Neural and motor basis of inter-individual interactions

The goal of my Ph.D. work was to investigate the behavioral markers and the brain activities responsible for the emergence of sensorimotor communication. Sensorimotor communication can be defined as a form of communication consisting into flexible exchanges based on bodily signals, in order to increase the efficiency of the inter-individual coordination. For instance, a soccer player carving his movements to inform another player about his intention. This form of interaction is highly dependent of the motor system and the ability to produce appropriate movements but also of the ability of the partner to decode these cues. To tackle these facets of human social interaction, we approached the complexity of the problem by splitting my research activities into two separate lines of research. First, we pursued the examination of motor-based humans’ capability to perceive and “read” other’s behaviors in focusing on single-subject experiment. The discovery of mirror neurons in monkey premotor cortex in the early nineties (di Pellegrino et al. 1992) motivated a number of human studies on this topic (Rizzolatti and Craighero 2004). The critical finding was that some ventral premotor neurons are engaged during visual presentation of actions performed by conspecifics. More importantly, those neurons were shown to encode also the actual execution of similar actions (i.e. irrespective of who the acting individual is). This phenomenon has been highly investigated in humans by using cortical and cortico-spinal measures (for review see, fMRI: Molenberghs, Cunnington, and Mattingley 2012; TMS: Naish et al. 2014; EEG: Pineda 2008). During single pulse TMS (over the primary motor cortex), the amplitude of motor evoked potentials (MEPs) provides an index of corticospinal recruitment. During action observation the modulation of this index follow the expected changes during action execution (Fadiga et al. 1995). However, dozens of studies have been published on this topic and revealed important inconsistencies. For instance, MEPs has been shown to be dependent on observed low-level motor features (e.g. kinematic features or electromyography temporal coupling; Gangitano, Mottaghy, and Pascual-Leone 2001; Borroni et al. 2005; Cavallo et al. 2012) as well as high level movement properties (e.g. action goals; Cattaneo et al. 2009; Cattaneo et al. 2013). Furthermore, MEPs modulations do not seem to be related to the observed effectors (Borroni and Baldissera 2008; Finisguerra et al. 2015; Senna, Bolognini, and Maravita 2014), suggesting their independence from low-level movement features. These contradictions call for new paradigms. Our starting hypothesis here is that the organization and function of the mirror mechanism should follow that of the motor system during action execution. Hence, we derived three action observation protocols from classical motor control theories: 1) The first study was motivated by the fact that motor redundancy in action execution do not allow the presence of a one-to-one mapping between (single) muscle activation and action goals. Based on that, we showed that the effect of action observation (observation of an actor performing a power versus a precision grasp) are variable at the single muscle level (MEPs; motor evoked potentials) but robust when evaluating the kinematic of TMS-evoked movements. Considering that movements are based on the coordination of multiple muscle activations (muscular synergies), MEPs may represent a partial picture of the real corticospinal activation. Inversely, movement kinematics is both the final functional byproduct of muscles coordination and the sole visual feedback that can be extracted from action observation (i.e. muscle recruitment is not visible). We conclude that TMS-evoked kinematics may be more reliable in representing the state of the motor system during action observation. 2) In the second study, we exploited the inter-subject variability inherent to everyday whole-body human actions, to evaluate the link between individual motor signatures (or motor styles) and other’s action perception. We showed no group-level effect but a robust correlation between the individual motor signature recorded during action execution and the subsequent modulations of corticospinal excitability during action observation. However, results were at odds with a strict version of the direct matching hypothesis that would suggest the opposite pattern. In fact, the more the actor’s movement was similar to the observer’s individual motor signature, the smaller was the MEPs amplitude, and vice versa. These results conform to the predictive coding hypothesis, suggesting that during AO, the motor system compares our own way of doing the action (individual motor signature) with the action displayed on the screen (actor’s movement). 3) In the third study, we investigated the neural mechanisms underlying the visual perception of action mistakes. According to a strict version of the direct matching hypothesis, the observer should potentially reproduce the neural activation present during the actual execution of action errors (van Schie et al. 2004). Here, instead of observing an increase of cortical inhibition, we showed an early (120 ms) decrease of intracortical inhibition (short intracortical inhibition) when a mismatch was present between the observed action (erroneous) and the observer’s expectation. As proposed by the predictive coding framework, the motor system may be involved in the generation of an error signal potentially relying on an early decrease of intracortical inhibition within the corticomotor system. The second line of research aimed at the investigation of how sensorimotor communication flows between agents engaged in a complementary action coordination task. In this regard, measures of interest where related to muscle activity and/or kinematics as the recording of TMS-related indexes would be too complicated in a joint-action scenario. 1) In the first study, we exploited the known phenomenon of Anticipatory Postural Adjustments (APAs). APAs refers to postural adjustments made in anticipation of a self- or externally-generated disturbance in order to cope for the predicted perturbation and stabilize the current posture. Here we examined how observing someone else lifting an object we hold can affect our own anticipatory postural adjustments of the arm. We showed that the visual information alone (joint action condition), in the absence of efference copy (present only when the subject is unloading by himself the object situated on his hand), were not sufficient to fully deploy the needed anticipatory muscular activations. Rather, action observation elicited a dampened APA response that is later augmented by the arrival of tactile congruent feedback. 2) In a second study, we recorded the kinematic of orchestra musicians (one conductor and two lines of violinists). A manipulation was added to perturb the normal flow of information conveyed by the visual channel. The first line of violinist where rotated 180°, and thus faced the second line. Several techniques were used to extract inter-group (Granger Causality method) and intra-group synchronization (PCA for musicians and autoregression for conductors). The analyses were directed to two kinematic features, hand and head movements, which are central for functionally different action. The hand is essential for instrumental actions, whereas head movements encode ancillary expressive actions. During the perturbation, we observed a complete reshaping of the whole patterns of communication going in the direction of a distribution of the leadership between conductor and violinists, especially for what regards head movements. In fact, in the perturbed condition, the second line acts as an informational hub connecting the first line to the conductor they no longer can see. This study evidences different forms of communications (coordination versus synchronization) flowing via different channels (ancillary versus instrumental) with different time-scales.