Mapping model units to visual neurons reveals population code for social behaviour.

The rich variety of behaviours observed in animals arises through the interplay between sensory processing and motor control. To understand these sensorimotor transformations, it is useful to build models that predict not only neural responses to sensory input1-5 but also how each neuron causally contributes to behaviour6,7. Here we demonstrate a novel modelling approach to identify a one-to-one mapping between internal units in a deep neural network and real neurons by predicting the behavioural changes that arise from systematic perturbations of more than a dozen neuronal cell types. A key ingredient that we introduce is 'knockout training', which involves perturbing the network during training to match the perturbations of the real neurons during behavioural experiments. We apply this approach to model the sensorimotor transformations of Drosophila melanogaster males during a complex, visually guided social behaviour8-11. The visual projection neurons at the interface between the optic lobe and central brain form a set of discrete channels12, and prior work indicates that each channel encodes a specific visual feature to drive a particular behaviour13,14. Our model reaches a different conclusion: combinations of visual projection neurons, including those involved in non-social behaviours, drive male interactions with the female, forming a rich population code for behaviour. Overall, our framework consolidates behavioural effects elicited from various neural perturbations into a single, unified model, providing a map from stimulus to neuronal cell type to behaviour, and enabling future incorporation of wiring diagrams of the brain15 into the model.

Hue selectivity from recurrent circuitry in Drosophila.

In the perception of color, wavelengths of light reflected off objects are transformed into the derived quantities of brightness, saturation and hue. Neurons responding selectively to hue have been reported in primate cortex, but it is unknown how their narrow tuning in color space is produced by upstream circuit mechanisms. We report the discovery of neurons in the Drosophila optic lobe with hue-selective properties, which enables circuit-level analysis of color processing. From our analysis of an electron microscopy volume of a whole Drosophila brain, we construct a connectomics-constrained circuit model that accounts for this hue selectivity. Our model predicts that recurrent connections in the circuit are critical for generating hue selectivity. Experiments using genetic manipulations to perturb recurrence in adult flies confirm this prediction. Our findings reveal a circuit basis for hue selectivity in color vision.

Dissection of the Head of a Live Fly for Functional Imaging.

In this protocol, we outline procedures to mount the fly and to open up the head cuticle to expose the optic lobes for in vivo imaging. The fly is first inserted into a custom-made fly chamber in which the fly's head is stabilized on a piece of aluminum foil. Once the fly is mounted in the chamber, its head cuticle is removed, exposing the optic lobe for recording. The brain tissues (above the foil), including the optic lobes, should be bathed in fly saline. Meanwhile, the eyes (below the foil) are kept dry to receive light stimuli during the recording. A considerable level of expertise and hand dexterity is required to handle a small animal such as a fly, especially when opening its head capsule without damaging the brain tissue. This expertise should be gained through mindful repetition of the protocol. With appropriate preparation and skills, the success rate for this procedure can be >95%. Using this protocol, it is possible to record ultraviolet (UV)-sensing photoreceptors, which have long visual fibers that terminate at the medulla (the second optic neuropil). Depending on the visual neurons of interest, some modifications to fly mounting might be needed.

Neural organization of the third optic neuropil, the lobula, in the highly visual semiterrestrial crab Neohelice granulata.

The visual neuropils (lamina, medulla, and lobula complex) of malacostracan crustaceans and hexapods have many organizational principles, cell types, and functional properties in common. Information about the cellular elements that compose the crustacean lobula is scarce especially when focusing on small columnar cells. Semiterrestrial crabs possess a highly developed visual system and display conspicuous visually guided behaviors. In particular, Neohelice granulata has been previously used to describe the cellular components of the first two optic neuropils using Golgi impregnation technique. Here, we present a comprehensive description of individual elements composing the third optic neuropil, the lobula, of that same species. We characterized a wide variety of elements (140 types) including input terminals and lobula columnar, centrifugal, and input columnar elements. Results reveal a very dense and complex neuropil. We found a frequently impregnated input element (suggesting a supernumerary cartridge representation) that arborizes in the third layer of the lobula and that presents four variants each with ramifications organized following one of the four cardinal axes suggesting a role in directional processing. We also describe input elements with two neurites branching in the third layer, probably connecting with the medulla and lobula plate. These facts suggest that this layer is involved in the directional motion detection pathway in crabs. We analyze and discuss our findings considering the similarities and differences found between the layered organization and components of this crustacean lobula and the lobula of insects.

The pigment-dispersing factor neuronal network systematically grows in developing honey bees.

The neuropeptide pigment-dispersing factor (PDF) plays a prominent role in the circadian clock of many insects including honey bees. In the honey bee brain, PDF is expressed in about 15 clock neurons per hemisphere that lie between the central brain and the optic lobes. As in other insects, the bee PDF neurons form wide arborizations in the brain, but certain differences are evident. For example, they arborize only sparsely in the accessory medulla (AME), which serves as important communication center of the circadian clock in cockroaches and flies. Furthermore, all bee PDF neurons cluster together, which makes it impossible to distinguish individual projections. Here, we investigated the developing bee PDF network and found that the first three PDF neurons arise in the third larval instar and form a dense network of varicose fibers at the base of the developing medulla that strongly resembles the AME of hemimetabolous insects. In addition, they send faint fibers toward the lateral superior protocerebrum. In last larval instar, PDF cells with larger somata appear and send fibers toward the distal medulla and the medial protocerebrum. In the dorsal part of the medulla serpentine layer, a small PDF knot evolves from which PDF fibers extend ventrally. This knot disappears during metamorphosis and the varicose arborizations in the putative AME become fainter. Instead, a new strongly stained PDF fiber hub appears in front of the lobula. Simultaneously, the number of PDF neurons increases and the PDF neuronal network in the brain gets continuously more complex.

Functional analysis of Pdp1 and vrille in the circadian system of a cricket.

Circadian rhythms are generated by a circadian clock for which oscillations are based on the rhythmic expression of the so-called clock genes. The present study investigated the role of Gryllus bimaculatus vrille (Gb'vri) and Par domain protein 1 (Gb'Pdp1) in the circadian clock of the cricket Gryllus bimaculatus. Structural analysis of Gb'vri and Gb'Pdp1 cDNAs revealed that they are a member of the bZIP transcription factors. Under light/dark cycles (LD) both genes were rhythmically expressed in the clock tissue, the optic lobes, whereas the rhythm diminished under constant darkness (DD). Gb'vri and Gb'Pdp1 mRNA levels were significantly reduced by RNA interference (RNAi) of Gb'Clk and Gb'cyc, suggesting they are controlled by Gb'CLK/Gb'CYC. RNAi of Gb'vri and Gb'Pdp1 had little effect on locomotor rhythms, although their effects became visible when treated together with Gb'cycRNAi. The average free-running period of Gb'vriRNAi/Gb'cycRNAi crickets was significantly shorter than that of Gb'cycRNAi crickets. A similar period shortening was observed also when treated with Gb'Pdp1RNAi/Gb'cycRNAi. Some Gb'Pdp1RNAi/Gb'cycRNAi crickets showed rhythm splitting into two free-running components with different periods. Gb'vriRNAi and Gb'Pdp1RNAi treatments significantly altered the expression of Gb'Clk, Gb'cyc, and Gb'tim in LD. These results suggest that Gb'vri and Gb'Pdp1 play important roles in cricket circadian clocks.

The optic lobe-pars intercerebralis axis is involved in circa'bi'dian rhythm of the large black chafer Holotrichia parallela.

The large black chafer Holotrichia parallela exhibits ~ 48-h circa'bi'dian rhythm. Although circabidian rhythm is suggested to involve the circadian clock, no physiological studies have been conducted to verify this involvement. We examined the effects of optic lobe or pars intercerebralis removal on the circabidian rhythm. After removing both optic lobes, all beetles lost their circabidian rhythms (N = 25), but all beetles exhibited circabidian rhythm after removing unilateral optic lobe (N = 18). However, 22% of the latter group exhibited day switching. After removal of the pars intercerebralis, 26.3% beetles showed arrhythmic patterns (N = 19). The number of paraldehyde fuchsin-stained pars intercerebralis cells in the arrhythmic group was significantly reduced compared to in the intact and sham-operated groups. The activity in the pars intercerebralis-removed beetles was significantly higher than that in the control groups. The results show that the optic lobe and at least part of the pars intercerebralis are necessary for circabidian rhythm, and bilateral optic lobes are necessary to maintain regularity of the two-day rhythm in H. parallela. This suggests that a neural circuit of circadian clock cells in the optic lobe to pars lateralis might be evolutionally conserved and used also for the generation of circabidian rhythm.

Direction Selective Neurons Responsive to Horizontal Motion in a Crab Reflect an Adaptation to Prevailing Movements in Flat Environments.

All animals need information about the direction of motion to be able to track the trajectory of a target (prey, predator, cospecific) or to control the course of navigation. This information is provided by direction selective (DS) neurons, which respond to images moving in a unique direction. DS neurons have been described in numerous species including many arthropods. In these animals, the majority of the studies have focused on DS neurons dedicated to processing the optic flow generated during navigation. In contrast, only a few studies were performed on DS neurons related to object motion processing. The crab Neohelice is an established experimental model for the study of neurons involved in visually-guided behaviors. Here, we describe in male crabs of this species a new group of DS neurons that are highly directionally selective to moving objects. The neurons were physiologically and morphologically characterized by intracellular recording and staining in the optic lobe of intact animals. Because of their arborization in the lobula complex, we called these cells lobula complex directional cells (LCDCs). LCDCs also arborize in a previously undescribed small neuropil of the lateral protocerebrum. LCDCs are responsive only to horizontal motion. This nicely fits in the behavioral adaptations of a crab inhabiting a flat, densely crowded environment, where most object motions are generated by neighboring crabs moving along the horizontal plane.SIGNIFICANCE STATEMENT Direction selective (DS) neurons are key to a variety of visual behaviors including, target tracking (preys, predators, cospecifics) and course control. Here, we describe the physiology and morphology of a new group of remarkably directional neurons exclusively responsive to horizontal motion in crabs. These neurons arborize in the lobula complex and in a previously undescribed small neuropil of the lateral protocerebrum. The strong sensitivity of these cells for horizontal motion represents a clear example of functional neuronal adaptation to the lifestyle of an animal inhabiting a flat environment.

Persistent Firing and Adaptation in Optic-Flow-Sensitive Descending Neurons.

A general principle of sensory systems is that they adapt to prolonged stimulation by reducing their response over time. Indeed, in many visual systems, including higher-order motion sensitive neurons in the fly optic lobes and the mammalian visual cortex, a reduction in neural activity following prolonged stimulation occurs. In contrast to this phenomenon, the response of the motor system controlling flight maneuvers persists following the offset of visual motion. It has been suggested that this gap is caused by a lingering calcium signal in the output synapses of fly optic lobe neurons. However, whether this directly affects the responses of the post-synaptic descending neurons, leading to the observed behavioral output, is not known. We use extracellular electrophysiology to record from optic-flow-sensitive descending neurons in response to prolonged wide-field stimulation. We find that, as opposed to most sensory and visual neurons, and in particular to the motion vision sensitive neurons in the brains of both flies and mammals, the descending neurons show little adaption during stimulus motion. In addition, we find that the optic-flow-sensitive descending neurons display persistent firing, or an after-effect, following the cessation of visual stimulation, consistent with the lingering calcium signal hypothesis. However, if the difference in after-effect is compensated for, subsequent presentation of stimuli in a test-adapt-test paradigm reveals adaptation to visual motion. Our results thus show a combination of adaptation and persistent firing in the neurons that project to the thoracic ganglia and thereby control behavioral output.

Binocular responsiveness of projection neurons of the praying mantis optic lobe in the frontal visual field.

Praying mantids are the only insects proven to have stereoscopic vision (stereopsis): the ability to perceive depth from the slightly shifted images seen by the two eyes. Recently, the first neurons likely to be involved in mantis stereopsis were described and a speculative neuronal circuit suggested. Here we further investigate classes of neurons in the lobula complex of the praying mantis brain and their tuning to stereoscopically-defined depth. We used sharp electrode recordings with tracer injections to identify visual projection neurons with input in the optic lobe and output in the central brain. In order to measure binocular response fields of the cells the animals watched a vertical bar stimulus in a 3D insect cinema during recordings. We describe the binocular tuning of 19 neurons projecting from the lobula complex and the medulla to central brain areas. The majority of neurons (12/19) were binocular and had receptive fields for both eyes that overlapped in the frontal region. Thus, these neurons could be involved in mantis stereopsis. We also find that neurons preferring different contrast polarity (bright vs dark) tend to be segregated in the mantis lobula complex, reminiscent of the segregation for small targets and widefield motion in mantids and other insects.

A minimal synaptic model for direction selective neurons in Drosophila.

Visual motion estimation is a canonical neural computation. In Drosophila, recent advances have identified anatomic and functional circuitry underlying direction-selective computations. Models with varying levels of abstraction have been proposed to explain specific experimental results but have rarely been compared across experiments. Here we use the wealth of available anatomical and physiological data to construct a minimal, biophysically inspired synaptic model for Drosophila's first-order direction-selective T4 cells. We show how this model relates mathematically to classical models of motion detection, including the Hassenstein-Reichardt correlator model. We used numerical simulation to test how well this synaptic model could reproduce measurements of T4 cells across many datasets and stimulus modalities. These comparisons include responses to sinusoid gratings, to apparent motion stimuli, to stochastic stimuli, and to natural scenes. Without fine-tuning this model, it sufficed to reproduce many, but not all, response properties of T4 cells. Since this model is flexible and based on straightforward biophysical properties, it provides an extensible framework for developing a mechanistic understanding of T4 neural response properties. Moreover, it can be used to assess the sufficiency of simple biophysical mechanisms to describe features of the direction-selective computation and identify where our understanding must be improved.

Back to the light, coevolution between vision and olfaction in the "Dark-flies" (Drosophila melanogaster).

Trade-off between vision and olfaction, the fact that investment in one correlates with decreased investment in the other, has been demonstrated by a wealth of comparative studies. However, there is still no empirical evidence suggesting how these two sensory systems coevolve, i.e. simultaneously or alternatively. The "Dark-flies" (Drosophila melanogaster) constitute a unique model to investigate such relation since they have been reared in the dark since 1954, approximately 60 years (~1500 generations). To observe how vision and olfaction evolve, populations of Dark-flies were reared in normal lighting conditions for 1 (DF1G) and 65 (DF65G) generations. We measured the sizes of the visual (optic lobes, OLs) and olfactory (antennal lobes, ALs) primary centres, as well as the rest of the brain, and compared the results with the original and its genetically most similar strain (Oregon flies). We found that, whereas the ALs decreased in size, the OLs (together with the brain) increased in size in the Dark-flies returned back to the light, both in the DF1G and DF65G. These results experimentally show that trade-off between vision and olfaction occurs simultaneously, and suggests that there are possible genetic and epigenetic processes regulating the size of both optic and antennal lobes. Furthermore, although the Dark-flies were able to mate and survive in the dark with a reduced neural investment, individuals being returned to the light seem to have been selected with reinvestment in visual capabilities despite a potential higher energetic cost.

Color vision in insects: insights from Drosophila.

Color vision is an important sensory capability that enhances the detection of contrast in retinal images. Monochromatic animals exclusively detect temporal and spatial changes in luminance, whereas two or more types of photoreceptors and neuronal circuitries for the comparison of their responses enable animals to differentiate spectral information independent of intensity. Much of what we know about the cellular and physiological mechanisms underlying color vision comes from research on vertebrates including primates. In insects, many important discoveries have been made, but direct insights into the physiology and circuit implementation of color vision are still limited. Recent advances in Drosophila systems neuroscience suggest that a complete insect color vision circuitry, from photoreceptors to behavior, including all elements and computations, can be revealed in future. Here, we review fundamental concepts in color vision alongside our current understanding of the neuronal basis of color vision in Drosophila, including side views to selected other insects.

Visual motion sensitivity in descending neurons in the hoverfly.

Many animals use motion vision information to control dynamic behaviors. For example, flying insects must decide whether to pursue a prey or not, to avoid a predator, to maintain their current flight trajectory, or to land. The neural mechanisms underlying the computation of visual motion have been particularly well investigated in the fly optic lobes. However, the descending neurons, which connect the optic lobes with the motor command centers of the ventral nerve cord, remain less studied. To address this deficiency, we describe motion vision sensitive descending neurons in the hoverfly Eristalis tenax. We describe how the neurons can be identified based on their receptive field properties, and how they respond to moving targets, looming stimuli and to widefield optic flow. We discuss their similarities with previously published visual neurons, in the optic lobes and ventral nerve cord, and suggest that they can be classified as target-selective, looming sensitive and optic flow sensitive, based on these similarities. Our results highlight the importance of using several visual stimuli as the neurons can rarely be identified based on only one response characteristic. In addition, they provide an understanding of the neurophysiology of visual neurons that are likely to affect behavior.

Optic lobe organization in stomatopod crustacean species possessing different degrees of retinal complexity.

Stomatopod crustaceans possess tripartite compound eyes; upper and lower hemispheres are separated by an equatorial midband of several ommatidial rows. The organization of stomatopod retinas is well established, but their optic lobes have been studied less. We used histological staining, immunolabeling, and fluorescent tracer injections to compare optic lobes in two 6-row midband species, Neogonodactylus oerstedii and Pseudosquilla ciliata, to those in two 2-row midband species, Squilla empusa and Alima pacifica. Compared to the 6-row species, we found structural differences in all optic neuropils in both 2-row species. Photoreceptor axons from 2-row midband ommatidia supply two sets of lamina cartridges; however, conspicuous spaces lacking lamina cartridges are observed in locations corresponding to where the cartridges of the upper four ommatidial rows of 6-row species would exist. The tripartite arrangement and enlarged projections containing fibers associated with the two rows of midband ommatidia can be traced throughout the entire optic lobe. However, 2-row species lack some features of medullar and lobular neuropils in 6-row species. Our results support the hypothesis that 2-row midband species are derived from a 6-row ancestor, and suggest specializations in the medulla and lobula found solely in 6-row species are important for color and polarization analysis.

Cellular and synaptic adaptations of neural circuits processing skylight polarization in the fly.

Specialized ommatidia harboring polarization-sensitive photoreceptors exist in the 'dorsal rim area' (DRA) of virtually all insects. Although downstream elements have been described both anatomically and physiologically throughout the optic lobes and the central brain of different species, little is known about their cellular and synaptic adaptations and how these shape their functional role in polarization vision. We have previously shown that in the DRA of Drosophila melanogaster, two distinct types of modality-specific 'distal medulla' cell types (Dm-DRA1 and Dm-DRA2) are post-synaptic to long visual fiber photoreceptors R7 and R8, respectively. Here we describe additional neuronal elements in the medulla neuropil that manifest modality-specific differences in the DRA region, including DRA-specific neuronal morphology, as well as differences in the structure of pre- or post-synaptic membranes. Furthermore, we show that certain cell types (medulla tangential cells and octopaminergic neuromodulatory cells) specifically avoid contacts with polarization-sensitive photoreceptors. Finally, while certain transmedullary cells are specifically absent from DRA medulla columns, other subtypes show specific wiring differences while still connecting the DRA to the lobula complex, as has previously been described in larger insects. This hints towards a complex circuit architecture with more than one pathway connecting polarization-sensitive DRA photoreceptors with the central brain.

Spectral response properties of higher visual neurons in Drosophila melanogaster.

The fruit fly Drosophila melanogaster can process chromatic information for true color vision and spectral preference. Spectral information is initially detected by a few distinct photoreceptor channels with different spectral sensitivities and is processed through the visual circuit. The neuroanatomical bases of the circuit are emerging. However, only little information is available in chromatic response properties of higher visual neurons from this important model organism. We used in vivo whole-cell patch-clamp recordings in response to monochromatic light stimuli ranging from 300 to 650 nm with 25-nm steps. We characterized the chromatic response of 33 higher visual neurons, including their general response type and their wavelength tuning. Color-opponent-type responses that had been typically observed in primates and bees were not identified. Instead, the majority of neurons showed excitatory responses to broadband wavelengths. The UV (300-375 nm) and middle wavelength (425-575 nm) ranges could be separated at the population level owing to neurons that preferentially responded to a specific wavelength range. Our results provide a first mapping of chromatic information processing in higher visual neurons of D. melanogaster that is a suitable model for exploring how color-opponent neural mechanisms are implemented in the visual circuits.

The Organization of the Second Optic Chiasm of the Drosophila Optic Lobe.

Visual pathways from the compound eye of an insect relay to four neuropils, successively the lamina, medulla, lobula, and lobula plate in the underlying optic lobe. Among these neuropils, the medulla, lobula, and lobula plate are interconnected by the complex second optic chiasm, through which the anteroposterior axis undergoes an inversion between the medulla and lobula. Given their complex structure, the projection patterns through the second optic chiasm have so far lacked critical analysis. By densely reconstructing axon trajectories using a volumetric scanning electron microscopy (SEM) technique, we reveal the three-dimensional structure of the second optic chiasm of Drosophila melanogaster, which comprises interleaving bundles and sheets of axons insulated from each other by glial sheaths. These axon bundles invert their horizontal sequence in passing between the medulla and lobula. Axons connecting the medulla and lobula plate are also bundled together with them but do not decussate the sequence of their horizontal positions. They interleave with sheets of projection neuron axons between the lobula and lobula plate, which also lack decussations. We estimate that approximately 19,500 cells per hemisphere, about two thirds of the optic lobe neurons, contribute to the second chiasm, most being Tm cells, with an estimated additional 2,780 T4 and T5 cells each. The chiasm mostly comprises axons and cell body fibers, but also a few synaptic elements. Based on our anatomical findings, we propose that a chiasmal structure between the neuropils is potentially advantageous for processing complex visual information in parallel. The EM reconstruction shows not only the structure of the chiasm in the adult brain, the previously unreported main topic of our study, but also suggest that the projection patterns of the neurons comprising the chiasm may be determined by the proliferation centers from which the neurons develop. Such a complex wiring pattern could, we suggest, only have arisen in several evolutionary steps.

Modality-Specific Circuits for Skylight Orientation in the Fly Visual System.

In the fly optic lobe, ∼800 highly stereotypical columnar microcircuits are arranged retinotopically to process visual information. Differences in cellular composition and synaptic connectivity within functionally specialized columns remain largely unknown. Here, we describe the cellular and synaptic architecture in medulla columns located downstream of photoreceptors in the dorsal rim area (DRA), where linearly polarized skylight is detected for guiding orientation responses. We show that only in DRA medulla columns both R7 and R8 photoreceptors target to the bona fide R7 target layer where they form connections with previously uncharacterized, modality-specific Dm neurons: two morphologically distinct DRA-specific cell types (termed Dm-DRA1 and Dm-DRA2) stratify in separate sublayers and exclusively contact polarization-sensitive DRA inputs, while avoiding overlaps with color-sensitive Dm8 cells. Using the activity-dependent GRASP and trans-Tango techniques, we confirm that DRA R7 cells are synaptically connected to Dm-DRA1, whereas DRA R8 form synapses with Dm-DRA2. Finally, using live imaging of ingrowing pupal photoreceptor axons, we show that DRA R7 and R8 termini reach layer M6 sequentially, thus separating the establishment of different synaptic connectivity in time. We propose that a duplication of R7→Dm circuitry in DRA ommatidia serves as an ideal adaptation for detecting linearly polarized skylight using orthogonal e-vector analyzers.

From skylight input to behavioural output: A computational model of the insect polarised light compass.

Many insects navigate by integrating the distances and directions travelled on an outward path, allowing direct return to the starting point. Fundamental to the reliability of this process is the use of a neural compass based on external celestial cues. Here we examine how such compass information could be reliably computed by the insect brain, given realistic constraints on the sky polarisation pattern and the insect eye sensor array. By processing the degree of polarisation in different directions for different parts of the sky, our model can directly estimate the solar azimuth and also infer the confidence of the estimate. We introduce a method to correct for tilting of the sensor array, as might be caused by travel over uneven terrain. We also show that the confidence can be used to approximate the change in sun position over time, allowing the compass to remain fixed with respect to 'true north' during long excursions. We demonstrate that the compass is robust to disturbances and can be effectively used as input to an existing neural model of insect path integration. We discuss the plausibility of our model to be mapped to known neural circuits, and to be implemented for robot navigation.