To the best of our knowledge, the first to describe the rules governing postural and voluntary movement control was Leonardo da Vinci (1452–1519), in his Libro A, Trattato della Pittura, which is now conserved within the Codex Urbinas Latinus. Leonardo wrote: “I say that if a motionless figure is poised on his feet, and his arm is extended in front of the chest, he will throw backward as much natural weight as the natural and accidental weight which he thrusts forward. And I say the same of each part that projects more than usual beyond the whole”. Moreover, Leonardo stated: “Never will a weight be lifted and carried by a man, without his extending outside himself more than as much weight as that which he wishes to lift, and he thrust it on the side opposite the one where he lifts the weight”. A couple of centuries later, Giovanni Alfonso Borrelli (1608–1679) followed the pathway of Leonardo in his De Motu Animalium, where he described some principles that govern the voluntary movement. In particular, Borelli considered the skeleton as multi-linked system of levers, which can maintain the body balance as long as the center of mass falls within the support base.

After the observation of Leonardo da Vinci and Giovanni Borelli, we had to wait until the end of the 19th century to first clinically observe the importance of correct postural tailoring during voluntary movements in order to execute a successful motor act. Joseph Babinski (1857–1932), in his De l’asynergie cérebelleuse, first described the lack of harmonious synergies in cerebellar patients; in particular, he observed the forward displacement of the knees and hips to compensate for the backward displacement of the trunk, neck and head when asking a subject to look upward. Babinski also observed that cerebellar patients, who were asked to perform the same looking upward task, were unable to coordinate the upper part of their body to their hip and lower limb, so that they usually fell. However, Babinski did not analyze the temporal relationship between the postural adjustments and the voluntary movement, and therefore he lacked to observe that also an anticipatory postural control, adjusted in a feed-forward manner, is needed when performing a voluntary movement.

In the 20th century, studies about the relationship between posture and voluntary movement stood on the shoulders of three giants of movement physiology: Charles Sherrington (1857–1952, Nobel Prize 1932), Walter Hess (1881–1973, Nobel Prize 1949) and Nikolai Bernstein (1896–1966, Stalin Prize for Science 1948). Sherrington sustained the idea of a dual coordinated control system: one for movement and one for posture. This idea was in agreement with the proposals of Hess, who held the view that without anticipation of postural adaptations (a component of his ereismatic-supporting function system), goal-directed movements were doomed to failure. On the other hand, Bernstein was the first to propose a unique motor command for both posture and voluntary movement: “movements are not chains of details but structures which are differentiated into details; they are structurally whole” (Bernstein, 1967). However his ideas did not reach western scientists until several years after his death since his articles and books were only recently translated into English.

These are the most frequently studied APAs. For example, when we intend to move a segment of the upper or the lower limb, a chain of muscular actions develops from that moving segment to the nearest fixation point. These actions precede the primary movement (in the range of tenths ms) and in many cases also the prime mover recruitment, the larger advance being observed in the muscles acting near the fixation point.

These inter-limb APAs have been documented for various movements such as: shoulder flexion and extension (Belen’kii et al., 1967; Lee, 1980; Clément et al., 1984; Horak et al., 1984; Bouisset and Zattara, 1987; Maki, 1993 in adults; Riach and Hayes, 1990 in children), shoulder lateral abduction (Aruin and Latash, 1995; Vernazza et al., 1996), elbow flexion (Friedli et al., 1984) and similar movements of the lower limb (Rogers and Pai, 1990; Do et al., 1991; Mouchnino et al., 1991; Nouillot et al., 1992). APAs also accompany movement involving the trunk, when bending it (Oddsson and Thorstensson, 1986; Crenna et al., 1987; Pedotti et al., 1989), during whole body reaching (Stapley et al., 1998), rising up on tiptoe (Houtz and Fischer, 1961; Lipshits et al., 1981; Nardone and Schieppati, 1988) and rocking on the heels (Nardone and Schieppati, 1988). In some experimental situations, it was asked to push or pull a handle with the upper limb, both in normal subjects (Cordo and Nashner, 1982; Lee et al., 1987) and in patients with Parkinson disease (Dick et al., 1986). Inter-limb APAs are also associated with movements involving a bimanual load-lifting task, when participants hold an object in one hand and the object is voluntarily lifted by the same person (active lifting) or by an experimenter (passive lifting). In the first case, the voluntary activation of the lifting hand is synchronous with an inhibitory APA in the contralateral biceps brachii (BB), thus avoiding an uncontrolled flexion of the elbow (Hugon et al., 1982) and preserving the upper limb posture. On the contrary, when the object is unexpectedly lifted by the experimenter, the BB inhibition starts about 50 ms after the unloading, showing that is it should have a reflex origin (postural reflex). The described experiment, known as “the barman task” allowed to make two conclusions: first, even a movement that does not lead to a whole body imbalance is preceded by APAs, aimed at preserving the single segment’s balance. Second, APAs may develop also in muscles that are not usually considered as postural muscles, such as muscles in the upper-limb. Moreover, since the degree of elbow rotation was shown to be lower when the subject unloaded the forearm by himself, the APA mechanism showed a greater efficacy in limb stabilization vs. the reflex mechanisms. The involvement of APAs in controlling the segmental stability was also confirmed for movements affecting the whole body balance (Patla et al., 2002).

Classically, APAs are known to be tuned depending on several kinematic aspects of the primary movement; specifically, the amplitude (Aruin and Shiratori, 2004), the speed (Horak et al., 1984; Shiratori and Aruin, 2007), the movement direction (Aruin and Latash, 1995; Pozzo et al., 2002) and also the mass of the moving segment (Friedli et al., 1984). Moreover, an increase of APAs was shown when the voluntary movement was performed against resistance (Baldissera et al., 2008).

In addition, the CNS is able to adapt inter-limb APAs to changes in the postural demand of the motor task (Belen’kii et al., 1967; Cordo and Nashner, 1982; Aruin and Shiratori, 2004; Shiratori and Aruin, 2007). Clear signs of this adaptative process have been observed even after the very first movement trial (Hall et al., 2010), although they usually develop within few movement repetitions.

Another feature of the APAs is that they overtly develop in those muscles which connect the moving segment(s) to the nearest fixation point (Belen’kii et al., 1967; Marsden et al., 1981; Cordo and Nashner, 1982; for a review see Massion, 1992), while they are correctly shaped, but kept subthreshold, in those muscles in which their action is useless (Baldissera et al., 2002). An indirect issue which is strictly correlated to the above observations is that APAs conform to the adequacy of the fixation point, being smaller and smaller or even disappearing when the fixation point does not guarantee the full discharge of the perturbation (Brenière et al., 1987 balance during locomotion; Nouillot et al., 1992 balance during unipedal stance; Dietz and Colombo, 1996 balance in water; Esposti and Baldissera, 2011 balance with two fixation chains).

As explained in the introduction, intra-limb APAs are those distributed to muscles of the same limb in which the movement occurs. The first evidence of such APAs dates to 1963, when Hopf and Hufschmidt (1963) observed that the anterior deltoid (AD) activity preceded the voluntary recruitment of BB driving elbow flexion. Later on, Aoki (1991) reported that wrist flexions, performed with the hand prone or supine, were accompanied by APAs in muscles acting at the elbow and that these APAs reverted their pattern in relation to the direction of the movement. In this regard, Chabran et al. (2001), observed also that when providing a support to the elbow the level of postural activity in shoulder muscles was deeply depressed, but the intra-limb APAs chronology remained unaffected. Examples of intra-limb APAs were also reported for shoulder and elbow movements by Almeida et al. (1995) and by Gribble and Ostry (1999).

More recently, Caronni and Cavallari (2009a,b) also reported that when only the index-finger is flexed, an APA chain develops in several upper-limb muscles to stabilize the segmental equilibrium of the arm. Indeed, these authors described that with the hand prone a brisk finger flexion was preceded by an excitatory burst in Extensor Carpi Radialis (ECR), Triceps Brachii (TB) and Superior Trapezius (ST), while Flexor Carpi Radialis (FCR), BB and AD showed a concomitant inhibition of their tonic activity (Figure 1A). The ECR-FCR coupled activities were appropriate for contrasting the by-product wrist flexion torque due to the voluntary contraction of the index-finger flexors. At the same time, the TB-BB and ST-AD actions contrasted the elbow and shoulder flexion torques produced by the upward perturbation that the index-finger flexion discharged on the metacarpophalangeal (MP) joint. Interestingly, when the hand posture was changed from prone to supine (Figure 1B), the APA pattern reverted in sign in the elbow and shoulder muscles but not in ECR and FCR. This is consistent with the fact that in this condition the perturbation produced on the MP joint was inverted, i.e., downward, but the by-product of recruiting the index-finger flexors was again a wrist flexion torque. These observations witness that APAs are versatile enough to adapt to the mechanical requirements of the postural context. Another proof that intra-limb APAs adapt to changes in the postural context within few trials comes from Bruttini et al. (2014) who illustrated that when tapping with the right index-finger, intra-limb APAs were deeply reduced (Figures 2A,B) if the right forearm was fixated at the wrist, i.e., closer to the voluntary moving segment. Moreover, Esposti et al. (2015) confirmed that also the intra-limb APAs adapt to changes in movement speed (Figure 2C).

In conclusion, intra-limb and inter-limb APAs seem to share similar control mechanisms. Indeed, both of them: (1) are distributed to several muscles creating a postural chain aiming to prevent the effects of the interaction torques generated by the voluntary movement; (2) revert in sign when movement direction is reverted; and (3) adapt to changes of the postural context within few trials of movement repetition. Thus, the central nervous system seems to use the same organization of the motor command for controlling both the segmental and the whole-body posture.

Several studies support the idea that Supplementary Motor Area (SMA) is involved in APAs generation. Severe APA impairments were observed in the bimanual load lifting (the so called barman task) when the load was held with the forearm contralateral to the lesioned SMA while no APA changes were observed in a patient with normal SMA, but suffering a complete callosal section (Massion et al., 1989; Viallet et al., 1992). This finding supports the idea that coordination between the posture and movement travels through subcortical level. Moreover, a 1-Hz repetitive Transcranial Magnetic Stimulation (TMS) on SMA, which induces an inhibitory effect, reduces the duration of APAs prior to stepping, without affecting their peak amplitude (Jacobs et al., 2009). More recently, Transcranial Direct Current Stimulation (tDCS) over the SMA has been shown to elicit a differential effect on the postural and the focal component of the movement (Figure 6), supporting the involvement of SMA in APA programming (Bolzoni et al., 2015).

The role of the Primary Motor Cortex in generating APAs has been shown by both human and animal studies. The stimulation of the primary motor cortex in the intact cats was indeed able to evoke movement in the contralateral side and APAs in the supporting limbs (Gahéry and Nieoullon, 1978), suggesting that the primary motor cortex in cats is able to control both the voluntary prime mover and the associated postural adjustments. Moreover, in cats, the dynamic of the discharge frequency in pyramidal neurons is time locked and directly proportional with the center of pressure displacement, a key parameter associated to APAs preserving the whole body equilibrium (Yakovenko and Drew, 2009). The role of the primary motor cortex in the anticipatory postural control was also observed in human studies. Indeed, Palmer et al. (1994) used the TMS to induce a silent period in either the left or the right M1 while the subject abducted his left arm, a movement which is known to be preceded by APAs in the contralateral Latissimus Dorsi muscle. Left M1 stimulation produced a delay of the APA onset, while the prime mover timing was unmodified; instead, right M1 stimulation just delayed the prime mover activation. Moreover, it has been shown for both inter- (Petersen et al., 2009) and intra-limb APAs (Caronni and Cavallari, 2009b) that when a postural muscle is at rest the spinal excitability is un-modulated during the whole period of motor preparation, while when that muscle is recruited in a postural chain the cortical motor action potentials are modulated according to the time course of APAs (Figure 7). From one perspective, this indicates that APAs processing involves M1 level, and from another this suggests that APAs and voluntary commands cannot be decoupled, since the former remains subliminal, i.e., it does not produce a mechanical effect, when the muscle is at rest. Although this may appear uneconomical, this result corroborates two previous observations in which the cortical excitability of hand movers has been shown to fluctuate under the threshold for motoneuronal firing when the ipsilateral foot was voluntarily oscillated (Baldissera et al., 2002) and in which the hidden effect developed in an overt APA when the hand was recruited in a postural act (Baldissera and Esposti, 2005).

Regarding Basal Ganglia, severe APAs impairment in patients with Parkinson’s disease were already observed by Viallet et al. (1987). More recently, anticipatory brain activity associated to a bimanual load-lifting task was localized in basal ganglia, SMA and thalamus, contra lateral to the load-bearing arm (Ng et al., 2013). It is worth noting that these areas are component nodes of the basal ganglia-thalamo-cortical motor network, which is known to be implicated in well-learned finger movements (Boecker et al., 1998). This indicates a superposition of the neural structures for APAs and those for voluntary motor command, and indirectly supports the view of a oneness of the motor command for both posture and primary movement.

The Cerebellum is deeply involved in APA regulation. Babinski (1899) reported that a cerebellar lesion disrupted the coordination between voluntary movement and equilibrium stabilization, indicating that the cerebellum is involved in postural control. This view agrees with the idea that the cerebellum contains forward internal models that could predict the consequences of an action (according to the perceived postural context) and can be used to overcome time delays associated with feedback control (Wolpert et al., 1998; Imamizu et al., 2000). Several studies on patients positively concluded for such an involvement. Indeed, patients suffering cerebellar lesions failed to show a normal anticipatory adjustment in grip force when lifting or moving an object (Müller and Dichgans, 1994; Babin-Ratté et al., 1999) and cerebellar lesions abolished APAs plasticity in a bimanual unloading task (Diedrichsen et al., 2005). Congruently, Diener et al. (1992) reported that cerebellar patients produce a normal pattern of APAs, but with abnormalities in their timing relationship with the onset of the prime mover. Finally, as shown in Figure 4, Bruttini et al. (2015) demonstrated that when performing a brisk index-finger flexion, cerebellar subjects showed a timing-disruption of intra-limb APAs, while their pattern was unmodified. Using functional magnetic resonance imaging, Schmitz et al. (2005) reported that APAs were associated with activation of sensorimotor areas, SMA and the cerebellum. On the contrary, Ng et al. (2013) found no evidence of cerebellar involvement during APAs using magnetoencephalography in a bimanual coordination task. Further support on the role of cerebellum in timing APAs come from animal studies. When reaching a water flask with the mouth, wild type mice showed a clear APA in hind-limb muscles, synchronous to neck muscles, while in transgenic mice with defective cerebellar Purkinje cells the hind-limb activity occurred markedly later than that in neck muscles (Yamaura et al., 2013).

Finally, Schepens and colleagues (Schepens and Drew, 2004; Schepens et al., 2008) emphasized the role of pontomedullary reticular formation (PMRF) in the coordination of posture and movement. In particular, they suggested that PMRF is a site of integration of signals from both cortical and subcortical structures, and that these signals ensure that APAs are appropriately scaled in time and magnitude to the intended movement, contributing to integrate the control of posture and movement, as also illustrated by Toussaint et al. (1998).

In summary, these anatomical and physiological studies show a large superimposition between neural structures governing the voluntary and the postural components of the movement, leaving the question open on whether these two processes are implemented separately or they are an expression of a unique posturo-focal command.

According to the classical view, the prime mover activity and the associated postural adjustments result from two different central commands (dual command hypothesis), which are independently dispatched to the prime mover and to the muscles generating the postural chain (Babinski, 1899; Hess, 1943; Cordo and Nashner, 1982; Brown and Frank, 1987). Conversely, a growing body of evidences favors the view that APAs and prime mover recruitment are both controlled by a shared motor command (Aruin and Latash, 1995; Stapley et al., 1998, 1999; Caronni and Cavallari, 2009b; Petersen et al., 2009). In this regard, it is interesting to mention the studies of Gritsenko et al. (2009) and Leonard et al. (2011), showing that when correcting an ongoing arm pointing movement, the CNS employs a predictive mode of postural control and consistently adapts the postural muscle activities before modifying the prime mover recruitment. These authors concluded that the postural corrections could be described as being a component of the voluntary movement, rather than ensuring the maintenance of equilibrium. Further, Bruttini et al. (2014) demonstrated that the APA chain associated to a voluntary movement cannot be decoupled from the command driving the focal movement. Indeed, when repeatedly trying to flex the index-finger under forearm ischemia, so that the voluntary command was normally dispatched but the prime mover not able to receive it, significant anticipatory adjustments were clearly visible in BB, TB and AD (Figure 8).

In this peculiar condition, in which no postural perturbation is generated, one would have expected that APAs were suppressed since unnecessary and uneconomical, unless the postural and the prime mover muscles obey a “shared” motor command. It is of course apparent that “ischemia” results does not allow to discern whether: (i) one specific neural structure generates both the postural and the voluntary commands; or (ii) two neural structures exist, one for the voluntary and the other for the postural commands, and one of the two acts as a master on the other; or (iii) a third neural structure exists, which simultaneously triggers the two slave “postural” and “voluntary” neural structures. However, from the point of view of the muscles (i.e., the only aspect that such experimental approach allowed to observe), these three ways of organizing the posturo-focal integration are all identical. Given this neural organization, it may be conceived that the system prefers an economy in computational terms, rather than in diminishing the number of activated neurons. In this case, one may imagine that any given voluntary command is associated with an arborized pattern of postural commands, forwarded to a number of possible fixation points. Thus, the most useful arborization for providing the actual support would be supraliminarly recruited, while the leftover branches would be still present but kept silent, i.e., under threshold. The amplification of APA transmission to certain targets, and the parallel attenuation towards other targets, could be automatically accomplished by servo-systems like PMRF neurons. This would effectively free the upper centers from the computational demand of adapting the APAs to the postural context. This view is also corroborated by the already cited works of Baldissera et al. (2002) and Baldissera and Esposti (2005), who showed that subliminal inter-limb APAs in a resting arm during foot movements become supraliminal when that segment is actively used for postural stabilization. Similar results were also shown for intra-limb APAs by Caronni and Cavallari (2009b).

The oneness of postural and voluntary command is also supported by the finding that the APA latency depends on the movement instruction, not on its actual velocity (Esposti et al., 2015). This conclusion was based on two observations. First, no correlation was found between APA latency and movement speed when subjects had to follow a go fast instruction (i.e., to move as fast as possible), nor when they had to go slow (at half their maximal speed). Second, the maximal speed was variable among subjects, so that minimum movement speeds within the range of go fast trials intermingled with maximal movement speeds within the range of go slow trials. Despite similar velocities, APAs were earlier in the former and less anticipated in the latter. Thus, APAs looked much more tailored on the expected perturbation than on the real one, favoring the idea of a functionally unique postural and focal command.