Motion Vision in Arthropods
Abstract and Keywords
Visual perception seems effortless to us, yet it is the product of elaborate signal processing in intricate brain circuits. Apart from vertebrates, arthropods represent another major animal group with sophisticated visual systems in which the underlying mechanisms can be studied. Arthropods feature identified neurons and other experimental advantages, facilitating an understanding of circuit function at the level of individual neurons and their synaptic interactions. Here, focusing on insect and crustacean species, we summarize and connect our current knowledge in four related areas of research: (1) elementary motion detection in early visual processing; (2) the detection of higher level visual features such as optic flow fields, small target motion and object distance; (3) the integration of such signals with other sensory modalities; and (4) state-dependent visual motion processing.
Vision requires animals to make sense of complex and fluctuating images projected from a three-dimensional environment via optics onto a two-dimensional sheet of photoreceptors. Once light signals are converted into neural activity patterns, visual features ultimately relevant for survival and reproductive success need to be extracted. Usually a considerable fraction of neural circuitry is devoted toward this end. A rich source of information is provided by visual motion, which is implicitly encoded in the temporal sequence of signals across the retinal array. Visual motion on the retina can have two different origins: It arises by the displacement of objects within the visual field of an observer, or it is generated by active or passive movement of the eyes of an observer relative to the environment. The former may hint toward other moving animals nearby, such as predators, prey, or mates, the detection of which is of obvious evolutionary significance. The latter provides useful feedback signals to control own movements and compensate for externally or internally generated perturbations of an intended trajectory. Moreover, information about the three-dimensional structure of the environment such as distance from objects can, in principle, be inferred from relative visual motion, that is, motion parallax.
Although visual motion provides a wealth of information, a fundamental problem is the inherent ambiguity, since all visual motion results in the sequential activation of photoreceptors. For instance, self-generated optic flow and motion signals from moving objects superimpose on the retina, a reliable segmentation of which can be vital. Many higher level features associated with visual motion arise sequentially in hierarchical processing layers. As a first step, motion signals are computed locally for many locations in visual space, a process termed “elementary motion detection.” It is a conceptually modest problem since local motion direction can be derived from simple spatiotemporal correlations. Yet how this is achieved by neurons and their interactions has been under intense investigation for more than 60 years. In subsequent stages, to infer the source of detected visual motion, several features of the spatiotemporal activity patterns such as velocity, size, level of coherency, and flow-field structure need to be taken into account. Such feature selectivity can be understood to result from neuronal properties and selective connections, which together constitute sophisticated matched filters, or templates, that signal the presence of signatures associated with certain sources. Segregation of self- and non-self-generated visual input can be further alleviated by conveying copies of motor commands to visual neurons such that visual feedback associated with an executed maneuver is effectively canceled early in sensory processing. Furthermore, information from other modalities is of fundamental importance for the interpretation of visual motion. For instance, mechanosensory signals provide additional input regarding changes in orientation and heading during flight. Interestingly, evidence accumulates that behavioral state influences visual processing to match it to the expected stimulus statistics. Motion vision can thus be studied at many stages, from sensory perception to the execution of motor routines.
However, to reach an integrative mechanistic understanding of visually guided behavior poses a considerable challenge, due to distributed yet interconnected and redundant circuitry. Arthropods hold great promise to achieve this goal because, first, their fixed eyes allow precisely controlled visual stimulation in intact animals; second, their small brains contain orders of magnitude fewer neurons than most vertebrates; third, neurons are usually uniquely identifiable, greatly facilitating the interpretation of experimental results across individuals; and last, transgenic approaches in Drosophila allow the functional manipulation of identified neurons in many ways with great specificity.
In this review, we will highlight our current knowledge of how arthropods make sense of spatiotemporal activity patterns on their retinae to detect visual motion and associated features most relevant to their ethology.
Overview of Arthropod Visual Systems
Sophisticated eyes have evolved in the bilaterian animal groups of vertebrates, arthropods, and cephalopods (Land & Nilsson, 2012). The two major types of eyes are the single-chambered or camera-type eyes found in cephalopods and vertebrates, and the compound eyes characteristic for most arthropods. Among the arthropods, insects and crustaceans are the dominant group for experimental vision research, because their hard exoskeleton facilitates fixation and stable recordings over long periods of time. However, single-unit recordings from the brain of a soft-bodied salticid spider have been achieved recently (Menda et al., 2014).
The basic feature of a typical arthropod visual system is its repetitive structure comprising multiple units (ommatidia) that are wrapped around the head in a hexagonal pattern and, each with its own optics, project light onto a convex array of photoreceptor cells (Fig. 1A) (Land & Nilsson, 2012). A defining characteristic of arthropod photoreceptors is the microvilli that employ a phophoinositoid signaling cascade downstream of Rhodopsin via phospholipase C (PLC). This biochemical pathway terminates in the opening of cation-permeable Trp and Trp-like channels to convert absorbed photons into neuronal depolarizations (Fain et al., 2010). Microvilli have been described as hemi-autonomous units that allow strong amplification, with the result that photoreceptors are highly sensitive and fast (Hardie & Juusola, 2015). How Trp channel opening is gated at the molecular level was enigmatic for a long time. However, recent surprising findings suggest, among others, two underlying key events: first, PLC-mediated hydrolysis of the phopholipid PIP2 releases a proton leading to acidification of the rhabdomeres (Huang et al., 2010); second, since PIP2 is an integral component of the cell membrane, its hydrolysis mediates a physical change of the cell membrane and thus a contraction of the photoreceptor (Hardie & Franze, 2012). Together, these studies suggest that Trp channel opening can indeed be triggered by a combination of photo-mechanical responses and cytosolic acidification upon PLC activation.
The part of the brain beneath the ommatidia devoted to the processing of visual information is called the “optic lobe” (Fig. 1B). It houses the separated neuropils called lamina, medulla, and lobula complex, the latter of which is subdivided into a lobula and a lobula plate in various arthropod groups such as flies, moths, and crabs (Strausfeld, 2005). Each neuropil is made up of retinotopically organized columnar units, corresponding to the ommatidial array, intersected by horizontal synaptic processing layers.
Depending on their spectral sensitivity, photoreceptor neurons convey signals to the lamina or medulla. Motion vision pathways are mainly supplied by lamina cells (but see Wardill et al., 2012), which receive histaminergic graded hyperpolarizing signals from photoreceptors of a single spectral type, called R1 to R6 in flies (Osorio & Bacon, 1994). Although in most insect species the photoreceptors in each ommatidium map onto one lamina column, the underlying connectivity in dipteran flies (suborder Brachycera) features a more complicated principle termed “neural superposition” (Fig. 1C): Each point in visual space is sampled by photoreceptors in different neighboring ommatidia, which converge onto the same lamina column at an amazing regularity (Braitenberg, 1967; Kirschfeld, 1967; Agi et al., 2014). This sophisticated design enhances sensitivity in dim light conditions by increasing photon capture without compromising spatial resolution.
Signals are temporally and spatially filtered by lamina cells, thereby enhancing contrast. Lamina cells feed into segregated parallel processing channels with different dynamics and contrast polarity preference in the next processing stage, the medulla. Signals are then further processed by lobula and lobula plate projection neurons, which signal higher level information about the visual scene such as small- and wide-field motion information to higher brain and motor centers. In Drosophila and other dipteran flies, comprehensive and detailed anatomical characterizations of most optic lobe neurons have been indispensable in guiding the functional exploration of motion vision circuits (Strausfeld, 1970; Fischbach & Dittrich, 1989; Bausenwein et al., 1992).
The Fundamental Problem and Implementation of Elementary Motion Detection
Photoreceptors signal local changes in light intensity and do not differentiate whether those arise by moving patterns or stationary luminance fluctuations. To detect the direction of visual motion, spatiotemporal correlations of photoreceptor signals across the retina have to be analyzed by downstream circuits. This fundamental operation is termed “elementary motion detection.” To explore mechanisms underlying elementary motion detection, a robust stereotypic behavior has been exploited as a readout, the optomotor response. Since moving animals receive self-generated optic flow on their retinae and use the associated visual feedback to stabilize an intended trajectory, wide-field rotational visual motion elicits behavioral steering responses syndirectional with the perceived motion direction. Such behavior is reliably evoked with marginal habituation in insects during tethered flight or walking. Hassenstein and Reichardt (1956) studied optomotor behavior of the beetle Chlorophanus viridis to conceive an algorithmic model for elementary motion detection based on signal correlation (Fig. 2A). During motion, adjacent image pixels change sequentially. The Hassenstein-Reichardt model reduces the signal delay between neighboring image points by asymmetric temporal filtering and generates direction selectivity by correlating the resulting signals. Certain features are predicted by this operation, such as temporal frequency tuning, contrast dependence, and velocity gain control, all of which can be experimentally reproduced (Borst, 2014). Owing to its fundamental nature for visual processing, the cellular implementation of this algorithm has received considerable interest yet remained elusive for a long time, due to complexity and dense packing of visual circuits that hampered manipulation and recordings of individual neuronal types. Within the last decade, however, Drosophila has emerged as an important species in which the cellular mechanisms can be studied in a systematic and robust fashion by way of genetic targeting (Borst, 2009; Venken et al., 2011). Given similar visual motion-mediated behaviors and homologous neural substrates for vision (Osorio & Bacon, 1994; Buschbeck & Strausfeld, 1996; Strausfeld, 2005), the underlying mechanisms are believed to be comparable across diverse arthropod groups.
To explore the cellular processing steps underlying elementary motion detection, the lamina neurons postsynaptic of photoreceptors represent a useful starting point (Fig. 2B). An important first step has been the demonstration that among those L1 and L2 together are necessary and sufficient for Drosophila’s directed behavioral response to moving gratings (Rister et al., 2007). By presenting moving edges instead of gratings in combination with genetic lamina cell block and direction-selective tangential cell electrophysiology, another study showed that L1 and L2 are not redundant but in fact feed into two motion pathways in which light increments (ON: L1) and decrements (OFF: L2) are processed in parallel (Joesch et al., 2010). This discovery has subsequently been corroborated and extended in various other studies (Clark et al., 2011; Eichner et al., 2011; Joesch et al., 2013; Silies et al., 2013; Tuthill et al., 2013). Which cells constitute the outputs of the two motion pathways? Small-field T4 and T5 cells have long been suspected because their axon terminals overlap with wide-field motion-sensitive tangential cells in the lobula plate (Fig. 2B). The two types differ in their dendritic location, suggesting a functional segregation: T4 receives input in the most proximal medulla layer while T5 cells have their dendrites confined to the lobula (Fig. 2C, D). By transgenic expression of the reporter GCaMP, which changes fluorescence upon changes in calcium concentration, their activity could be visualized using two-photon microscopy. This approach revealed that T4 responds to moving ON edges while T5 prefers moving OFF edges (Maisak et al., 2013). Moreover, the T4/T5 terminals in the lobula plate are strictly segregated according to their directional preference. Thus, they constitute a directional tuning map with the four lobula plate layers representing the four cardinal directions of motion (Fig. 2E, F). In summary, inputs and outputs of two elementary motion detection channels have been identified: Direction of moving light increments is computed between L1 and T4, while moving light decrements are processed between L2 and T5.
To establish experimentally the mechanisms that generate direction selectivity, as a first step, ideally each neuron type connecting lamina and T4/T5 cells needs to be scrutinized regarding its spatiotemporal receptive field and requirement for direction selectivity in downstream neurons or behavior. In this respect, dense reconstructions of medulla neurons and circuits have narrowed down the number of candidate neurons from ~60 per column to about two neuron types (Mi1, Tm3) providing major input to T4 (Takemura et al., 2013) and four neuron types (Tm1, Tm2, Tm4, Tm9) providing input to T5 (Shinomiya et al., 2014) (Fig. 2B). Moreover, these studies have measured spatial offsets relative to each other as predicted by motion detection models. For T4, a small retinotopic offset from Tm3 to Mi1 inputs matches the T4 preferred direction to some extent, as indicated by the axon terminal position in the four lobula plate layers (Takemura et al., 2013). The temporal dynamics of Mi1 and Tm3 were subsequently measured by whole-cell patch recordings (Behnia et al., 2014), revealing an average Mi1 peak time delayed by 18 ms compared to Tm3. Together with connectomics (Takemura et al., 2013), this finding suggested that Mi1 and Tm3 represent the postulated spatially offset and asymmetrically filtered input lines to T4. Importantly, this layout points towards a “null direction suppression,” as put forward by Barlow and Levick (1965), where direction selectivity is generated by a delayed inhibition shunting spatially offset fast excitation (Fig. 2A). However, in this particular case, the Barlow-Levick model seems incompatible with the finding that T4 and T5 cells produce direction selectivity by amplification of excitatory inputs (Fisher et al., 2015b). Moreover, while silencing Mi1 indeed abolishes direction selectivity in downstream neurons for all stimuli tested, Tm3 is dispensable at low to intermediate velocities (Ammer et al., 2015), suggesting that at least one other cell type is involved in motion detection in the ON pathway.
Other recent studies have focused on candidate OFF pathway medulla cells Tm1, Tm2, Tm4, and Tm9, all of which increase their activity in response to decreasing light intensity in a non-direction-selective way (Behnia et al., 2014; Meier et al., 2014; Fisher et al., 2015a; Serbe et al., 2016). Temporal differences between Tm1 and Tm2 responses have suggested a possible role in elementary motion detection at the level of T5 (Behnia et al., 2014). However, surprisingly, using T4/T5 calcium signals, tangential cell voltage, or behavior as readouts, all four cell types have been shown to be necessary to various degrees to generate OFF pathway-specific direction selectivity (Meier et al., 2014; Fisher et al., 2015a; Serbe et al., 2016). Although different temporal dynamics of the four cell types suggest some sort of functional specialization, this is not reflected in block phenotypes as a function of edge velocity (Fisher et al., 2015a; Serbe et al., 2016). Fisher et al. (2015a) reported an unexpected large receptive field of Tm9 (>60°) and suggested a potential gating mechanism for this cell rather than providing local signals for spatially offset nonlinear correlation. However, another study has systematically measured similarly sized small receptive fields for all four candidate cells (~5°), including Tm9, in agreement with local input in visual space (Serbe et al., 2016). Therefore, as in T4, it is currently unclear how the different medulla cells in the OFF pathway contribute to direction selectivity in T5 cells. Clearly, more experiments are required and alternative models need to be taken into account to reach an understanding of how direction-selective signals are generated in the dendrites of T4 and T5 cells.
In the following, we will provide an overview on circuits and behaviors associated with higher level visual motion processing and discuss how far the underlying mechanisms might rely on the outputs from elementary motion detection circuits as outlined earlier.
Wide-Field Optic Flow Processing
When an animal moves, it generates feedback signals detected by its sensory systems. Particularly for flying animals such as many insects feedback information is important in order to stabilize motor output, which is susceptible to internally (e.g., left-right motor imbalance) and externally generated (e.g., wind gusts) perturbations. Self-generated visual wide-field motion, called optic flow, is useful in this context because unintended translatory and rotatory components of the flight path can in principle be robustly inferred from it (Gibson, 1950; Karmeier et al., 2006) (Fig. 3A). In line with this idea, wide-field motion-sensitive visual neurons in the optic lobes and higher processing areas have been described in various arthropods, which could counteract unintended course deviations. These neurons are characterized by their direction-selective robust responses over large (>30°), often binocular receptive fields (flies: Bishop & Keehn, 1967; Dvorak et al., 1975; Hausen, 1976; Joesch et al., 2008; Krapp & Hengstenberg, 1996; Schnell et al., 2010; moths and butterflies: Collett & Blest, 1966; Rind, 1983; Ibbotson et al., 1991; Milde, 1993; bees: Kaiser & Bishop, 1970; DeVoe et al., 1982; Ibbotson & Goodman, 1990; Ibbotson, 1991; locusts: Kien, 1974; Rind, 1990; dragonfly: Olberg, 1981a; cockroach: Kathman et al., 2014; crabs: Horseman et al., 2011). In most cases, such neurons have been found to be motion-opponent, which means that they not only respond with excitation to their preferred but also with inhibition to the opposite or null direction.
How local motion information is integrated across large receptive fields is best understood in lobula plate tangential cells of flies (Egelhaaf et al., 2002; Borst et al., 2010; Fig. 2, 3B-D’). Tangential cells have large dendrites which receive excitatory synaptic inputs from local retinotopically arranged T4/T5 motion detectors (Schnell et al., 2012; Maisak et al., 2013; Hopp et al., 2014; Mauss et al., 2014). They can be broadly subdivided into (a) spiking heterolateral elements, subserving binocular integration (e.g., H1 and V1), and (b) lobula plate output neurons (e.g., HS and VS cells; Fig. 3B, D), conveying signals mainly via graded potentials (with superimposed irregular action potentials) to motor neurons and premotor descending neurons (Hausen, 1984; Hausen & Egelhaaf, 1989). Lobula plate tangential neurons can also be subdivided into horizontal and vertical cells, depending on their overall preferred direction of motion determined by the layer in which they receive input from T4/T5 cell terminals. For instance, HS cells with a preference for front-to-back motion ramify their dendrites in layer 1 (Fig. 3B’) (Hausen et al., 1980; Scott et al., 2002; Schnell et al., 2010), while Hx cells that prefer back-to-front motion confine their dendrites to layer 2 (Fig. 3C’) (Krapp & Hengstenberg, 1996; Wasserman et al., 2015). Likewise, layer 3 V2 (Hausen, 1976, 1984; Wertz et al., 2008) and layer 4 VS cells (Fig. 3D’) (Hausen et al., 1980; Scott et al., 2002; Joesch et al., 2008; Hopp et al., 2014; Mauss et al., 2015) are tuned to upward and downward flow, respectively. Like most wide-field direction-selective neurons, lobula plate tangential cells are motion-opponent, that is, receive oppositely tuned excitatory and inhibitory input. The origin of inhibitory null direction input was puzzling, since it requires activity of T4/T5 cells (Schnell et al., 2012), yet those are cholinergic and excitatory (Mauss et al., 2014; Shinomiya et al., 2014). Moreover, tangential cell dendrites generally do not overlap with T4/T5 terminals tuned to their null direction (but see later discussion). This gap in understanding was recently filled by the identification of bistratified glutamatergic lobula plate-intrinsic (LPi) neurons (Fig. 3E, E’), which receive T4/T5 input in one layer and convey an inhibitory signal to the neighboring layer, giving rise to null direction inhibition in tangential cells expressing glutamate-gated chloride channel α (Mauss et al., 2015) (Fig. 3F). The identification of LPi neurons allowed for establishing a fundamental role of motion-opponent integration in ensuring flow field selectivity. This is because flow fields containing opponent directions of motion, such as expansion during forward translation, elicit oppositely tuned excitatory and inhibitory signals impinging on different parts of the large tangential cell dendrite, resulting in response cancelation (Fig. 3F). Motion-opponent input thus makes tangential cells more selective for unidirectional optic flow. This principle might also apply to achieve selectivity to more complex flow fields, since in shore crabs, neurons selective to expanding and receding flow fields have also been shown to receive local null direction inhibition (Horseman et al., 2011).
Intriguingly, although fly tangential cells generally exhibit a global preferred direction, the spatial structure of their receptive fields can have additional rotational components, as shown in blowflies (Fig. 3G) (Krapp & Hengstenberg, 1996; Krapp et al., 1998). Tangential cells have thus been concluded to constitute effective matched filters to detect optic flow evoked by certain rotational maneuvers (Franz & Krapp, 2000). The rotational components have been proposed to arise by different complementary mechanisms (Borst et al., 2010; Borst & Weber, 2011), some of which are outlined here (Fig. 3H). First, while VS dendrites mostly arborize in downward-motion-selective layer 4, the dorsal subtrees of VS cells 7–10 extend through the lobula plate to layer 1, where they most likely receive additional retinotopic direction-selective signals from horizontal layer T4/T5 cells (Hengstenberg et al., 1982; Scott et al., 2002; Hopp et al., 2014). Interestingly, these inputs partly display unexpected upward selectivity (Hopp et al., 2014), probably arising from changes of the ommatidial grid orientation toward the eye perimeter (Petrowitz et al., 2000), suggesting that the geometrical arrangement of the ommatidia favors detection of rotational optic flow. Dorsal VS cell 7–10 dendrites have also been shown to receive horizontally tuned input from electrotonically coupled dCH neurons, which in turn obtain their horizontal tuning from contralateral H2 and ipsilateral HSN and HSE cells (Haag & Borst, 2007). Second, VS cells with preferred direction input at opposite azimuthal positions reciprocally inhibit each other via Vi and Vi2 interneurons. For instance, VS7-10 responding to downward motion at lateral positions are electrically connected to Vi. Vi then provides sign-inverting synaptic input to frontal VS1 (Haag & Borst, 2007). Thus, in addition to frontal downward selectivity through local motion detectors, VS1 obtains also sign-inverted input in the lateral visual field and therefore presumably a rotational flow-field selectivity. VS cell receptive fields are further broadened by electric coupling of similarly tuned cells (Haag & Borst, 2004; Elyada et al., 2009), which together with reciprocal inhibition via Vi increases response robustness to fluctuating input from patchy natural scenes (Cuntz et al., 2007). VS cells also receive fast input from three light-sensing dorsal organs, the ocelli. Those provide little spatial detail, but their responses are modulated by high contrast between sky and ground, for instance during body rotation. Such modulations likely contribute to the VS’s cells rotational flow field tuning, though their integration with compound eye input remains to be investigated (Parsons et al., 2010).
Binocular integration can also be observed at the level of the tangential cell network presumed to control course around the vertical body axis (Hausen, 1982; Horstmann et al., 2000; Haag & Borst, 2001; Krapp et al., 2001; Farrow et al., 2006). Important neural elements are the unilateral wide-field horizontal system (HS) cells, which receive elementary motion detector signals tuned to front-to-back motion and bilateral H1 and H2 cells, receiving unilateral oppositely tuned (i.e., back to front) local inputs (Fig. 3I). Such monocularly integrated local signals are ambiguously evoked by yaw rotation or forward/sideward translation. However, two of the HS cells (HSN and HSE) receive contralateral input from H1 and H2, presumably making them more selective to yaw rotation causing front-to-back optic flow on the ipsilateral eye, and back-to-front optic flow on the contralateral eye (Karmeier et al., 2006). Binocular input is also integrated at the level of H2 itself, since H2 and HSE cells are electrically coupled. The graded membrane potential of HSE thereby facilitates H2’s spiking response to rotational optic flow (Farrow et al., 2006).
Thus, a combination of wide-field local motion detector integration in specific lobula plate layers, motion opponency mediated by inhibitory layer interactions, and selective lateral connectivity and binocular integration among tangential cells construct sophisticated matched filters. Those are suited to robustly detect self-generated wide-field pattern motion associated with most flight maneuvers. Interestingly, accumulation of calcium signals in HS tangential cell terminals rather than the more transient membrane voltage precisely matches simultaneously observed optomotor steering behavior, providing a possible mechanism for temporal integration (Schnell et al., 2014). A plausible role of tangential cells is thus to supply self-motion evoked visual feedback to counteract unintended course deviations. Direct evidence for this notion, however, is scarce. For instance, regarding evidence for necessity, a fruit fly mutant (ombH31) lacking tangential cells shows a strong reduction in optomotor turning reactions (Heisenberg et al., 1978). Likewise, optomotor responses in larger flies are similarly compromised, when HS cell axons are cut (Hausen & Wehrhahn, 1983) or HS/VS cell precursors are laser ablated (Geiger & Nässel, 1981). As a complementary experiment in support of sufficiency, Blondeau (1981) has achieved course control maneuvers by electrically stimulating lobula plate neurons. Recently, genetic targeting of a variant of the light-sensitive cation channel Channelrhodopsin-2 (C128S) to HS cells in Drosophila has made it possible to stimulate those noninvasively and much more selectively by focal blue light illumination delivered unilaterally (Haikala et al., 2013). As a consequence, in line with HS cells’ assumed role to counteract visually perceived rotation, tethered flies turn their head and adjust their wingbeat asymmetrically, indicative of a flight-turning response.
Lobula plate tangential cells can thus be understood as part of a visual reflex circuit that detects and counteracts unintended course deviations from optic flow by supplying error signals to motor control centers. This poses the question how this reflex is suppressed in flies in order to perform voluntary turns. Going back to ideas originally developed by Helmholtz, Holst and Mittelstaedt (1950) have proposed that a copy of the motor command (“efference copy”) is used to cancel the expected sensory input (“reafference”) associated with the executed behavior. A recent study has explored the idea that this could be achieved already at the level of the tangential cells (Kim et al., 2015). Indeed, internally generated turns, that is, saccades, of flies in tethered flight coincide with brief changes in the HS cells’ membrane potential (Schnell et al., 2014). In a series of experiments Kim et al. (2015) could show that saccade-related membrane potential changes had an appropriate sign, latency, and amplitude to cancel HS cell responses to visual motion. Thus, it appears that in tangential cells both visual input and signals related to voluntary motor commands are integrated in an opponent way so as to suppress the visual feedback responses expected from intended course deviations. It will be interesting to confirm directly in closed-loop experiments whether HS cell responses to self-generated image shifts are indeed selectively canceled. Furthermore, reafferent feedback depends on duration and magnitude of a saccade and also on the structure of the environment. Thus, the question arises whether motor-related inputs are modulated by these internal and external parameters, requiring a sophisticated control circuit.
Small-Field Object Detection
The detection of moving objects is relevant for inter- and intraspecific behaviors such as prey capture, escape, and mating (Nordström, 2012; Olberg, 2012). For instance, male and, to a lesser degree, female flies engage in pursuit upon detecting other flies in their visual field. This behavior consists of fast maneuvers with many rapid changes of flight course and is observed in several fly species in the context of territorial conflicts and intersexual chasing (Fig. 4A) (Collett & Land, 1978). For houseflies, a response delay of an impressive 30 ms was derived from cross-correlating angular orientation and angular velocity of leading and chasing fly, respectively (Land & Collett, 1974; Wehrhahn et al., 1982). Predatory insects such as dragonflies and killer flies rely on small object pursuit in order to capture other flying insects (Fig. 4B) (Olberg et al., 2000; Wardill et al., 2015). Blowflies pursue objects by fixating their retinal image in the frontal visual field, with retinal position and size being largely sufficient to control turning and thrust (Land & Collett, 1974; Boeddeker & Egelhaaf, 2003; Boeddeker et al., 2003). Other insects such as hoverflies and dragonflies are capable of intercepting the target course (Fig. 4C). This is thought to be implemented by stabilization of targets on the retina in a zone of increased acuity in a reactive way to maintain a constant bearing, resulting in a collision course (Collett & Land, 1978; Olberg et al., 2000, 2007; Gonzalez-Bellido et al., 2013). However, dragonfly (Plathemis lydia) prey capture has revealed an additional sophisticated level of predictive control that involves an internal representation of the target trajectory and own body movements (Mischiati et al., 2015). Continuous visual tracking thus seems to be unnecessary for the interception of a prey flying at a constant speed and heading. Small objects may also evoke escape behavior, as described for Drosophila (Maimon et al., 2008). The task of detecting and tracking small moving objects must rely on robust mechanisms since it has to be fast and successful in natural, usually densely cluttered environments. This is particularly challenging when self-motion evoked optic flow and small object motion superimpose on the retina. Neurons selectively responding to small object motion can be described by their size tuning, their receptive field size and location, and whether or not they respond to small-field motion in a direction-selective way. Such neurons have been identified in various fly species and dragonflies (Olberg, 1981a; Egelhaaf, 1985a; O’Carroll, 1993; Nordström et al., 2006; Kim et al., 2015).
In principle, local motion detectors such as the T4 and T5 neurons encode the retinal position and motion direction of small objects, but they are also excited by wide-field motion. Hence, further processing is required to make a neuron selective for small moving objects. A well-studied example of neurons selective for small moving objects is the feature detection (FD) cells identified in Calliphora. These neurons are characterized by rather large receptive fields (~60°–120°), yet they respond predominantly to relatively small moving features (~10°) and not to wide-field motion (Egelhaaf, 1985a). In addition, FD cells are direction selective. For instance, FD1 and FD4 prefer front-to-back motion, while FD2 and FD3 are tuned to the opposite direction. The main dendritic arborization of all FD cells resides in the lobula plate, and they are therefore expected to spatially integrate the outputs from arrays of T4/T5 local motion detectors (Maisak et al., 2013). Based on pharmacological and laser ablation experiments, FD1’s small-field selectivity has been concluded to arise by inhibitory actions from a binocular motion-sensitive vCH cell (Warzecha et al., 1993). vCH obtains its motion sensitivity from electrically coupled HS cells, leading to a spatially blurred motion image (Cuntz et al., 2003). Motion contrast and thereby small-field selectivity in FD cells is presumably enhanced by subtracting vCH’s blurred motion image from its own local motion detector input (Fig. 4D). In summary, FD cell small-field motion detection can be understood as the integration of local excitatory motion detector outputs and spatially pooled and thus size-dependent inhibitory signals. Based on comparison of FD cell signals with behavioral responses of tethered flying flies, FD cells were suggested to underlie figure-ground discrimination, for instance in the context of chasing conspecifics (Egelhaaf, 1985b).
In hoverflies and dragonflies, small target motion detector (STMD) neurons have been identified with exquisitely small size tuning of less than 3° (Fig. 4E, F) (O’Carroll, 1993; Nordström et al., 2006). While some STMDs respond in a directional-selective way (Fig. 4G, H), similar to FD cells, others respond vigorously to motion in any direction. Remarkably, large grating patterns fail to evoke either excitation or inhibition of STMDs, and those even respond to small object motion when the background moves syndirectionally (Nordström et al., 2006; Barnett et al., 2007), ruling out background subtraction by wide-field inhibitory input as described earlier for FD cells. How then does robust rejection of background motion by STMD neurons arise? Given that moving dark objects against a bright background locally produces an OFF signal from the leading edge followed by an ON signal from the trailing edge, a model (termed elementary small target motion detector, or ESTMD model) has been put forward to capture this feature by nonlinearly correlating a delayed OFF with an undelayed ON signal (Fig. 4I) (Wiederman et al., 2008, 2013). Sharp spatial tuning is achieved by lateral inhibition of neighboring ON or OFF signals before the correlation stage. Various predictions of the model have been experimentally confirmed (Nordström, 2012). For instance, the velocity optimum of STMDs increases with increasing bar width in agreement with the temporal filters of the model (Fig. 4J) (Geurten et al., 2007). Furthermore, STMD responses to small moving targets are strongly inhibited by distractor targets moving at ~3° separation, providing evidence for the assumed lateral inhibitory interactions (Fig. 4K) (Bolzon et al., 2009). In contrast to a Reichardt-like motion detector, however, the underlying local operation takes into account signals from the same point in space. The ESTMD thus does not produce direction-selective responses.
Recently, small-field-selective, movement-detecting neurons have also been identified in Drosophila among optic-glomeruli interneurons (OGINs) in the lateral protocerebrum (Kim et al., 2015). OGIN response properties are reminiscent of hover- and dragonfly STMDs: They preferentially respond to small moving objects (in a nondirectional way) across wide receptive fields but not wide-field optic flow, suggesting that they contribute to object detection. Given direction-unselective responses of small-field-selective, movement-detecting neurons in Drosophila and other species, the question arises to what extent directional motion information is actually required for tracking small objects. In the context of stripe fixation behavior in tethered walking Drosophila, the requirement of the elementary motion detector neurons T4 and T5 has been tested. Genetically silencing T4 and T5 eliminates all turning responses to wide-field motion, demonstrating their absolute requirement for optomotor behavior. Interestingly, flies were still able to orient toward vertical stripes though at a reduced performance level. These results suggest the existence of a position and a motion system acting in parallel to mediate efficient object tracking (Bahl et al., 2013).
Extracting Distance Information From Visual Motion
Information about the three-dimensional structure of the environment is important for all moving animals, particularly so for fast flying insects, in order to avoid crashes and permit save landings. Most arthropods with fixed-focus compound eyes of low spatial resolution, small interocular separation, and binocular overlap usually cannot rely on stereopsis (except e.g., mantises; Nityananda et al., 2016) or on accommodation to visually estimate distance, as vertebrates such as primates do. However, distance information can also be inferred by motion parallax from monocular visual motion during translation, since the retinal velocity of moving features projected onto the eye is inversely proportional to the distance of the feature: Close objects move fast while distant objects move slowly (Gibson, 1950). Honey bees exploit this simple relationship between distance and retinal velocity, for instance, to control altitude and flight deceleration during landing, this way achieving a smooth touchdown (Lehrer et al., 1988; Srinivasan et al., 2000, 2001). Likewise, walking fruit flies infer object distance based on azimuthal translation of its retinal projection, that is, motion parallax, rather than changes in size (Fig. 5A). Stereopsis is not involved (Schuster et al., 2002). As another example, locusts perform a characteristic peering behavior consisting of translational head movements to estimate the distance to a jumping target from motion parallax (Sobel, 1990). Finally, hawk moths hover in front of flowers to feed and use expanding and receding optic flow as cues to maintain a constant distance (Farina et al., 1994).
In contrast to translational flow, no spatial information can be obtained from rotational optic flow, since objects move at the same retinal velocity irrespective of their distance from the observer. Therefore, flies and other insects presumably exhibit a saccadic flight (short turns) and gaze (forward flight) strategy in order to facilitate spatial vision during the intersaccadic intervals (Hateren & Schilstra, 1999; Mronz & Lehmann, 2008; Egelhaaf et al., 2012, 2014). Elementary motion detector arrays in flies have been discussed as possible neural substrates to construct contrast-weighted nearness maps from the environmental retinal projections based on motion parallax during intersaccadic flight (Egelhaaf et al., 2014; Schwegmann et al., 2014). However, it is currently unclear how such spatial motion information might be read out by downstream neurons and in how far orientation behavior indeed relies on such detailed motion maps.
Fast estimation of distance information is particularly vital in the context of obstacle collision avoidance and escape from rapidly approaching predators. During relative object approach on a direct collision course, its retinal projection expands nonlinearly with decreasing distance (Fig. 5B). For constant approach velocities, the time course of angular size and angular velocity can be fully characterized by the object size-to-speed ratio, given by l/|v| with units of time (with l denoting object half size and |v| approach velocity) (Fotowat & Gabbiani, 2011). l/|v| is equivalent to the time-to-collision when the stimulus subtends 90° on the retina. Broadly speaking, given a similar l/|v| ratio, small and slowly approaching objects appear similar on the retina to large and fast objects. Object approach can thus be effectively simulated by patterns radially expanding on the retina, termed looming stimuli. Depending on the behavioral state and the retinal position of a looming stimulus, animals may respond differently. For instance, stationary fruit flies and locusts exhibit an escape jump biased away from the focus of expansion (Santer et al., 2005; Card & Dickinson, 2008). The time point of this response prior to simulated collision is linearly related to l/|v|, meaning that it occurs with a fixed delay after a certain angular size, or threshold, has been reached (Fotowat & Gabbiani, 2007; Fotowat et al., 2009). Likewise, when tethered or freely flying Drosophila are confronted with a lateral looming stimulus or expanding flow field, they execute a maneuver to avoid the anticipated collision (Tammero & Dickinson, 2002; Tammero et al., 2004; Muijres et al., 2014). However, presenting looming or an expanding flow field frontally to flying flies elicits a landing response, involving a stereotyped sequence of leg extensions, shift of the wingbeat plane, and reduction in thrust (Goodman, 1960; Braitenberg & Ferretti, 1966; Borst, 1986; Borst & Bahde, 1988a; Tammero & Dickinson, 2002; Schilling & Borst, 2015).
Looming-sensitive neurons have been found in moths, crabs, locusts, and flies (Schlotterer, 1977; Rind & Simmons, 1992; Wicklein & Strausfeld, 2000; Oliva et al., 2007; de Vries & Clandinin, 2012; Oliva & Tomsic, 2014), as well as various vertebrate species (Fotowat & Gabbiani, 2011). Looming sensors are usually large wide-field neurons, which respond independent of position, shape, and contrast to expanding visual stimuli, and much less so to translational or rotational optic flow. A particularly well-studied example in locusts is a pair of giant visual interneurons (lobula giant movement detector, LGMD), which receive visual input through three distinct dendritic fields: a large dendrite obtaining excitatory retinotopic input from the entire visual hemisphere, and two smaller dendrites receiving nonretinotopic inhibitory inputs (Fig. 5C) (Schlotterer, 1977; Rind & Simmons, 1992; Hatsopoulos et al., 1995; Gabbiani et al., 1999, 2002; Peron et al., 2009; Fotowat & Gabbiani, 2011). Interestingly, matching the corresponding escape behavior, the peak firing rate is reached before anticipated collision across a wide range of l/|v| (>5 ms), indicating that LGMD does not merely track edge velocity Θ’ or acceleration (Fig. 5D, E). Instead, peak firing preceding collision can be explained at the level of the LGMD by taking into account an additional inhibitory signal related to angular size Θ. Multiplying Θ’ with a negative exponential of Θ can capture the time course of LGMD firing in response to a looming stimulus and predicts that the neuron’s peak response occurs at a fixed delay after the retinal projection exceeds an angular threshold (Hatsopoulos et al., 1995; Gabbiani et al., 1999). What is the biophysical basis for such a multiplicative operation? A plausible mechanism supported by experimental data is a linear dendritic integration of excitatory and inhibitory synaptic inputs related to Θ’ and Θ, respectively, encoded logarithmically, with subsequent exponentiation via voltage-gated channels at the spike-initiating zone (Gabbiani et al., 2002).
The fact that LGMD receives excitatory inputs related to angular velocity on its large dendritic fan suggested that those signals might be generated by a correlation-type elementary motion detector, predicting a nonlinear direction-selective enhancement of sequentially activated neighboring inputs. To test this idea, Jones and Gabbiani (2010) designed a sophisticated apparatus to deliver light stimuli independently to individual ommatidia. Interestingly, no nonlinear enhancement of adjacent inputs was found, speaking against a correlation mechanism. Instead, the response latency decreased with increasing photoreceptor activation speed throughout the visual pathway. This latency decrease favors synchronization of spatially integrated and sequentially activated signals evoked by accelerating edges, and thus increases LGMD’s sensitivity to looming retinal projections of approaching objects.
In the semiterrestrial crab Neohelice (previously Chasmagnathus) granulata, neurons have been identified, termed monostratified lobula giants (MLG1), that resemble the LGMD neuron in many aspects, suggesting similar mechanisms underlying their feature selectivity. For instance, MLG1 neurons are rather weakly activated by unidirectional wide-field motion but vigorously spike in response to looming stimuli, during which their peak firing rates precede the time point of simulated collision (Medan et al., 2007; Oliva & Tomsic, 2014). A linear relationship between peak firing rate and l/|v| ratio suggests that MLG1 neurons, like LGMD, act as angular threshold detectors. However, in contrast to single bilateral LGMDs, MLG1 neurons in crabs exist in 16 bilateral pairs with smaller receptive fields. A recent study showed that the 16 MLG1 neuron receptive fields in each eye map the 360° azimuthal space (Medan et al., 2015). This is interesting, because in contrast to ballistic fly and locust escape jumps, crabs escape in a more controlled fashion on the ground, dynamically regulating run direction and velocity during object approach. The population activity of MLG1 neurons is thus well suited to control a directed escape. Indeed, such a role is supported by their activity closely matching the observed behavior across a variety of conditions (Oliva et al., 2007; Sztarker & Tomsic, 2008; Hemmi & Tomsic, 2012).
In Drosophila, a heterogeneous group of approximately five so-called FOMA-1 neurons has been identified (de Vries & Clandinin, 2012) that also selectively increase firing in response to a looming stimulus. According to the hallmarks of looming detection, the responses were related linearly to the object’s size-to-speed ratio as well as position, polarity, and global luminance invariant. Furthermore, genetic silencing and optogenetic activation of FOMA-1 neurons suggested that their activity is necessary and sufficient to elicit escape behavior (de Vries & Clandinin, 2012). The fact that some FOMA-1 neurons at least partly obtain synaptic input in the lobula plate suggests the interesting possibility that in Drosophila, unlike in locusts, spatial integration of elementary motion detector signals (i.e., from T4/T5 neurons) might underlie looming detection. This notion is substantiated by the fact that both the landing and avoidance responses of fruit flies depend on spatial wavelength, velocity, and contrast of the stimulus, much like elementary motion detectors (Borst & Bahde, 1986, 1988b; Duistermars et al., 2007). With the identification of a genetic driver line selective for T4/T5 elementary motion detectors in Drosophila, it became possible to test this idea directly by silencing T4/T5 and probing tethered flight behavior in response to looming stimuli (Schilling & Borst, 2015). While in control flies avoidance and landing responses could be evoked readily, depending on the position of the stimulus, all looming-evoked behavior was effectively abolished in T4/T5 block flies, demonstrating their role in looming detection (Fig. 5F, G). Looming detection in flies and locusts might thus be implemented by different mechanisms. It will be interesting to investigate how neurons downstream of T4/T5 in the lobula plate such as the FOMA-1 neurons integrate local motion information in order to signal expanding and not other flow fields. Looming-sensitive neurons are expected to integrate excitatory signals from at least two layers representing opposite directions of motion (Fig. 5H). However, they should be unresponsive to single layer activation, as it occurs, for example, during rotation around the vertical or longitudinal body axis.
Eventually, to control behavior, visual information must be conveyed from the optic lobes to motor centers coordinating muscle contractions, taking into account sensory input from other modalities. Signals are reformatted into motor coordinates, and motor programs are being gated and selected, based on available sensory information and internal parameters such as state and previous experience (Huston & Jayaraman, 2011). Regarding multimodal behavioral control, in flies at least three modalities underlie reflexive course stabilizations (Fig. 6A): (1) wide-field optic flow, as outlined earlier; (2) three single lens defocused light-sensing organs, the ocelli, obtain coarse information about light intensities in many pterygote insects. Ocelli are implicated to control gaze stabilization and flight control by detecting roll and pitch motion resulting from displacement of the horizon (Stange et al., 2002; Krapp & Wicklein, 2008; Parsons et al., 2010); (3) small club-shaped appendages in the third thoracic segment, called halteres, beat in antiphase to wings during flight and are deflected by Coriolis forces associated with body rotations (Pringle, 1948). Complementary to feedback from compound eyes and ocelli, albeit in a higher dynamic angular velocity range, haltere-associated sensory neurons are implicated in corrective steering reflexes (Sherman & Dickinson, 2004). Course deviations are thus encoded in signals from compound eyes, ocelli, and halteres in a complementary robust fashion in different dynamic angular velocity ranges, from rather slow flow field detectors to fast haltere-mediated feedback. Moreover, mechanosensory neurons associated with antennae are activated by wind and thus contribute to stable course control (Fuller et al., 2014). Antennae can also be deflected by Corriolis forces and thus signal body rotations. This might be particularly important for flight stability in insects without halteres such as moths (Sane et al., 2007).
Descending neurons of the cervical connective are of particular interest to study sensorimotor transformations, since they represent the sole link between visual feature detectors and thoracic motor centers controlling walking and flight. In addition, neck motorneurons obtain direct input from tangential cells or descending neurons (Fig. 6B). Such higher order neurons may signal related external events, detected through multiple (multimodal) sensory channels, or they may signal different events calling for similar behavioral responses. They thus represent an intersection between sensory and motor systems, and they may accordingly fit more into one or the other (Olberg, 1981b).
Several studies in flies have characterized the responses of neck motor and descending neurons to wide-field optic flow, which arise by selective input from combinations of lobula plate tangential cells (Milde et al., 1987; Strausfeld et al., 1987; Haag et al., 2007, 2010; Huston & Krapp, 2008; Wertz et al., 2008, 2009a, 2009b, 2012; Kauer et al., 2015). For instance, descending neurons of the ocellar and vertical system (DNOVS) 1 and 2 receive ipsilateral electrotonic input from specific VS cells tuned to similar axes of rotation (Haag et al., 2007; Wertz et al., 2008). The tuning width of DNOVS cells to their preferred axis of rotation did not differ from their main VS cell input. However, data from using more naturalistic inhomogeneous flow fields for stimulation suggest that convergence supports a more robust and smooth representation of rotational optic flow compared to their inputs, the VS cells, which exhibited higher fluctuations in their responses (Fig. 6C) (Cuntz et al., 2007; Wertz et al., 2009a). Descending and neck motor neurons usually receive additional contralateral direction-selective inputs (Fig. 6D) (Huston & Krapp, 2008). At the level of DNOVS2, inputs from both hemispheres interact in a nonlinear way such that simultaneous stimulation evokes greater responses than the linear sum of each individual response (Wertz et al., 2008). Such binocular integration is thought to greatly enhance selectivity to visual flow fields associated with particular flight maneuvers.
Some descending and motor neurons in the fly have been found to exhibit rather subtle subthreshold direction-selective graded potential changes. However, vigorous spiking responses of similar directional tuning can be observed in the same neurons when halteres are beating or the Johnston organ of the antennae is stimulated by air puffs (Fig. 6E) ( Huston & Krapp, 2009; Haag et al., 2010), suggesting a powerful gating mechanism. Importantly, while a gating effect does arise by passive haltere movement, activity increase in neck motorneurons precedes active haltere movement. Together with other evidence, this suggested an additional internally generated signal associated with behavioral state (Haag et al., 2010), a subject covered in more detail later.
In dragonflies (Anax junius), descending neurons have been broadly classified into object and self-movement detectors (Olberg, 1981a). Interestingly, object movement detectors only responded to visual input, while self-movement detectors could be activated by a broad range of multimodal input, such as air puffs directed to the front and rear head and neck, as well as passive and active movement of head and abdomen, respectively (Fig. 6F) (Olberg, 1981b). This presumably reflects the predominant significance of distant visual cues for prey tracking, while reflexive optomotor behavior exploits additional mechanosensory and proprioceptive input for robust flight stabilization. Eight pairs of target-selective descending neurons (Fig. 6G) have been studied in detail in Anax, as well as in another species (Libellula luctuosa) (Olberg, 1986; Frye & Olberg, 1995; Gonzalez-Bellido et al., 2013). They are thought to receive (indirect) inputs from small target detector neurons in the optic lobe, endowing them with a preference for moving small targets with short latencies (~30 ms). Importantly, while all of these descending neurons supply the thoracic ganglia that control wing muscles, each one is unique with respect to receptive field, preferred target size, and directional tuning (Fig. 6G). Together, they sample the dorso-frontal visual space such that their combined activity encodes a population vector accurately indicating location and direction of moving targets (Gonzalez-Bellido et al., 2013). This suggests that only few descending neurons may control visually guided prey interception. How their joint activity is read out to steer wing movements remains to be investigated.
A bilateral pair of descending neurons termed giant fibers (Power, 1948) has been implicated in fly escape responses to visual stimuli (Fig. 6H, I) (Allen et al., 2006). A recent study in Drosophila has been able to relate their visual response properties and behavioral role in unprecedented detail (Reyn et al., 2014). By performing patch-clamp recordings in animals that were restrained but able to move legs and wings, the authors could show in a subset of individual trials that looming stimuli trigger one or two individual spikes in the giant fiber and an escape response (Fig. 6J). However, flight initiation in response to looming could also be observed without a giant fiber response. Interestingly, flies may exhibit a short duration (<7 ms) escape resulting in an initially uncontrolled flight segment, or a long duration (~ 7–25 ms) involving a preceding wing elevation and more stable initial flight. Exploiting genetic strategies to cell-specifically silence and activate the giant fiber could firmly establish their causal role in mediating short but not long mode escape (Fig. 6K) (Reyn et al., 2014). These and other data (Fotowat et al., 2009) support the existence of at least two escape pathways with different activation thresholds and neural delays, whose interactions regulate innate escape responses in a surprisingly flexible manner.
A link between looming-sensitive neurons and leg motorneurons in locust is provided by the descending contralateral movement detector (DCMD). The DCMD neuron receives direct input from the previously highlighted optic lobe looming-sensitive LGMD neuron via a strong synapse, so that every presynaptic spike elicits a postsynaptic spike. In the thorax, DCMD provides direct and indirect input to motorneurons innervating flexors and extensors of the hind leg tibia. The co-contraction of flexors and extensors stores energy in elastic joint elements, which is suddenly released by inhibition of flexor motorneurons followed by unlocking and extension of the tibia. Using a sophisticated telemetry system allowing simultaneous neural and muscle recording in unrestrained locusts, Fotowat et al. (2011) have addressed the question of how looming stimuli-evoked spiking activity in DCMD might translate into coordinated motor activities underlying escape jumps. The initial co-contraction was found to be highly correlated with a threshold in the DCMD spiking activity, whereas take-off followed the DCMD peak firing rate with a fixed delay. Furthermore, the study uncovered a neural correlate for behavioral variability, explaining the occurrence of an escape jump by the total number of spikes following co-contraction onset. Thus, different aspects of the underlying motor program appear to be dictated by different attributes of the spiking activity of individual descending neurons (Fotowat & Gabbiani, 2011; Fotowat et al., 2011).
In summary, studies on descending neurons in insects uncovered a relatively small number of feature detectors, providing a glimpse of how visual motion information is compressed and conveyed to motor centers to produce robust yet flexible behavior. In the future, it will be interesting to fully characterize their multimodal input-output functions and how they impinge on central pattern generators controlling complex motor programs.
State-Dependent Visual Processing
Although at this point the impression might arise that sensory information is processed in arthropods in a static and stereotypic fashion, this is not quite the case. State-dependent modulation of visual processing has been observed over different time scales. For instance, responses of lobula neurons to visual threat stimuli are subject to seasonal variation in crabs (Sztarker & Tomsic, 2008). As another example, signaling of object approach by DCMD neuron is markedly different in gregarious versus solitary locusts (Schistocerca gregaria). Strong habituation, reflected by reduced peak firing rate and overall number of spikes, is observed only in solitarious individuals (Matheson et al., 2004). Furthermore, motion-sensitive wide-field neurons in honeybees (Kaiser & Steiner-Kaiser, 1983) and blowflies (Bult et al., 1991) show a strong circadian modulation in their spiking responses to pattern movement, and in bees during nights an ongoing elevated activity was observed following brief arousal stimuli.
Recently, various studies have studied the effect of locomotor state on visual motion processing by recording the activity of optic flow-sensitive neurons in tethered yet behaving flies (flying or walking on an air-suspended ball) (Chiappe et al., 2010; Maimon et al., 2010; Jung et al., 2011; Suver et al., 2012; Tuthill et al., 2014). In tethered flying Drosophila, the electrical activity of lobula plate tangential cells of the vertical system has been recorded by whole-cell patch-clamp (Fig. 7A) (Maimon et al., 2010). Two striking effects in behaving animals became apparent: first, baseline activity was tonically increased; and second, the responses to large moving gratings doubled, that is, increased in gain (Fig. 7B). Different time courses of the cessation of these effects at the offset of flight, with the gain boost decaying more slowly, suggested different underlying mechanisms. A plausible source for the baseline depolarization is an increase in excitatory synaptic input, as indicated in tangential cells during flight by a decrease in membrane resistance and an increase in voltage and current fluctuations in a hyperpolarized condition. A marked gain change of Drosophila tangential cell responses has also been observed in walking versus quiescent flies (Chiappe et al., 2010). Another interesting effect at the level of tangential cells associated with behavioral state is an increased sensitivity to visual motion at higher temporal frequencies, shown in tethered walking and flying Drosophila and flying blowfly Lucilia (Fig. 7C) (Chiappe et al., 2010; Jung et al., 2011; Suver et al., 2012). These findings provide an explanation for inconsistencies in the literature between temporal frequency optima of up to 10 Hz obtained for the optomotor response and ~1 Hz measured physiologically in tangential cells in usually restrained animals. The discrepancy can thus be readily explained by different behavioral state.
The biogenic amine octopamine has been implicated in controlling behavior-related arousal and locomotor states by neuromodulatory action (Bacon et al., 1995). Furthermore, extensive arborizations of octopaminergic neurons can be found in the optic lobes of insects (Fig. 7D) (Sinakevitch & Strausfeld, 2006; Busch et al., 2009). Octopamine has therefore been a promising candidate to mediate the observed effects. Indeed, application of the octopamine agonist chlordimeform in flies at rest can mimic various behavior-associated activity changes of tangential cells such as an increased baseline level, visual motion response gain, and shift of the tuning toward higher velocities (Fig. 7C) (Longden & Krapp, 2009, 2010; Jung et al., 2011). Direct evidence for activity-dependent modulation of visual motion-sensitive neurons by octopaminergic neurons has been recently provided by exploiting genetically targeted manipulation in Drosophila (Fig. 7E) (Suver et al., 2012). First, activity of optic lobe octopaminergic neurons was increased during tethered flight, as measured by two-photon calcium imaging in the lobula plate and beyond; second, thermogenetic activation of octopaminergic neurons in quiescent flies mimicked the visual boost in tangential cells observed during flight; and third, genetic silencing octopaminergic neurons by expression of the inwardly rectifying potassium channel Kir2.1 abolished the flight-mediated response boost in tangential cells during preferred direction motion. However, the source and role of the other physiological manifestation of behavioral state, that is, the tonic baseline shift in tangential cells, is still unclear (Suver et al., 2012). The aforementioned findings can be interpreted such that the mechanisms underlying sensory feature extraction are tuned by neuromodulation and potential other mechanisms in a dynamic and state-dependent fashion such as to match the range of expected stimulus statistics and reciprocally save resources during periods of quiescence.
Conclusion and Future Directions
Motion vision has emerged as a classic subject of sensory physiology with broad implications for neural processing in general. Such general aspects of motion vision circuits pertain, for instance, functional asymmetric wiring, spatiotemporal signal filtering, coincidence detection by nonlinear enhancement, gain control, feature detection, efference copy, and behavioral state modulation by biogenic amines. Due to their experimental accessibility, arthropods have been a valuable animal group to elucidate the underlying mechanisms at the level of individually identifiable neurons and their synaptic interactions. Such work has revealed how diverse higher level visual features are being extracted from shared lower level inputs in successive processing stages.
Although the mechanisms discovered in arthropods are interesting in themselves, comparisons can be drawn to other animal groups such as vertebrates. Despite obvious differences in neural structure, some amazing functional similarities have been uncovered. As one striking example, both in insects and vertebrates, visual motion information is computed in parallel pathways processing increments and decrements of light (Borst & Helmstaedter, 2015). At higher processing levels, neurons with similar properties selective for various optic flow fields, looming stimuli, and small moving objects have been identified in arthropods and vertebrates alike (e.g., reviewed in Nordström & O’Carroll, 2009; Borst & Euler, 2011; Fotowat & Gabbiani, 2011). Reflected by mechanistic similarities, these and other examples highlight a common demand of any visual system to effectively extract similar fundamental features from complex and time-varying neural signals derived from a two-dimensional sensor. In Drosophila, the small-field motion detector neurons T4 and T5, first described 100 years ago (Cajal & Sánchez, 1915), have been found to be of fundamental importance for most behaviors studied in the context of motion vision, such as optomotor behavior, collision avoidance, and object tracking (Bahl et al., 2013; Schilling & Borst, 2015). Whether in vertebrates feature detectors in the cortex rely on direction-selective ganglion cells in the retina in a similar way remains to be established (Cruz-Martín et al., 2014).
Regarding future work, neither vertebrate nor invertebrate model systems yet offer sufficient insight as to how the most basic motion vision operation—elementary motion detection—is achieved biophysically, although significant progress seems within reach. Given distributed yet interconnected and redundant neural circuitry, a challenge in the decades to come will be to identify the mechanisms that enable robust and dynamic representations of animal–environment interactions, taking into account multimodal sensory and self-generated feedback information. A useful concept in this context is that of a matched filter (Wehner, 1987). With limited computing power, sensory systems are not so much concerned with comprehensively analyzing the entire sensory space but are rather shaped by ecology through evolution to fulfil particular purposes. Such adaptations are found in sensory receptor neurons up to higher level feature detectors. For instance, both the geometrical arrangement of ommatidia (Petrowitz et al., 2000) and the connectivity of flow field–selective interneurons (Krapp & Hengstenberg, 1996) can be interpreted to favor the detection of visual feedback cues associated with body rotations. To identify yet higher level multimodal integrators, the trick will be to find the right feature key for selective activation in the enormous multimodal parameter space. Since behavior heavily feeds back on sensory input, it will be inevitable to establish mechanisms and causal roles of individual neurons and circuits in, ideally unrestrained, behaving animals. Experimental advantages of arthropods with uniquely identifiable neurons and particularly Drosophila with ever-improving genetic control over neural structure and activity will likely play an important role in this quest.
We are grateful to Alexander Arenz for critically reading the manuscript and Stefan Prech for artwork.
Agi, E., Langen, M., Altschuler, S. J., Wu, L. F., Zimmermann, T., & Hiesinger, P. R. (2014). The evolution and development of neural superposition. Journal of Neurogenetics, 28, 216–232.Find this resource:
Allen, M. J., Godenschwege, T. A., Tanouye, M. A., & Phelan, P. (2006). Making an escape: Development and function of the Drosophila giant fibre system. Seminars in Cell Development Biology, 17, 31–41.Find this resource:
Ammer, G., Leonhardt, A., Bahl, A., Dickson, B. J, & Borst, A. (2015). Functional specialization of neural input elements to the Drosophila ON motion detector. Current Biology, 25, 2247–2253.Find this resource:
Bacon, J. P., Thompson, K. S, & Stern, M. (1995). Identified octopaminergic neurons provide an arousal mechanism in the locust brain. Journal of Neurophysiology, 74, 2739–2743.Find this resource:
Bahl, A., Ammer, G., Schilling, T, & Borst, A. (2013). Object tracking in motion-blind flies. Nature Neuroscience, 16, 730–738.Find this resource:
Barlow, H. B, & Levick, W. R. (1965). The mechanism of directionally selective units in rabbit’s retina. Journal of Physiology, 178, 477–504.Find this resource:
Barnett, P. D., Nordström, K., & O’Carroll, D. C. (2007). Retinotopic organization of small-field-target-detecting neurons in the insect visual system. Current Biology, 17, 569–578.Find this resource:
Bausenwein, B., Dittrich, A. P., & Fischbach, K. F. (1992). The optic lobe of Drosophila melanogaster. II. Sorting of retinotopic pathways in the medulla. Cell Tissue Research, 267, 17–28.Find this resource:
Behnia, R., Clark, D. A., Carter, A. G., Clandinin, T. R., & Desplan, C. (2014). Processing properties of ON and OFF pathways for Drosophila motion detection. Nature, 512, 427–430.Find this resource:
Bishop, L. G., & Keehn, D. G. (1967). Neural correlates of the optomotor responses in the fly. Kybernetik, 3, 288–295.Find this resource:
Blondeau, J. (1981). Electrically evoked course control in the fly Calliphora erythrocephala. Journal of Experimental Biology, 92, 143–153.Find this resource:
Boeddeker, N., & Egelhaaf, M. (2003). Steering a virtual blowfly: Simulation of visual pursuit. Proceedings of Biological Science, 270, 1971–1978.Find this resource:
Boeddeker, N., Kern, R., & Egelhaaf, M. (2003). Chasing a dummy target: Smooth pursuit and velocity control in male blowflies. Proceedings of Biological Science, 270, 393–399.Find this resource:
Bolzon, D. M., Nordström, K., & O’Carroll, D. C. (2009). Local and large-range inhibition in feature detection. Journal of Neuroscience, 29, 14143–14150.Find this resource:
Borst, A. (1986). Time course of the houseflies’ landing response. Biology Cybernetics, 54, 379–383.Find this resource:
Borst, A. (2009). Drosophila’s view on insect vision. Current Biology, 19, R36–R47.Find this resource:
Borst, A. (2014). Neural circuits for elementary motion detection. Journal of Neurogenetics, 28, 361–373.Find this resource:
Borst, A., & Bahde, S. (1986). What kind of movement detector is triggering the landing response of the housefly? Biology Cybernetics, 55, 59–69.Find this resource:
Borst, A., & Bahde, S. (1988a). Spatio-temporal integration of motion. Naturwissenschaften, 75, 265–267.Find this resource:
Borst, A., & Bahde, S. (1988b). Visual information processing in the fly’s landing system. Journal of Comparative Physiology A, 163, 167–173.Find this resource:
Borst, A., & Euler, T. (2011). Seeing things in motion: Models, circuits, and mechanisms. Neuron, 71, 974–994.Find this resource:
Borst, A., Haag, J., & Reiff, D. F. (2010). Fly motion vision. Annual Review of Neuroscience, 33, 49–70.Find this resource:
Borst, A., & Helmstaedter, M. (2015). Common circuit design in fly and mammalian motion vision. Nature Neuroscience, 18, 1067–1076.Find this resource:
Borst, A., & Weber, F. (2011). Neural action fields for optic flow based navigation: A simulation study of the fly lobula plate network. PLoS ONE, 6, e16303.Find this resource:
Braitenberg, V. (1967). Patterns of projection in the visual system of the fly. I. Retina-lamina projections. Experimental Brain Research, 3, 271–298.Find this resource:
Braitenberg, V., & Ferretti, C. T. (1966). Landing reaction of musca domestica induced by visual stimuli. Naturwissenschaften, 53, 155–155.Find this resource:
Bult, R., Schuling, F. H., & Mastebroek, H. A. K. (1991). Circadian inputs influence the performance of a spiking, movement-sensitive neuron in the visual system of the blowfly. Journal of Biology Rhythms, 6, 55–69.Find this resource:
Busch, S., Selcho, M., Ito, K., & Tanimoto, H. (2009). A map of octopaminergic neurons in the Drosophila brain. Journal of Comparative Neurology, 513, 643–667.Find this resource:
Buschbeck, E. K., & Strausfeld, N. J. (1996). Visual motion-detection circuits in flies: Small-field retinotopic elements responding to motion are evolutionarily conserved across taxa. Journal of Neuroscience, 16, 4563–4578.Find this resource:
Cajal, S. Y., & Sánchez, D. Y. (1915). [Contribution to the knowledge of the nerve centers of insects. Part I. Retina and optic centers.] Trabajos del Laboratorio de Investigaciones Biologicas de la Universidad de Madrid (pp. 1–164). (Imprenta de Hijos de Nicholas Moja, Madrid).Find this resource:
Card, G., & Dickinson, M. H. (2008). Visually mediated motor planning in the escape response of Drosophila. Current Biology, 18, 1300–1307.Find this resource:
Chiappe, M. E., Seelig, J. D., Reiser, M. B., & Jayaraman, V. (2010). Walking modulates speed sensitivity in Drosophila motion vision. Current Biology, 20, 1470–1475.Find this resource:
Clark, D. A., Bursztyn, L., Horowitz, M. A., Schnitzer, M. J., & Clandinin, T. R. (2011). Defining the computational structure of the motion detector in Drosophila. Neuron, 70, 1165–1177.Find this resource:
Collett, T. S., & Blest, A. D. (1966). Binocular, directionally selective neurones, possibly involved in the optomotor response of insects. Nature, 212, 1330–1333.Find this resource:
Collett, T. S., & Land, M. F. (1978). How hoverflies compute interception courses. Journal of Comparative Physiology A, 125, 191–204.Find this resource:
Cruz-Martín, A., El-Danaf, R. N., Osakada, F., Sriram, B., Dhande, O. S., Nguyen, P. L., Callaway, E. M., Ghosh, A., & Huberman, A. D. (2014). A dedicated circuit links direction-selective retinal ganglion cells to the primary visual cortex. Nature, 507, 358–361.Find this resource:
Cuntz, H., Haag, J., & Borst, A. (2003). Neural image processing by dendritic networks. Proceedings of the National Academy of Sciences USA, 100, 11082–11085.Find this resource:
Cuntz, H., Haag, J., Forstner, F., Segev, I., & Borst, A. (2007). Robust coding of flow-field parameters by axo-axonal gap junctions between fly visual interneurons. Proceedings of the National Academy of Sciences USA, 104, 10229–10233.Find this resource:
DeVoe, R. D., Kaiser, W., Ohm, J., & Stone, L. S. (1982). Horizontal movement detectors of honeybees: Directionally-selective visual neurons in the lobula and brain. Journal of Comparative Physiology A, 147, 155–170.Find this resource:
de Vries, S. E. J., & Clandinin, T. R. (2012). Loom-sensitive neurons link computation to action in the Drosophila visual system. Current Biology, 22, 353–362.Find this resource:
Duistermars, B. J., Chow, D. M., Condro, M., & Frye, M. A. (2007). The spatial, temporal and contrast properties of expansion and rotation flight optomotor responses in Drosophila. Journal of Experimental Biology, 210, 3218–3227.Find this resource:
Dvorak, D. R., Bishop, L. G., & Eckert, H. E. (1975). On the identification of movement detectors in the fly optic lobe. Journal of Comparative Physiology A, 100, 5–23.Find this resource:
Egelhaaf, M. (1985a). On the neuronal basis of figure-ground discrimination by relative motion in the visual system of the fly. II. Figure-detection cells, a new class of visual interneurons. Biology Cybernetics, 52, 195–209.Find this resource:
Egelhaaf, M. (1985b). On the neuronal basis of figure-ground discrimination by relative motion in the visual system of the fly. III. Possible input circuitries and behavioural significance of the FD-cells. Biology Cybernetics, 52, 267–280.Find this resource:
Egelhaaf, M., Boeddeker, N., Kern, R., Kurtz, R., & Lindemann, J. P. (2012). Spatial vision in insects is facilitated by shaping the dynamics of visual input through behavioral action. Frontiers in Neural Circuits, 6, 108.Find this resource:
Egelhaaf, M., Kern, R., Krapp, H. G., Kretzberg, J., Kurtz, R., & Warzecha, A.-K. (2002). Neural encoding of behaviourally relevant visual-motion information in the fly. Trends in Neurosciences, 25, 96–102.Find this resource:
Egelhaaf, M., Kern, R., & Lindemann, J. P. (2014). Motion as a source of environmental information: a fresh view on biological motion computation by insect brains. Frontiers in Neural Circuits, 8, 127.Find this resource:
Eichner, H., Joesch, M., Schnell, B., Reiff, D. F., & Borst, A. (2011). Internal structure of the fly elementary motion detector. Neuron, 70, 1155–1164.Find this resource:
Elyada, Y. M., Haag, J., & Borst, A. (2009). Different receptive fields in axons and dendrites underlie robust coding in motion-sensitive neurons. Nature Neuroscience, 12, 327–332.Find this resource:
Fain, G. L., Hardie, R., & Laughlin, S. B. (2010). Phototransduction and the evolution of photoreceptors. Current Biology, 20, R114–R124.Find this resource:
Farina, W. M., Varjú, D., & Zhou, Y. (1994). The regulation of distance to dummy flowers during hovering flight in the hawk moth Macroglossum stellatarum. Journal of Comparative Physiology A, 174, 239–247.Find this resource:
Farrow, K., Haag, J., & Borst, A. (2006). Nonlinear, binocular interactions underlying flow field selectivity of a motion-sensitive neuron. Nature Neuroscience, 9, 1312–1320.Find this resource:
Fischbach, K. F., & Dittrich, A. P. M. (1989). The optic lobe of Drosophila melanogaster. I. A Golgi analysis of wild-type structure. Cell Tissue Research, 258, 441–475.Find this resource:
Fisher, Y. E., Leong, J. C. S., Sporar, K., Ketkar, M. D., Gohl, D. M., Clandinin, T. R., & Silies, M. (2015a). A class of visual neurons with wide-field properties is required for local motion detection. Current Biology, 25, 3178–3189.Find this resource:
Fisher, Y. E., Silies, M., & Clandinin, T. R. (2015b). Orientation selectivity sharpens motion detection in Drosophila. Neuron, 88, 390–402.Find this resource:
Fotowat, H., Fayyazuddin, A., Bellen, H. J., & Gabbiani, F. (2009). A novel neuronal pathway for visually guided escape in Drosophila melanogaster. Journal of Neurophysiology, 102, 875–885.Find this resource:
Fotowat, H., & Gabbiani, F. (2007). Relationship between the phases of sensory and motor activity during a looming-evoked multistage escape behavior. Journal of Neuroscience, 27, 10047–10059.Find this resource:
Fotowat, H., & Gabbiani, F. (2011). Collision detection as a model for sensory-motor integration. Annual Reviews in Neuroscience, 34, 1–19.Find this resource:
Fotowat, H., Harrison, R. R., & Gabbiani, F. (2011). Multiplexing of motor information in the discharge of a collision detecting neuron during escape behaviors. Neuron, 69, 147–158.Find this resource:
Franz, M. O., & Krapp, H. G. (2000). Wide-field, motion-sensitive neurons and matched filters for optic flow fields. Biology Cybernetics, 83, 185–197.Find this resource:
Frye, M. A. (2010). Multisensory systems integration for high-performance motor control in flies. Current Opinions in Neurobiology, 20, 347–352.Find this resource:
Frye, M. A., & Olberg, R. M. (1995). Visual receptive field properties of feature detecting neurons in the dragonfly. Journal of Comparative Physiology A, 177, 569–576.Find this resource:
Fuller, S. B., Straw, A. D., Peek, M. Y., Murray, R. M., & Dickinson, M. H. (2014). Flying Drosophila stabilize their vision-based velocity controller by sensing wind with their antennae. Proceedings of the National Academy of Sciences USA, 111, 1182–1191.Find this resource:
Gabbiani, F., Krapp, H. G., Koch, C., & Laurent, G. (2002). Multiplicative computation in a visual neuron sensitive to looming. Nature, 420, 320–324.Find this resource:
Gabbiani, F., Krapp, H. G., & Laurent, G. (1999). Computation of object approach by a wide-field, motion-sensitive neuron. Journal of Neuroscience, 19, 1122–1141.Find this resource:
Geiger, G., & Nässel, D. R. (1981). Visual orientation behaviour of flies after selective laser beam ablation of interneurones. Nature, 293, 398–399.Find this resource:
Geurten, B. R. H., Nordström, K., Sprayberry, J. D. H., Bolzon, D. M., & O’Carroll, D. C. (2007). Neural mechanisms underlying target detection in a dragonfly centrifugal neuron. Journal of Experimental Biology, 210, 3277–3284.Find this resource:
Gibson, J. J. (1950). The perception of the visual world. Boston, MA: Houghton Mifflin Company.Find this resource:
Gonzalez-Bellido, P. T., Peng, H., Yang, J., Georgopoulos, A. P., & Olberg, R. M. (2013). Eight pairs of descending visual neurons in the dragonfly give wing motor centers accurate population vector of prey direction. Proceedings of the National Academy of Sciences USA, 110, 696–701.Find this resource:
Goodman, L. J. (1960). The landing responses of insects. Journal of Experimental Biology, 37, 854–878.Find this resource:
Haag, J., & Borst, A. (2001). Recurrent network interactions underlying flow-field selectivity of visual interneurons. Journal of Neuroscience, 21, 5685–5692.Find this resource:
Haag, J., & Borst, A. (2004). Neural mechanism underlying complex receptive field properties of motion-sensitive interneurons. Nature Neuroscience, 7, 628–634.Find this resource:
Haag, J., & Borst, A. (2007). Reciprocal inhibitory connections within a neural network for rotational optic-flow processing. Frontiers in Neuroscience, 1, 111–121.Find this resource:
Haag, J., Wertz, A., & Borst, A. (2007). Integration of lobula plate output signals by DNOVS1, an identified premotor descending neuron. Journal of Neuroscience, 27, 1992–2000.Find this resource:
Haag, J., Wertz, A., & Borst, A. (2010). Central gating of fly optomotor response. Proceedings of the National Academy of Sciences USA, 107, 20104–20109.Find this resource:
Haikala, V., Joesch, M., Borst, A., & Mauss, A. S. (2013). Optogenetic control of fly optomotor responses. Journal of Neuroscience, 33, 13927–13934.Find this resource:
Hardie, R. C., & Franze, K. (2012). Photomechanical responses in Drosophila photoreceptors. Science, 338, 260–263.Find this resource:
Hardie, R. C., & Juusola, M. (2015). Phototransduction in Drosophila. Current Opinions in Neurobiology, 34, 37–45.Find this resource:
Hassenstein, B., & Reichardt, W. (1956). Systemtheoretische Analyse der Zeit-, Reihenfolgen- und Vorzeichenauswertung bei der Bewegungsperzeption des Rüsselkäfers Chlorophanus. Zeitschrift für Naturforschung, 11b, 513–524.Find this resource:
Hateren, J., & Schilstra, C. (1999). Blowfly flight and optic flow. II. Head movements during flight. Journal of Experimental Biology, 202 (Pt 11), 1491–1500.Find this resource:
Hatsopoulos, N., Gabbiani, F., & Laurent, G. (1995). Elementary computation of object approach by wide-field visual neuron. Science, 270, 1000–1003.Find this resource:
Hausen, K. (1976). Functional characterization and anatomical identification of motion sensitive neurons in the lobula plate of the blowfly Calliphora erythrocephala. Zeitschrift für Naturforschung, 31c, 629–633.Find this resource:
Hausen, K. (1982). Motion sensitive interneurons in the optomotor system of the fly. Biology Cybernetics, 46, 67–79.Find this resource:
Hausen, K. (1984). The lobula-complex of the fly: Structure, function and significance in visual behaviour. In M. A. Ali (Ed.), Photoreception and vision in invertebrates (pp. 523–559). New York, NY: Plenum Press.Find this resource:
Hausen, K., & Egelhaaf, M. (1989). Neural mechanisms of visual course control in insects. In D. G. Stavenga et al. (Eds.), Facets of vision (pp. 391–424). Berlin, Germany: Springer-Verlag.Find this resource:
Hausen, K., & Wehrhahn, C. (1983). Microsurgical lesion of horizontal cells changes optomotor yaw responses in the blowfly Calliphora erythrocephala. Proceedings of Biological Sciences, 219, 211–216.Find this resource:
Hausen, K., Wolburg-Buchholz, W., & Ribi, W. A. (1980). The synaptic organization of visual interneurons in the lobula complex of flies. A light and electron microscopical study using silver-intensified cobalt-impregnations. Cell Tissue Research, 208, 371–387.Find this resource:
Heisenberg, M., Wonneberger, R., & Wolf, R. (1978). Optomotor-blindH31—a Drosophila mutant of the lobula plate giant neurons. Journal of Comparative Physiology A, 124, 287–296.Find this resource:
Hemmi, J. M., & Tomsic, D. (2012). The neuroethology of escape in crabs: from sensory ecology to neurons and back. Current Opinion in Neurobiology, 22, 194–200.Find this resource:
Hengstenberg, R., Hausen, K., & Hengstenberg, B. (1982). The number and structure of giant vertical cells (VS) in the lobula plate of the blowfly Calliphora erythrocephala. Journal of Comparative Physiology A, 149, 163–177.Find this resource:
Holst von, E., & Mittelstaedt, H. (1950). Das Reafferenzprinzip. Wechselwirkungen zwischen Zentralnervensystem und Peripherie. Naturwissenschaften, 20, 464–476.Find this resource:
Hopp, E., Borst, A., & Haag, J. (2014). Subcellular mapping of dendritic activity in optic flow processing neurons. Journal of Comparative Physiology A, 200, 359–370.Find this resource:
Horseman, B. G., Macauley, M. W. S., & Barnes, W. J. P. (2011). Neuronal processing of translational optic flow in the visual system of the shore crab Carcinus maenas. Journal of Experimental Biology, 214, 1586–1598.Find this resource:
Horstmann, W., Egelhaaf, M., & Warzecha, A. K. (2000). Synaptic interactions increase optic flow specificity. European Journal of Neuroscience, 12, 2157–2165.Find this resource:
Huang, J., Liu, C.-H., Hughes, S. A., Postma, M., Schwiening, C. J., & Hardie, R. C. (2010). Activation of TRP channels by protons and phosphoinositide depletion in Drosophila photoreceptors. Current Biology, 20, 189–197.Find this resource:
Huston, S. J., & Jayaraman, V. (2011). Studying sensorimotor integration in insects. Current Opinion in Neurobiology, 21, 527–534.Find this resource:
Huston, S. J., & Krapp, H. G. (2008). Visuomotor transformation in the fly gaze stabilization system. PLoS Biology, 6, e173.Find this resource:
Huston, S. J., & Krapp, H. G. (2009). Nonlinear integration of visual and haltere inputs in fly neck motor neurons. Journal of Neuroscience, 29, 13097–13105.Find this resource:
Ibbotson, M. R. (1991). Wide-field motion-sensitive neurons tuned to horizontal movement in the honeybee, Apis mellifera. Journal of Comparative Physiology A, 168, 91–102.Find this resource:
Ibbotson, M. R., & Goodman, L. J. (1990). Response characteristics of four wide-field motion-sensitive descending interneurones in Apis Melifera. Journal of Experimental Biology, 148, 255–279.Find this resource:
Ibbotson, M. R., Maddess, T., & DuBois, R. (1991). A system of insect neurons sensitive to horizontal and vertical image motion connects the medulla and midbrain. Journal of Comparative Physiology A, 169, 355–367.Find this resource:
Joesch, M., Plett, J., Borst, A., & Reiff, D. F. (2008). Response properties of motion-sensitive visual interneurons in the lobula plate of Drosophila melanogaster. Current Biology, 18, 368–374.Find this resource:
Joesch, M., Schnell, B., Raghu, S. V., Reiff, D. F., & Borst, A. (2010). ON and OFF pathways in Drosophila motion vision. Nature, 468, 300–304.Find this resource:
Joesch, M., Weber, F., Eichner, H., & Borst, A. (2013). Functional specialization of parallel motion detection circuits in the fly. Journal of Neuroscience, 33, 902–905.Find this resource:
Jones, P. W., & Gabbiani, F. (2010). Synchronized neural input shapes stimulus selectivity in a collision-detecting neuron. Current Biology, 20, 2052–2057.Find this resource:
Jung, S. N., Borst, A., & Haag, J. (2011). Flight activity alters velocity tuning of fly motion-sensitive neurons. Journal of Neuroscience, 31, 9231–9237.Find this resource:
Kaiser, D. W., & Bishop, L. G. (1970). Directionally selective motion detecting units in the optic lobe of the honeybee. Journal of Comparative Physiology A, 67, 403–413.Find this resource:
Kaiser, W., & Steiner-Kaiser, J. (1983). Neuronal correlates of sleep, wakefulness and arousal in a diurnal insect. Nature, 301, 707–709.Find this resource:
Karmeier, K., van Hateren, J. H., Kern, R., & Egelhaaf, M. (2006). Encoding of naturalistic optic flow by a population of blowfly motion-sensitive neurons. Journal of Neurophysiology, 96, 1602–1614.Find this resource:
Kathman, N. D., Kesavan, M., & Ritzmann, R. E. (2014). Encoding wide-field motion and direction in the central complex of the cockroach Blaberus discoidalis. Journal of Experimental Biology, 217, 4079–4090.Find this resource:
Kauer, I., Borst, A., & Haag, J. (2015). Complementary motion tuning in frontal nerve motor neurons of the blowfly. Journal of Comparative Physiology A, 201, 411–426.Find this resource:
Kien, J. (1974). Sensory integration in the locust optomotor system—II: Direction selective neurons in the circumoesophageal connectives and the optic lobe. Vision Research, 14, 1255–1268.Find this resource:
Kim, A. J., Fitzgerald, J. K., & Maimon, G. (2015). Cellular evidence for efference copy in Drosophila visuomotor processing. Nature Neuroscience, 18, 1247–1255.Find this resource:
Kirschfeld, K. (1967). Die Projektion der optischen Umwelt auf das Raster der Rhabdomere im Komplexauge von Musca. Experimental Brain Research, 3, 248–270.Find this resource:
Krapp, H. G., & Hengstenberg, R. (1996). Estimation of self-motion by optic flow processing in single visual interneurons. Nature, 384, 463–466.Find this resource:
Krapp, H. G., Hengstenberg, R., & Egelhaaf, M. (2001). Binocular contributions to optic flow processing in the fly visual system. Journal of Neurophysiology, 85, 724–734.Find this resource:
Krapp, H. G., Hengstenberg, B., & Hengstenberg, R. (1998). Dendritic structure and receptive-field organization of optic flow processing interneurons in the fly. Journal of Neurophysiology, 79, 1902–1917.Find this resource:
Krapp, H. G., & Wicklein, M. (2008). Central processing of visual information in insects. In A. I. Basbaum et al. (Eds.), The senses: A comprehensive reference (pp. 131–204). San Diego, CA: Academic Press.Find this resource:
Land, M. F., & Collett, T. S. (1974). Chasing behaviour of houseflies (Fannia canicularis). Journal of Comparative Physiology A, 89, 331–357.Find this resource:
Land, M. F., & Nilsson, D.-E. (2012). Animal eyes. Oxford, UK: Oxford University Press.Find this resource:
Lehrer, M., Srinivasan, M. V., Zhang, S. W., & Horridge, G. A. (1988). Motion cues provide the bee’s visual world with a third dimension. Nature, 332, 356–357.Find this resource:
Longden, K. D., & Krapp, H. G. (2009). State-dependent performance of optic-flow processing interneurons. Journal of Neurophysiology, 102, 3606–3618.Find this resource:
Longden, K. D., & Krapp, H. G. (2010). Octopaminergic modulation of temporal frequency coding in an identified optic flow-processing interneuron. Frontiers in System Neuroscience, 4, 153.Find this resource:
Maimon, G., Straw, A. D., & Dickinson, M. H. (2008). A simple vision-based algorithm for decision making in flying Drosophila. Current Biology, 18, 464–470.Find this resource:
Maimon, G., Straw, A. D., & Dickinson, M. H. (2010). Active flight increases the gain of visual motion processing in Drosophila. Nature Neuroscience, 13, 393–399.Find this resource:
Maisak, M. S., Haag, J., Ammer, G., Serbe, E., Meier, M., Leonhardt, A., Schilling, T., Bahl, A., Rubin, G. M., Nern, A., Dickson, B. J., Reiff, D. F., Hopp, E., & Borst, A. (2013). A directional tuning map of Drosophila elementary motion detectors. Nature, 500, 212–216.Find this resource:
Matheson, T., Rogers, S. M., & Krapp, H. G. (2004). Plasticity in the visual system is correlated with a change in lifestyle of solitarious and gregarious locusts. Journal of Neurophysiology, 91, 1–12.Find this resource:
Mauss, A. S., Meier, M., Serbe, E., & Borst, A. (2014). Optogenetic and pharmacologic dissection of feedforward inhibition in Drosophila motion vision. Journal of Neuroscience, 34, 2254–2263.Find this resource:
Mauss, A. S., Pankova, K., Arenz, A., Nern, A., Rubin, G. M., & Borst, A. (2015). Neural circuit to integrate opposing motions in the visual field. Cell, 162, 351–362.Find this resource:
Medan, V., Berón De Astrada, M., Scarano, F., & Tomsic, D. (2015). A network of visual motion-sensitive neurons for computing object position in an arthropod. Journal of Neuroscience, 35, 6654–6666.Find this resource:
Medan, V., Oliva, D., & Tomsic, D. (2007). Characterization of lobula giant neurons responsive to visual stimuli that elicit escape behaviors in the crab Chasmagnathus. Journal of Neurophysiology, 98, 2414–2428.Find this resource:
Meier, M., Serbe, E., Maisak, M. S., Haag, J., Dickson, B. J., & Borst, A. (2014). Neural circuit components of the Drosophila OFF motion vision pathway. Current Biology, 24, 385–392.Find this resource:
Menda, G., Shamble, P. S., Nitzany, E. I., Golden, J. R., & Hoy, R. R. (2014). Visual perception in the brain of a jumping spider. Current Biology, 24, 2580–2585.Find this resource:
Milde, J. J. (1993). Tangential medulla neurons in the moth Manduca sexta. Structure and responses to optomotor stimuli. Journal of Comparative Physiology A, 173, 783–799.Find this resource:
Milde, J. J., Seyan, H. S., & Strausfeld, N. J. (1987). The neck motor system of the fly Calliphora erythrocephala. Journal of Comparative Physiology A, 160, 225–238.Find this resource:
Mischiati, M., Lin, H.-T., Herold, P., Imler, E., Olberg, R., & Leonardo, A. (2015). Internal models direct dragonfly interception steering. Nature, 517, 333–338.Find this resource:
Mronz, M., & Lehmann, F.-O. (2008). The free-flight response of Drosophila to motion of the visual environment. Journal of Experimental Biology, 211, 2026–2045.Find this resource:
Muijres, F. T., Elzinga, M. J., Melis, J. M., & Dickinson, M. H. (2014). Flies evade looming targets by executing rapid visually directed banked turns. Science, 344, 172–177.Find this resource:
Nityananda, V., Tarawneh, G., Rosner, R., Nicolas, J., Crichton, S., & Read, J. (2016). Insect stereopsis demonstrated using a 3D insect cinema. Scientific Reports, 6, 18718.Find this resource:
Nordström, K. (2012). Neural specializations for small target detection in insects. Current Opinion in Neurobiology, 22, 272–278.Find this resource:
Nordström, K., Barnett, P. D., & O’Carroll, D. C. (2006). Insect detection of small targets moving in visual clutter. PLoS Biology, 4, e54.Find this resource:
Nordström, K., & O’Carroll, D. C. (2009). Feature detection and the hypercomplex property in insects. Trends in Neurosciences, 32, 383–391.Find this resource:
O’Carroll, D. (1993). Feature-detecting neurons in dragonflies. Nature, 362, 541–543.Find this resource:
Olberg, R. M. (1981a). Object- and self-movement detectors in the ventral nerve cord of the dragonfly. Journal of Comparative Physiology A, 141, 327–334.Find this resource:
Olberg, R. M. (1981b). Parallel encoding of direction of wind, head, abdomen, and visual pattern movement by single interneurons in the dragonfly. Journal of Comparative Physiology A, 142, 27–41.Find this resource:
Olberg, R. M. (1986). Identified target-selective visual interneurons descending from the dragonfly brain. Journal of Comparative Physiology A, 159, 827–840.Find this resource:
Olberg, R. M. (2012). Visual control of prey-capture flight in dragonflies. Current Opinion in Neurobiology, 22, 267–271.Find this resource:
Olberg, R. M., Seaman, R. C., Coats, M. I., & Henry, A. F. (2007). Eye movements and target fixation during dragonfly prey-interception flights. Journal of Comparative Physiology A, 193, 685–693.Find this resource:
Olberg, R. M., Worthington, A. H., & Venator, K. R. (2000). Prey pursuit and interception in dragonflies. Journal of Comparative Physiology A, 186, 155–162.Find this resource:
Oliva, D., Medan, V., & Tomsic, D. (2007). Escape behavior and neuronal responses to looming stimuli in the crab Chasmagnathus granulatus (Decapoda: Grapsidae). Journal of Experimental Biology, 210, 865–880.Find this resource:
Oliva, D., & Tomsic, D. (2014). Computation of object approach by a system of visual motion-sensitive neurons in the crab Neohelice. Journal of Neurophysiology, 112, 1477–1490.Find this resource:
Osorio, D., & Bacon, J. P. (1994). A good eye for arthropod evolution. Bioessays, 16, 419–424.Find this resource:
Parsons, M. M., Krapp, H. G., & Laughlin, S. B. (2010). Sensor fusion in identified visual interneurons. Current Biology, 20, 624–628.Find this resource:
Peron, S. P., Jones, P. W., & Gabbiani, F. (2009). Precise subcellular input retinotopy and its computational consequences in an identified visual interneuron. Neuron, 63, 830–842.Find this resource:
Petrowitz, R., Dahmen, H., Egelhaaf, M., & Krapp, H. G. (2000). Arrangement of optical axes and spatial resolution in the compound eye of the female blowfly Calliphora. Journal of Comparative Physiology A, 186, 737–746.Find this resource:
Power, M. E. (1948). The thoracico‐abdominal nervous system of an adult insect, Drosophila melanogaster. Journal of Comparative Neurology, 88, 347–409.Find this resource:
Pringle, J. W. S. (1948). The gyroscopic mechanism of the halteres of Diptera. Philosophical Transactions of the Royal Society B, 233, 347–384.Find this resource:
Ready, D. F., Hanson, T. E., & Benzer, S. (1976). Development of the Drosophila retina, a neurocrystalline lattice. Development Biology, 53, 217–240.Find this resource:
Reyn von, C. R., Breads, P., Peek, M. Y., Zheng, G. Z., Williamson, W. R., Yee, A. L., Leonardo, A., & Card, G. M. (2014). A spike-timing mechanism for action selection. Nature Neuroscience, 17, 962–970.Find this resource:
Rind, F. C. (1983). A directionally sensitive motion detecting neurone in the brain of a moth. Journal of Experimental Biology, 102, 253–271.Find this resource:
Rind, F. C. (1990). A directionally selective motion-detecting neurone in the brain of the locust: physiological and morphological characterization. Journal of Experimental Biology, 149, 1–19.Find this resource:
Rind, F. C., & Simmons, P. J. (1992). Orthopteran DCMD neuron: a reevaluation of responses to moving objects. I. Selective responses to approaching objects. Journal of Neurophysiology, 68, 1654–1666.Find this resource:
Rister, J., Pauls, D., Schnell, B., Ting, C.-Y., Lee, C.-H., Sinakevitch, I., Morante, J., Strausfeld, N. J., Ito, K., & Heisenberg, M. (2007). Dissection of the peripheral motion channel in the visual system of Drosophila melanogaster. Neuron, 56, 155–170.Find this resource:
Sane, S. P., Dieudonné, A., Willis, M. A., Daniel, T. L. (2007). Antennal mechanosensors mediate flight control in moths. Science. 315, 863–866.Find this resource:
Santer, R. D., Yamawaki, Y., Rind, F. C., & Simmons, P. J. (2005). Motor activity and trajectory control during escape jumping in the locust Locusta migratoria. Journal of Comparative Physiology A, 191, 965–975.Find this resource:
Schilling, T., & Borst, A. (2015). Local motion detectors are required for the computation of expansion flow-fields. Open Biology, 4, 1105–1108.Find this resource:
Schlotterer, G. R. (1977). Response of the locust descending movement detector neuron to rapidly approaching and withdrawing visual stimuli. Canadian Journal of Zoology, 55, 1372–1376.Find this resource:
Schnell, B., Joesch, M., Forstner, F., Raghu, S. V., Otsuna, H., Ito, K., Borst, A., & Reiff, D. F. (2010). Processing of horizontal optic flow in three visual interneurons of the Drosophila brain. Journal of Neurophysiology, 103, 1646–1657.Find this resource:
Schnell, B., Raghu, S. V., Nern, A., & Borst, A. (2012). Columnar cells necessary for motion responses of wide-field visual interneurons in Drosophila. Journal of Comparative Physiology A, 198, 389–395.Find this resource:
Schnell, B., Weir, P. T., Roth, E., Fairhall, A. L., & Dickinson, M. H. (2014). Cellular mechanisms for integral feedback in visually guided behavior. Proceedings of the National Academy of Sciences USA, 111, 5700–5705.Find this resource:
Schuster, S., Strauss, R., & Götz, K. G. (2002). Virtual-reality techniques resolve the visual cues used by fruit flies to evaluate object distances. Current Biology, 12, 1591–1594.Find this resource:
Schwegmann, A., Lindemann, J. P., & Egelhaaf, M. (2014). Depth information in natural environments derived from optic flow by insect motion detection system: A model analysis. Frontiers in Comparative Neuroscience, 8, 994.Find this resource:
Scott, E. K., Raabe, T., & Luo, L. (2002). Structure of the vertical and horizontal system neurons of the lobula plate in Drosophila. Journal of Comparative Neurology, 454, 470–481.Find this resource:
Serbe, E., Meier, M., Leonhardt, A., & Borst, A. (2016). Comprehensive characterization of the major presynaptic elements to the Drosophila OFF motion detector. Neuron, 89, 1–13.Find this resource:
Sherman, A., & Dickinson, M. H. (2004). Summation of visual and mechanosensory feedback in Drosophila flight control. Journal of Experimental Biology, 207, 133–142.Find this resource:
Shinomiya, K., Karuppudurai, T., Lin, T.-Y., Lu, Z., Lee, C.-H., & Meinertzhagen, I. A. (2014). Candidate neural substrates for off-edge motion detection in Drosophila. Current Biology, 24, 1062–1070.Find this resource:
Silies, M., Gohl, D. M., Fisher, Y. E., Freifeld, L., Clark, D. A., & Clandinin, T. R. (2013). Modular use of peripheral input channels tunes motion-detecting circuitry. Neuron, 79, 111–127.Find this resource:
Sinakevitch, I., & Strausfeld, N. J. (2006). Comparison of octopamine-like immunoreactivity in the brains of the fruit fly and blow fly. Journal of Comparative Neurology, 494, 460–475.Find this resource:
Sobel E. C. (1990). The locust’s use of motion parallax to measure distance. Journal of Comparative Physiology A, 167, 579–588.Find this resource:
Srinivasan, M. V., Zhang, S., & Chahl, J. S. (2001). Landing strategies in honeybees, and possible applications to autonomous airborne vehicles. Biology Bulletin, 200, 216–221.Find this resource:
Srinivasan, M. V., Zhang, S. W., Chahl, J. S., Barth, E., & Venkatesh, S. (2000). How honeybees make grazing landings on flat surfaces. Biology Cybernetics, 83, 171–183.Find this resource:
Stange, G., Stowe, S., Chahl, J. S., & Massaro, A. (2002). Anisotropic imaging in the dragonfly median ocellus: A matched filter for horizon detection. Journal of Comparative Physiology A, 188, 455–467.Find this resource:
Strausfeld, N. J. (1970). The optic lobes of Diptera. Philosophical Transactions of the Royal Society B, 258, 135–223.Find this resource:
Strausfeld, N. J. (2005). The evolution of crustacean and insect optic lobes and the origins of chiasmata. Arthropod Structure & Development, 34, 235–256.Find this resource:
Strausfeld, N. J., Seyan, H. S., & Milde, J. J. (1987). The neck motor system of the fly Calliphora erythrocephala. Journal of Comparative Physiology A, 160, 205–224.Find this resource:
Suver, M. P., Mamiya, A., & Dickinson, M. H. (2012). Octopamine neurons mediate flight-induced modulation of visual processing in Drosophila. Current Biology, 22, 2294–2302.Find this resource:
Sztarker, J., & Tomsic, D. (2008). Neuronal correlates of the visually elicited escape response of the crab Chasmagnathus upon seasonal variations, stimuli changes and perceptual alterations. Journal of Comparative Physiology A, 194, 587–596.Find this resource:
Takemura, S.-Y., Bharioke, A., Lu, Z., Nern, A., Vitaladevuni, S., Rivlin, P. K., … Chklovskii, D. B. (2013). A visual motion detection circuit suggested by Drosophila connectomics. Nature, 500, 175–181.Find this resource:
Tammero, L. F., & Dickinson, M. H. (2002). Collision-avoidance and landing responses are mediated by separate pathways in the fruit fly, Drosophila melanogaster. Journal of Experimental Biology, 205, 2785–2798.Find this resource:
Tammero, L. F., Frye, M. A., & Dickinson, M. H. (2004). Spatial organization of visuomotor reflexes in Drosophila. Journal of Experimental Biology, 207, 113–122.Find this resource:
Tuthill, J. C., Nern, A., Holtz, S. L., Rubin, G. M., & Reiser, M. B. (2013). Contributions of the 12 neuron classes in the fly lamina to motion vision. Neuron, 79, 128–140.Find this resource:
Tuthill, J. C., Nern, A., Rubin, G. M., & Reiser, M. B. (2014). Wide-field feedback neurons dynamically tune early visual processing. Neuron, 82, 887–895.Find this resource:
Venken, K. J. T., Simpson, J. H., & Bellen, H. J. (2011). Genetic manipulation of genes and cells in the nervous system of the fruit fly. Neuron, 72, 202–230.Find this resource:
Wardill, T. J., Knowles, K., Barlow, L., Tapia, G., Nordström, K., Olberg, R. M., & Gonzalez-Bellido, P. T. (2015). The killer fly hunger games: Target size and speed predict decision to pursuit. Brain Behavior Evolution, 86, 28–37.Find this resource:
Wardill, T. J., List, O., Li, X., Dongre, S., McCulloch, M., Ting, C.-Y., … Juusola, M. (2012). Multiple spectral inputs improve motion discrimination in the Drosophila visual system. Science, 336, 925–931.Find this resource:
Warzecha, A. K., Egelhaaf, M., & Borst, A. (1993). Neural circuit tuning fly visual interneurons to motion of small objects. I. Dissection of the circuit by pharmacological and photoinactivation techniques. Journal of Neurophysiology, 69, 329–339.Find this resource:
Wasserman, S. M., Aptekar, J. W., Lu, P., Nguyen, J., Wang, A. L., Keleş, M. F., … Frye, M. A. (2015). Olfactory neuromodulation of motion vision circuitry in Drosophila. Current Biology, 25, 467–472.Find this resource:
Wehner, R. (1987). “Matched filters”—neural models of the external world. Journal of Comparative Physiology A, 161, 511–531.Find this resource:
Wehrhahn, C., Poggio, T., & Bülthoff, H. (1982). Tracking and chasing in houseflies (Musca). Biology Cybernetics, 45, 123–130.Find this resource:
Wertz, A., Borst, A., & Haag, J. (2008). Nonlinear integration of binocular optic flow by DNOVS2, a descending neuron of the fly. Journal of Neuroscience, 28, 3131–3140.Find this resource:
Wertz, A., Gaub, B., Plett, J., Haag, J., & Borst, A. (2009a). Robust coding of ego-motion in descending neurons of the fly. Journal of Neuroscience, 29, 14993–15000.Find this resource:
Wertz, A., Haag, J., & Borst, A. (2009b). Local and global motion preferences in descending neurons of the fly. Journal of Comparative Physiology A, 195, 1107–1120.Find this resource:
Wertz, A., Haag, J., & Borst, A. (2012). Integration of binocular optic flow in cervical neck motor neurons of the fly. Journal of Comparative Physiology A, 198, 655–668.Find this resource:
Wicklein, M., & Strausfeld, N. J. (2000). Organization and significance of neurons that detect change of visual depth in the hawk moth Manduca sexta. Journal of Comparative Neurology, 424, 356–376.Find this resource:
Wiederman, S. D., Shoemaker, P. A., & O’Carroll, D. C. (2008). A model for the detection of moving targets in visual clutter inspired by insect physiology. PLoS ONE, 3, e2784.Find this resource:
Wiederman, S. D., Shoemaker, P. A., & O’Carroll, D. C. (2013). Correlation between OFF and ON channels underlies dark target selectivity in an insect visual system. Journal of Neuroscience, 33, 13225–13232.Find this resource: