This study introduces a low-cost, open source, immersive Virtual Reality (VR) system designed to investigate real-time social interactions and collective behaviors in fish. Understanding collective animal behaviors, such as schooling in fish, presents significant observational challenges due to rapid and complex interactions. To address these difficulties, we developed an innovative closed-loop VR environment allowing precise control and measurement of interactions between real and virtual fish. This setup incorporates high-speed 3D tracking, real-time visual feedback, and automated data processing, creating realistic and interactive scenarios. We present experimental results, obtained with Rummy-nose tetras ( Hemigrammus rhodostomus ), showing that freely moving real fish consistently adjusted their speed, depth, and spatial positioning to follow virtual fish effectively. Fish matched moderate virtual speeds comfortably, struggled slightly at higher speeds, and actively maintained vertical positioning to sustain group cohesion. Fish also displayed significant environmental awareness, adjusting their proximity to the tank walls while maintaining visual contact with virtual fish. The implementation of this VR system provides critical insights into sensory and cognitive processes underlying collective movement and decision-making in fish. By revealing both flexibility and constraints in fish behavior, our findings underscore the potential of VR technology to study social interactions in controlled yet realistic settings. Ultimately, this VR system advances our understanding of collective animal behaviors and lays a foundation for future studies across different species, enhancing insights into the neural and behavioral mechanisms of group dynamics. ### Competing Interest Statement The authors have declared no competing interest. Agence Nationale de la Recherche, ANR-20-CE45-0006-1 Agencia Estatal de Investigación, PID2020-115088RB-I00

Bees are among the master navigators of the insect world. Despite impressive advances in robot navigation research, the performance of these insects is still unrivaled by any artificial system in terms of training efficiency and generalization capabilities, particularly considering the limited computational capacity. On the other hand, computational principles underlying these extraordinary feats are still only partially understood. The theoretical framework of reinforcement learning (RL) provides an ideal focal point to bring the two fields together for mutual benefit. In particular, we analyze and compare representations of space in robot and insect navigation models through the lens of RL, as the efficiency of insect navigation is likely rooted in an efficient and robust internal representation, linking retinotopic (egocentric) visual input with the geometry of the environment. While RL has long been at the core of robot navigation research, current computational theories of insect navigation are not commonly formulated within this framework, but largely as an associative learning process implemented in the insect brain, especially in the mushroom body (MB). Here we propose specific hypothetical components of the MB circuit that would enable the implementation of a certain class of relatively simple RL algorithms, capable of integrating distinct components of a navigation task, reminiscent of hierarchical RL models used in robot navigation. We discuss how current models of insect and robot navigation are exploring representations beyond classical, complete map-like representations, with spatial information being embedded in the respective latent representations to varying degrees.

Abstract Automated cameras (including camera traps) are an established observation tool, allowing, for example the identification of behaviours and monitoring without harming organisms. However, limitations including imperfect detection, insufficient data storage and power supply restrict the use of camera traps, making inexpensive and customizable solutions desirable. We describe a camera system and evaluation toolset based on Raspberry Pi computers and YOLOv5 that can overcome those shortcomings with its modular properties. We facilitate the set‐up and modification for researchers via detailed step‐by‐step guides. A customized camera system prototype was constructed to monitor fast‐moving organisms on a continuous schedule. For testing and benchmarking, we recorded mason bees (Osmia cornuta) approaching nesting aids on 20 sites. To efficiently process the extensive video material, we developed an evaluation toolset utilizing the convolutional neural network YOLOv5 to detect bees in the videos. In the field test, the camera system performed reliably for more than a week (2 h per day) under varying weather conditions. YOLOv5 detected and classified bees with only 775 original training images. Overall detection reliability varied depending on camera perspective, site and weather conditions, but a high average detection precision (78%) was achieved, which was confirmed by a human observer (80% of algorithm‐based detections confirmed). The customized camera system mitigates several disadvantages of commercial camera traps by using interchangeable components and incorporates all major requirements a researcher has for working in the field including moderate costs, easy assembly and an external energy source. We provide detailed user guides to bridge the gap between ecology, computer science and engineering.

Many insects can use the polarization of the skylight as a navigational cue. As shown previously, freely walking Drosophila orient along the e-vector of linearly polarized UV light presented both dorsally and ventrally. We are interested in the neuronal mechanisms leading to this behavior, and specifically how the central complex and its inputs are involved. We investigated the behavior of flies exposed to linearly polarized near-UV light (400 nm) presented dorsally. Flies walked freely in a circular, flat arena surrounded by a heat barrier. Using the GAL4-UAS genetic system, we drove the expression of the potassium inward rectifier KIR2.1 to perturb each of several different neuron types of the polarization vision pathway. Perturbing EPG compass neurons in the central complex slightly weakened average alignment and increased its variability. On the other hand, when two different GAL4 lines driving expression in the ER4m ring neurons, identified by connectomics as the major polarization inputs to the fly central complex, were perturbed, the alignment strength increased. A similar effect was observed when the inputs to ER4m, the TuBua neurons, were perturbed. We did not predict EPG and ER4m perturbations to cause opposite effects. Further investigation would be required to understand the physiological mechanisms of these contradictory behavioral effects. ### Competing Interest Statement The authors have declared no competing interest.

How do brains—biological or artificial—respond and adapt to an ever-changing environment? In a recent meeting, experts from various fields of neuroscience and artificial intelligence met to discuss internal world models in brains and machines, arguing for an interdisciplinary approach to gain deeper insights into the underlying mechanisms.

Insects have important roles globally in ecology, economy, and health, yet our understanding of their behavior remains limited. Bees, for example, use vision and a tiny brain to find flowers and return home, but understanding how they perform these impressive tasks has been hampered by limitations in recording technology. Here, we present Fast Lock-On (FLO) tracking. This method moves an image sensor to remain focused on a retroreflective marker affixed to an insect. Using paraxial infrared illumination, simple image processing can localize the sensor location of the insect in a few milliseconds. When coupled with a feedback system to steer a high-magnification optical system to remain focused on the insect, a high-spatiotemporal resolution trajectory can be gathered over a large region. As the basis for several robotic systems, we show that FLO is a versatile idea that can be used in combination with other components. We demonstrate that the optical path can be split and used for recording high-speed video. Furthermore, by flying an FLO system on a quadcopter drone, we track a flying honey bee and anticipate tracking insects in the wild over kilometer scales. Such systems have the capability to provide higher-resolution information about insects behaving in natural environments and as such will be helpful in revealing the biomechanical and neuroethological mechanisms used by insects in natural settings.

Like many other animals, insects are capable of returning to previously visited locations using path integration, which is a memory of travelled direction and distance. Recent studies suggest that Drosophila can also use path integration to return to a food reward. However, the existing experimental evidence for path integration in Drosophila has a potential confound: pheromones deposited at the site of reward might enable flies to find previously rewarding locations even without memory. Here, we show that pheromones can indeed cause naïve flies to accumulate where previous flies had been rewarded in a navigation task. Therefore, we designed an experiment to determine if flies can use path integration memory despite potential pheromonal cues by displacing the flies shortly after an optogenetic reward. We found that rewarded flies returned to the location predicted by a memory-based model. Several analyses are consistent with path integration as the mechanism by which flies returned to the reward. We conclude that although pheromones are often important in fly navigation and must be carefully controlled for in future experiments, Drosophila may indeed be capable of performing path integration.

Tracking mosquitoes in real time, as opposed to recording video files and performing the tracking step later, is useful for two reasons. The first is efficiency. Real-time tracking requires less storage because video images do not need to be saved and followed by a tracking step. The second is that tracking data can be used to interact with the animal in some way, such as triggering the approach of a looming object. In this protocol, we discuss the use of Braid, free software for performing real-time, multicamera, multianimal tracking. We describe a setup with four cameras capable of tracking the three-dimensional (3D) position of mosquitoes at 100 frames per second in a volume of 30 cm × 30 cm × 60 cm with millimeter accuracy. The specific hardware configuration is flexible and can be substituted using different or additional components to adjust the tracking parameters as needed.

RT-qPCR-based diagnostic tests play important roles in combating virus-caused pandemics such as Covid-19. However, their dependence on sophisticated equipment and the associated costs often limits their widespread use. Loop-mediated isothermal amplification after reverse transcription (RT-LAMP) is an alternative nucleic acid detection method that overcomes these limitations. Here, we present a rapid, robust, and sensitive RT-LAMP-based SARS-CoV-2 detection assay. Our 40-min procedure bypasses the RNA isolation step, is insensitive to carryover contamination, and uses a colorimetric readout that enables robust SARS-CoV-2 detection from various sample types. Based on this assay, we have increased sensitivity and scalability by adding a nucleic acid enrichment step (Bead-LAMP), developed a version for home testing (HomeDip-LAMP), and identified open-source RT-LAMP enzymes that can be produced in any molecular biology laboratory. On a dedicated website, rtlamp.org (DOI: 10.5281/zenodo.6033689), we provide detailed protocols and videos. Our optimized, general-purpose RT-LAMP assay is an important step toward population-scale SARS-CoV-2 testing.

Mosquitoes track odors, locate hosts, and find mates visually. The color of a food resource, such as a flower or warm-blooded host, can be dominated by long wavelengths of the visible light spectrum (green to red for humans) and is likely important for object recognition and localization. However, little is known about the hues that attract mosquitoes or how odor affects mosquito visual search behaviors. We use a real-time 3D tracking system and wind tunnel that allows careful control of the olfactory and visual environment to quantify the behavior of more than 1.3 million mosquito trajectories. We find that CO 2 induces a strong attraction to specific spectral bands, including those that humans perceive as cyan, orange, and red. Sensitivity to orange and red correlates with mosquitoes’ strong attraction to the color spectrum of human skin, which is dominated by these wavelengths. The attraction is eliminated by filtering the orange and red bands from the skin color spectrum and by introducing mutations targeting specific long-wavelength opsins or CO 2 detection. Collectively, our results show that odor is critical for mosquitoes’ wavelength preferences and that the mosquito visual system is a promising target for inhibiting their attraction to human hosts.

Flying insects have evolved the ability to evade looming objects, such as predators and swatting hands. This is particularly relevant for blood-feeding insects, such as mosquitoes that routinely need to evade the defensive actions of their blood hosts. To minimize the chance of being swatted, a mosquito can use two distinct strategies-continuously exhibiting an unpredictable flight path or maximizing its escape maneuverability. We studied how baseline flight unpredictability and escape maneuverability affect the escape performance of day-active and night-active mosquitoes (Aedes aegypti and Anopheles coluzzii, respectively). We used a multi-camera high-speed videography system to track how freely flying mosquitoes respond to an event-triggered rapidly approaching mechanical swatter, in four different light conditions ranging from pitch darkness to overcast daylight. Results show that both species exhibit enhanced escape performance in their natural blood-feeding light condition (daylight for Aedes and dark for Anopheles). To achieve this, they show strikingly different behaviors. The enhanced escape performance of Anopheles at night is explained by their increased baseline unpredictable erratic flight behavior, whereas the increased escape performance of Aedes in overcast daylight is due to their enhanced escape maneuvers. This shows that both day and night-active mosquitoes modify their flight behavior in response to light intensity such that their escape performance is maximum in their natural blood-feeding light conditions, when these defensive actions by their blood hosts occur most. Because Aedes and Anopheles mosquitoes are major vectors of several deadly human diseases, this knowledge can be used to optimize vector control methods for these specific species.

We introduce a new approach to reduce uncorrelated background signals from fluorescence imaging data, using real-time subtraction of background light. Our method allows removing image artifacts that cause signal clipping in a conventional imaging setup.

Article Figures and data Abstract Introduction Results Discussion Materials and methods Data availability References Decision letter Author response Article and author information Metrics Abstract Many animals have large visual fields, and sensory circuits may sample those regions of visual space most relevant to behaviours such as gaze stabilisation and hunting. Despite this, relatively small displays are often used in vision neuroscience. To sample stimulus locations across most of the visual field, we built a spherical stimulus arena with 14,848 independently controllable LEDs. We measured the optokinetic response gain of immobilised zebrafish larvae to stimuli of different steradian size and visual field locations. We find that the two eyes are less yoked than previously thought and that spatial frequency tuning is similar across visual field positions. However, zebrafish react most strongly to lateral, nearly equatorial stimuli, consistent with previously reported spatial densities of red, green, and blue photoreceptors. Upside-down experiments suggest further extra-retinal processing. Our results demonstrate that motion vision circuits in zebrafish are anisotropic, and preferentially monitor areas with putative behavioural relevance. Introduction The layout of the retina, and the visual system as a whole, evolved to serve specific behavioural tasks that animals perform to survive in their respective habitats. A well-known example is the position of the eyes in the head which varies between hunting animals (frontal eyes) and animals that frequently need to avoid predation (lateral eyes) (Cronin et al., 2014). Hunting animals keep the prey within particular visual field regions to maximise behavioural performance (Hoy et al., 2016; Bianco et al., 2011; Yoshimatsu et al., 2020). To avoid predation, however, it is useful to observe a large proportion of visual space, especially those regions in which predators are most likely to appear (Smolka et al., 2011; Zhang et al., 2012). The ecological significance of visual stimuli thus depends on their location within the visual field, and it is paralleled by non-uniform processing channels across the retina. This non-uniformity manifests as an area centralis or a fovea in many species, which is a region of heightened photoreceptor density in the central retina and serves to increase visual performance in the corresponding visual field regions. Photoreceptor densities put a direct physical limit on performance parameters such as spatial resolution (Haug et al., 2010; Merigan and Katz, 1990). In addition to these restrictions mediated by the peripheral sensory circuitry, an animal's use of certain visual field regions is also affected by behaviour-specific neural pathways, for example pathways dedicated to feeding and stabilisation behaviour. The retinal and extra-retinal circuit anisotropies can in turn effect a dependence of behavioural performance on visual field location (Hoy et al., 2016; Bianco et al., 2011; Murasugi and Howard, 1989; Shimizu et al., 2010; Yang and Guo, 2013; Zimmermann et al., 2018; Baden et al., 2013). Investigating behavioural performance limits and non-uniformities can offer insights into the processing capabilities and ecological adaptations of animal brains, especially if they can be studied and quantitatively understood at each processing step. The larval zebrafish is a promising vertebrate organism for such an endeavour, since its brain is small and a wide array of experimental techniques are available (Baier and Scott, 2009; McLean and Fetcho, 2011). Zebrafish are lateral-eyed animals and have a large visual field, which increases during early development and in individual 4-day-old larvae has been reported to cover about 163° per eye (Easter and Nicola, 1996), although little is known about interindividual variability. Their retina contains four different cone photoreceptor types (Nawrocki et al., 1985), each distributed differently across the retina. UV photoreceptors are densest in the ventro-temporal retina (area temporalis ventralis), whereas the red, green, and blue photoreceptors cover more central retinal regions (Zimmermann et al., 2018). Zebrafish larvae perform a wide range of visually mediated behaviours, ranging from prey capture (Trivedi and Bollmann, 2013; Mearns et al., 2020) and escape behaviour (Heap et al., 2018) to stabilisation behaviour (Kubo et al., 2014; Orger et al., 2008); however, the importance of stimulus location within the visual field for the execution of the respective behaviours has only recently been recognized and is still not well understood (Hoy et al., 2016; Zimmermann et al., 2018; Mearns et al., 2020; Kist and Portugues, 2019; Wang et al., 2020; Johnson et al., 2020; Lagogiannis et al., 2020). During visually mediated stabilisation behaviours, such as optokinetic and optomotor responses, animals move their eyes and bodies, respectively, in order to stabilise the retinal image and/or the body position relative to the visual surround. The optokinetic response (OKR) consists of reflexively executed stereotypical eye movements, in which phases of stimulus 'tracking' (slow phase) are interrupted by quick phases (Figure 1a). In the quick phases, eye position is reset by a saccade in the direction opposite to stimulus motion. In humans, optokinetic responses are strongest in the central visual field (Howard and Ohmi, 1984). Furthermore, lower visual field locations of the stimulus evoke stronger OKR than upper visual field locations, which likely represents an adaptation to the rich optic flow information available from the structures on the ground in the natural environments of primates (Murasugi and Howard, 1989; Hafed and Chen, 2016). Figure 1 with 2 supplements see all Download asset Open asset Presenting visual stimuli across the visual field. (a) When presented with a horizontal moving stimulus pattern, zebrafish larvae exhibit optokinetic response (OKR) behaviour, where eye movements track stimulus motion to minimise retinal slip. Its slow phase is interrupted by intermittent saccades, and even if only one eye is stimulated (solid arrow), the contralateral eye is indirectly yoked to move along (dashed arrow). (b) Often, experiments on visuomotor behaviour such as OKR sample only a small part of the visual field, whether horizontally or vertically. As different spatial directions may carry different behavioural importance, an ideal stimulation setup should cover all or most of the animal's visual field. For zebrafish larvae, this visual field can be represented by an almost complete unit sphere. (c) We arranged 232 LED tiles with 64 LEDs each across a spherical arena, such that 14,484 LEDs (green dots) covered nearly the entire visual field. (d) The same individual positions, shown in geographic coordinates. Each circle represents a single LED. Each cohesive group of eight-by-eight circles corresponds to the 64 LEDs contained in a single tile. (e) To identify LED and stimulus locations, we use Up-East-North geographic coordinates: Azimuth α describes the horizontal angle, which is zero in front of the animal and, when seen from above, increases for rightward position. Elevation β refers to the vertical angle, which is zero throughout the plane containing the animal, and positive above. (f) The spherical arena is covered in flat square tiles carrying 64 green LEDs each. (g) Its structural backbone is made of a 3D-printed keel and ribs. Left and right hemispheres were constructed as separate units. (h) Across 85-90% of the visual field, we can then present horizontally moving bar patterns of different location, frequency and size to evoke OKR. Figure 1—source data 1 SCAD files for 3D-printing the arena scaffold. https://cdn.elifesciences.org/articles/63355/elife-63355-fig1-data1-v2.zip Download elife-63355-fig1-data1-v2.zip In zebrafish larvae, OKR behaviour has been used extensively to assess visual function in genetically altered animals (Brockerhoff et al., 1995; Muto et al., 2005; Neuhauss et al., 1999). OKR tuning to velocity, frequency, and contrast of grating stimuli has been measured (Clark, 1981; Cohen et al., 1977; Huang and Neuhauss, 2008; Rinner et al., 2005), and, more recently, zebrafish are used as a model for investigating vertebrate sensorimotor transformations (Kubo et al., 2014; Portugues et al., 2014). While zebrafish can distinguish rotational from translational optic flow to evoke appropriate optokinetic and optomotor responses (Kubo et al., 2014; Naumann et al., 2016; Wang et al., 2019), it is still unclear which regions of the visual field zebrafish preferentially observe in these behaviours. The aquatic lifestyle, in combination with the preferred swimming depths (Lindsey et al., 2010), might cause the lower visual field to contain less relevant information when compared to terrestrial animals. This in turn might result in behavioural biases to other – more informative – visual field regions. A corresponding systematic behavioural quantification in zebrafish or other aquatic species, which would relate OKR behaviour to naturally occurring motion statistics and the underlying neuronal representations in retina and retino-recipient brain structures, has been prevented by technical limitations. Specifically, little is known about (i) the dependence of OKR gain on stimulus location or (ii) on stimulus sizes, (iii) possible interactions between stimulus location, size and frequency, (iv) putative asymmetries between the left and right hemispheres of the visual field, and (v) the relationship between a putative dependence of OKR on stimulus location and zebrafish retinal architecture. In other species with large visual fields, such as Drosophila, full-surround stimulation setups have been designed and used successfully (Reiser and Dickinson, 2008; Kim et al., 2017; Maisak et al., 2013), but to date, none has been used for fish. This is at least partly due to their aquatic environment and the associated difficulties regarding the refraction of stimulus light at the air-water interface. Such distortions of shape can be partially compensated by pre-emptively altering the shape of the stimulus. However, using regular computer screens or video projection, the resulting luminance profiles remain anisotropic unless compensated, potentially biasing the response toward brighter locations. Additionally, most stimulus arenas cannot easily be combined with the recording of neural activity, for example, via calcium imaging, as stimulus light and calcium fluorescence overlap in both the spectral and time domains. These challenges must be overcome to enable full-field visual stimulation in zebrafish neurophysiology experiments (Figure 1b). At least one existing solution for stimulating and tracking freely moving zebrafish supports unusually large stimuli, although despite its versatility, it still only covers part of the visual field and does not address the remaining issues of total internal reflection and the interoperability with laser-scanning microscopes (Stowers et al., 2017 ). Here, we present a novel visual stimulus arena for aquatic animals, which covers almost the entire surround of the animal, and use it to characterise the anisotropy of the zebrafish OKR across different visual field locations as well as the tuning to stimulus size, spatial frequency and leftside versus rightside stimulus locations. We find that the OKR is mostly symmetric across both eyes and driven most strongly by lateral stimulus locations. These stimulus locations approximately correspond to a retinal region of increased photoreceptor density. By rotating the experimental setup and/or the animal, our control experiments revealed that additional extra-retinal determinants of OKR drive exist as well. Our characterisation of OKR drive across the visual field will help inform bottom-up models of the vertebrate neural pathways underlying the optokinetic response and other visual behaviour. Results Spherical LED arena allows presentation of stimuli across the visual field By combining 3D printing with electronic solutions developed in Drosophila vision research, we constructed a spherical stimulus arena containing 14,848 individual LEDs covering over 90% of the visual field of zebrafish larvae (Figure 1c, Materials and methods section on coverage). Using infrared illumination via an optical pathway coupled into the sphere (Figure 1—figure supplement 1a–b), we tracked eye movements of larval zebrafish during presentation of visual stimuli (Dehmelt et al., 2018). To avoid stimulus aberrations at the air-to-water interface, we designed a nearly spherical glass bulb containing fish and medium (Figure 1—figure supplement 1c–d). With this design, stimulus light from the surrounding arena is virtually not refracted (light is orthogonal to the air-to-water interface) and reaches the eyes of the zebrafish larva in a straight line. Thus, no geometric corrections are required during stimulus design (Source code 1), and stimulus luminance is expected to be nearly isotropic across the visual field. We additionally designed the setup to minimise visual obstruction and developed a new embedding technique to immobilise the larva at the tip of a narrow glass triangle (see Materials and methods). In almost all possible positions, fish can thus perceive stimuli without interference (Wang et al., 2021). The distance between most of the adjacent LED pairs is smaller than the photoreceptor spacing in the larval retina (Haug et al., 2010; Tappeiner et al., 2012), resulting in a good spatial resolution across the majority of the spherical arena surface (Figure 1—figure supplement 2, see Materials and methods section on resolution). As flat square LED tiles cannot be perfectly arranged on a spherical surface (Supplementary file 1A), small triangular gaps are unavoidable. More importantly, several gaps in LED coverage, resulting from structural elements of the arena, were restricted mainly to the back, the top, and bottom of the animal. The 'keel' behind and in front of the fish supports the horizontal 'ribs', and the circular openings in the top and bottom accommodate the optical path for eye tracking or scanning microscopy (also see discussion of arena geometry in our data repository [Dehmelt et al., 2020; https://gin.g-node.org/Arrenberg_Lab/spherical_arena]). Stimulus position dependence of the optokinetic response Horizontally moving vertical bars reliably elicit OKR in zebrafish larvae (Beck et al., 2004). We used a stimulus which rotated clock- and counter-clockwise with a sinusoidal velocity pattern (velocity amplitude 12.5 degree/s, frequency of the velocity envelope 0.1 Hz, spatial frequency 0.06 cycles/degree, Figure 2a). OKR performance was calculated by measuring the amplitude of the resulting OKR slow-phase eye movements after the saccades had been removed (Figure 2b, Figure 2—figure supplement 1, Source code 1C, Materials and methods). The OKR gain then corresponds to the speed of the slow-phase eye movements divided by the speed of the stimulus (which is equivalent to the ratio of the eye position and stimulus position amplitudes). In addition to traditional full-field stimulation, our arena can display much smaller stimuli in different parts of the visual field. These novel stimuli evoked reliable OKR even at remote stimulus locations (Figure 2c, Figure 2—figure supplement 2), and thus allowed us to investigate previously inaccessible behavioural parameters. We excluded any trials from data analysis that showed other behaviours (e.g. drifts and ongoing spontaneous eye movements) superimposed on OKR (Figure 2—figure supplement 2). In addition to the characteristic eye traces (Figure 2c), Bode plots of the magnitude and phase shift relative to the stimulus qualitatively match previously reported zebrafish response to full-field stimulation (Figure 5—figure supplement 1, Beck et al., 2004), further confirming that the behaviour we observe is indeed OKR. Figure 2 with 2 supplements see all Download asset Open asset OKR gain is inferred from a piece-wise fit to the slow phase of tracked eye movements. (a) We present a single pattern of horizontally moving bars to evoke OKR and crop it by superimposing a permanently dark area of arbitrary size and shape (left). Its velocity follows a sinusoidal time course, repeating every 10 s for a total of 100 s for each stimulus phase (right). (b) OKR gain is the amplitude of eye movement (green trace) relative to the amplitude of the sinusoidal stimulus (grey trace). The OKR gain is often well below 1, e.g. for high stimulus velocities as used here (up to 12.5°/s). (c) Even small stimuli are sufficient to elicit reliable OKR, although gains are low if stimuli appear in a disfavoured part of the visual field. Shown here are responses to a whole-field stimulus (top) and to a disk-shaped stimulus in the extreme upper rear (bottom, Figure 2—figure supplement 2). To quantify position tuning, we cropped the presented gratings (Figure 2a) to a disk-shaped area of constant size, centred on one of 38 nearly equidistant parts of the visual field (Figure 3a, Supplementary File 1B, Video 1, Figure 3—video 1). The distribution of positions was symmetric between the left and right, upper and lower, as well as front and rear hemispheres, with some stimuli falling right on the edge between two hemispheres. As permanent asymmetries in a stimulus arena or in its surroundings could affect OKR gain, we repeated our experiments in a second group of larvae after rotating the arena by 180 degrees (Figure 4a–b), then matched the resulting pairs of OKR gains during data analysis (see Materials and methods, Figure 4e–f). Any remaining asymmetries in the OKR distributions should result from biological lateralisation, if indeed there exists such consistent lateralisation across individuals. Figure 3 with 6 supplements see all Download asset Open asset OKR gain depends on stimulus location. (a) The stimulus is cropped to a disk-shaped area 40 degrees in diameter, centred on one of 38 nearly equidistant locations (Supplementary file 1B) across the entire visual field (left), to yield 38 individual stimuli (right). (b–d) Dots reveal the location of stimulus centres D1-D38. Their colour indicates the average OKR gain across individuals and trials, corrected for external asymmetries. Surface colour of the sphere displays the discretely sampled OKR data filtered with a von Mises-Fisher kernel, in a logarithmic colour scale. Top row: OKR gain of the left eye (b), right eye (d), and the merged data including only direct stimulation of either eye (c), shown from an oblique, rostrodorsal angle. Bottom row: same, but shown directly from the front. OKR gain is significantly higher for lateral stimulus locations and lower across the rest of the visual field. The spatial distribution of OKR gains is well explained by the bimodal sum of two von-Mises Fisher distributions. (e) Mercator projections of OKR gain data shown in panels (b–d). White and grey outlines indicate the area covered by each stimulus type. Numbers indicate average gain values for stimuli centred on this location. Red dots show mean eye position during stimulation. Dashed outline and white shading on panels (b, d, e) indicate indirect stimulation via yoking, that is, stimuli not directly visible to either the left or right eye. Data from n = 7 fish for the original configuration and n = 5 fish for the rotated arena. Figure 3—source data 1 Numerical data and graphical elements of Figure 3b–d. https://cdn.elifesciences.org/articles/63355/elife-63355-fig3-data1-v2.mat Download elife-63355-fig3-data1-v2.mat Figure 3—source data 2 Numerical data and graphical elements of Figure 3e. https://cdn.elifesciences.org/articles/63355/elife-63355-fig3-data2-v2.mat Download elife-63355-fig3-data2-v2.mat Figure 4 with 3 supplements see all Download asset Open asset The OKR is biased towards upper environmental elevations irrespective of fish orientation. (a) Regular arena setup. (b) Arena can be tilted 180 degrees so front and rear, upper and lower LED positions are swapped. The bulb holder moves accordingly, so from the perspective of the fish, left and right, upper and lower LEDs are swapped. (c) Upside-down embedding setup. (d) Setup with inverted optical path, including illumination. (e–h) Results in body-centred coordinates, where positive elevations refer to dorsal positions, for the four setups shown in (a–d). As in Figure 3b–e, colour indicates the discretely sampled OKR data filtered with a von Mises-Fisher kernel and follows a logarithmic colour scale. (g,h) Experiments with presentation of a less regularly distributed set of stimuli, cropped to disks of 64 degrees polar angle instead of the 40 degrees used in (a,b,e,f). (g) Fish embedded upside-down exhibit a slight preference for stimuli below the body-centred equator, that is, positions slightly ventral to their body axis. (h) Fish embedded upright, as in (a). To account for environmental asymmetries such as arena anisotropies, we combined the data underlying (e) and (f) to obtain Figure 3b–e (see Materials and methods). Data from (e) n = 7, (f) n = 5, (g) n = 3, (h) n = 10 fish. Figure 4—source data 1 Numerical data and graphical elements of Figure 4e. https://cdn.elifesciences.org/articles/63355/elife-63355-fig4-data1-v2.mat Download elife-63355-fig4-data1-v2.mat Figure 4—source data 2 Numerical data and graphical elements of Figure 4f. https://cdn.elifesciences.org/articles/63355/elife-63355-fig4-data2-v2.mat Download elife-63355-fig4-data2-v2.mat Figure 4—source data 3 Numerical data and graphical elements of Figure 4g. https://cdn.elifesciences.org/articles/63355/elife-63355-fig4-data3-v2.mat Download elife-63355-fig4-data3-v2.mat Figure 4—source data 4 Numerical data and graphical elements of Figure 4h. https://cdn.elifesciences.org/articles/63355/elife-63355-fig4-data4-v2.mat Download elife-63355-fig4-data4-v2.mat Video 1 Download asset This video cannot be played in place because your browser does support HTML5 video. You may still download the video for offline viewing. Download as MPEG-4 Download as WebM Download as Ogg Repulsion-based algorithm to numerically distribute stimulus centres across a sphere surface (Source code 1B), with gradual convergence on a logarithmic timescale (blue to green). A subset of stimulus centres is held on the equator at zero elevation (orange). To overcome our spatially discrete sampling, we then fit our data with a bimodal function comprised of two Gaussian-like two-dimensional distributions on the stimulus sphere surface (see Materials and methods, Source code 1D), to determine the location of highest OKR gain evoked by ipsilateral stimuli and contralateral stimuli, respectively. We observed significantly higher OKR gains in response to nearly lateral stimuli, and lower gains across the rest of the visual field (Figure 3b–e). OKR was strongest for stimuli near an azimuth of −82.5 degrees and an elevation of 5.1 degrees for the left side (in body-centred coordinates), as well as 81.7 and 1.6 degrees for the right side – slightly rostral of the lateral meridian, and slightly above the equator. In the nasal visual field (binocular overlap) the OKR gain was relatively high, but still lower than for lateral visual field locations. Note that due to the fast stimulus speeds, the absolute slow phase eye velocities were high, while the OKR gain was relatively low. We chose such high stimulus speeds to minimise the experimental recording time needed to obtain reliable OKR measurements for each visual field location. As our stimulus arena is not completely covered by LEDs (Figure 1c, Figure 1d), some areas remain permanently dark. These could interfere with the perception of stimuli presented on adjacent LEDs. This is especially relevant as LED coverage is almost perfect for some stimulus positions (near the equator), whereas the size of triangular holes increases at others (towards the poles). We thus performed control experiments comparing the OKR gain evoked by a stimulus in a densely covered part of the arena to the OKR gain evoked by same stimulus, but in the presence of additional dark triangular patches (Figure 3—figure supplement 1a). We found no significant difference in OKR gain (Figure 3—figure supplement 1c, t-test, p>0.05). Additionally, we performed another series of control experiments using a dark shape mimicking the dark structural elements, the front 'keel' of the arena (Figure 3—figure supplement 1b). Again, we found no difference in OKR gain (Figure 3—figure supplement 1c, t-test, p>0.05), and thus ruled out that position dependence data was corrupted by incomplete LED coverage. Since the eyes were moving freely in our experiments, the range of eye positions during OKR, or so-called beating field (Schaerer and Kirschfeld, 2000), could have changed with stimulus position. We found that animals instead maintained similar median horizontal eye positions (e.g. left eye: −83.7 ± 1.8 degrees, right eye: 80.3 ± 1.9 degrees, average median ± standard deviation of medians, n = 7 fish, Figure 3—figure supplement 2) even for the most peripheral stimulus positions. A priori, it is unclear whether the sampling preference originates from the peculiarities of the sensory periphery in the eye, or the behavioural relevance inferred by central brain processing. The former would establish stimulus preference based on its position relative to the eye and, by extension, its representation on specific parts of the retina (i.e. an eye-centered reference frame). The latter would establish stimulus preference in body- or head-centered reference frames or based on stimulus positions relative to environment (a world-centered reference frame). A world-centered reference frame is useful, if stimuli in different environmental locations have different behavioural relevance (such as a predator approaching from the water surface). To start distinguishing these possible effects in the context of OKR, as well as to reveal any stimulus asymmetries accidentally introduced during the experiment, we performed control experiments with larvae embedded upside-down (i.e. with their dorsum towards the lower pole of the arena, Figure 4c). Note that such an upside-down state can occur when the fish loses balance in natural behaviour and would normally be counteracted by the animals' self-righting reflex. As a result of upside-down embedding, world-centred and fish-centred coordinate systems were no longer identical, in that 'up' in one is 'down' in the other, and 'left' becomes 'right'. See Supplementary file 1C for a summary of these changes across experiments, and Materials and methods for notes on head- and retina-centred coordinate systems. To facilitate comparisons across embedding types, all positions from here onwards are given relative to the visual field, and thus in fish-centred coordinates. Unexpectedly, the elevation of highest OKR gains relative to the fish changed from slightly above to slightly below the equator of the visual field when comparing upright to inverted fish (Figure 4e, Figure 4g): When upright, azimuths and fish-centred elevations of the peaks of the best fits to data were −78.7° and 8.2° for the left eye, as well as 81.8° and 3.1° for the right eye. When inverted, −88.0° and 1.6° for the left eye, as well as 82.8° and −15.5° for the right eye. These numbers were obtained from the gains of those eyes to which any given stimulus was directly visible. The results from Figure 4e–f were combined to correct Figure 3 for external asymmetries (Materials and methods); this is why the best-fit position reported for Figure 4e alone differs slightly from that reported above for Figure 3b–e. Because the set of visual stimuli presented to inverted fish stemmed from an earlier stimulus protocol with less even sampling of the visual field, a slight scaling of azimuths and elevations is expected. The consistent reduction of the elevation, however, is not. We performed a permutation test in which embedding-direction labels were randomly swapped while stimulus-location labels were maintained, and the Gaussian-type fit to data was then repeated on each permuted dataset. This test confirmed that fish preferred upward (in the environmental reference frame) rather than dorsalward elevations (p<0.05, Source code 1E, Figure 4—figure supplement 3). While the fit peaks capture its centres of mass, the apparent maxima of the OKR spatial distribution are even further apart, with their elevation flipping signs from the dorsal to the ventral hemisphere (shown in Figure 4e,g). Adjustment by the fish of its vertical resting eye position between the upright and inverted body positions would have been a simple potential explanation for this result. However, time-lapse frontal microscopy images (Materials and methods) ruled this out, since for both upside-up and upside-down embedding the eyes were inclined by an average of about four degrees towards the dorsum (3.5 ± 1.0° for the left eye, 4.9 ± 0.8° for the right eye, mean ± s.e.m., Figure 3—figure supplement 3). We also tested the influence of camera and infrared light (840 nm) positions (Figure 4d) – which in either case should have been invisible to the fish (Shcherbakov et al., 2013) – and found that they could indeed not explain the observed differences (Figure 4h). In summary, the body-centred preferred location only flipped from slightly dorsal to slightly ventral in upside-down embedded fish (Figure 4g), and thus remained virtually unchanged in environmental coordinates across all control experiments. Eye-, head-, or body-centred reference frames are therefor

Digital photography and videography provide rich data for the study of animal behavior and are consequently widely used techniques. For fixed, unmoving cameras there is a resolution versus field-of-view tradeoff and motion blur smears the subject on the sensor during exposure. While these fundamental tradeoffs with stationary cameras can be sidestepped by employing multiple cameras and providing additional illumination, this may not always be desirable. An alternative that overcomes these issues of stationary cameras is to direct a high-magnification camera at an animal continually as it moves. Here, we review systems in which automatic tracking is used to maintain an animal in the working volume of a moving optical path. Such methods provide an opportunity to escape the tradeoff between resolution and field of view and also to reduce motion blur while still enabling automated image acquisition. We argue that further development will be useful and outline potential innovations that may improve the technology and lead to more widespread use.