A New Perspective on Predictive Motor Signaling

Current Biology

Review A New Perspective on Predictive Motor Signaling Hans Straka1, John Simmers2,*, and Boris P. Chagnaud1 1Department

Biology II, Ludwig-Maximilians-University Munich, Grosshaderner Str. 2, 82152 Planegg, Germany
gratives d’Aquitaine, CNRS UMR 5287, 33076 Bordeaux, France of Bordeaux, Institut de Neurosciences Cognitives et Inte *Correspondence: [email protected] https://doi.org/10.1016/j.cub.2018.01.033

2University

Adaptive behavior relies on complex neural processing in multiple interacting networks of both motor and sensory systems. One such interaction employs intrinsic neuronal signals, so-called ‘corollary discharge’ or ‘efference copy’, that may be used to predict the sensory consequences of a specific behavioral action, thereby enabling self-generated (reafferent) sensory information and extrinsic (exafferent) sensory inflow to be dissociated. Here, by using well-established examples, we seek to identify the distinguishing features of corollary discharge and efference copy within the framework of predictive motor-to-sensory system coordination. We then extend the more general concept of predictive signaling by showing how neural replicas of a particular motor command not only inform sensory pathways in order to gate reafferent stimulation, but can also be used to directly coordinate distinct and otherwise independent behaviors to the original motor task. Moreover, this motor-to-motor pairing may additionally extend to a gating of sensory input to either or both of the coupled systems. The employment of predictive internal signaling in such motor systems coupling and remote sensory input control thus adds to our understanding of how an organism’s central nervous system is able to coordinate the activity of multiple and generally disparate motor and sensory circuits in the production of effective behavior. Introduction Animal behavior depends on the activation of centrally-generated motor programs that range from the production of simple reflexes to complex, voluntary movements mediated by multiple and often distributed muscle groups. A consequence of all such motor actions is the activation of one or many sensory systems, with so-called ‘reafference’ arising from self-movement potentially combining with externally-derived ‘exafferent’ sensory inputs (Figure 1). In the absence of appropriate counter-measures, a misinterpreted perception of the environment would occur because the animal is unable to distinguish between sensory inflow resulting from external events and information due to stimulation by the self-generated motor action itself. The activation of reafferent sensory signals can arise either directly, for example from the proprioceptive sensing of muscle movement, or indirectly, for example from a displacement of sensory epithelia or of an entire sense organ relative to the self-stimulus source. How, for example, do primates suppress the perception of self-produced retinal image motion during saccadic eye movements so that the scanned visual world is perceived as stationary? Similarly, how can electrogenic fish continuously extract information about their electrical sensory environment despite undulatory displacements of their electroreceptors during swimming movements? In order for animals to predict the sensory consequences of their own behavior and thereby enable behaviorally relevant stimuli to be processed, motor systems implement intrinsic neural representations of the actual movement command, originally designated ‘corollary discharge’ [1] or ‘efference copies’ [2] and are terms still widely used by investigators. These predictive signals are forwarded to various levels of sensory processing where they serve to attenuate or eliminate the self-generated reafferent component (Figure 1). The employment of such a predictive,

feed-forward mechanism in sensory processing is widespread throughout the animal kingdom, with examples ranging from visuo-motor processing in flies [3], electroreception in fish [4], body motion detection in amphibians [5] and primates [6] and vocalization in amphibians, fish, birds and mammals [7] to the inability to tickle oneself [8,9], inner speech production [10] and motor learning in humans [11]. Moreover, a failure to disambiguate self- from externally-induced sensory input may be causal to neuropathologies such as schizophrenia [12]. Whereas previous studies of this type of neural signaling have almost exclusively focused on motor predictions within the context of individual sensorimotor systems, recent studies have shown that feed-forward copies of a behavioral command may also impinge on other motor behaviors. Such motor-tomotor coupling in which a neural representation of one motor output may directly influence distinct and otherwise independent motor control systems thus widens the extent to which predictive signaling is employed in nervous system function. In this review, we first define our view of ‘corollary discharge’ and ‘efference copy’ as two closely related, yet distinguishable, constituents of predictive signaling generally with respect to origin, signal content and function. By drawing on several key examples, we will underline these differences, but also the overall difficulties in making a clear functional separation, by briefly surveying existing knowledge on the traditional role of these feed-forward motor signals in motor-to-sensory processing. We then present new evidence for predictive motor-tomotor signaling and describe how this coordinating influence is also able to instruct remote sensory pathways about the reafferent consequences of movement. With our conceptual approach, we ultimately aim at synthesizing the various interactive strategies within the more general framework of motor prediction.

R232 Current Biology 28, R232–R243, March 5, 2018 ª 2018 Elsevier Ltd.

Current Biology

Review Efference copy / Corollary discharge Motor system

Sensory system

Prediction of sensory feedback Reafference Actual sensory feedback

Motor command Behavior

Exafference Environment

External sensory stimuli Current Biology

Figure 1. The interaction between motor and sensory systems during the generation of behavior. Efference copy or corollary discharge, as internal predictors of the sensory consequences of a given behavior, differentially influence the processing of reafferent and exafferent sensory information.

Corollary Discharge versus Efference Copy Various aspects of the anatomical organization and role of corollary discharges/efference copies in motor-to-sensory prediction have been reviewed previously [13–19]. A common unifying principle is that these internal forward signals influence sensory processing by gating particular maladaptive constituents of the sensory inflow during a motor action. In several systems, the gating of sensory inflow occurs directly at the periphery to potentially prevent an over-stimulation of the primary sensor itself. During the escape behavior of crayfish, for example, the corollary discharge acts to prevent mechanosensory hair cells from becoming habituated during a tailflip so that they maintain their responsiveness [20,21]. In swimming dogfish, rhythmic sinusoidal movements of the trunk and tail produce water turbulences that consequently activate the hair cells of lateral line neuromasts, the mechanosensors for water motion along the body in fish and aquatic amphibians. During dogfish swimming, corollary discharges of the locomotor commands directly attenuate lateral line hair cell sensory signal transduction, presumably avoiding reafferent saturation [22,23]. A direct cancellation of sensory reafference during selfmotion, however, can also occur at the level of the spinal cord through so-called ‘primary afferent depolarization’ (PAD), involving a presynaptic depolarization of muscle and cutaneous afferents at the first central synapse [24,25]. This presynaptic inhibition, which to a large extent arises from the movement-producing neural circuitry itself, thus plays a powerful role in controlling the synaptic effectiveness of the sensory feedback signals expected from the actual motor activity [25]. Such a mechanism is widespread amongst locomotor systems in particular, ranging

from vertebrate swimming [26] to insect flying [27] and limbbased locomotion (for example [28,29]). In most known cases, however, the efference copy/corollary discharge signal is integrated with reafferent sensory input at one or more central levels along the ascending sensory processing pathway (Figure 2A). For example, during chirping behavior in crickets, one of the best understood systems of motor-to-sensory prediction at the cellular level, an identified interneuron rhythmically inhibits (both pre-synaptically and post-synaptically) central auditory signal processing in time with the animal’s own singing. As a consequence, central auditory sensitivity is diminished only during each phase of the selfgenerated chirp cycle, so that the peripheral sensor remains receptive to the auditory environment [16,30–32]. In weakly electric fish, neural motor copies of their own electric organ discharge are subtracted from concurrently occurring electroreceptive sensory inputs within the cerebellum-like electrosensory lobe, thereby enabling detection of electrical sensory cues from the environment [33]. Thus, the target site for corollary discharge/efference copy influences on sensory signal processing can vary substantially between different systems, with the functional impact ranging from compensation or attenuation to a complete cancellation of the reafferent signal component [18,34,35]. Although currently used interchangeably by many researchers, corollary discharge, as the term implies, was initially defined as a motor activity-related representation that reports to other brain areas about an impending motor act, without defining the specific temporal patterning of that signal [1] (reviewed in [18]; Figure 2A, left and 2B). On the other hand, an efference copy signal sent to an adjacent sensory pathway was defined as a close replica of the actual output command conveyed to a target effector (Figure 2A, right and 2C) [2]. In most cases, the exact source of efference copy signals in lower motor pathway stages remains unknown. As a direct projection pathway from motoneurons would require collateral axonal branching to distant targets that is unlikely and hitherto unknown (at least in vertebrates), the efferent signal presumably emanates from the local premotor/motor circuitry that ultimately generates the final output command, rather than directly from the motoneurons themselves (Figure 2C). This specificity therefore differs from the more generalized origin of corollary discharge that can arise from different levels of a descending motor control pathway (Figure 2B), from local circuits in cortical and subcortical nuclei to brainstem and spinal motor circuitry responsible for driving a given behavior. Such intrinsic predictive signals are thus likely to have a temporal structure that differs from that of the actual movement command [18]. Moreover, unlike efference copies, which tend to preferentially target early stages in the sensory processing pathway (Figure 2A,C), the influence of corollary discharge, a less specifically patterned version of predictive signaling, can be exerted at any level in the motor planning-to-execution hierarchy (Figure 2A,B). In the following we have adhered, where possible, to these distinctions in describing ways in which feedforward motor predictions conveyed within motor-to-sensory (or even motor-to-motor; see below) circuits ensure the unambiguous perception of sensory inputs and thereby ultimately the production of effective behavior. Current Biology 28, R232–R243, March 5, 2018 R233

Current Biology

Review A

COROLLARY DISCHARGE CNS

MOTOR PATHWAY

Figure 2. The differences in pathway substrates for corollary discharge and efference copy signaling.

EFFERENCE COPY CNS

Motor representation

CNS

SENSORY PATHWAY

MOTOR PATHWAY

Motor replica

Motor output

Sensory input

B

Motor output

C Higher order motor circuitry

Higher order sensory circuitry

Higher order motor circuitry

-

Corollary discharge Local motor circuitry

-

-

Higher order sensory circuitry

Local sensory circuitry

Local motor circuitry

LEI

-

First order nuclei

Local sensory circuitry

(A) Organizational principles of predictive motorto-sensory signaling mediated by corollary discharges (light red shading, left) and efference copies (green shading, right). In each case, the distribution of color represents the different hierarchical levels at which predictive signals originate (left and right descending arrows) and are processed (middle ascending arrow) in motor and corresponding sensory pathways. See text for further explanation. (B) Corollary discharge pathways (violet dashed lines) predominantly connect higher order motor and sensory circuitries. (C) Efference copy pathways (violet dashed lines) predominantly connect local motor circuits with peripheral sense organs (green ellipse) and/or first order sensory nuclei (yellow ellipse). LEI, LII, hypothetical local excitatory and inhibitory interneurons; MN, motoneuron.

(Figure 3C), thus preventing the perception of illusionary object motion during a Efference saccade. Motor copy command Gaze and Posture Stabilization Reafference In an analogous manner to the visual system, detection of head/body motion Behavior Behavior also requires a reliable reference frame to Exafference dissociate vestibular inputs arising from Environment Environment passively-induced perturbations of head/ body position from active voluntary moveCurrent Biology ments such as during self-generated head turns or locomotion in primates [13,38,39]. During passive head/body motion, retinal image slip and postural Corollary Discharge and Reafferent Gating Saccadic Eye Movements imbalances are counteracted by an activation of vestibulo-ocular A classical and often referenced example of a potential reaffer- and/or vestibulo-spinal reflexes, which result from the transformaence-derived ambiguity in sensory perception is retinal image tion of vestibular sensory signals into spatio-temporally appromotion detection [36] during eye/head/body movements. In the priate extraocular and spinal motor commands. These reflexive reabsence of an appropriate reference frame, the brain is inca- sponses ensure stabilization of both posture and eye position pable of deciding if a perceived image motion on the retina is relative to the external world [40–42]. During actively generated head turns in monkeys, a predicdue to the movement of an object in the external world or is caused by movement of the eyes relative to the object that tive discharge originating from premotor/motor areas that remains stationary. A particularly well-studied example is visual activates neck muscles informs central vestibular sensorimotor image processing during saccadic eye movements in pri- circuits about the impending head movement, thereby assisting mates [37]. A fast redirecting eye displacement from one point in disambiguating self-induced from passively-induced head of focus to another (Figure 3A, upper image) causes a corre- movements [13]. Whereas the discharge of vestibulo-spinal spondingly rapid movement of the visual scene across the neurons is robustly modulated during passive head moveretina, which would be perceived as blurred (Figure 3A, lower ments, reafferent vestibular signals conveyed by these deimage). This potentially observable image motion, however, is scending pathway neurons are strongly attenuated during an outcome of the saccadic eye movement, not of actual scene active head rotations [14,38]. The underlying cellular mechamotion, and is not perceived (yellow framed segments in nism is based on a gating of vestibular afferent inputs specifically in vestibulo-spinal neurons and potentially involving local Figure 3A, lower image). The underlying disambiguating mechanism takes advantage inhibitory circuits as shown in frogs [43], which could cause of neuronal copies of the extraocular motor commands that vestibular input cancellation [6,14,44]. This reduction ensures encode the overall spatio-temporal parameters of the saccade a faithful execution of the voluntary movement, which would (Figure 3C). These corollary discharge signals, which are otherwise be prevented or at least impaired by reflexive signals generated in the superior colliculus (SC), are forwarded to cen- causing inappropriate postural reactions [14]. A comparable tral visual processing areas (Figure 3B), where they are inte- mechanism is implemented in primate gaze control where grated with incoming visual inputs to generate a modified visual a representation of the neck muscle motor command is percept through a transient shunting of visual image processing conveyed to the vestibular nuclei in order to suppress reafferent MN

R234 Current Biology 28, R232–R243, March 5, 2018

LII

Sensory periphery

Current Biology

Review A

B

FEF

Thalamus

V1 SC

LGN Mot

Motor control center C

Visual input Modified visual output

CD

Motor command

Current Biology

Figure 3. Role of corollary discharge in saccadic suppression of visual perception in primates. (A) During a saccade from left to right (yellow boxed squares), image perception does not appear blurred (upper photo), even though the saccadic eye motion results in a relative displacement of fixated object location, which would otherwise cause image blurring on the retina (bottom image). (B) Brain neural circuitry for visual signal processing (orange), saccade motor command generation (blue) and corollary discharge (CD) influence on target image processing. Adapted from [19]. (C) Schematized pathways underlying the subtractive influence of CD on visual inputs during a saccade, causing a temporal suppression of visual perception. FEF, frontal eye field; LGN, lateral geniculate nucleus; Mot, extraocular motoneurons; SC, superior colliculus; V1, visual area 1.

responses, thereby enabling an effective voluntary redirection of gaze [38]. Vocal Behavior Another motor behavior that is intimately linked to corollary discharge control of sensory processing is vocalization (for

example [45–48]), as this behavior unavoidably leads to auditory reafferent signals. To decrease unwanted vocal reafference, predictive vocal signals inform neurons along the peripheral and central auditory pathway about the occurrence of self-generated sound. A specific central site at which an attenuation of these reafferent signals occurs is the nucleus of the lateral lemniscus, as shown in echo-locating bats during supersonic call emission [48]. In monkeys, extracellular recordings in the hindbrain revealed vocal–audio interactions just prior to, and during, selfvocalization [49,50]. Similarly, activity in marmoset and squirrel monkey cortical auditory neurons is suppressed during vocal activity [45,51]. However, this suppression only blocks an expected neuronal discharge modulation, thereby allowing unexpected alterations in the self-generated sound to be detected [46]. In all these cases, the neuronal signal used to attenuate auditory signal processing [46,52,53] is a corollary discharge that derives from the production of vocal patterns along descending hierarchical motor stages. Furthermore, corollary discharges to auditory areas from vocal motor centers are not only important in the suppression of reafferent stimulation but also contribute to vocal learning [54]. A further example of reafferent sound suppression is the cancellation of licking-evoked auditory responsiveness in mice [55]. Similar to intended vocalization, licking behavior generates reafferent auditory stimuli, although here the reafference is suppressed in the dorsal cochlear nucleus, one of the first stages in the auditory processing pathway. Moreover, the synaptic and circuit mechanisms by which corollary discharges are conveyed from the motor to auditory cortices to modulate sensory processing have been investigated in freely-behaving mice [56]. During self-motion such as locomotion, head movements and grooming, the motor-related signal, which precedes movement onset, is conveyed via long-range motor-to-auditory circuitry to activate local interneurons that in turn inhibit auditory cortical excitatory neurons and thus suppress sound-evoked responses during subsequent movement. The biomedical relevance of such detailed findings are potentially important, as a dysfunction in motor-toauditory circuitry and corollary discharge failure is implicated in a variety of abnormal human hearing conditions, including auditory hallucinations, a major symptom of schizophrenia [12,57]. Efference Copy and Reafferent Gating In contrast to corollary discharges, we consider an efference copy as a replica (complete or partial) of the final motor circuit output that produces a distinct behavior. By definition, these replicas are forwarded to other neuronal targets without themselves being directly involved in generating that behavior. Nonetheless, the exact content of the copy signal can vary, ranging from a complete mirror image to a motor signature that preferentially encodes one or more temporal features of the corresponding motor command (for example [6,58–60]). Efference copies exert various effects on associated sensory processing, ranging from an attenuation of signal encoding at the periphery (for example [5]) to a complete central cancellation of self-produced sensory input [34,35]. In the following, we will briefly illustrate examples of the neuronal signature conveyed by efference copies and describe how these firing patterns affect signal processing at ascending sensory pathway stages that are relatively close to the motor command (Figure 2A,C). Current Biology 28, R232–R243, March 5, 2018 R235

Current Biology

Review Figure 4. Role of efference copy signaling in electroreception by weakly electric fish.

A

(A,B) Schematic silhouettes of an electrogenic fish with a putative electric field (red/yellow colored ellipse) in a stationary condition (A) and during swimming (B). Subtraction of the correlate of neural activity that generates the electric field (A, middle) or the locomotor command during swimming (B, middle) from the sensory encoded electric field (A,B, left) allows extracting the position of unpredictable objects in the environment (A, right) and minimizing self motion-related disturbances of the electric field (B, right). (C) Schematic of electrosensory circuitry underlying the subtractive interaction between electromotor and locomotor efference copies (blue) with reafferent and exafferent sensory signals (red), enabling the relevant sensory signal to be extracted.

B

C

Efference copies of electromotor + locomotor commands

Subtraction Effective sensory signal

Reafferent + exafferent sensory inputs from electroreceptors

Deciphering the temporal pattern of a particular efference copy requires an understanding of how the dynamic structure of the motor output is conferred by local motor circuits. In most known cases, however, the pattern of the predictive discharge is only revealed by the functional impact on its sensory targets. Thus, the relationship between an efference copy and its underlying motor source can in general only be indirectly inferred, as can the assignment of its origin to a particular cellular or circuit component of the local motor command-producing R236 Current Biology 28, R232–R243, March 5, 2018

network. Despite this handicap, there are a number of motor-to-sensory systems in which the discharge composition and functional role of efference copies has been able to be defined. A major contributor to this advance, and to identifying predictive motor signaling in general, has been the development of semiintact and reduced in vitro experimental preparations in which the still-viable central nervous system is isolated to varying degrees from the sensory periphery, muscle effectors, or both (see below). The absence of muscle movements and/or sensory feedback stimulation has allowed studying more directly the central nervous production of motoneuronal commands, which in such preparations can be recorded from the disconnected cranial or spinal nerves as so called ‘fictive’ behavior. Electroreception in Mormyrid Fish Weakly electric fish navigate through their environment by actively generating electric fields and monitoring alterations in the latter induced by conductive or nonconductive objects in the surrounding environment (Figure 4A, left). However, self-generated electric fields directly activate electroreceptors along the animal’s body, thereby masking the detection Current Biology of usually small external perturbations (grey colored area in Figure 4A, left). To extract these small perturbations from the overall sensory signal, electromotor efference copies that reflect mirror images of the unperturbed electric field (Figure 4A, middle) are subtracted from the combined reafferent and exafferent signal in the electrosensory lobe, a hindbrain nucleus dedicated to the processing of electrosensory inputs [33,61,62]. This subtraction unmasks the exafferent sensory component, thus resulting in a representation uniquely of the electrical disturbance of the self-generated electric field by the nearby object (Figure 4A, right).

Current Biology

Review Electrosensory probing of the environment is in addition impaired by locomotor activity (Figure 4B, left) or respirationrelated movements of the opercula covering the gills. In Mormyriform electrogenic fish, which mostly generate forward propulsion by body/tail undulations, locomotor movements cause rhythmic horizontal undulations of the animal’s electroreceptors, which deteriorate the sensing of electric field perturbations by modulating the field’s strength and spatial magnitude [59]. In order to compensate for these tail undulation-induced disturbances, efference copies of the motor command that drives alternating axial muscle contractions (Figure 4B, middle) are subtracted from the combined reafferent and exafferent sensory signal within the electrosensory lateral line lobe circuitry (Figure 4C). This, in turn, renders the sensory signal processing network optimally susceptible for the detection of electric field perturbations (Figure 4B, right). A similar subtractive computation is performed with motor efference copies associated with the generation of gill ventilation-related opercular movements [63,64]. Thus, efference copies of the movement command output, together with proprioceptive inputs, generate central adaptive filters that reduce self-body-motion-derived or breathing-induced spatial alterations in electroreceptor signaling. Vocalizing Behavior The most detailed understanding of the neuronal computation used to decrease reafferent signaling is undoubtedly the reafferent suppression that occurs during cricket stridulating behavior [16,31,32,65]. An interneuron that is not implicated in song production, but is concurrently activated by the stridulating motor command, directly blocks the presynaptic terminals of auditory afferents and central postsynaptic neurons while leaving the auditory periphery functionally unimpaired [16,31,32]. In vocalizing fish, hindbrain efferent neurons project to the sacculus, the main auditory organ in fish, and decrease the sensitivity of both hair cells and afferent fibers [66]. The efference copy transmitted by these neurons, which carries predominantly call duration information originating from the vocal pattern generator, [67,68], is thus likely to cause a decrease in auditory sensitivity during vocal activity. A similar peripheral effect on auditory signal encoding occurs during vocal motor behavior in bats. Cochlear microphonic recordings revealed a suppression of auditory activity before, during and immediately after self-generated vocalization, which is mediated by hindbrain signals originating from the medial olivo-cochlear bundle [69]. Locomotion A similar modulation of peripheral mechanoreceptor sensitivity occurs during locomotion, as reported in a number of aquatic anamniotes. In swimming dogfish, spinal locomotor efference copies are conveyed to lateral line hair cells [70], and locomotor efference copies with similar dynamic signatures are transmitted to both the vestibular and lateral line sensory peripheries of swimming Xenopus laevis tadpoles [5]. In Xenopus, by inscribing different parameters of the propulsive output commands including burst frequency and amplitude, and the duration of a given swim episode, the spinal copies cause a strong attenuation of head/body motion-induced encoding in lateral line sensory afferents during locomotor activity [5]. In contrast, the efference copies differentially affect movement-induced firing in individual sensory afferents of the vestibular system, although the overall responsiveness of these afferents to head/body

movement during swimming is also strongly reduced during swimming. The differential impact of the same locomotor efference copy signal on the two mechanosensory systems exemplifies a modality-specific control of sensory stimulus encoding, corresponding to differing bilateral (vestibular) versus unilateral (lateral line) principles of sensory signal processing [5,71,72]. Despite these differences, however, the marked attenuation of responsiveness to head/body motion and hydrodynamic stimuli by hair cells/afferent fibers in both systems is commensurate with an adaption of sensory processing to altered stimulus statistics during locomotion. Another example of locomotor-related reafferent suppression is the cancellation of optic flow detection during changes in flight direction induced by saccadic body movements in Drosophila [73]. A differentiation of self-generated reafferent optic flow from that resulting from external stimuli, such as through sudden displacement by wind gusts, is accomplished by body motion-activating neurons that send efference copies to specialized visual processing neurons, which in turn initiate compensatory motor responses [74]. These efference copies cancel the optic flow input and thereby prevent the generation of optomotor reflex behavior normally required for the maintenance of flight stability [3]. Efference Copy and Motor Systems Coupling As discussed above, predictive signaling is essential for distinguishing sensory inflow caused by unpredictable external influences from reafferent stimulation resulting from an animal’s own actions [75], thereby enabling a cancellation of the selfgenerated component. In this view, predictive signaling has been traditionally and almost exclusively defined as a central nervous process involved in the encoding of motor information with its impact uniquely confined to closely associated sensory-processing targets (for example see [16]). A further defining feature within this motor source–sensory target framework is that the copy signaling pathways must convey information about a movement without causing the actual movement. Extending this idea, we propose that the communication of local motor circuits with other neuronal sites via internal replicas of motor commands is not restricted to influencing the processing of sensory signals activated by the motor action. Rather, internal motor predictions can be additionally engaged in coordinating distributed central circuits of different motor systems that are otherwise functionally and anatomically unrelated. On this basis, therefore, internal predictive neural signaling through efference copies can be seen to have a potentially wider significance than the exclusive transmission of motor command replicas to motor action-related sensory-processing pathways, in that it may also be involved in the central nervous coordination between multiple complex behaviors. It should be re-emphasized, however, that this broader definition to encompass motor systems coupling does not include efference copy signaling that contributes to shaping the initial movement command itself. A classical example of the latter, which we exclude here, is the Renshaw cell-mediated recurrent inhibition of mammalian spinal motoneurons [76], even though, sensu stricto, the local axon collateral circuitry involved in this process does convey a direct copy of motoneuronal output. Current Biology 28, R232–R243, March 5, 2018 R237

Current Biology

Review A

Figure 5. Schematics depicting two possible strategies for the central coupling of different motor behaviors.

B Higher order motor circuitry

Local motor circuitry 1

Higher order motor circuitry 1

Local motor circuitry 1

Local motor circuitry 2

(A) Co-recruitment of local motor circuits 1 and 2 by common higher order commands. (B) Recruitment of local motor circuit 2 (red arrow) by efference copy signaling from local motor circuit 1.

Local motor circuitry 2

signals are conveyed directly via ascending long-projection pathways to the parafacial respiratory group (pFRG), one of the two oscillatory brainstem netBehavior 1 Behavior 1 works that contribute to breathing rhythm generation [81]. By employing Behavior 2 Behavior 2 isolated brainstem-spinal cord preparaCurrent Biology tions of neonatal rats (and therefore in the absence of all sensory feedback influences), the locomotor efference copy Nor do we include coordinating mechanisms originating from increases the frequency of respiratory motor activity by a spehigher brain centers that send parallel corollary signals to down- cific augmentation of the excitability of pFRG pre-inspiratory stream motor systems (Figure 5A), such as the predictive and neurons [80]. This intrinsic motor-to-motor coupling presumably motor learning processes that enable coordinating left/right satisfies the increase in oxygen demand caused by accelerated hand and foot pedal movements while playing the piano [11,77]. locomotor movements, and with a much faster onset than the Rather, our definition applies specifically to the capability of respiratory adjustments conferred by feedback from oxygen replicate forward signals emanating from the local motor com- sensors, thus facilitating the immediate and subsequent susmand circuitry for one behavior to directly influence one or tained supply of oxygen to the body. more local motor circuits responsible for another, different Locomotion — Tentacle Retraction behavior (Figure 5B). In the following, we will first illustrate Another example of locomotor-to-motor coupling is found besuch efference-copy-mediated coordination of motor systems, tween the spinal swim-generating and hindbrain trigeminal motor followed by the presentation of evidence that this motor-to-mo- circuitries that control movements of the paired mechanosensitor coupling can also be employed in reducing reafferent sensory tive tentacles (similar to fish barbels) in Xenopus tadpoles [82]. input generated by the motor system originally responsible for A subset of trigeminal motoneurons, which receive spinal efference copy input about the duration, frequency and strength of the coupling. Animal locomotion is a complex behavior that requires the co- ongoing locomotor movements, drive tentacle retractions that ordinated activation of a large number of limb and body muscles. are likely to reduce the hydrodynamic resistance and/or prevent As a consequence, the locomotor act in itself generates probably reafferent sensory stimulation of the tentacle during undulatory the greatest diversity and amplitude of reafferent sensory infor- swimming movements [82]. Although yet to be established mation amongst the repertoire of an animal’s motor behaviors. experimentally, other related examples of locomotor-to-motor Because of the critical role that locomotion plays in animal coupling potentially mediated by efference copies include the survival — finding food, mating, escaping danger, migrating — retraction of limbs or external gills during the switch to undulatory it is perhaps unsurprising that the neural networks producing lo- tail-based swimming in the salamander and axolotl [83,84]. comotor commands often also employ efference copy signaling Locomotion — Gaze Stabilization to ensure that an organism’s other motor systems remain adap- One of the best elucidated examples of efference copy-meditively coordinated with this dominant motor act. To date, a num- ated motor-to-motor coupling is the role that spinal locomotor ber of examples of locomotor coupling with other motor systems central pattern generator (CPG) circuitry plays in controlling gaze stabilization. Irrespective of their mode of locomotion, all via intrinsic signaling pathways have been identified. animals require a concurrent activation of gaze-stabilizing eye Locomotion — Respiration A well-established example of such efference-mediated movements because of the disruptive effect that the resultant coupling is the interaction between mammalian locomotor and head/body motion has on visual perception. In Xenopus tadrespiratory motor systems, which are otherwise distinct and poles, propulsive swimming movements consist of caudal to functionally independent. This coupling relationship is readily rostral waves of axial body motion that are produced by alterobservable in our everyday lives when steady running or strong nating left–right contractions of tail muscles [85]. This undulatory exercise is associated with an increase in our respiratory rhythm bending of the body/tail, which is a common feature of locomoin order to satisfy the augmented oxygen demand (for example tion in many limbless aquatic and terrestrial vertebrates, gener[78,79]). A major neuronal substrate for this coordination is ates lateral displacements of the head (Figure 6A) that cause intrinsic efference copy signaling from the locomotor rhythm- retinal image drift with a resultant degradation of visual informagenerating centers in the lumbar spinal cord [80]. The copy tion processing. In order to maintain visual acuity, retinal image Motor command 1

Motor command 2

Motor command 1

R238 Current Biology 28, R232–R243, March 5, 2018

Motor command 2

Current Biology

Review B

Eye position (°)

Head position (°)

A

C

E

Head

20 0

Eye muscle

-20

VOR

Eye

20

Ves Eff Ocular Mot

0

Ves Nuc

Ves Aff

-20

0

0.2

0.4

0.6

Efference copy

Time (s) D

Inner ear

ri-MR LR

MR

le-MR

MR

LR

ri-LR Contraction

le-LR

Loco CPG

Contraction

le-VR ri-VR Current Biology

Figure 6. Efference copy-mediated gaze stabilization during locomotion in Xenopus tadpoles. (A) Schematics of a tadpole depicting swimming-related right left head/body oscillations (red arrows) and concurrent eye movements (black arrows). (B) Monitor of eye (black) and head position (red) during swimming showing the phase opposition of movements (dashed vertical lines). (C) Putative visual scene during locomotion in the absence (upper, blurred image) and presence (lower, sharp image) of gaze-stabilizing eye movements. (D) Schematized rhythmic burst discharge in bilateral medial (MR, yellow) and lateral rectus (LR, violet) extraocular motor nerves and left and right spinal ventral roots (le/ri-VR, green/blue) during fictive swimming. Bursts in le-LR and ri-MR nerves are coincident with bursts in the right VR (blue boxed area) and vice versa for antagonistic ri-LR and le-MR nerves in the opposite swim half-cycle (green boxed area). This complies with eye movements that are oppositely-directed to head displacements resulting from right–left tail muscle contractions during undulatory swimming (see tadpole drawings). (E) Schematic drawing of the hindbrain/spinal cord depicting efference copy-mediated coupling between the spinal locomotor central pattern generator (Loco CPG) and extraocular motoneurons (Ocular Mot) and inner ear vestibular efferent neurons (Ves Eff). Motion-derived vestibular afferent (Ves Aff) signals are integrated in the vestibular nuclei (Ves Nuc) and conveyed to extraocular motoneurons by vestibulo-ocular reflex (VOR) pathways. Adapted from [5].

drift is minimized by counteractive image-stabilizing eye motion that is traditionally attributed to ocular reflexes [40,41]. However, recent studies have found that, further to such reactive gaze-stabilizing processes, intrinsic feed-forward copies of the spinal motor commands responsible for generating locomotor movements are used as an internal prediction of the disruptive consequences of head/body movements for visual processing [86–88]. Central to this efference-copy-driven gaze-stabilizing process is that relatively stereotyped rhythmic locomotion produces highly predictable visual perturbations reflecting both the timing and amplitude of ongoing head/body displacement [89]. During tadpole swimming, oscillatory head movements are accompanied by oppositely-directed conjugate eye movements (Figure 6A,B) [87] and instead of blurring of the visual field (Figure 6C, upper), these ensure stable image position on the retina (Figure 6C, lower). Experiments conducted on isolated brainstem/spinal cord preparations in the absence of mechanosensory inputs confirmed that these counteractive eye movements are indeed driven by locomotor efference copies from the tadpole’s spinal cord [86,87]. Specifically, during spontaneous fictive swimming in vitro, the extraocular motor nerves of synergistic pairs of horizontal eye muscles express locomotor-timed burst discharge that is spatio-temporally appropriate

for offsetting swimming-related head displacements in vivo (Figure 6D) [86,87]. Similarly, in elasmobranch and teleost fish, earlier behavioral evidence has suggested that intrinsic signals associated with rhythmic tail undulation, rather than those arising from sensory systems, may coordinate compensatory eye movements with lateral head oscillations during swimming [90,91]. A recent behavioral study [92] has also found evidence for a coupling between locomotor and eye movements in the shiner surfperch, which often uses synchronous pectoral fin beating during locomotion. Here, strong propulsive thrust during each pectoral fin abduction is accompanied by synchronized eye movements that potentially reduce image slip on the retina. In swimming tadpoles, the spino-extraocular coupling is spatially-specific (it activates lateral and medial rectus eye muscles only) and corresponds closely to the plane specificity of the horizontal angular vestibulo-ocular reflex (VOR) [93], and thus is consistent with the production of conjugate eye movements that would counteract horizontal head displacements [87]. The locomotor efference copies are conveyed directly from the spinal CPG circuitry to brainstem extraocular motor targets by a population of segmentally-iterated neurons in the rostral cord region (Figure 6E), without the intervention of other supraspinal areas Current Biology 28, R232–R243, March 5, 2018 R239

Current Biology

Review such as the midbrain reticular formation, cerebellum or even the vestibular nuclei [87,88]. Significantly, these spino-extraocular coupling pathways are altered during metamorphosis when the animal’s locomotor strategy changes progressively from larval tail-based swimming to rhythmic, bilaterally-synchronous hindlimb-kicking for generating forward propulsion [85]. In accordance with the functional necessity for retinal image stabilization in adult frogs, the gradual alterations in efference copy pathway trajectories and their signaling patterns throughout the metamorphic transition continue to match image stabilizing requirements as one locomotor mechanism emerges and replaces the other [88,94]. Given that spinal efference copies access the ocular motor centers via pathways that bypass the vestibular nuclei, how does this internal prediction of locomotor movement interact with reafferent feedback signaling by VORs? Experiments in isolated spinal cord/brainstem preparations of larval Xenopus with intact labyrinths indicated that the horizontal angular VOR-derived influence on extraocular activity is selectively suppressed during swimming activity, whereas other VORs, such as those arising from the utricle, remained unaffected [87]. Although the site(s) at which this selective gating of the horizontal angular VOR during locomotion remains to be determined, as seen above, one already identified gating process occurs peripherally: namely, the efference copy-driven gain reduction of mechanosensory transmission from hair cells onto first-order vestibular afferents in the inner ear (Figure 6E; see above) [5]. This selective cancellation of reafferent information in Xenopus locomotor-gaze control invites comparison with the more traditional examples of efference copy-mediated reafferent suppression described above. In both cases, the internal signals provide predictive information that preempts the slower engagement of movement sensing pathways and provides an estimate of the expected sensory consequences of a behavioral action. However, there are two fundamental differences. First, the latter situation involves a single motor system and the employment of efference copies in processing information in sensory pathways associated with that particular system, whereas the second case involves the coupling between two independent motor systems that are each capable of separate behavior. Second, in the typical motor-tosensory condition, reafferent suppression allows exafferent inputs to be unambiguously perceived, whereas in Xenopus locomotor gaze control, the efference copy emerging from the first motor act (locomotion) is not just used to appropriate the second behavior (compensatory eye movements) to the former’s own actions: additionally, it cancels the subordinate behavior’s normal sensory inputs, presumably to maintain effective eye rotation amplitudes, which would be excessively large if the VOR and efference copy signals were to be integrated additively. From an evolutionary perspective, finally, it is possible that the employment of efference copy signaling in gaze control during locomotion is representative of an ancestral condition where an intrinsic spino-extraocular influence was potentially available during undulatory swimming of vertebrate ancestors, long before body-motion sensing vestibular endorgans had evolved [95]. Co-employment of Predictive Signaling So far, we have considered motor-to-sensory and motor-tomotor predictive signaling separately, although it is important R240 Current Biology 28, R232–R243, March 5, 2018

to realize that this distinction cannot be made either on the basis of organism or even a particular behavioral task. Instead, the two types of interaction may operate in various combinations to influence the same behavioral processes within the same animal. In weakly electric fish, for example, to ensure effective electroreceptive sensing during locomotion, efference copies originating from the electromotor system (Figure 4A) [33] and those arising from spinal swim circuits (Figure 4B) [59] are likely to act simultaneously on electrosensory lobe neurons, where the expected reafferent components of both the electric organ discharge and the alteration in the electric field produced by swim-related tail bending are subtracted. Another example of conjointly employed predictive signaling occurs during vocal–auditory interactions. While the mammalian middle ear reflex involves a motor-to-motor efference copy that, by increasing tension of the stapedius muscle, decreases the gain of acoustic amplification in the middle ear during vocal behavior, concurrent motor-to-sensory efference copies transmitted by olivo-cochlear neurons to the peripheral hair cell sensors decrease transduction sensitivity [96]. These peripheral modifications leading to altered sound perception are further complemented by central effects in which corollary discharges modulate the activity of neurons in higher order auditory circuits that filter unwanted reafferent stimulation. Similarly, multiple predictive signaling occurs during larval frog locomotion as described above. While a motor-to-motor efference copy drive (from spinal cord to ocular motor centers) is used to reduce retinal image slip and stabilize gaze during swimming (Figure 6A–C), a parallel motor-to-sensory copy (from spinal cord to peripheral inner ear hair cells) decreases the sensitivity of vestibular motion detection. Consequently, the sensing of self-generated movement, both by gaze control circuitry and by the spinal locomotor CPG itself, becomes attenuated. In addition to these parallel motor-to-motor and motor-to-sensory influences, preliminary evidence indicates that spinal locomotor efference copies are also conveyed concurrently to supraspinal respiratory centers [97], presumably to adjust branchial ventilation rates to the increase in oxygen demand. These examples of different predictive signaling processes and their impact at multiple sensory and motor levels of the nervous system thus further exemplify the diversity with which both efference copies and corollary discharges can be employed to reduce reafferent influences on individual motor acts, and are undoubtedly multiply involved in controlling other behaviors. Conclusion Predictive signaling provides the basis for differentiating reafferent from exafferent sensory inputs, and therefore constitutes a crucial aspect of central nervous system operation. While we have gained a reasonable understanding of the functional impact of predictive motor signaling on a number of sensory and motor systems, detailed knowledge of how such intrinsic signals are encoded is thus far mostly lacking. Furthermore, a designation of corollary discharge or efference copy signaling as distinct mechanisms is in many cases difficult to establish, indicating that unifying the two coupling processes under the general concept of ‘predictive signaling’ may ultimately have a greater relevance. Indeed, to date in only a few motor systems has the identity and the firing patterns of the neurons that convey predictive signals

Current Biology

Review been elucidated. Questions therefore arise as to whether this mediating role is restricted to specialized cell populations, or whether axon collaterals of neurons involved in generating the actual behavioral command might indeed in some cases be responsible for conveying the copy signal. How does the predictive signal affect sensory or motor processing at the cellular level? Are there general excitatory or inhibitory mechanisms involved, like shunting inhibition or discharges that cause primary afferent depolarization, as in the cricket, and to what extent is neuronal processing in cortical and subcortical circuits subject to reconfiguration by state-dependent feed-forward signaling [98]? Knowledge about such fundamental processes would provide important insights into the advantages, for example, of reafferent gating at peripheral as compared to central nervous system levels. There is thus still much to learn about the role of predictive signaling in shaping sensory perception and motor function. ACKNOWLEDGEMENTS This work was funded by the French ‘Agence Nationale de la Recherche’ (ANR-08-BLAN-0145-01, ‘Galodude’), the German Science Foundation (CRC 870 and STR 478/3-1) and the German Federal Ministry of Education and Research under the grant code 01 EO 0901 and 01 GQ 1407. We are grateful to the Bayerisch-Franzo¨sisches Hochschulzentrum (BFHZ) for financial support, and the Center for Advanced Studies at the Ludwig-Maximilians-University Munich for a LMUCAS Visiting Fellowship to JS. We also thank Drs. Harold Zakon and Joel Glover for very helpful discussions and comments on the manuscript.

REFERENCES 1. Sperry, R.W. (1950). Neural basis of the spontaneous optokinetic response produced by visual inversion. J. Comp. Physiol. Psychol. 43, 482–489.

13. Cullen, K.E. (2004). Sensory signals during active versus passive movement. Curr. Opin. Neurobiol. 14, 698–706. 14. Cullen, K.E. (2011). The neural encoding of self-motion. Curr. Opin. Neurobiol. 21, 587–595. 15. Cullen, K.E. (2014). The neural encoding of self-generated and externally applied movement: implications for the perception of self-motion and spatial memory. Front. Integr. Neurosci. 7, 108. 16. Poulet, J.F., and Hedwig, B. (2006). The cellular basis of a corollary discharge. Science 311, 518–522. 17. Crapse, T.B., and Sommer, M.A. (2008a). Corollary discharge circuits in the primate brain. Curr. Opin. Neurobiol. 18, 552–557. 18. Crapse, T.B., and Sommer, M.A. (2008b). Corollary discharge across the animal kingdom. Nat. Rev. Neurosci. 9, 587–600. 19. Wurtz, R.H. (2008). Neuronal mechanisms of visual stability. Vision Res. 48, 2070–2089. 20. Krasne, F.B., and Bryan, J.S. (1973). Habituation: regulation through presynaptic inhibition. Science 182, 590–592. 21. Edwards, D.H., Heitler, W.J., and Krasne, F.B. (1999). Fifty years of a command neuron: the neurobiology of escape behavior in the crayfish. Trends Neurosci. 22, 153–161. 22. Russell, I.J. (1971). The role of the lateral-line efferent system in Xenopus laevis. J. Exp. Biol. 54, 621–641. 23. Roberts, B.L., and Russell, I.J. (1972). The activity of lateral-line efferent neurones in stationary and swimming dogfish. J. Exp. Biol. 57, 435–448. 24. Rudomin, P., and Schmidt, R.F. (1999). Presynaptic inhibition in the vertebrate spinal cord revisited. Exp. Brain Res. 129, 1–37. 25. Rudomin, P. (2009). In search of lost presynaptic inhibition. Exp. Brain Res. 196, 139–151. 26. El Manira, A., Tegner, J., and Grillner, S. (1997). Locomotor-related presynaptic modulation of primary afferents in the lamprey. Eur. J. Neurosci. 9, 696–705.

2. von Holst, E., and Mittelstaedt, H. (1950). Das Reafferenzprinzip. Naturwissenschaften 37, 464–476.

27. Bu¨schges, A., and Wolf, H. (1999). Phase-dependent presynaptic modulation of mechanosensory signals in the locust flight system. J. Neurophysiol. 81, 959–962.

3. Kim, A.J., Fitzgerald, J.K., and Maimon, G. (2015). Cellular evidence for efference copy in Drosophila visuomotor processing. Nat. Neurosci. 18, 1247–1255.

28. Gossard, J.P., Cabelguen, J.M., and Rossignol, S. (1991). An intracellular study of muscle primary afferents during fictive locomotion in the cat. J. Neurophysiol. 65, 914–926.

4. Requarth, T., and Sawtell, N.B. (2011). Neural mechanisms for filtering self-generated sensory signals in cerebellum-like circuits. Curr. Opin. Neurobiol. 21, 602–608.

29. Clarac, F., Cattaert, D., and Le Ray, D. (2000). Central control components of a ‘simple’ stretch reflex. Trends Neurosci. 23, 199–208.

5. Chagnaud, B.P., Banchi, R., Simmers, J., and Straka, H. (2015). Spinal corollary discharge modulates motion sensing during vertebrate locomotion. Nat. Commun. 6, 7982. 6. Roy, J.E., and Cullen, K.E. (2001). Selective processing of vestibular reafference during self-generated head motion. J. Neurosci. 21, 2131–2142. 7. Kelley, D.B., and Bass, A.H. (2010). Neurobiology of vocal communication: mechanisms for sensorimotor integration and vocal patterning. Curr. Opin. Neurobiol. 20, 748–753. 8. Blakemore, S.J., Wolpert, D., and Frith, C. (1998). Central cancellation of self-produced tickle sensation. Nat. Neurosci. 1, 635–640. 9. Blakemore, S.J., Wolpert, D., and Frith, C. (2000). Why can’t you tickle yourself. NeuroReport 11, 11–16. 10. Whitford, T.J., Jack, B.N., Pearson, D., Griffiths, O., Luque, D., Harris, A.W.F., Spencer, K.M., and Le Pelley, M.E. (2017). Neurophysiological evidence of efference copies to inner speech. eLife 6, e28197. 11. Wolpert, D.M., and Flanagan, J.R. (2001). Motor prediction. Curr. Biol. 11, R729–732. 12. Pynn, L.K., and DeSouza, J.F. (2013). The function of efference copy signals: implications for symptoms of schizophrenia. Vision Res. 76, 124–133.

30. Poulet, J.F., and Hedwig, B. (2002). A corollary discharge maintains auditory sensitivity during sound production. Nature 418, 872–876. 31. Poulet, J.F., and Hedwig, B. (2003a). Corollary discharge inhibition of ascending auditory neurons in the stridulating cricket. J. Neurosci. 23, 4717–4725. 32. Poulet, J.F., and Hedwig, B. (2003b). A corollary discharge mechanism modulates central auditory processing in singing crickets. J. Neurophysiol. 89, 1528–1540. 33. Bell, C.C. (1979). Central nervous system physiology of electroreception, a review. J. Physiol. (Paris) 75, 361–379. 34. Bell, C.C. (1981). An efference copy which is modified by reafferent input. Science 214, 450–453. 35. Bell, C.C. (2001). Memory-based expectations in electrosensory systems. Curr. Opin. Neurobiol. 11, 481–487. 36. Walls, G.L. (1962). The evolutionary history of eye movements. Vis. Res. 2, 69–80. 37. Wurtz, R.H., and Sommer, M.A. (2004). Identifying corollary discharges for movement in the primate brain. Prog. Brain Res. 144, 47–60. 38. Cullen, K.E. (2012). The vestibular system: multimodal integration and encoding of self-motion for motor control. Trends Neurosci. 35, 185–196.

Current Biology 28, R232–R243, March 5, 2018 R241

Current Biology

Review 39. Carriot, J., Jamali, M., Chacron, M.J., and Cullen, K.E. (2017). The statistics of the vestibular input experienced during natural self-motion differ between rodents and primates. J. Physiol. 595, 2751–2766.

61. Sawtell, N.B., Williams, A., and Bell, C.C. (2005). From sparks to spikes: information processing in the electrosensory systems of fish. Curr. Opin. Neurobiol. 15, 437–443.

40. Straka, H., and Dieringer, N. (2004). Basic organization principles of the VOR: lessons from frogs. Prog. Neurobiol. 73, 259–309.

62. Heiligenberg, W., and Bastian, J. (1984). The electric sense of weakly electric fish. Annu. Rev. Physiol. 46, 561–583.

41. Angelaki, D.E., and Cullen, K.E. (2008). Vestibular system: the many facets of a multimodal sense. Annu. Rev. Neurosci. 31, 125–150.

63. Bodznick, D., and Montgomery, J.C. (1992). Suppression of ventilatory reafference in the elasmobranch electrosensory system: Medullary neuron receptive fields support a common mode ejection mechanism. J. Exp. Biol. 171, 127–137.

42. Straka, H., Zwergal, A., and Cullen, K.E. (2016). Vestibular animal models: contributions to understanding physiology and disease. J. Neurol. 263 (Suppl 1 ), S10–S23. 43. Straka, H., and Dieringer, N. (2000). Convergence pattern of uncrossed excitatory and inhibitory semicircular canal-specific inputs onto secondorder vestibular neurons of frogs. Organization of vestibular side loops. Exp. Brain Res. 135, 462–473.

64. Montgomery, J.C., and Bodznick, D. (1993). Hindbrain circuitry mediating common mode suppression of ventilatory reafference in the electrosensory system of the little skate Raja erinacea. J. Exp. Biol. 183, 203–216. 65. Poulet, J.F., and Hedwig, B. (2007). New insights into corollary discharges mediated by identified neural pathways. Trends Neurosci. 30, 14–21.

44. Roy, J.E., and Cullen, K.E. (2004). Dissociating self-generated from passively applied head motion: neural mechanisms in the vestibular nuclei. J. Neurosci. 24, 2102–2111.

66. Boyle, R., Rabbitt, R.D., and Highstein, S.M. (2009). Efferent control of hair cell and afferent responses in the semicircular canals. J. Neurophysiol. 102, 1513–1525.

45. Eliades, S.J., and Wang, X. (2003). Sensory-motor interaction in the primate auditory cortex during self-initiated vocalizations. J. Neurophysiol. 89, 2194–2207.

67. Chagnaud, B.P., and Bass, A.H. (2013). Vocal corollary discharge communicates call duration to vertebrate auditory system. J. Neurosci. 33, 18775–18780.

46. Eliades, S.J., and Wang, X. (2008). Neural substrates of vocalization feedback monitoring in primate auditory cortex. Nature 453, 1102–1106.

68. Weeg, M.S., Land, B.R., and Bass, A.H. (2005). Vocal pathways modulate efferent neurons to the inner ear and lateral line. J. Neurosci. 25, 5967– 5974.

47. Ford, J.M., Roach, B.J., Faustman, W.O., and Mathalona, D.H. (2008). Out-of-synch and out-of-sorts: Dysfunction of motor-sensory communication in schizophrenia. Biol. Psychiat. 63, 736–743. 48. Suga, N., and Shimozawa, T. (1974). Site of neural attenuation of responses to self-vocalized sounds in echolocating bats. Science 183, 1211–1213. 49. Hage, S.R., and Ju¨rgens, U. (2006). Localization of a vocal pattern generator in the pontine brainstem of the squirrel monkey. Eur. J. Neurosci. 23, 840–844. 50. Hage, S.R., and Ju¨rgens, U. (2007). The olivocochlear system takes part in audio-vocal interaction. In Hearing: From Sensory Processing to Perception, B. Kollmeier, G. Klump, V. Hohmann, U. Langemann, M. Mauermann, S. Uppenkamp, and J. Verhey, eds. (New York: Springer), pp. 191–197. 51. Mu¨ller-Preuss, P., and Ploog, D. (1981). Inhibition of auditory cortical neurons during phonation. Brain Res. 215, 61–76. 52. Suga, N., and Schlegel, P. (1972). Neural attenuation of responses to emitted sounds in echolocating bats. Science 177, 82–84. 53. Ford, J.M., Roach, B.J., and Mathalon, D.H. (2010). Assessing corollary discharge in humans using noninvasive neurophysiological methods. Nat. Protoc. 5, 1160–1168. 54. Roberts, T.F., Hisey, E., Tanaka, M., Kearney, M.G., Chattree, G., Yang, C.F., Shah, N.M., and Mooney, R. (2017). Identification of a motor-to-auditory pathway important for vocal learning. Nat. Neurosci. 20, 978. 55. Singla, S., Dempsey, C., Warren, R., Enikolopov, A.G., and Sawtell, N.B. (2017). A cerebellum-like circuit in the auditory system cancels responses to self-generated sounds. Nat. Neurosci. 20, 943–950. 56. Schneider, D.M., Nelson, A., and Mooney, R. (2014). A synaptic and circuit basis for corollary discharge in the auditory cortex. Nature 513, 189. 57. Ford, J.M., and Mathalon, D.H. (2005). Corollary discharge dysfunction in schizophrenia: can it explain auditory hallucinations? Int. J. Psychophysiol. 58, 179–189.

69. Goldberg, R.L., and Henson, O.W., Jr. (1998). Changes in cochlear mechanics during vocalization: evidence for a phasic medial efferent effect. Hear Res. 122, 71–81. 70. Russell, I.J., and Roberts, B.L. (1972). Inhibition of spontaneous lateral-line activity by efferent nerve stimulation. J. Exp. Biol. 57, 77–82. 71. Chagnaud, B.P., and Coombs, S. (2013). Information encoding and processing by the peripheral lateral line system. In The Lateral Line System. Springer Handbook of Auditory Research, Vol. 48, S. Coombs, H. Bleckmann, R. Fay, and A. Popper, eds. (New York: Spinger), pp. 151–194. 72. Bleckmann, H., and Mogdans, J. (2013). Central processing of lateral line information. In The Lateral Line System. Springer Handbook of Auditory Research, Vol. 48, S. Coombs, H. Bleckmann, R. Fay, and A. Popper, eds. (New York: Spinger), pp. 253–280. 73. Collett, T.S., and Land, M.F. (1975). Visual control of flight behaviour in the hoverfly Syritta pipiens L. J. Comp. Physiol. A 99, 1–66. 74. Kim, A.J., Fenk, L.M., Lyu, C., and Maimon, G. (2017). Quantitative predictions orchestrate visual signaling in Drosophila. Cell 168, 280–294. 75. Brooks, J.X., Carriot, J., and Cullen, K.E. (2015). Learning to expect the unexpected: rapid updating in primate cerebellum during voluntary selfmotion. Nat. Neurosci. 18, 1310–1317. 76. Eccles, J.C., Eccles, R.M., Iggo, A., and Lundberg, A. (1961). Electrophysiological investigations on Renshaw cells. J. Physiol. 159, 461–478. 77. Flanagan, J.R., Vetter, P., Johansson, R.S., and Wolpert, D.M. (2003). Prediction precedes control in motor learning. Curr. Biol. 13, 146–150. 78. Bramble, D., and Carrier, D. (1983). Running and breathing in mammals. Science 219, 251–256. 79. Funk, G., Valenzuela, I.I., and Milsom, W. (1997). Energetic consequences of coordinating wingbeat and respiratory rhythms in birds. J. Exp. Biol. 200, 915–920.

58. Chagnaud, B.P., and Bass, A.H. (2011). Vocalization frequency and duration are coded in separate hindbrain nuclei. Nat. Commun. 2, 346.

80. Le Gal, J.-P., Juvin, L., Cardoit, L., Thoby-Brisson, M., and Morin, D. (2014). Remote control of respiratory neural network by spinal locomotor generators. PLoS ONE 9, e89670.

59. Requarth, T., and Sawtell, N.B. (2014). Plastic corollary discharge predicts sensory consequences of movements in a cerebellum-like circuit. Neuron 82, 896–907.

81. Onimaru, H., and Homma, I. (2003). A novel functional neuron group for respiratory rhythm generation in the ventral medulla. J. Neurosci. 23, 1478–1486.

60. Roy, J.E., and Cullen, K.E. (1998). A neural correlate for vestibulo-ocular reflex suppression during voluntary eye-head gaze shifts. Nat. Neurosci. 1, 404–410.

€nzi, S., Banchi, R., Straka, H., and Chagnaud, B.P. (2015). Locomotor 82. Ha corollary activation of trigeminal motoneurons: coupling of discrete motor behaviors. J. Exp. Biol. 218, 1748–1758.

R242 Current Biology 28, R232–R243, March 5, 2018

Current Biology

Review 83. D’Aouˆt, K., and Aerts, P. (1997). Kinematics and efficiency of steady swimming in adult axolotls (Ambystoma mexicanum). J. Exp. Biol. 200, 1863– 1871. 84. Fish, F.E. (1984). Kinematics of undulatory swimming in the American alligator. Copeia 4, 839–843. 85. Combes, D., Merrywest, S.D., Simmers, J., and Sillar, K.T. (2004). Developmental segregation of spinal networks driving axial- and hindlimb-based locomotion in metamorphosing Xenopus laevis. J. Physiol. 559, 17–24. 86. Combes, D., Le Ray, D., Lambert, F.M., Simmers, J., and Straka, H. (2008). An intrinsic feed-forward mechanism for vertebrate gaze stabilization. Curr. Biol. 18, R241–243. 87. Lambert, F.M., Combes, D., Simmers, J., and Straka, H. (2012). Gaze stabilization by efference copy signaling without sensory feedback during vertebrate locomotion. Curr. Biol. 22, 1649–1658. 88. von Uckermann, G., Le Ray, D., Combes, D., Straka, H., and Simmers, J. (2013). Spinal efference copy signaling and gaze stabilization during locomotion in juvenile Xenopus frogs. J. Neurosci. 33, 4253–4264. 89. Chagnaud, B.P., Simmers, J., and Straka, H. (2012). Predictability of visual perturbation during locomotion: implications for corrective efference copy signaling. Biol. Cybern. 106, 669–679. 90. Harris, A.J. (1965). Eye movements of the dogfish, Squalus acanthias L. J. Exp. Biol. 43, 107–130.

91. Easter, S.S., Jr., and John, P.R. (1974). Horizontal compensatory eye movements in goldfish (Carassius auratus). J. Comp. Physiol. A 92, 37–57. 92. Mandecki, J.L., and Domenici, P. (2015). Eye movements are coordinated with pectoral fin beats during locomotion in a marine teleost fish. J. Exp. Biol. 218, 1122–1125. 93. Robinson, D.A. (1982). The use of matrices in analyzing the three-dimensional behavior of the vestibulo-ocular reflex. Biol. Cybern. 46, 53–66. 94. von Uckermann, G., Lambert, F.M., Combes, D., Straka, H., and Simmers, J. (2016). Adaptive plasticity of retinal image stabilization during locomotion in developing Xenopus. J. Exp. Biol. 219, 1110–1121. 95. Beisel, K.W., Wang-Lundberg, Y., Maklad, A., and Fritzsch, B. (2005). Development and evolution of the vestibular sensory apparatus of the mammalian ear. J. Vestib. Res. 15, 225–241. 96. Glowatzki, E., and Fuchs, P.A. (2000). Cholinergic synaptic inhibition of inner hair cells in the neonatal mammalian cochlea. Science 288, 2366–2368. 97. Combes, D., Merlet, L., Thoby-Brisson, M., Morin, D., and Simmers, J. (2015). Central coupling between locomotion and respiration in the metamorphosing frog. Soc. Neurosci. Abstr. 41, 798.07. 98. Pakan, J.M., Lowe, S.C., Dylda, E., Keemink, S.W., Currie, S.P., Coutts, C.A., and Rochefort, N.L. (2016). Behavioral-state modulation of inhibition is context-dependent and cell type specific in mouse visual cortex. eLife 5, e14985.

Current Biology 28, R232–R243, March 5, 2018 R243