Towering models of brain functioning propose that the brain is a predictive machine aiming at updating internal models of the environmental causes of sensory inflow in order to minimize perceptuo-motor errors (Friston, 2010). Predictive models are fundamental when dealing with a dynamic environment such as when interacting with other individuals by predicting their behavior.
On one side, fronto-parietal cortical and cerebellar systems are thought to support predictions about the consequences of one's own as well as others' movements (Kilner et al., 2007; Sokolov et al., 2017). Crucially, given the cerebellar function in monitoring and predicting sensorimotor events (Leggio et al., 2011; Peterburs & Desmond, 2016), the role of cortico-cerebellar connections in supporting interactive behaviors is of great interest but still largely unknown (Van Overwalle et al., 2014).
On the other side, interacting with others also depends on the reactivity of the autonomic system which modulates arousal and readiness to control one's behavior. Among the available measures of autonomic functioning, heart rate variability (HRV) indexes the degree to which the cardiac activity can be modulated to meet changing situational and emotional demands (Thayer & Lane, 2009). Since neuroimaging studies point to an HRV-related cortico-cerebellar network (Kumral et al., 2019), we plan to study for the first time in healthy individuals and cerebellar patients whether HRV mediates the reactivity to prediction errors during interpersonal interactions.
The current project aims at advancing the understanding of cortico-cerebellar networks supporting interpersonal interactions by integrating motion kinematics, autonomic monitoring, functional brain imaging and brain lesion approaches in healthy individuals and cerebellar patients engaged in realistic interpersonal interactions in virtual reality.
The study has the ambitious objective to go substantially beyond the current state of the art in the following innovative ways: 1) being the first using motion kinematics to study motor functions of cerebellar patients engaged in interactive scenarios; 2) moving the study of cortico-cerebellar anatomo-functional connectivity from passive observation and individual action execution to realistic interpersonal interactions; 3) combining autonomic nervous system monitoring, motion kinematics, virtual reality and advanced neuroimaging techniques and analyses.
Studying patients with cerebellar impairments will allow to verify: 1) whether cerebellar patients' motor deficits are increased or reduced when dealing with a realistic interpersonal interaction compared to individual action performance, 2) whether cortico-cerebellar anatomo-functional disconnections affect individual's ability to predict and adjust to a partner behavior in interactive conditions, 3) whether HRV is mediated by cortico-cerebellar anatomo-functional connectivity and how does it impact the ability to interact with a partner in case of prediction violations.
This will have an important outcome for cerebellar patients. Indeed, our published data on apraxics (Candidi et al., 2017) and preliminary data on Parkinson patients (Era et al., in prep) support the unexpected notion that predictive and dynamic interactions with a partner may be beneficial for motor performance possibly because these conditions activate a larger cortico-subcortical and cerebellar network compared to those where individuals act alone. This perspective may suggest that interactive situations are relevant to facilitate rehabilitation of motor functions in a wide variety of pathologies implying motor impairments.
In this context, the project is entirely innovative because it aims at establishing whether cortico-cerebellar networks are essential for these benefits. A recent rehabilitation proposal (Butti et al., 2020) is coherent with the idea that an observed action prediction task (i.e. social prediction training) implemented in a virtual environment may facilitate children and adolescents affected by congenital cerebellar malformation in the recovery of higher-order abilities, such as cognitive processing of social stimuli.
Two more aspects need to be underlined.
The hypothesis that the HRV is causally associated to individuals' behavior to cope with interpersonal motor interactions by making individuals' reactivity to prediction errors better is entirely new and timely as indicated by recent reviews and perspective articles (Owens et al., 2018). Similarly, the idea that HRV is mediated by cortico-cerebellar networks will benefit from clarifying the dynamics of cortico-cerebellar responses associated to HRV changes during interpersonal interaction.
Although functional neuroimaging has been extensively used to describe the cortical regions involved in reach-to-grasp actions, only few studies have explored grasping in an interpersonal context by comparing performed actions similar or dissimilar to the observed ones (Newman-Norlund et al., 2007; Ocampo et al., 2011). However, to our knowledge no study has examined the effect of (violations of) predictions of others' actions in an interpersonal context. Furthermore, the study of the cortico-cerebellar dynamic functional connectivity underpinnings of interpersonal interactions is innovative and offers the possibility to better qualify the sensorimotor predictions and monitoring function performed by the cerebellum for the ability to efficiently behave in social interactions.
The population characteristics set possible limits in drawing conclusions about the cognitive components involved in the tasks. Over and above cognitive processing, cerebellar patients show a sensorimotor deficit that affects their general performance. Comparing RT measures between patients and non-neurological subjects has been a challenge for neuropsychology. The mean performance of the slow group on a variety of tasks regressed on that of matched controls shows slopes greater than 1 indicating that as task difficulty increases patients are increasingly impaired. Models have been proposed to partial out the specific effects due to the experimental manipulations from a general slowing factor (RAM; Faust et al., 1999). Analysis of the relation between the conditions mean and the SDs enable to calculate the motor execution time in each group independently from the cognitive component involved in the task (DEM; Myerson et al., 2003). Taken together, these models have proven useful in the context of cross-linguistic studies of reading (Marinelli et al., 2014), in interpreting dyslexics children's behavior (Martelli et al. 2014) and the slowness of traumatic brain injury patients in decision making tasks (Puopolo, et al., 2013). Here, these models will provide a crucial tool for interpreting the behavioral responses of cerebellar patients.