Talk: Grip-specific dynamics are not shared between action and observation in the frontoparietal co… скачать в хорошем качестве

Talk: Grip-specific dynamics are not shared between action and observation in the frontoparietal co…

Speaker: James Goodman, German Primate Center (grid.418215.b) Title: Grip-specific dynamics are not shared between action and observation in the frontoparietal cortical grasping network Emcee: Keerthi Krishnan Backend host: Richard Lange Details: https://neuromatch.io/abstract?submis... Twitter:   / @jimmy_m_goode   Presented during Neuromatch Conference 3.0, Oct 26-30, 2020. Summary: Neurons in the frontoparietal cortical grasping network respond during both the execution and observation of action. This observation has been used to support numerous theories linking motor control and cognition but remains poorly understood at the population level. To address this, we use chronic multi-electrode arrays to record simultaneously from neuronal populations in the macaque primary motor (M1), premotor (F5), and posterior parietal (AIP) cortices as animals either grasp a variety of different objects or observe a human subject doing the same. We find a surprising lack of functional neuronal classes in any of these cortical areas: Instead, neurons in each area span a unimodal spectrum of sensitivities to observed action, with some neurons in F5 and AIP even preferentially active during observation. The lack of clear neuron classes suggests the link between action and observation may indeed be better understood in the framework of neural manifolds. Using demixed principal component analysis (dPCA), we find evidence for dimensions of neural activity shared between action and observation contexts, which decrease in prominence from AIP to F5 to M1. However, this activity does not include a shared representation of grip type, suggesting a lack of common representation of the specific details of the observed action. Linear classifiers confirm the lack of congruent grip representation across all dimensions of neural activity. Finally, we show that dynamical system models of the data fail to offer a unified explanation of the disparate representations of grip types across action and observation contexts. Activity during observation therefore lacks the precise grip-specific representations and dynamics seen during movement. This may reflect a non-specific mental rehearsal of movement that specifically avoids grip-potent neural dimensions to avoid making unwanted movements.