The kinematic signature of voluntary actions

Individual differences in cooperative and competitive play strategies

INTRODUCTION

Cooperation and competition are common in social interactions. It is not clear how individual differences in personality may predict performance strategies under these two contexts. We evaluated whether instructions to play cooperatively and competitively would differentially affect dyads playing a Pong video game. We hypothesized that instructions to play cooperatively would result in lower overall points scored and differences in paddle control kinematics relative to when participants were instructed to play competitively. We also predicted that higher scores in prosociality and Sportspersonship would be related to better performance during cooperative than competitive conditions.

METHODS

Pairs of participants played a Pong video game under cooperative and competitive instructions. During competitive trials, participants were instructed to score more points against one another to win the game. During the cooperative trials, participants were instructed to work together to score as few points against one another as possible. After game play, each participant completed surveys so we could measure their trait prosociality and Sportspersonship.

RESULTS

Condition was a significant predictor of where along the paddle participants hit the ball, which controlled ball exit angles. Specifically, during cooperation participants concentrated ball contacts on the paddle towards the center to produce more consistent rebound angles. We found a significant correlation of Sex and the average points scored by participants during cooperative games, competitive games, and across all trials. Sex was also significantly correlated with paddle kinematics during cooperative games. The overall scores on the prosociality and Sportspersonship surveys were not significantly correlated with the performance outcomes in cooperative and competitive games. The dimension of prosociality assessing empathic concern was significantly correlated with performance outcomes during cooperative video game play.

DISCUSSION

No Sportspersonship survey score was able to predict cooperative or competitive game performance, suggesting that Sportspersonship personality assessments are not reliable predictors of cooperative or competitive behaviors translated to a virtual game setting. Survey items and dimensions probing broader empathic concern may be more effective predictors of cooperative and competitive performance during interactive video game play. Further testing is encouraged to assess the efficacy of prosocial personality traits as predictors of cooperative and competitive video game behavior.

Predicting and understanding human action decisions during skillful joint-action using supervised machine learning and explainable-AI

This study investigated the utility of supervised machine learning (SML) and explainable artificial intelligence (AI) techniques for modeling and understanding human decision-making during multiagent task performance. Long short-term memory (LSTM) networks were trained to predict the target selection decisions of expert and novice players completing a multiagent herding task. The results revealed that the trained LSTM models could not only accurately predict the target selection decisions of expert and novice players but that these predictions could be made at timescales that preceded a player's conscious intent. Importantly, the models were also expertise specific, in that models trained to predict the target selection decisions of experts could not accurately predict the target selection decisions of novices (and vice versa). To understand what differentiated expert and novice target selection decisions, we employed the explainable-AI technique, SHapley Additive explanation (SHAP), to identify what informational features (variables) most influenced modelpredictions. The SHAP analysis revealed that experts were more reliant on information about target direction of heading and the location of coherders (i.e., other players) compared to novices. The implications and assumptions underlying the use of SML and explainable-AI techniques for investigating and understanding human decision-making are discussed.

Motor invariants in action execution and perception

The nervous system is sensitive to statistical regularities of the external world and forms internal models of these regularities to predict environmental dynamics. Given the inherently social nature of human behavior, being capable of building reliable predictive models of others' actions may be essential for successful interaction. While social prediction might seem to be a daunting task, the study of human motor control has accumulated ample evidence that our movements follow a series of kinematic invariants, which can be used by observers to reduce their uncertainty during social exchanges. Here, we provide an overview of the most salient regularities that shape biological motion, examine the role of these invariants in recognizing others' actions, and speculate that anchoring socially-relevant perceptual decisions to such kinematic invariants provides a key computational advantage for inferring conspecifics' goals and intentions.

Mirror neuron brain regions contribute to identifying actions, but not intentions

Previous studies have struggled to determine the relationship between mirror neuron brain regions and two distinct “action understanding” processes: identifying actions and identifying the intentions underlying those actions. This may be because the identification of intentions from others' actions requires an initial action identification process. Disruptive transcranial magnetic stimulation was administered to left inferior frontal gyrus (lIFG) during a novel cognitive task to determine which of these “action understanding” processes is subserved by mirror neuron brain regions. Participants identified either the actions performed by observed hand actions or the intentions underlying those actions. The extent to which intention identification was disrupted by lIFG (vs. control site) stimulation was dependent on the level of disruption to action identification. We subsequently performed functional magnetic resonance imaging during the same task. During action identification, responses were widespread within mirror neuron areas including lIFG and inferior parietal lobule. However, no independent responses were found in mirror neuron brain regions during intention identification. Instead, responses occurred in brain regions associated with two distinct mentalizing localizer tasks. This supports an account in which mirror neuron brain regions are involved in an initial action identification process, but the subsequent identification of intentions requires additional processing in mentalizing brain regions.

Identification of a Brain Network Underlying the Execution of Freely Chosen Movements

Action selection refers to the decision regarding which action to perform in order to reach a desired goal, that is, the "what" component of intention. Whether the action is freely chosen or externally instructed involves different brain networks during the selection phase, but it is assumed that the way an action is selected should not influence the subsequent execution phase of the same movement. Here, we aim to test this hypothesis by investigating whether the modality of movement selection influences the brain networks involved during the execution phase of the movement. Twenty healthy volunteers performed a delayed response task in an event-related functional magnetic resonance imaging design to compare freely chosen and instructed unimanual or bimanual movements during the execution phase. Using activation analyses, we found that the pre-supplementary motor area (preSMA) and the parietal and cerebellar areas were more activated during the execution phase of freely chosen as compared to instructed movements. Connectivity analysis showed an increase of information flow between the right posterior parietal cortex and the cerebellum for freely chosen compared to instructed movements. We suggest that the parieto-cerebellar network is particularly engaged during freely chosen movement to monitor the congruence between the intentional content of our actions and their outcome.

Kinematic-Based Classification of Social Gestures and Grasping by Humans and Machine Learning Techniques

The affective motion of humans conveys messages that other humans perceive and understand without conventional linguistic processing. This ability to classify human movement into meaningful gestures or segments plays also a critical role in creating social interaction between humans and robots. In the research presented here, grasping and social gesture recognition by humans and four machine learning techniques (k-Nearest Neighbor, Locality-Sensitive Hashing Forest, Random Forest and Support Vector Machine) is assessed by using human classification data as a reference for evaluating the classification performance of machine learning techniques for thirty hand/arm gestures. The gestures are rated according to the extent of grasping motion on one task and the extent to which the same gestures are perceived as social according to another task. The results indicate that humans clearly rate differently according to the two different tasks. The machine learning techniques provide a similar classification of the actions according to grasping kinematics and social quality. Furthermore, there is a strong association between gesture kinematics and judgments of grasping and the social quality of the hand/arm gestures. Our results support previous research on intention-from-movement understanding that demonstrates the reliance on kinematic information for perceiving the social aspects and intentions in different grasping actions as well as communicative point-light actions.

Grasping Social Apathy: The Role of Reach-To-Grasp Action Kinematics for the Assessment of Social Apathy in Mild Neurocognitive Disorders

BACKGROUND

Social apathy, a reduction in initiative in proposing or engaging in social activities or interactions, is common in mild neurocognitive disorders (MND). Current apathy assessment relies on self-reports or clinical scales, but growing attention is devoted to defining more objective, measurable and non-invasive apathy proxies.

OBJECTIVE

In the present study we investigated the interest of recording action kinematics in a social reach-to-grasp task for the assessment of social apathy.

METHODS

Thirty participants took part in the study: 11 healthy controls (HC; 6 females, mean age = 68.3±10.5 years) and 19 subjects with MND (13 females, mean age = 75.7±6.3 years). Based on the Diagnostic Criteria for Apathy, MND subjects were classified as socially apathetic (A-MND, N = 9) versus non-apathetic (NA-MND, N = 10). SensRing, a ring-shaped wearable sensor, was placed on their index finger, and subjects were asked to reach and grasp a can to place it into a cup (individual condition) and pass it to a partner (social condition).

RESULTS

In the reach-to-grasp phase of the action, HC and NA-MND showed different acceleration and velocity profiles in the social versus individual condition. No differences were found for A-MND.

CONCLUSION

Previous studies showed the interest of recording patients' level of weekly motor activity for apathy assessment. Here we showed that a 10-min reach-to-grasp task may provide information to differentiate socially apathetic and non-apathetic subjects with MND, thus providing a tool easily usable in the clinical practice. Future studies with a bigger sample are needed to better characterize these findings.

Test-retest repeatability reveals a temporal kinematic signature for an upper limb precision grasping task in adults

Hand-eye coordination skills, such as reaching and grasping, are fundamentally important for the performance of most daily activities. Upper limb kinematics recorded by motion tracking systems provide detailed insight into the central nervous system control of movement planning and execution. For example, kinematic metrics can reveal deficits in control, and compensatory neuromotor strategies in individuals with neuropathologies. However, the clinical utility of kinematic metrics is currently limited because their psychometric properties, such as test-retest repeatability, have not been well characterized. Therefore, the purpose of this study was to examine the degree of repeatability of spatiotemporal kinematic metrics and determine which, if any, measures form a kinematic signature for a precision grasping task. Healthy adults (n = 40) were tested on two occasions separated by 5-10 days on a bead threading task consisting of reaching and precision grasping. Results showed good test-retest repeatability for reach peak velocity, reach and grasp durations, whereas poor to moderate reliability was observed for measures of spatial precision and maximum grip aperture. In addition, analysis showed that reliable estimates of kinematic metrics can be obtained using 10 trials. Overall, our results indicate that reach peak velocity and temporal metrics form a stable characteristic, or a kinematic signature, of individual performance on a standardized bead threading task. These findings suggest potential utility in applying kinematic metrics for clinical assessment of upper limb reaching tasks.

Highs and Lows in Motor Control Development

Motor control is classically described as relying on two components: anticipatory control (feedforward processing) and online control (feedback processing). Here we aimed to unveil the developmental steps of both feedback and feedforward control in 5-10 years old children, using a simple and ecological task. We manipulated object's weight in a reach-to-displace paradigm. When the weight was known before lifting it, anticipatory processes were quantifiable during the reaching phase. Conversely, an unknown weight triggered online corrections during the displacing phase. Movement kinematics revealed that children anticipate this objet property as young as 5 y-o. This anticipation becomes adequate around 7 y-o and is paralleled by poor online corrections. This simple yet relevant paradigm should allow quantifying deviations from neurotypical patterns in disorders of motor control.

Thirst for Intention? Grasping a Glass Is a Thirst-Controlled Action

Every day and every hour, we feel we perform numerous voluntary actions, i.e., actions under the control of our will. Individual’s ability to initiate goal-directed movement is classically described as a hierarchical motor organization, from an intentional module, mostly considered as a black box, to muscular activity supporting action execution. The general focus is usually set on the triggering of action by intention, which is assumed to be the only entry to the action cascade, rather than on the preceding formation of intentions. If intentions play a key role in the specification of movement kinematic parameters, it remains largely unknown whether unconscious cognitive processes may also affect action preparation and unfolding. Recently, a seemingly irrelevant variable, thirst, was shown to modulate a simple arbitrary action such as key-pressing. Thirsty individuals were shown to produce stronger motor inhibition in no-go trials when a glass of water was present. In the present experiment, we intended to explore whether motor inhibition operates not only upstream from the action cascade but may also affect the unfolding of reaching movements, i.e., at a lower-level control. Thirsty vs. non-thirsty control subjects were asked to reach and grasp green (go trial) or red glasses (no-go trial) filled with either water or transparent gel wax with a central candlewick. Thirsty subjects were faster to initiate actions toward the water glasses. They also exhibited an earlier maximal grip aperture and a global reduction of movement time which was mostly explained by a shortening of deceleration time. The deceleration phase was correlated with individual’s thirst rating. In addition, no-go trial toward a glass of water tended to inhibit the next movement toward a glass filled with gel wax. Thus, our results show that an unintentional influence of an internal state can reorganize voluntary action structure not only at the decision-making level but also at the level of motor control. Although subjects explicitly paid more attention and were more cautious to glasses filled with water, they reported no explicit sensation of an increased urge to grasp it, further suggesting that these effects are controlled by covert mechanisms.

Disarming the gunslinger effect: Reaction beats intention for cooperative actions

According to the famous physicist Niels Bohr, gunfights at high noon in Western movies not only captivate the cinema audience but also provide an accurate illustration of a psychophysical law. He suggested that willed actions come with slower movement execution than reactions, and therefore that a film's hero is able to get the upper hand even though the villain normally draws first. A corresponding "gunslinger effect" has been substantiated by empirical studies. Because these studies used a markedly competitive setting, however, it is currently unclear whether the gunslinger effect indeed reflects structural differences between willed actions and reactive movements, or whether it is a by-product of the competitive setting. To obtain bullet-proof evidence for a true reactive advantage, we investigated willed and reactive movements during a cooperative interaction of two participants. A pronounced reactive advantage emerged, indicating that two independent systems indeed control willed and reactive movements.

Freely chosen and instructed actions are terminated by different neural mechanisms revealed by kinematics-informed EEG

Neurophysiological accounts of human volition are dominated by debates on the origin of voluntary choices but the neural consequences that follow such choices remain poorly understood. For instance, could one predict whether or not an action was chosen voluntarily based only on how that action is motorically executed? We investigated this possibility by integrating scalp electroencephalograms and index-finger accelerometer recordings acquired while people chose between pressing a left or right button either freely or as instructed by a visual cue. Even though freely selected and instructed actions were executed with equal vigor, the timing of the movement to release the button was comparatively delayed for freely selected actions. This chronometric difference was six-times larger for the β-oscillations over the sensorimotor cortex that characteristically accompany an action's termination. This surprising modulation of an action's termination by volition was traceable to volition-modulated differences in how the competing yet non-selected action was represented and regulated.

Effects of intentional movement preparation on response times to symbolic and imitative cues

Speeded responses to an external cue are slower when the cue interrupts preparation to perform the same or a similar action in a self-paced manner. To explore the mechanism underlying this 'cost of intention', we examined whether the size of the cost is influenced by the nature of the external cue. Specifically, we assessed whether the cost of intention is different for movements made in response to an imitative cue (an on-screen hand movement) compared to those made in response to a symbolic cue. Consistent with previous reports, externally cued responses were significantly slower on trials where participants were preparing to perform an internally driven movement later in the trial. Also as predicted, simple response times to the imitative cue were faster than those made to the symbolic cue. Critically, the cost of intention was similar for each cue type, suggesting that preparing an intentional action influenced responses cued by the symbolic and imitative cues to a similar degree. These findings suggest that the nature of the external cue does not influence the response time delay associated with concurrent intentional preparation. Together with previous findings, the results of the current study shed further light on the potential mechanisms underlying the cost of intention.

Automatically Characterizing Sensory-Motor Patterns Underlying Reach-to-Grasp Movements on a Physical Depth Inversion Illusion

Recently, movement variability has been of great interest to motor control physiologists as it constitutes a physical, quantifiable form of sensory feedback to aid in planning, updating, and executing complex actions. In marked contrast, the psychological and psychiatric arenas mainly rely on verbal descriptions and interpretations of behavior via observation. Consequently, a large gap exists between the body's manifestations of mental states and their descriptions, creating a disembodied approach in the psychological and neural sciences: contributions of the peripheral nervous system to central control, executive functions, and decision-making processes are poorly understood. How do we shift from a psychological, theorizing approach to characterize complex behaviors more objectively? We introduce a novel, objective, statistical framework, and visuomotor control paradigm to help characterize the stochastic signatures of minute fluctuations in overt movements during a visuomotor task. We also quantify a new class of covert movements that spontaneously occur without instruction. These are largely beneath awareness, but inevitably present in all behaviors. The inclusion of these motions in our analyses introduces a new paradigm in sensory-motor integration. As it turns out, these movements, often overlooked as motor noise, contain valuable information that contributes to the emergence of different kinesthetic percepts. We apply these new methods to help better understand perception-action loops. To investigate how perceptual inputs affect reach behavior, we use a depth inversion illusion (DII): the same physical stimulus produces two distinct depth percepts that are nearly orthogonal, enabling a robust comparison of competing percepts. We find that the moment-by-moment empirically estimated motor output variability can inform us of the participants' perceptual states, detecting physiologically relevant signals from the peripheral nervous system that reveal internal mental states evoked by the bi-stable illusion. Our work proposes a new statistical platform to objectively separate changes in visual perception by quantifying the unfolding of movement, emphasizing the importance of including in the motion analyses all overt and covert aspects of motor behavior.