Our aim is to analyze the neural dynamics in primate brain, associated to complex visuomotor tasks, requiring action execution in different contexts or mere action observation. The cortical activity will be recorded from different areas of the fronto-parietal network (dorsal premotor and posterior parietal cortex) of the macaque brain, during the animal task performance. To this aims two macaques will be trained to perform identical isometric actions under different social contexts (individually or jointly with a mate), or to merely observe the motor output of the isometric action performed by a conspecific, consisting in a cursor's motion on a screen.
Neural processes at the population level, will be studied by making use mainly of a dimensionality-reduction technique (Principal Component Analysis) applied to the neuronal responses. Previous studies of neural population dynamics of cell activity from monkey motor areas during visuomotor task suggests that it mostly represents the generation and timing of motor behavior. Here will move a step forward, adding to this approach new behavioral contexts, to shed new light on dynamical features of neural processes subtending higher-order visuomotor behavior.
The novelty of this project is manifold. First, there are not other studies in the literature, besides those performed in our lab, in which the single-unit and multi-unit activity is recorded from brains of pairs of interacting monkey in tasks similar to those proposed in this project . Our multi-task approach allows unique comparisons of the neural dynamics across several cognitive-motor conditions. In particular, we intend to compare: i) similar motor tasks performed in two different contexts (e.g. individual vs joint-behavior); ii) action execution vs action observation; iii) same type of action performed in different directions (e.g. individual force application in eight directions); and iv) observing the consequences of the other partner¿s action, i.e. a visual cursor's motion, toward different locations.
Second, to address our question we will adopt a modern class of statistical methods consisting in a dimensionality reduction technique, well-suited for analysing neu¬ral population activity. By making use of this technique, we go beyond the traditional methods adopted to analyse the pattern of neural activation, usually analysed at single-unit level, in relation to the movement parameters or more generally to the task-related variables. Dimensionality reduction methods (Cunningham and Yu, 2014) provide instead low-dimensional representations starting from high-dimensional data as those associated to the activity of an elevated number of neurons. These new representations have been proved to reveal, several feature of interest in the data, otherwise hidden. These features might be those related to the neural dynamics which might be shared and preserved when the population activity is studied in different tasks with different behavioral demands.
Previous studies of neural population dynamics in motor and premotor cortex have already used dimensionality reduction (like PCA). They found that the main components are related to dynamical aspects of the task. In Yu et al. (Yu et al., 2009) and following studies (Kaufman et al., 2016; Lara et al., 2018), the interpretation of dynamical aspects of neural space is that their pattern is ¿presumably related to generating the arm movement and it is thus sensible that it is time-locked to movement onset¿. The pattern of neural dynamics has been interpreted as related to state changes and movement timing, rather than movement type. However, in the above studies, the results were obtained only from neural data collected during motor performance in a classical reaching paradigm. Therefore the proposed interpretation might have been influences by the nature of the adopted experimental paradigm, since they did not analysed the dynamics associated to behavioral conditions not requiring movement generation, as during mere action observation.
We will then introduce new behavioral contexts which offer the opportunity to study the dynamical aspects of neural space not only for identical actions under different cognitive contexts (ACT-SOLO vs ACT-JA) but above all in more abstract conditions, such as during the mere observation of a moving cursor.
Furthermore, our saccadic task used as a control task, will add new information to previous findings, allowing to explore the putative influence of oculomotor behavior which can critically impacts the neural dynamics of premotor and parietal neural activity associated to motor behavior. In all previous studies on neural dynamics applied to visuomotor tasks, the potential effect of eye movements has not been documented and apparently neglected.
No investigation is currently available on social predictive probabilistic coding during orienting of attention: the data collected in the present project will literally open a new window on this unexplored issue.