail motion was assessed by measuring the number of pixels corresponding to tail movement in each video frame during imaging sessions. We processed each video frame using three consecutive functions from the OpenCV library: First, we applied a uniform Gaussian blur to each frame (cv2.GaussianBlur). Second, we dilated the resulting image (cv2.MORPH_OPEN). Third, we used the KNN background subtractor function to identify and sum moving foreground pixels that corresponded to the moving tail (cv2.createBackgroundSubtractorKNN). To compare movement across videos, we z-scored the tail movement of each video comparing each frame to a 15 s rolling mean. Time points with a z-score exceeding 2 were classified as motion. We averaged significant movement across fish using a 500-ms bin, and each trial was categorized as a motion trial if the mean of significance movement during stimulus presentation exceeds zero.

Statistics

Groups were tested for normality using the Shapiro–Wilk test. Non-parametric tests (Mann–Whitney U, Kruskal–Wallis H) were conducted if the Shapiro–Wilk P < 0.05 for any group. Otherwise, parametric tests were used. Post hoc tests were conducted if main effect was P < 0.05. All post hoc tests (Dunn’s test for non-parametric data, or t-tests for parametric data) were corrected for multiple comparisons with a Bonferroni or Holm correction. For nested data (neurons within imaging session within experimental groups), we determined the significance of pairwise comparisons using linear mixed-effects models, with imaging session identity (id) as a random variable. Exact tests and P values are reported in Supplementary Table 1.

Reporting summary

Further information on research design is available in the Nature Portfolio Reporting Summary linked to this article.

Data are available at https://doi.org/10.6084/m9.figshare.30223705 (ref. ⁷¹).  Source data are provided with this paper.

1. Krause, J. & Ruxton, G. D. Living in Groups (Oxford Univ. Press, 2002).

2. Sumpter, D. J. T. The principles of collective animal behaviour. Philos. Trans. R. Soc. B 361, 5–22 (2006).

   Article  CAS  Google Scholar 

3. Couzin, I. D. Collective cognition in animal groups. Trends Cogn. Sci. 13, 36–43 (2009).

   Article  PubMed  Google Scholar 

4. Couzin, I. D., Krause, J., Franks, N. R. & Levin, S. A. Effective leadership and decision-making in animal groups on the move. Nature 433, 513–516 (2005).

   Article  ADS  CAS  PubMed  Google Scholar 

5. Strandburg-Peshkin, A. et al. Visual sensory networks and effective information transfer in animal groups. Curr. Biol. 23, R709–R711 (2013).

   Article  CAS  PubMed  PubMed Central  Google Scholar 

6. Rosenthal, S. B., Twomey, C. R., Hartnett, A. T., Wu, H. S. & Couzin, I. D. Revealing the hidden networks of interaction in mobile animal groups allows prediction of complex behavioral contagion. Proc. Natl Acad. Sci. USA 112, 4690–4695 (2015).

   Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

7. Nagy, M., Akos, Z., Biro, D. & Vicsek, T. Hierarchical group dynamics in pigeon flocks. Nature 464, 890–893 (2010).

   Article  ADS  CAS  PubMed  Google Scholar 

8. Sumpter, D., Buhl, C., Biro, D. & Couzin, I. Information transfer in moving animal groups. Theory Biosci. 127, 177–186 (2008).

   Article  PubMed  Google Scholar 

9. Gallup, A. C. et al. Visual attention and the acquisition of information in human crowds. Proc. Natl Acad. Sci. USA 109, 7245–7250 (2012).

   Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

10. Bain, N. & Bartolo, D. Dynamic response and hydrodynamics of polarized crowds. Science 363, 46–49 (2019).

    Article  ADS  CAS  PubMed  Google Scholar 

11. Hein, A. M. Ecological decision-making: From circuit elements to emerging principles. Curr. Opin. Neurobiol. 74, 102551 (2022).

    Article  CAS  PubMed  Google Scholar 

12. Yu, J.-H., Napoli, J. L. & Lovett-Barron, M. Understanding collective behavior through neurobiology. Curr. Opin. Neurobiol. 86, 102866 (2024).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

13. Schulze, L. et al. Transparent Danionella translucida as a genetically tractable vertebrate brain model. Nat. Methods 15, 977–983 (2018).

    Article  CAS  PubMed  Google Scholar 

14. Britz, R., Conway, K. W. & Rüber, L. The emerging vertebrate model species for neurophysiological studies is Danionella cerebrum, new species (Teleostei: Cyprinidae). Sci. Rep. 11, 18942 (2021).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

15. Zada, D. et al. Development of neural circuits for social motion perception in schooling fish. Curr. Biol. 34, 3380–3391.e5 (2024).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

16. Isa, T., Marquez-Legorreta, E., Grillner, S. & Scott, E. K. The tectum/superior colliculus as the vertebrate solution for spatial sensory integration and action. Curr. Biol. 31, R741–R762 (2021).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

17. Knudsen, E. I. Neural circuits that mediate selective attention: a comparative perspective. Trends Neurosci. 41, 789–805 (2018).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

18. Kappel, J. M. et al. Visual recognition of social signals by a tectothalamic neural circuit. Nature 608, 146–152 (2022).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

19. MacIver, M. A. & Finlay, B. L. The neuroecology of the water-to-land transition and the evolution of the vertebrate brain. Philos. Trans. R. Soc. B 377, 20200523 (2022).

    Article  Google Scholar 

20. Procaccini, A. et al. Propagating waves in starling, Sturnus vulgaris, flocks under predation. Anim. Behav. 82, 759–765 (2011).

    Article  Google Scholar 

21. Herbert-Read, J. E., Buhl, J., Hu, F., Ward, A. J. W. & Sumpter, D. J. T. Initiation and spread of escape waves within animal groups. R. Soc. Open Sci. 2, 140355 (2015).

    Article  ADS  PubMed  PubMed Central  Google Scholar 

22. Poel, W. et al. Subcritical escape waves in schooling fish. Sci. Adv. 8, eabm6385 (2022).

    Article  PubMed  PubMed Central  Google Scholar 

23. Hein, A. M., Gil, M. A., Twomey, C. R., Couzin, I. D. & Levin, S. A. Conserved behavioral circuits govern high-speed decision-making in wild fish shoals. Proc. Natl Acad. Sci. USA. 115, 12224–12228 (2018).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

24. Ward, A. J. W., Herbert-Read, J. E., Sumpter, D. J. T. & Krause, J. Fast and accurate decisions through collective vigilance in fish shoals. Proc. Natl Acad. Sci. USA. 108, 2312–2315 (2011).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

25. Lima, S. L. Collective detection of predatory attack by social foragers: fraught with ambiguity? Anim. Behav. 50, 1097–1108 (1995).

    Article  Google Scholar 

26. Davidson, J. D. et al. Collective detection based on visual information in animal groups. J. R. Soc. Interface. 18, 20210142 (2021).

    Article  PubMed  PubMed Central  Google Scholar 

27. Pacher, K. et al. Better and faster decisions by larger fish shoals in the wild. Sci. Adv. 11, eadt8600 (2025).

    Article  ADS  PubMed  PubMed Central  Google Scholar 

28. Sten, T. H., Li, R., Otopalik, A. & Ruta, V. Sexual arousal gates visual processing during Drosophila courtship. Nature 595, 549–553 (2021).

    Article  ADS  Google Scholar 

29. Cowley, B. R. et al. Mapping model units to visual neurons reveals population code for social behaviour. Nature 629, 1100–1108 (2024).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

30. Forli, A. & Yartsev, M. M. Hippocampal representation during collective spatial behaviour in bats. Nature 621, 796–803 (2023).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

31. Akinrinade, I. et al. Evolutionarily conserved role of oxytocin in social fear contagion in zebrafish. Science 379, 1232–1237 (2023).

    Article  ADS  CAS  PubMed  Google Scholar 

32. Puri, P. et al. Collective motion in Danionella emerges from discrete copying interactions. Phys. Rev. Lett. 137, 098403 (2026).

    Article  PubMed  Google Scholar 

33. Hoffmann, M., Henninger, J., Veith, J., Richter, L. & Judkewitz, B. Blazed oblique plane microscopy reveals scale-invariant inference of brain-wide population activity. Nat. Commun. 14, 8019 (2023).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

34. Bass, A. H. & Perelmuter, J. T. Danionella fishes. Nat. Methods 21, 1767–1769 (2024).

    Article  CAS  PubMed  Google Scholar 

35. Henninger, J. et al. Brain-wide hierarchical and sexually dimorphic tuning for social vocalizations. Preprint at bioRxiv https://doi.org/10.64898/2026.03.04.709502 (2026).

36. Branco, T. & Redgrave, P. The neural basis of escape behavior in vertebrates. Annu. Rev. Neurosci. 43, 417–439 (2020).

    Article  CAS  PubMed  Google Scholar 

37. Escobedo, R. et al. A data-driven method for reconstructing and modelling social interactions in moving animal groups. Philos. Trans. R. Soc. B 375, 20190380 (2020).

    Article  CAS  Google Scholar 

38. Lin, G. et al. Experimental evidence of stress-induced critical state in schooling fish. PRX Life 3, 033018 (2025).

    Article  ADS  Google Scholar 

39. Veith, J. et al. The mechanism for directional hearing in fish. Nature 631, 118–124 (2024).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

40. Romey, W. L., Smith, A. L. & Buhl, C. Flash expansion and the repulsive herd. Anim. Behav. 110, 171–178 (2015).

    Article  Google Scholar 

41. Storms, R. F., Carere, C., Zoratto, F. & Hemelrijk, C. K. Complex patterns of collective escape in starling flocks under predation. Behav. Ecol. Sociobiol. 73, 10 (2019).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

42. Hamilton, W. D. Geometry for the selfish herd. J. Theor. Biol. 31, 295–311 (1971).

    Article  ADS  CAS  PubMed  Google Scholar 

43. Hein, A. M. et al. The evolution of distributed sensing and collective computation in animal populations. eLife 4, e10955 (2015).

    Article  PubMed  PubMed Central  Google Scholar 

44. McKee, A. & McHenry, M. J. The strategy of predator evasion in response to a visual looming stimulus in zebrafish (Danio rerio). Integr. Org. Biol. 2, obaa023 (2020).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

45. Dunn, T. W. et al. Neural circuits underlying visually evoked escapes in larval zebrafish. Neuron 89, 613–628 (2016).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

46. Fahimipour, A. K. et al. Wild animals suppress the spread of socially transmitted misinformation. Proc. Natl Acad. Sci. USA 120, e2215428120 (2023).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

47. Hemelrijk, C. K., Costanzo, A., Hildenbrandt, H. & Carere, C. Damping of waves of agitation in starling flocks. Behav. Ecol. Sociobiol. 73, 125 (2019).

    Article  Google Scholar 

48. Krauzlis, R. J., Bogadhi, A. R., Herman, J. P. & Bollimunta, A. Selective attention without a neocortex. Cortex 102, 161–175 (2018).

    Article  PubMed  Google Scholar 

49. Larsch, J. & Baier, H. Biological motion as an innate perceptual mechanism driving social affiliation. Curr. Biol. 28, 3523–3532.e4 (2018).

    Article  CAS  PubMed  Google Scholar 

50. Lifshitz, I. et al. Distinct distributed neural dynamics predict pallium-dependent social approach. Nat. Commun. 17, 4848 (2026).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

51. Solié, C. et al. Superior colliculus to VTA pathway controls orienting response and influences social interaction in mice. Nat. Commun. 13, 817 (2022).

    Article  ADS  PubMed  PubMed Central  Google Scholar 

52. Lu, X. et al. Detecting biological motion signals in human and monkey superior colliculus: a subcortical-cortical pathway for biological motion perception. Nat. Commun. 15, 9606 (2024).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

53. Anneser, L. et al. The neuropeptide Pth2 dynamically senses others via mechanosensation. Nature 588, 653–657 (2020).

    Article  ADS  CAS  PubMed  Google Scholar 

54. Penalva-Tena, A. et al. Oxytocin-mediated social preference and socially reinforced reward learning in the miniature fish Danionella cerebrum. Curr. Biol. 35, 363–372.e3 (2025).

    Article  CAS  PubMed  Google Scholar 

55. Baden, T. The vertebrate retina: a window into the evolution of computation in the brain. Curr. Opin. Behav. Sci. 57, 101391 (2024).

    Article  Google Scholar 

56. Heap, L. A. L., Vanwalleghem, G., Thompson, A. W., Favre-Bulle, I. A. & Scott, E. K. Luminance changes drive directional startle through a thalamic pathway. Neuron 99, 293–301.e4 (2018).

    Article  CAS  PubMed  Google Scholar 

57. Salay, L. D., Ishiko, N. & Huberman, A. D. A midline thalamic circuit determines reactions to visual threat. Nature 557, 183–189 (2018).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

58. Jourjine, N. & Hoekstra, H. E. Expanding evolutionary neuroscience: insights from comparing variation in behavior. Neuron 109, 1084–1099 (2021).

    Article  CAS  PubMed  Google Scholar 

59. Kadobianskyi, M., Schulze, L., Schuelke, M. & Judkewitz, B. Hybrid genome assembly and annotation of Danionella translucida. Sci. Data 6, 156 (2019).

    Article  PubMed  PubMed Central  Google Scholar 

60. Kadobianskyi, M. et al. Multimodal reference brain atlas of adult Danionella cerebrum. Preprint at bioRxiv https://doi.org/10.64898/2026.03.09.710483 (2026).

61. Atabay, K. D. et al. Whole-body single-cell atlas of an adult vertebrate in homeostasis and regeneration. Preprint at bioRxiv https://doi.org/10.64898/2026.02.03.703562 (2026).

62. Zada, D., Kadobianskyi, M., Judkewitz, B. & Lovett-Barron, M. Convergent thyroid-ATPase interactions regulate collective behavior in Danionella. Cell Rep. 45, 116730 (2025).

    Article  PubMed  PubMed Central  Google Scholar 

63. Lee, T. J. & Briggman, K. L. Visually guided and context-dependent spatial navigation in the translucent fish Danionella cerebrum. Curr. Biol. 33, 5467–5477.e4 (2023).

    Article  CAS  PubMed  Google Scholar 

64. Rajan, G. et al. Evolutionary divergence of locomotion in two related vertebrate species. Cell Rep. 38, 110585 (2022).

    Article  CAS  PubMed  Google Scholar 

65. Lopes, G. et al. Bonsai: an event-based framework for processing and controlling data streams. Front. Neuroinform. 9, 7 (2015).

    Article  PubMed  PubMed Central  Google Scholar 

66. Lopes, G. et al. Creating and controlling visual environments using BonVision. eLife 10, e65541 (2021).

    Article  ADS  CAS  PubMed  PubMed Central  Google Scholar 

67. Pereira, T. D. et al. SLEAP: A deep learning system for multi-animal pose tracking. Nat. Methods 19, 486–495 (2022).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

68. Pachitariu, M. et al. Suite2p: beyond 10,000 neurons with standard two-photon microscopy. Preprint at bioRxiv https://doi.org/10.1101/061507 (2016).

69. Stringer, C. et al. Rastermap: a discovery method for neural population recordings. Nat. Neurosci. 28, 201–212 (2025).

    Article  CAS  PubMed  Google Scholar 

70. Pachitariu, M. & Stringer, C. Cellpose 2.0: how to train your own model. Nat. Methods 19, 1634–1641 (2022).

    Article  CAS  PubMed  PubMed Central  Google Scholar 

71. Yu., J. H., Meyerhof, G. T., Milan, J., Napoli, J. L. & Lovett-Barron, M. Neuronal detection of social actions directs collective escape behavior. Figshare https://doi.org/10.6084/m9.figshare.30223705 (2026).

Download references

Authors and Affiliations

1. Department of Neurobiology, School of Biological Sciences. University of California, San Diego, La Jolla, CA, USA

   Jo-Hsien Yu  (游若嫺), Geoff T. Meyerhof, Jimjohn Milan, Julia L. Napoli & Matthew Lovett-Barron

Contributions

J.-H.Y., G.T.M. and M.L.-B. designed experiments. J.-H.Y., G.T.M. and J.M. conducted behavioural experiments. J.-H.Y. conducted imaging experiments. J.-H.Y., J.M., G.T.M., J.L.N. and M.L.-B. contributed software and analysed data. J.-H.Y. and M.L.-B. wrote the paper, with assistance from all authors.

Corresponding author

Correspondence to Matthew Lovett-Barron.

Competing interests

The authors declare no competing interests.

Peer review information

Nature thanks Ethan Scott who co-reviewed with Conrad Lee; Julie Semmelhack who co-reviewed with Peixiong Zhao; and Guy Theraulaz for their contribution to the peer review of this work. Peer reviewer reports are available.

Open Access This article is licensed under a Creative Commons Attribution 4.0 International License, which permits use, sharing, adaptation, distribution and reproduction in any medium or format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons licence, and indicate if changes were made. The images or other third party material in this article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the material. If material is not included in the article’s Creative Commons licence and your intended use is not permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from the copyright holder. To view a copy of this licence, visit http://creativecommons.org/licenses/by/4.0/.

Reprints and permissions