Visual motion processing in flying insects is strongly modulated by behavioural state, yet the mechanisms by which mechanosensory feedback contributes to this modulation remain poorly understood. Here we show that wing mechanosensation alone is sufficient to modulate a subset of optic flow-sensitive descending neurons (WFDNs) in butterflies. Airflow stimulation of the wings, mimicking flight conditions, increased baseline firing rates and reduced response latencies in WFDNs, without altering response gain or temporal frequency tuning. These effects indicate that mechanosensory modulation acts through mechanisms distinct from those governing other state-dependent changes in visual processing. Consistent with this interpretation, previous work has shown that baseline firing and response latency can be rapidly modulated, whereas gain changes arise through slower neuromodulatory pathways. Mechanosensory modulation was cell-type specific: the horizontally tuned WFDNL neuron was consistently affected across individuals, whereas other WFDN types were largely insensitive, likely reflecting the unilateral airflow stimuli used here. Interestingly, WFDNL modulation arose exclusively from mechanosensory input from the proximal area of the wing, not from distal wing deformation. This suggests that descending visual pathways require only coarse gain or excitability modulation rather than detailed information about wing shape or strain, leaving fast reflexive control of distal wing deformation to local ganglionic circuits. Together, our results demonstrate that wing mechanosensation selectively modulates visual descending pathways by altering excitability and timing rather than visual feature encoding, supporting the existence of multiple, parallel mechanisms for state-dependent visual modulation during flight. ### Competing Interest Statement The authors have declared no competing interest. European Research Council, https://ror.org/0472cxd90, 804315 Biotechnology and Biological Sciences Research Council, https://ror.org/00cwqg982, BB/R002509/1, BB/X002276/1 United States Air Force Office of Scientific Research, https://ror.org/011e9bt93, FA8655-23-1-7049

Insects achieve agile flight using a sensor-rich control architecture whose embodiment eliminates the need for complex computation. For example, their visual systems are tuned to detect the optic flow associated with specific self-motions, but what functional principle does this tuning embed, and how does it facilitate motor control? Here, we tested the hypothesis that evolution cotunes physics and physiology by aligning an insect's sensors to its dynamically important modes of self-motion. Specifically, we show that the spatial tuning of the blowfly motion vision system maximizes the open-loop Hankel singular values, which quantify the flow of signal energy from gust disturbances and control inputs to sensor outputs, jointly optimizing observability and controllability. This evolutionary principle differs from the conventional engineering-design paradigm of optimizing state estimation, with implications for robotic systems combining high performance with minimal actuator usage.

Previous experiments using a biohybrid fly-robotic-interface (FRI) revealed that the neural activity of an identified directional-selective cell in the blowfly motion vision system indicates proximity to the walls of an experimental area. For a constant turning radius of the FRI along an oscillatory trajectory, the spike rate of the H1-cell was inversely proportional to wall distance. We have now implemented a distance estimation algorithm on a small 2-wheeled robot, based on a parsimonious model of directional motion detection in flies and tested its performance under different conditions. Applied to monocular and binocular image motion captured with an on-board video camera, the algorithm enabled the avoidance of collisions at a success rate of 93.5

Insects have evolved over hundreds of millions of years to achieve exceptional agility, energy efficiency, and manoeuvrability. A key factor in their control capabilities is the use of efference copy mechanisms, which enable fast, robust control while maintaining sensors within their optimal operating range. Previous work has demonstrated the advantages of the efference copy-based fully-separable degrees of freedom (FSDoF) control architecture in single-input-single-output systems. However, insect flight control must handle multiple different time delays within distinct sensory and motor pathways. This paper presents a framework for designing FSDoF controllers, that phenomenologically model efference copies, for multi-input-multi-output (MIMO) systems with multiple different time delays (MDTDs). We provide a structured method for implementing FSDoF control on MDTD systems and benchmark its performance against model predictive control (MPC). Our results show that FSDoF achieves shorter settling times, smoother state transitions, lower actuation costs, and lower cross-coupling errors across a range of tested dynamics. Our findings highlight the potential of FSDoF control for biological and bio-inspired systems that operate under multiple different time-delayed dynamics.

Insects demonstrate remarkable agility in flight despite constant changes in flight dynamics throughout their lives. However, it is unclear whether such resilience is conferred via purely feedback control or whether adaptive feedforward control mechanisms are present.  This study examines whether adaptive feedforward control mechanisms are present in Drosophila melanogaster flight, by comparing the free-flight trajectories with and without wing damage and antennae ablation. Flies with partial wing excisions exhibited increased flight speeds in the dark compared to intact-wing controls. Upon exposure to visual contrast in light, the clipped-wing flies reduced their speed comparable to that of the control group flies. Notably, the lower speed persisted upon returning to the dark, indicating an enduring change to the flight controller. To discern between feedforward adaptation and a change in mechanosensory feedback gains, we replicated the experiment after ablating the antennal arista, the primary mechanosensors for sensing airspeed. Although flies with ablated antennae flew with greater variance in speed, they displayed a parallel trend in mean speed adaptation: increased speed in the dark, compensation in the light, and sustained lower speed in subsequent dark conditions. This consistent pattern strongly supports the involvement of adaptive feedforward control rather than the adjustment of mechanosensory feedback gains. Our investigation unveils an adaptive strategy in D. Melanogaster flight, illustrating its ability to set flight speed through adaptive feedforward control mechanisms. ### Competing Interest Statement The authors have declared no competing interest.

Color provides an important visual dimension for object detection and classification. In most animals, color and motion vision are largely separated throughout early stages of visual processing. However, accumulating evidence indicates crosstalk between chromatic and achromatic pathways. Here, we investigate the spectral sensitivity of the motion-vision pathway at the level of pre-motor descending neurons (DNs) in two butterfly species with different retinal compositions and wing coloration. Butterflies engage in fast, agile flight within often colorful visual ecologies, which may heighten evolutionary pressure to integrate color and motion vision. Indeed, we observed a separation of spectral sensitivities that matches the functional properties of butterfly DNs, such that wide-field, optic flow-sensitive DNs involved in stabilization reflexes have effective broadband spectral responses, while target-selective DNs involved in target tracking are comparatively narrowband and match conspecific wing coloration. Our findings demonstrate the spectral tuning of motion vision within a pre-motor neuronal bottleneck that controls behavior. Video abstract

In this article we explore the benefits of matching sensing characteristics to actuation and dynamics in the context of spatially distributed sensorimotor architectures, motivated by recently discovered connections in blowfly flight physics and visual physiology. Within the proposed framework, we present novel semidefinite programs with linear matrix inequality constraints which yield directions encoded in the sensory output that maximize the smallest unstable Hankel singular value of the system. This is a coordinate-invariant metric that minimizes the control energy required to stabilize an unstable system and maximizes the achievable robustness to unstructured additive uncertainty over all possible controllers. We also reformulate the problem to achieve a prescribed speed of response, which can be applied to stable and unstable systems. We adapt a maximally robust controller synthesis method from previous work which provides a tool for validation. We additionally present an H ∞ controller formulation which allows for a trade-off between minimization of actuator effort and robustness versus disturbance rejection and tracking capability, providing design flexibility over the maximally robust controller.

Biological systems have evolved to perform high-speed voluntary movements whilst maintaining robustness and stability. This paper examines a control architecture based on the principles of efference copies found in insect sensorimotor control which we call the fully-separable-degrees-of-freedom (FSDoF) controller. Within a control engineering framework, we benchmark the advantages of this control architecture against two common engineering control schemes: a pure feedback (PFB) controller and a Smith predictor (SP). Our study identifies three advantages of the FSDoF for biology. It is advantageous in controlling systems with sensor delays, and it can effectively handle noise. Thirdly, it allows biological sensors to increase their operating range. We evaluate the robustness of the FSDoF controller and show that it achieves improved performance with equal stability margins and robustness. Finally, we discuss variations of the FSDoF which theoretically provide the same performance.

Color provides an important dimension for object detection and classification. In most animals, color- and motion-vision are largely separated throughout early stages of visual processing. However, accumulating evidence indicates crosstalk between chromatic and achromatic pathways. Here we investigate the spectral sensitivity of the butterfly motion-vision pathway at the level of pre-motor descending neurons (DNs), which connect the brain to thoracic motor centres. Butterflies engage in fast agile flight within often-colorful visual ecologies, which may heighten evolutionary pressure to integrate color- and motion-vision. Indeed, we observed a separation of spectral sensitivities that matches the functional properties of butterfly DNs, such that wide-field optic flow-sensitive DNs involved in stabilisation reflexes have effectively broadband spectral responses, whilst target-selective DNs involved in target-tracking are comparatively narrowband and match conspecific wing coloration. Our findings demonstrate an integration of color- and motion-vision within a pre-motor neuronal bottleneck that controls behavior. ### Competing Interest Statement The authors have declared no competing interest. Air Force Office of Scientific Research, FA8655-23-1-7049 Biotechnology and Biological Sciences Research Council, BB/X002276/1

Time-invariant controllers can lead to suboptimal reference tracking for robotic systems faced with variable loads and dynamics. To mitigate this, adaptive control techniques such as model reference adaptive control (MRAC) and adaptive incremental nonlinear dynamic inversion (INDI) adapt to varying dynamics. However, these adaptive control methods require partial knowledge of the system dynamics to obtain the sensitivity derivatives/Jacobian-the direction in which to adapt. The prior knowledge required by these methods can increase the development/design time and limit controller flexibility when faced with highly variable dynamics. Here, inspired by insect control and cerebellum learning systems, we propose an 'implicit' adaptive feedforward (IAFF) control architecture. In contrast to other methods, it does not require any prior knowledge of the system dynamics, and can be deployed in a plug-and-play manner. IAFF uses one filter to learn the sensitivity derivatives and another filter to generate the feedforward actuation command. We demonstrate this adaptive controller in simulations and on a Parrot Mambo mini-drone. We observe that from random initial filter weights, the controller improved the drone's altitude reference tracking capabilities within 30 seconds, and position (pitch and roll) control within minutes. The low computational cost allowed the real-time learning algorithm to be executed on the onboard Arm A9 microprocessor. Copyright (c) 2025 The Authors. This is an open access article under the CC BY-NC-ND license (https://creativecommons.org/licenses/by-nc-nd/4.0/)

Collision avoidance in flying insects is mostly based on visual motion cues such as retinal image expansion or the relative magnitude of retinal image shifts. In earlier studies, we found that the activity of an identified visual interneuron (H1-cell) in a fly mounted on a bio-hybrid fly robot interface (FRI) was modulated by the robot's turning radius and the distance to the walls of an experimental arena. To characterise the neural mechanisms underlying visual distance estimation we set up a virtual reality environment (FlyVR) that enabled us to reproduce the input to the motion vision pathway experienced by flies on the FRI and record the H1-cell activity without modulations by other sensory modalities. After establishing a qualitative alignment of the results obtained on the FRI and our FlyVR system, we now address the outstanding question of whether the distance-dependent modulation of the H1-cell activity depends on the velocity of the FRI. Our results suggest that at a fixed turning radius within the range tested the robot velocity hardly affects the H1-cell spike rate. The functional significance of this surprising result is discussed as well as further analysis steps to elucidate the neural computations involved.

Biological sensors have evolved to act as matched filters that respond preferentially to the stimuli they expect to receive during ecologically relevant tasks [[1][1]–[3][2]]. For instance, insect visual systems are tuned to detect stimuli ranging from the small-target motion of mates [[4][3], [5][4]] or prey [[6][5], [7][6]] to the polarization pattern of the sky [[1][1], [8][7]]. In flies, individually identified neurons called lobula plate tangential cells respond to optic flow fields matched to specific self-motions [[9][8], [10][9]], forming the output layer of what is presently nature’s best-understood deep convolutional neural network [[11][10]–[13][11]]. But what functional principle does their tuning embed, and how does this aid motor control? Here we test the hypothesis that evolution co-tunes physics and physiology by aligning the preferred directions of an animal’s sensors to the most dynamically-significant directions of its motor system [[14][12]]. We build a state-space model of blowfly flight by combining visual electrophysiology, synchrotron-based X-ray microtomography, high-speed videogrammetry, and computational fluid dynamics. We then apply control-theoretic tools to show that the tuning of the fly’s widefield motion vision system maximizes the flow of energy from control inputs and disturbances to sensor outputs, rather than optimizing state estimation as is the conventional approach to sensor placement in engineering [[15][13], [16][14]]. We expect the same functional principle to apply across sensorimotor systems in other organisms, with implications for the design of novel control architectures for robotic systems combining high performance with low computational load and low power consumption.### Competing Interest StatementThe authors have declared no competing interest. [1]: #ref-1 [2]: #ref-3 [3]: #ref-4 [4]: #ref-5 [5]: #ref-6 [6]: #ref-7 [7]: #ref-8 [8]: #ref-9 [9]: #ref-10 [10]: #ref-11 [11]: #ref-13 [12]: #ref-14 [13]: #ref-15 [14]: #ref-16

Insects’ ability to know the velocity they are flying is of interest to both biologists and roboticists. While the Reichardt detector is one of the most prominent models for insect motion vision, it has various limitations when extracting image velocity in a natural environment. Here we demonstrate a method for estimating image velocity by weighting the outputs of a population of Reichardt detectors where individual detectors are tuned to different temporal frequencies. By providing stimuli of different spatial frequencies and velocities, we then perform a convex optimisation on each average output to find weights. We show that when the weighted detector arrays are provided with different stimuli, the output reasonably approximates image velocity. Our results have implications for power-limited autonomous systems and suggest a potential mechanism for insect motion vision.

The ability of animals and robots to move through a given environment without colliding with any obstacles requires a robust distance estimation mechanism. Previous electrophysiological studies and work using a biohybrid fly-robot-interface (FRI) suggest that fly directional-selective interneurons may be involved in the neural control of collision-avoidance behaviour. We have set up a virtual reality (FlyVR) environment and studied the blowfly’s H1-cell, an interneuron analyzing visual wide-field motion, to access its distance-dependent responses that was discovered using the FRI. The results gathered under open-loop FlyVR conditions are in qualitative agreement with open- and closed-loop data obtained on the FRI. They suggest that the capability of flies to estimate distance may depend on the animal’s specific movement trajectory in combination with the receptive field properties of the H1-cell. Our findings in the fly motion vision pathway may inform the design of energy-efficient collision avoidance strategies for autonomous robotic systems.

The optokinetic nystagmus is a gaze-stabilizing mechanism reducing motion blur by rapid eye rotations against the direction of visual motion, followed by slower syndirectional eye movements minimizing retinal slip speed. Flies control their gaze through head turns controlled by neck motor neurons receiving input directly, or via descending neurons, from well-characterized directional-selective interneurons sensitive to visual wide-field motion. Locomotion increases the gain and speed sensitivity of these interneurons, while visual motion adaptation in walking animals has the opposite effects. To find out whether flies perform an optokinetic nystagmus, and how it may be affected by locomotion and visual motion adaptation, we recorded head movements of blowflies on a trackball stimulated by progressive and rotational visual motion. Flies flexibly responded to rotational stimuli with optokinetic nystagmus-like head movements, independent of their locomotor state. The temporal frequency tuning of these movements, though matching that of the upstream directional-selective interneurons, was only mildly modulated by walking speed or visual motion adaptation. Our results suggest flies flexibly control their gaze to compensate for rotational wide-field motion by a mechanism similar to an optokinetic nystagmus. Surprisingly, the mechanism is less state-dependent than the response properties of directional-selective interneurons providing input to the neck motor system.

Polarisation vision is commonplace among invertebrates; however, most experiments focus on determining behavioural and/or neurophysiological responses to static polarised light sources rather than moving patterns of polarised light. To this end, we designed a polarisation stimulation device based on superimposing polarised and non-polarised images from two projectors, which can display moving patterns at frame rates exceeding invertebrate flicker fusion frequencies. A linear polariser fitted to one projector enables moving patterns of polarised light to be displayed, whilst the other projector contributes arbitrary intensities of non-polarised light to yield moving patterns with a defined polarisation and intensity contrast. To test the device, we measured receptive fields of polarisation sensitive Argynnis paphia butterfly photoreceptors for both non-polarised and polarised light. We then measured local motion sensitivities of the optic flow-sensitive lobula plate tangential cell H1 in Calliphora vicina blowflies under both polarised and non-polarised light, finding no polarisation sensitivity in this neuron.

SUMMARYGaze stabilization reflexes reduce motion blur and simplify the processing of visual information by keeping the eyes level. These reflexes typically depend on estimates of the rotational motion of the body, head, and eyes, acquired by visual or mechanosensory systems. During rapid movements, there can be insufficient time for sensory feedback systems to estimate rotational motion, requiring additional mechanisms. Solutions to this common problem are likely to be adapted to an animal’s behavioral repertoire. Here, we examine gaze stabilization in three families of dipteran flies, each with distinctly different flight behaviors. Through frequency response analysis based on tethered-flight experiments, we demonstrate that fast roll oscillations of the body lead to a stable gaze in hoverflies, whereas the reflex breaks down at the same speeds in blowflies and horseflies. Surprisingly, the high-speed gaze stabilization of hoverflies does not require sensory input from the halteres, their low-latency balance organs. Instead, we show how the behavior is explained by a hybrid control system that combines a sensory-driven, active stabilization component mediated by neck muscles, and a passive component which exploits physical properties of the animal’s anatomy—the mass and inertia of its head. This adaptation requires hoverflies to have specializations of the head-neck joint that can be employed during flight. Our comparative study highlights how species-specific control strategies have evolved to support different visually-guided flight behaviors.SIGNIFICANCE STATEMENTAcross the animal kingdom, reflexes are found which stabilize the eyes to reduce the impact of motion blur on vision—analogous to the image stabilization functions found in modern cameras. These reflexes can be complex, often combining predictions about planned movements with information from multiple sensory systems which continually measure self-motion and provide feedback. The processing of this information in the nervous system incurs time delays which impose limits on performance when fast stabilization is required. Hoverflies overcome the limitations of sensory-driven stabilization reflexes by exploiting the passive stability provided by the head during roll perturbations with particularly high rotational kine-matics. Integrating passive and active mechanisms thus extends the useful range of vision and likely facilitates distinctive aspects of hoverfly flight.

The goal of this article is to complement the comprehensive account Spatial Vision in Arthorpods provided by Wehner R. (1981) in the Handbook of Sensory Physiology. We will confine ourselves to review a selection of examples where over the last decades an integrated approach of quantitative behavioral studies in combination with electrophysiology and modeling best illustrates neural mechanisms underlying visually guided behavior. Many of these studies have been successfully carried out in a variety of flying insects, namely flies.

Effective visuomotor coordination is a necessary requirement for the survival of many terrestrial, aquatic, and aerial animal species. We studied the kinematics of aerial pursuit in the blowfly Lucilia sericata using an actuated dummy as target for freely flying males. We found that the flies perform target tracking in the horizontal plane and target interception in the vertical plane. Our behavioural data suggest that the flies' trajectory changes are a controlled combination of target heading angle and of the rate of change of the bearing angle. We implemented control laws in kinematic models and found that the contributions of proportional navigation strategy are negligible. We concluded that the difference between horizontal and vertical control relates to the difference in target heading angle the fly keeps constant: 0° in azimuth and 23° in elevation. Our work suggests that male Lucilia control both horizontal and vertical steerings by employing proportional controllers to the error angles. In horizontal plane, this controller operates at time delays as small as 10 ms, the fastest steering response observed in any flying animal, so far.