Research in the last decade has undermined the idea of perception as a continuous process, providing strong empirical support for its rhythmic modulation. More recently, it has been revealed that the ongoing motor processes influence the rhythmic sampling of sensory information. In this review, we will focus on a growing body of evidence suggesting that oscillation-based mechanisms may structure the dynamic interplay between the motor and sensory system and provide a unified temporal frame for their effective coordination. We will describe neurophysiological data, primarily collected in animals, showing phase-locking of neuronal oscillations to the onset of (eye) movements. These data are complemented by novel evidence in humans, which demonstrate the behavioral relevance of these oscillatory modulations and their domain-general nature. Finally, we will discuss the possible implications of these modulations for action–perception coupling mechanisms.
MOVEMENT-LOCKED SYNCHRONIZATION OF NEURONAL OSCILLATIONS
Research in the last decade has provided a solid empirical basis for the longstanding postulate that perception may not be continuous, but rather modulated, in a rhythmic fashion (VanRullen, 2016; Valera, Toro, John, & Schwartz, 1981; Harter, 1967). Cyclic fluctuations of neuronal excitability inherently structure the way the brain samples and processes the external inputs over time. This rhythmic neural machinery does not act as a passive sensory filter but flexibly tunes sensory processing in space and time to optimize it. The ongoing dynamics in brain oscillatory states is subjected to both bottom–up influences as well as top–down control. For example, oscillations can be synchronized or phase reset by exogenous cues. This unmasks cue-locked rhythmicity in behavioral performance (e.g., detection and RTs), with visual sampling alternating across attended locations and objects (Fiebelkorn, Pinsk, & Kastner, 2018; Helfrich et al., 2018; Jia, Liu, Fang, & Luo, 2017; Drewes, Zhu, Wutz, & Melcher, 2015; Huang, Chen, & Luo, 2015; Landau, Schreyer, van Pelt, & Fries, 2015; Song, Meng, Chen, Zhou, & Luo, 2014; Holcombe & Chen, 2013; Landau & Fries, 2012). On the other hand, in the presence of temporally predictable sensory stimuli, brain oscillations actively adjust to align in time the high (low) excitability states with the expected events, leading to sensory enhancement (suppression) of relevant (irrelevant) stimuli and consequent behavioral benefits (Samaha, Bauer, Cimaroli, & Postle, 2015; Morillon, Hackett, Kajikawa, & Schroeder, 2015; Bonnefond & Jensen, 2012; Cravo, Rohenkohl, Wyart, & Nobre, 2011; Stefanics et al., 2010; Lakatos et al., 2009; Schroeder & Lakatos, 2009).
Natural stimuli often contain temporal regularities, and our brain exploits them predictively via ongoing oscillatory phase alignment (e.g., see Zoefel, 2018; Morillon & Baillet, 2017; Vander Ghinst et al., 2016; Besle et al., 2011; Saleh, Reimer, Penn, Ojakangas, & Hatsopoulos, 2010; Lakatos, Karmos, Mehta, Ulbert, & Schroeder, 2008). However, the temporal structure of the input arriving to our sensory organs is also shaped by our own movement. We actively collect, rather than passively register, sensory information, and we do so by constantly moving our receptors (Gibson, 1962). The visual function, which is tightly coupled with the incessant movement of the eyes, exemplifies the notion of “active sensing.”
That movement participates strongly to the sensory function is mostly evident in “exploratory” behaviors—that is, motor actions aimed at gathering sensory information—some of which are peculiar to the animal kingdom, such as sniffing and whisking. Intriguingly, these behaviors often display a rhythmic component. (Micro)saccadic eye movements, for example, are naturally performed at a rate of ∼2–3 Hz (Rucci, Ahissar, & Burr, 2018). The analogy between the overt rhythmicity of motor behavior and the covert rhythmicity of attentional sampling is appealing, and some authors have suggested that they may rely on similar neuronal mechanisms (Helfrich et al., 2018; Schroeder, Wilson, Radman, Scharfman, & Lakatos, 2010). Motor-related signals (e.g., corollary discharge) are available before the actual execution of a movement and may thus serve as endogenous predictive cues to inform the sensory systems about the upcoming inputs. Traditionally, these anticipatory signals have been conceived instrumental to counteract the disruptive side effects of movement on perception (Crapse & Sommer, 2008). For example, it is well known that they allow to filter out spurious self-generated signals by selective sensory suppression and may participate to the mechanism mediating perceptual stability by updating and remapping spatial information across movements (Binda & Morrone, 2018; Burr & Morrone, 2011; Medendorp, 2011; Ross, Morrone, Goldberg, & Burr, 2001; Diamond, Ross, & Morrone, 2000). A corollary discharge signal may also operate a momentary boost of perceptual sensitivity to optimize processing of the new sensory inflow brought about by the movement itself (Schroeder et al., 2010; Melloni, Schwiedrzik, Rodriguez, & Singer, 2009). This perceptual enhancement\suppression, similar to that of attentional origin but time-locked to movement onset, might be achieved through the active modulation of neuronal oscillations.
Growing electrophysiological evidence, mostly deriving from monkey studies, shows that eye movements are accompanied by complex changes in oscillatory activity. These modulations affect multiple brain sites, including low- and higher order visual areas (V1, V2, V4, STS), as well as memory-related structures (e.g., hippocampus), and involve a wide range of frequencies (Lowet et al., 2018; Neupane, Guitton, & Pack, 2017; Staudigl, Hartl, Noachtar, Doeller, & Jensen, 2017; Lowet, Roberts, Bosman, Fries, & De Weerd, 2016; Brunet et al., 2015; Hoffman et al., 2013; Jutras, Fries, & Buffalo, 2013; Bartlett, Ovaysikia, Logothetis, & Hoffman, 2011; Ito, Maldonado, Singer, & Grün, 2011; Bosman, Womelsdorf, Desimone, & Fries, 2009; Rajkai et al., 2008). The majority of these studies analyze the immediate postfixational epoch, showing a phase reset of low-frequency (delta\theta\alpha) oscillations, high-frequency (gamma) power modulations, and an increase in spike-field coherence. However, these effects may not be caused by the movement itself but by the sensory consequences of the movement, such as the image refresh associated with the eye movement. Only few evidence points to a direct involvement of motor signals, like the demonstration of persistent oscillatory modulations when the eyes move on a homogenous screen (Ito et al., 2011) or in complete darkness (Rajkai et al., 2008). Yet, the relative contribution of motor- and sensory-driven signals is difficult to disentangle in many cases, especially when the frequency of the neuronal rhythm undergoing phase modulations matches closely the saccadic rate. In this case, the periodically evoked transients could easily be mistaken for ongoing oscillations (Deouell, 2016). In support of the motor-driven account, one early study, recording local field potentials in monkeys engaged in free viewing of natural images, reported that alpha–beta oscillatory phases were more strongly locked to the onset of saccades compared with fixations (Ito et al., 2011). However, as for the majority of the studies (Hoffman et al., 2013; Jutras et al., 2013; Bartlett et al., 2011; Rajkai et al., 2008), phase concentration was confined only to the postmovement epoch. One exception to this finding is provided by a recently published study by Staudigl and colleagues (2017), who collected both intracranial and magnetoencephalography data in humans during free exploration and memorization of visual images. The two data sets consistently show that alpha oscillations in visual and memory-related structures are phase-locked to saccade onset already 250 msec before the (self-initiated) movement. More interestingly, alpha phase-locking to saccades was stronger during visual scan of items that were subsequently remembered as opposed to forgotten, suggesting its functional relevance for the encoding, not just the sampling, of visual information (Staudigl et al., 2017).
The possibility that phase alignment of slow rhythms could precede and actually predict movement onset has also been put forward by Bosman and colleagues (2009) for microsaccades, although conclusive empirical proof is still lacking. The microsaccadic delta\theta-band rhythm (∼2–4 Hz) periodically shapes local and interareal gamma-band synchronization in early visual areas (Lowet et al., 2016, 2018), indicating that it may have a fundamental role in regulating the information flow across the visual circuitry.
As outlined above, modulations of oscillatory activity at the time of eye movements are being increasingly documented by the rapid accumulation of data coming from electrophysiological recordings in animals. These studies, however, can hardly offer evidence of whether these modulations do actually bear any relevance for perception. More recent works, mostly behavioral but also neurophysiological, specifically address this issue in humans and show that movement-locked oscillations are indeed perceptually relevant, anticipatory, and most importantly, not an exclusive property of the oculomotor system.
NEURONAL OSCILLATIONS SYNCHRONIZE PERCEPTION AND ACTION
In a series of studies, participants have been asked to perform a dual task: They had to execute a self-paced movement and, at the same time, to discriminate\detect a near-threshold visual stimulus, which was briefly flashed at unpredictable times relative to movement performance. By using a time-resolved approach, visual perception was probed over a long time window surrounding movement execution, allowing to reveal possible oscillatory traces in perceptual performance already during the motor planning phase (see Figure 1).
Benedetto and Morrone (2017) had participants performing continuous, slowly paced saccades (∼1 saccade every 3 sec) between two fixed targets and probed visual contrast discrimination in-between movements. They found delta-band oscillations (∼2–3 Hz) in visual performance, which are time-locked to saccade onset, beginning ∼1 sec before and continuing up to 1 sec after it (see Figure 2). This unveiled an ongoing perceptual rhythm that clearly outlasts the short-lived neuronal modulations observed in the monkey electrophysiological recordings. Given the long intersaccade interval (∼3 sec), the authors can also firmly exclude that the premovement rhythmicity reflects postmovement modulation due to the preceding saccade. This adds compelling evidence to the neurophysiological literature by suggesting that eye movements are effectively coupled to an ongoing visual delta rhythm. Interestingly, saccadic visual suppression and postfixational enhancement are embedded within the perceptual oscillation—that is, they are both part of its phasic modulations—opening the possibility that the ongoing oscillation determines the time of the transient phenomena (Benedetto & Morrone, 2017). Nonetheless, they report that the rhythmic modulation in visual perception slightly changes its dominant frequency from the pre- (∼3 Hz) to the postsaccadic (∼2 Hz) epoch, indicating that saccade execution may introduce a discontinuity in the oscillatory dynamics (Figure 2).
In a similar experiment, Hogendoorn (2016) suggested instead that saccades do not reset the phase of visual oscillations: The presaccadic phase (and frequency) is in fact preserved after the eye movement, although for a shorter time, compared with what was reported by Benedetto & Morrone (2017) (∼500 msec compared with ∼1 sec). The across-movement phase preservation leads Hogendoorn to speculate that eye movements, rather than playing an active role, may be themselves constrained by the phase of an ongoing rhythm, presumably of attentional origin (Hogendoorn, 2016). Both studies report saccade-related behavioral rhythmicity to be confined within the delta-band (2–4 Hz; see also Wutz, Muschter, van Koningsbruggen, Weisz, & Melcher, 2016, for consistent results, further discussed below). Higher frequency (alpha\beta) oscillations of behavioral performance (visual RTs) after a microsaccadic movement have also been reported recently (Bellet, Chen, & Hafed, 2017). Interestingly, this alpha\beta oscillatory period that follows a microsaccade is further modulated by a slower oscillatory dynamics, alternating between visual hemifields at a rate of ∼2.5 Hz, with the initial hemifield preference being coherent with the direction of the microsaccadic movement. This alternate hemifield switching recalls the antiphasic fluctuation of spatial attention between different visual locations and objects (Re, Inbar, Richter, & Landau, 2019; Fiebelkorn et al., 2018; Helfrich et al., 2018; Jia et al., 2017; Fiebelkorn, Saalmann, & Kastner, 2013; Landau & Fries, 2012). Recent monkey (Fiebelkorn et al., 2018) and human intracranial (Helfrich et al., 2018) data provide converging evidence that the rhythmic sampling of visual spatial locations is shaped by multiplexed oscillations across the frontoparietal network, with higher frequency modulations being coupled to a lower, theta-band, neuronal rhythm. Altogether, this evidence points back to the long-debated link between attention and eye movements (Smith & Schenk, 2012; Rizzolatti, Riggio, Dascola, & Umiltá, 1987) and raises the question of whether the saccadic initiation might actually be dictated by a covert attentional rhythm (Helfrich, 2018; Helfrich et al., 2018), as suggested also by Hogendoorn (2016). More recently, Fiebelkorn and Kastner (2019) have proposed a model that aims at reconciling attention-based sensory sampling and eye movements control within a unified view. According to their proposal, two opposite states would alternate at a theta rhythm (Fiebelkorn & Kastner, 2019). A given phase of this theta rhythm would be associated with increased perceptual sensitivity (at the attended location) and concomitant motor suppression. As shown by the same authors (Fiebelkorn et al., 2018; Helfrich et al., 2018), this theta phase is coupled, at the behavioral level, with improved perceptual performance and, at the neural level, with high gamma power in parietal (LIP)/sensory areas and high beta power in motor-related (FEF) areas, which index activity enhancement and suppression, respectively. The opposite theta phase would instead promote eye movements initiation and, at the same time, dampen sensory activity. Consistently, motor areas are released from beta-band suppression, whereas sensory areas are inhibited by alpha-band synchronization, which could explain the decline in perceptual performance (Fiebelkorn, Pinsk, & Kastner, 2019; Fiebelkorn et al., 2018). Within this model, sensory (sampling) and motor (exploratory eye movements) processes would be boosted at opposite phases of a common theta rhythm. This provides a possible explanation for the oscillations in human visual sensitivity in synchrony with saccadic eye movements onset described above (Benedetto & Morrone, 2017; Hogendoorn, 2016; Wutz et al., 2016).
Remarkably, oscillations in visual perception are not solely observed with movements of the oculomotor effector, which, by imposing a displacement of the visual receptor (the retina), is by necessity both anatomically and functionally integrated with the visual system. Tomassini, Spinelli, Jacono, Sandini, and Morrone (2015) asked participants to perform a (self-initiated) reaching movement with the right hand while monitoring two different spatial locations for the appearance of an unpredictable, low-contrast, visual target. Visual performance for both locations shows rhythmic, theta-band (3–7 Hz) periodicity that is time-locked to the hand movement. Like for saccades (Benedetto & Morrone, 2017), the observed action-locked perceptual rhythmicity emerges long before movement onset, suggesting an automatic coupling between visual processing and motor planning (Tomassini et al., 2015). In a follow-up EEG experiment, Tomassini, Ambrogioni, Medendorp, and Maris (2017) reveal the neurophysiological underpinnings of this coupling, showing that action planning is accompanied by an endogenous phase adjustment of perceptually relevant neuronal oscillations (see Figure 3). The authors found evidence of two distinct epochs in which theta (∼4 Hz) phases are both predictive of visual perception (for later presented stimuli) and consistently aligned to the (future) hand movement: an early epoch, around 1.5 sec before motor execution, and a later epoch, starting at 0.5 sec and peaking at movement onset (see Figure 3). Despite sharing the same spectral specificity, these two visuomotor rhythms have different topographies (frontocentral and occipitoparietal in the early and late epochs, respectively), and most importantly, they have independent predictive power for perception, suggesting that they might reflect two distinct processes, which are initiated at different times during movement preparation (Tomassini et al., 2017). This depicts a more complex picture compared with what was previously provided by the purely behavioral studies: Multiple oscillatory signals are coupled to both motor and visual performance with varying temporal dynamics and spatial distribution, and they might therefore play distinct sensorimotor functions.
Overall, the current evidence demonstrates that visual rhythms are not only phase reset by external, attentional-capturing cues, but they can be locked to internally generated motor events even of nonocular nature, with functional consequences for perception. Yet, this putative visuomotor oscillatory coupling exhibits complex temporal, spatial, and spectral features.
Some variability has been reported in the exact modulation frequency across studies and participants. Research on attention mechanisms has proposed that an 8–10 Hz visual sampling rhythm is divided (cycle-by-cycle) across space so that each location is subsampled at a rate that scales inversely with the total number of attended locations. Theta range (4–5 Hz) rhythmicity in visual sampling is, in fact, primarily reported when two different locations\objects are simultaneously attended (Re et al., 2019; Fiebelkorn et al., 2013, 2018; Helfrich et al., 2018; Jia et al., 2017; Landau et al., 2015; Landau & Fries, 2012). Apparently, the movement-locked effects do not comply with this rule. Despite differences in visual task (detection\discrimination or segregation\integration), number of locations to be attended (one\two\multiple), and eccentricity of the stimuli (foveal\peripheral), saccade-locked (Benedetto & Morrone, 2017; Hogendoorn, 2016; Wutz et al., 2016) and hand-locked (Tomassini et al., 2015, 2017; Benedetto, Spinelli, & Morrone, 2016) rhythmicity has been generally observed in the delta (∼2–4 Hz; with one exception, see Bellet et al., 2017) and theta (∼4–6 Hz) range, respectively. Saccadic scan of the world is typically performed at a rate of about two to three saccades per second (Rucci et al., 2018; Morrone & Burr, 2009; Findlay & Gilchrist, 2008). One can argue that the perceptual modulations merely reflect the inherent sampling frequency imposed by the oculomotor system. In other words, perceptual periodicities might match the preferred frequency of the effectors involved in the sensorimotor behavior. Alternatively, as already mentioned, delta\theta-band rhythmicity may reflect a common attention-based clocking mechanism, which governs jointly both perceptual sensitivity and movement initiation (Fiebelkorn et al., 2019).
However, both these lines of reasoning do not fit equally well the case of hand movements, which, in contrast to eye movements, do not show any clear temporal organization in natural behavior and do not (at least anatomically) mediate the actual sampling of visual information.
Many factors may indeed contribute to the frequency variability, including individual specificities (Benedetto, Lozano-Soldevilla, & VanRullen, 2018; Gulbinaite, van Viegen, Wieling, Cohen, & VanRullen, 2017; Ho, Leung, Burr, Alais, & Morrone, 2017; Benedetto et al., 2016; Samaha & Postle, 2015; Tomassini et al., 2015; Fiebelkorn et al., 2013) and task difficulty (Babu Henry Samuel, Wang, Hu, & Ding, 2018; Chen, Wang, Wang, Tang, & Zhang, 2017). A fascinating field of future investigation will be understanding whether this diversity, both across tasks and subjects, effectively indexes functional differentiation. This will probably also help in gaining important insights into the functional relevance of this phenomenon and its possible context- and task-dependent modulation.
To clarify the functional role of this sensorimotor synchronization mechanism, it is also important to consider whether it is domain-general. The existing literature addressing the rhythmic nature of perception has been traditionally focused on vision (see VanRullen, 2016). Nevertheless, a few studies have reported oscillatory modulations also in tactile (Baumgarten, Schnitzler, & Lange, 2015; Ai & Ro, 2014) and auditory (Ho et al., 2017; Hickok, Farahbod, & Saberi, 2015) perception, suggesting that rhythmicity may be a general (amodal) organizing principle. However, the strength of the oscillatory modulations varies from study to study, and it has been particularly challenging to demonstrate their presence for audition. Changes in the experimental conditions (as the inclusion of acoustic noise or binaural presentation) may—in fact—mask or disrupt the perceptual oscillation (VanRullen, Zoefel, & Ilhan, 2014; Zoefel & Heil, 2013). Evidence of phase modulations are obtained also in cross-modal and multisensory studies, reinforcing the suggestion that neural oscillations may play a role in synchronizing signal processing between different sensory modalities (Mercier et al., 2015; Romei, Gross, & Thut, 2012; Lakatos, Chen, O'Connell, Mills, & Schroeder, 2007).
So far, action-locked perceptual oscillations have only been reported for visual stimuli, although for movements executed by different effectors (eyes, hands). The effector independence strongly points to a mechanism that transcends the anatomofunctional links between the sensory and motor systems involved. Nevertheless, it does not exclude that this phenomenon may be a peculiarity of vision. Indeed, the sensory modalities largely differ in their anatomofunctional interplay with the motor system. Because of the anatomical co-localization of sensors (retina, skin) and effectors (eyes, limbs), both vision and somatosensation own deep functional interconnections with the oculomotor and skeletomotor system, respectively. Differently, audition does not share the relevant sensory organ with any effector, and for this reason, it is rather independent from overt “sensory-gathering” motor routines. Yet, despite audition being less obviously coupled with the motor system than vision and somatosensation, many pieces of evidence have uncovered a substantial motor contribution to the neural processing of auditory (e.g., Morillon & Baillet, 2017) and, in particular, speech (e.g., Park, Ince, Schyns, Thut, & Gross, 2015; D'Ausilio et al., 2009) information.
Extending the investigation of the present phenomenon to other nonvisual domains could thus prove a valuable tool to identify the anatomical and/or functional architectures that possibly constrain its implementation and eventually clarify the specific computational/functional needs it might fulfill.
In the following sections, we will discuss different hypotheses on how this sensory–motor oscillatory coupling might be achieved, with primary reference to the visual modality, and what functions it might subserve.
ACTION–PERCEPTION COUPLING: POSSIBLE MECHANISMS AND FUNCTIONAL ROLE
The new evidence gathered in humans has revealed some key features of the movement-locked neuronal and corresponding visual rhythmicity, which prompt to partly revise the potential functional role of these modulations (Schroeder et al., 2010; Melloni et al., 2009). First, this phenomenon is not exclusive of the oculomotor system, but it also involves the hand effector, which is anatomically and functionally decoupled from the visual receptor. This means that the oscillatory coupling is established independently of any intrinsic link between the motor and the sensory system. Second, it is observed for arbitrary sensory–motor contingencies, that is, concurrently performed, but functionally unrelated, visual and motor tasks. Third, it is not a transient phenomenon, but an anticipatory, long-term modulation that precedes movement onset by ∼1 sec. This finding excludes that the final motor output (muscle contraction or reafference from body motion) may be itself the true synchronizing event and instead points to a hidden endogenous source. Finally, it occurs in the absence of any overt rhythmicity, either in motor behavior or in sensory stimuli. Hand movements (e.g., reaching, grasping, object manipulation), different from other “sensory-gathering” motor behaviors (e.g., visual exploration, locomotion, whisking, sniffing, licking), do not even manifest rhythmic patterns in natural scenarios.
Altogether, this suggests that the functional significance of this mechanism may go far beyond the idea that motor signals merely “support” the sensory function by promoting timely suppression and enhancement of sensory excitability to enable perceptual stability (Melloni et al., 2009).
A first step toward understanding the genesis and, possibly, the functional role of this phenomenon requires answering the following questions: When exactly during the cascade of neural events leading up to action execution is this visuomotor synchronization achieved? And what mechanism is responsible for this synchronization?
The studies already described have started to address these questions by delineating the spatiotemporal and spectral features of the visuomotor oscillatory coupling, but they did not reveal the directionality of this coupling: Is it the motor activity that drives the rhythmic visual modulation at the time of action planning\performance, or conversely, is it an intrinsic visual rhythm that yokes motor activity, cyclically dictating the probability of spontaneous movement initiation? Alternatively, are both motor and visual processes regulated by a shared rhythmic source?
Evidence suggests that motor functions, like sensory ones, are governed by underlying rhythmic processes indexing fluctuating states of neuronal excitability. Voluntary movements tend to be timed according to preferential phases of the ongoing oscillations (Bates, 1951). Behavioral studies have reported nonuniform, periodic, distributions of movement onset times (Dehaene, 1993; Treisman, Faulkner, & Naish, 1992; White & Harter, 1969; Latour, 1967) and rhythmic fluctuations of RTs following sensory cues (Huang et al., 2015; Song et al., 2014). This evidence is corroborated by neurophysiological findings showing systematic associations between the phase (and amplitude) of neuronal oscillations within the theta\alpha range and response speed (Kienitz et al., 2018; Drewes & VanRullen, 2011; Hamm, Dyckman, Ethridge, McDowell, & Clementz, 2010; Bollimunta, Chen, Schroeder, & Ding, 2008; Kirschfeld, 2008; Linkenkaer-Hansen, Nikulin, Palva, Ilmoniemi, & Palva, 2004).
However, given that the movements are externally triggered, it is difficult to disentangle whether the rhythmical modulation in RTs originates from oscillations in the sensory system (i.e., at the input stage) or in the motor system (i.e., at the output stage).
By probing directly corticospinal excitability, TMS studies provide more compelling evidence that neuronal oscillations actually entail cyclic modulations of the motor system excitability. Indeed, motor evoked potentials are modulated not only by the amplitude (İşcan, Schurger, Vernet, Sitt, & Valero-Cabré, 2018; Keil et al., 2014; Schulz, Übelacker, Keil, Müller, & Weisz, 2014; Mäki & Ilmoniemi, 2010) but also by the pre-TMS phase of ongoing oscillations at both peripheral (i.e., muscle activity; Keil et al., 2014, van Elswijk et al., 2010) and cortical level (Khademi, Royter, & Gharabaghi, 2018; Berger, Minarik, Liuzzi, Hummel, & Sauseng, 2014; Keil et al., 2014). The use of neuromodulation techniques, such as transcranial alternating current stimulation, in combination with TMS, have further shown that entrained, in addition to ongoing, beta-band oscillations affect motor evoked potentials size in a phase-dependent manner (Schilberg et al., 2018; Guerra et al., 2016; Nakazono, Ogata, Kuroda, & Tobimatsu, 2016; Raco, Bauer, Tharsan, & Gharabaghi, 2016).
Interestingly, the ongoing activity within the motor system, even at its most peripheral level (i.e., the muscles), shows early modulations by sensory stimulation. Visual stimuli, for example, elicit time-locked recruitment of neck and upper-limb muscles at very short latencies (<100 msec; Gu, Wood, Gribble, & Corneil, 2016; Pruszynski et al., 2010; Corneil, Olivier, & Munoz, 2004) and reset the phase of low-frequency muscle oscillations (Wood, Gu, Corneil, Gribble, & Goodale, 2015). Multiphasic event-related responses to salient auditory and somatosensory stimuli can be detected in the motor output (i.e., in the force produced by isometric contraction), and these responses are coupled in time and amplitude to corresponding EEG evoked potentials to the same stimuli (Novembre et al., 2018). Notably, the stimulus-locked responses observed in the force are neither startle-like nor reflexive, suggesting a flexible, context-dependent, sensorimotor “resonance” mechanism (Novembre et al., 2018).
Sensory stimuli not only affect peripheral motor activity but also modulate phase dynamics in an extended sensorimotor cortical network (Hirvonen, Monto, Wang, Palva, & Palva, 2018; Lobier, Palva, & Palva, 2018; Mercier et al., 2015; Besle et al., 2011; Bressler, Coppola, & Nakamura, 1993). Interestingly, multimodal stimuli promote stronger local as well as interregional low-frequency (delta\theta) phase synchronization between sensory and motor areas compared with unimodal stimuli (Mercier et al., 2015), and the strength of the sensorimotor coupling positively scales with response speed (Hirvonen et al., 2018; Mercier et al., 2015). A recent study on monkeys further shows that the oscillatory synchronization between motor and somatosensory areas is subject to learning-dependent plasticity (Arce-McShane, Ross, Takahashi, Sessle, & Hatsopoulos, 2016), reinforcing the idea that phase alignment may be a general strategy to establish effective neuronal information transfer (Palva & Palva, 2018; Fries, 2015; Womelsdorf & Fries, 2006; Engel, Fries, & Singer, 2001).
Another set of evidence showing perception-to-action oscillatory modulations stems from motor priming paradigms. Visual stimuli can automatically activate specific motor plans; this immediate motor facilitation of the primed action is, however, shortly replaced in time by its inhibition (Sumner & Brandwood, 2008; Eimer & Schlaghecken, 2003), with RTs for two competing actions fluctuating with antiphasic theta-band periodicity for almost 1 sec (Huang et al., 2015).
Whereas all these pieces of evidence suggest that sensory stimuli and, in particular, visual stimuli can trigger changes in motor-related oscillatory activity and sensorimotor phase synchronization (and corresponding motor performance), a recent study shows that the reverse is also true. Tomassini and D'Ausilio (2018) characterized the spectrotemporal dynamics of visual perception ensuing from externally controlled activation of the somatomotor hand system, which allowed bypassing of the endogenous component involved in action programming and execution (Tomassini & D'Ausilio, 2018). To this aim, they exploited the peripheral stimulation of the median nerve: In fact, despite not targeting specifically and uniquely the motor cortex, this stimulation activates with nearly the same latency both the somatosensory and motor subdivisions of the hand, which are known to constitute a single functional unit in the control of movement (e.g., Lemon, 2008). They observed strong and long-lasting (1 sec) alpha oscillations in visual perception following the median nerve stimulation. Phase-locking of visual rhythms is thus not conditional on the formulation of an intention to move or of a motor plan but may also ensue from the passive recruitment of the somatomotor system.
We have seen multifaceted findings showing that the sensory and motor systems exert reciprocal influences through modulation of the ongoing brain dynamics.
The evidence available so far is not conclusive with regard to the origin of the visuomotor synchronization and its underlying brain circuitry. However, a few clues suggest that motor-related activity might play a key role. Notably, visuomotor synchronization (as indexed by movement-locked rhythmicity in perception) seems to be automatically established whether movement is freely initiated (Bellet et al., 2017; Benedetto & Morrone, 2017; Tomassini et al., 2015), internally timed (Tomassini et al., 2017), or externally triggered (Hogendoorn, 2016; Wutz et al., 2016), that is, in conditions demanding very different mechanisms for the control of movement onset. Any phasic modulation of spontaneous movement initiation due to ongoing fluctuations in motor excitability should be independent of the task-specific constraints placed on movement timing. However, in the study by Tomassini et al. (2017), it is shown that the visuomotor oscillatory coupling evolves with a different temporal dynamics when participants are asked to perform the movement at two different time intervals (1.5 and 2.3 sec) after cue presentation. This difference (which is not a trivial consequence of cue presentation) strongly suggests that theta oscillatory dynamics may be itself a constituent part of movement preparation, at least of the neuronal process that is specifically involved in the active (task-related) control of movement timing. The rhythmic coupling between the visual and motor system is thus not invariant to the current motor state, suggesting that it may structure dynamically the functional interplay between the two systems, enabling the timely incorporation of sensory information within the ongoing motor plan.
MULTIPLE CLOCKS, ONE TIME?
Natural behavior commonly relies on multimodal sensorimotor loops, whereby the sensory stream of information continuously updates the current motor programming, and the motor system, in turn, generates predictions about the upcoming sensory data. Multimodal signals, however, travel at different speeds along the sensory pathways. Alongside, the perceived timing of sensory events varies across modalities and features of the sensory stimulation (Burr, Cicchini, Arrighi, & Morrone, 2011; Harrington, Castillo, Fong, & Reed, 2011; Tomassini, Gori, Burr, Sandini, & Morrone, 2011; Johnston, Arnold, & Nishida, 2006; Kanai, Paffen, Hogendoorn, & Verstraten, 2006). How does the brain deal with temporally inconsistent signals and achieve an efficient sensorimotor control?
Besides affecting sensitivity, it is known that brain oscillatory activity is strictly related to the temporal features of the stimuli (Wiener, Parikh, Krakow, & Coslett, 2018; Milton & Pleydell-Pearce, 2016; Baumgarten et al., 2015; Cecere, Rees, & Romei, 2015; Kononowicz & van Rijn, 2015; Samaha & Postle, 2015; Kösem, Gramfort, & van Wassenhove, 2014; Parker, Chen, Kingyon, Cavanagh, & Narayanan, 2014). For instance, recent evidence points to a direct mapping between neuronal oscillations and temporal windows of integration\segregation (Ronconi & Melcher, 2017; Ronconi, Oosterhof, Bonmassar, & Melcher, 2017). Interestingly, the probability of either integrating or segregating two stimuli alternates as a function of the phase of neuronal rhythms (Ronconi et al., 2017), and these rhythms are aligned to saccadic fixation onset (Wutz et al., 2016). This latter evidence suggests that movement-related modulations of oscillatory activity may also have an impact on the temporal processing of the stimuli.
Multiple lines of evidence strongly suggest that the motor system is critically involved in time-keeping functions (e.g., Merchant & Yarrow, 2016), as attested by its consistent recruitment in both implicit and explicit temporal tasks (Merchant, Harrington, & Meck, 2013; Wiener, Turkeltaub, & Coslett, 2010). It has been shown that motor areas effectively tune the ongoing activity of sensory areas to incoming rhythmical inputs, yielding perceptual improvements (Morillon & Baillet, 2017). The motor system can thus convey temporal predictions to the sensory system through rhythmic top–down modulation (Morillon et al., 2015; Arnal & Giraud, 2012; Schubotz, 2007). Movement also has a powerful influence on perceived time. Visual timing, for example, is compressed and\or dilated around the execution of both eye (Binda, Cicchini, Burr, & Morrone, 2009; Morrone, Ross, & Burr, 2005) and hand movements (Tomassini, Vercillo, Torricelli, & Morrone, 2018; Tomassini & Morrone, 2016; Hagura, Kanai, Orgs, & Haggard, 2012; Park, Schlag-Rey, & Schlag, 2003; Haggard, Clark, & Kalogeras, 2002). This movement-related temporal modulation has been observed also for other sensory domains, such as for tactile stimuli, suggesting that it is a general phenomenon (Tomassini, Gori, Baud-Bovy, Sandini, & Morrone, 2014; Tomassini, Gori, Burr, Sandini, & Morrone, 2012; Yarrow & Rothwell, 2003).
One recent study by Benedetto, Burr, and Morrone (2018) further shows that audiovisual temporal order judgments undergo rhythmic fluctuations (at around 7–8 Hz) time-locked to a button press action (Figure 4). The modulation does not affect temporal sensitivity (precision of the judgments), but the temporal bias, that is, whether the visual stimulus is perceived ahead or behind of the auditory stimulus (in line with the neuronal phase effects on audiovisual perceived simultaneity reported by Ikumi, Torralba, Ruzzoli, & Soto-Faraco, 2019; Kösem et al., 2014). This suggests that processing resources and\or speed may rhythmically alternate between the visual and auditory modalities, and this alternation may be temporally synchronized to the ongoing motor processing. The motor system seems to be able to orchestrate the processing of multiple signals that are relayed at different speeds along the CNS by exerting endogenous control over the brain oscillatory dynamics.
The existence of a mechanism that keeps perception and action finely synchronized is suggested by a recent study by Tomassini et al. (2018). The authors assessed interval estimation for a brief visual stimulus (150 msec) that was shown (at random times) while participants were performing rhythmic finger tapping (at 1 Hz). Perceived visual time undergoes distortions, which are locked to the motor acts; time is compressed close to the onset of finger taps and expanded in-between successive taps. Remarkably, the temporal dynamics of these perceptual distortions scales linearly with the timing of the motor tapping, so that maximal time expansion is always experienced at the center of the intertap interval, independent of the natural (trial-by-trial) variability in the tapping rate (see Figure 4). Perceptual time is thus anchored to the internally dictated rhythm of motor production. These results indicate that even if the sensory and motor clocks might be distinct, their functioning is nevertheless strictly coupled.
The movement-locked rhythmicity in perception may thus be the by-product of an oscillation-based mechanism whereby the dynamics of sensory processes can be plastically scaled to be synchronized with the ongoing motor processes, effectively closing up the sensorimotor loop.
Growing evidence shows that brain oscillatory dynamics is anticipatorily phase-locked to movement onset (Staudigl et al., 2017; Tomassini et al., 2017; Popovych et al., 2016) and that this very same activity affects perceptual (Tomassini et al., 2017) as well as memory (Staudigl et al., 2017) performances, that is, it is behaviorally relevant. Strikingly, similar movement-locked rhythmicity in behavioral performance is reported irrespective of the effector, movement type and task demand; for example, with saccades (Benedetto & Morrone, 2017; Hogendoorn, 2016; Wutz et al., 2016) and microsaccades (Bellet et al., 2017), reaching movements (Tomassini et al., 2015), isometric contraction (Tomassini et al., 2017), button press (Zhang, Morrone, & Alais, 2019; Benedetto et al., 2016), as well as with freely initiated (Bellet et al., 2017; Benedetto & Morrone, 2017; Tomassini et al., 2015), internally timed (Tomassini et al., 2017), externally triggered movements (Hogendoorn, 2016; Wutz et al., 2016) and also exogenous activation of the somatomotor hand system (Tomassini & D'Ausilio, 2018).
These findings outline a new and promising view on the intimate interconnection between sensory and motor functions and its possible neurophysiological substrate.
The evidence available so far is restricted to low-level, near-threshold, sensory tasks (e.g., visual contrast detection) and arbitrary sensory–motor contingencies (concurrently performed, but functionally decoupled, sensory and motor tasks). Although this points to an automatic and domain-general mode of operation of sensorimotor systems, which may subserve core functions, its general relevance and functional significance are still not clear. Indeed, if oscillatory mechanisms do shape the sensorimotor information flow, their functional impact should be amplified for more complex behaviors that truly depend on fast and accurate information exchange between the sensory and motor system for their successful accomplishment.
The study of the role of oscillatory mechanisms in action–perception coupling is at its beginning, and understanding the degree of (domain) specificity, plasticity, and context dependency of these mechanisms is only one of the many aspects that call for further investigation.
The existing evidence certainly reveals that perception and action are inherently coupled even when their coordination is not directly enforced by the task at hand. Their study cannot thus prescind from a unified perspective whereby perception is conceived as a sensorimotor phenomenon for which attempting to identify univocally the origin, within either the motor or the sensory system, is most probably an intrinsically ill-posed problem.
This work was supported by the European Research Council (ERC) under the European Union's Horizon 2020 research and innovation programme (grant agreement no. 801715 - PUPILTRAITS), by the ERC-FP7 ECSPLAIN (grant agreement no. 338866) to M. C. M., and by the ERC/ADV GenPercept (grant agreement no. 832813). A. T. acknowledges the support of EnTimeMent “ENtrainment and synchronization at multiple TIME scales in the MENTal foundations of expressive gesture” (H2020-FETPROACT-824160).
Reprint requests should be sent to Alice Tomassini, Center for Translational Neurophysiology of Speech and Communication, Istituto Italiano di Tecnologia, Via Fossato di Mortara, 17-19, 44121 Ferrara, Italy, or via e-mail: email@example.com.
This article is part of a Special Focus deriving from a symposium at the 2018 annual meeting of Cognitive Neuroscience Society, entitled, “Oscillatory mechanisms underlying human perception and attention.”