Neural coding of grasp force planning and control in macaque areas AIP, F5, and M1

Abstract

Much research has been done in the last decades to decipher how the brain controls grasping movements. The anterior intraparietal area (AIP), the hand area of the ventral premotor cortex (F5), and the hand area of the primary motor cortex (M1) have been identified as essential cortical areas for the control of hand shape. However, much less is known about how neurons from these areas code another essential parameter of grasping actions, grasp force. Especially, the role of the higher order areas F5 and AIP in this process remains elusive. This study aims to address the lack of knowledge about the neural coding of grasp force planning and control in these areas. To achieve this, we trained two macaque monkeys (Macaca mulatta) on a delayed grasping task with two grip types (whole-hand grip or precision grip) and three different levels of force (0-12 N). While the monkeys performed the task, we recorded the activity of singleand multi-units from AIP, F5, and M1. We calculated the percentage of grip type and force tuned units (cluster-based permutation test) and calculated the amount of variance explained by grip type and force for the population of units of each brain area (demixed principal component analysis). We show here, for the first time, the modulation of single AIP neurons to grasp force. Furthermore, we confirm and extend previous findings that showed such neural modulation in F5 and M1. Surprisingly, the percentages of units responding to grasp force control in AIP and F5 were not much less than M1 and similar to the amount of units responding to grip type. In F5, the amount of variance explained by grasp force was almost 16 as high as that explained by grip type. In AIP and M1, grip type clearly explained more variance than grasp force, but also in these areas, the amount of variance explained by force was sufficient to reliably decode the force conditions. We also found strong neural modulation to grasp force conditions before movement onset in F5, which possibly represents a role of this area for grasp force planning. In AIP, grasp force planning activity was found only in one of both monkeys, and as expected, not in M1 (checked only in one animal). Lastly, we found that, although force tuning was influenced by grip type in some units, only a small fraction of the population variance in each area contained mixed selectivity for grip type and force. Information about grasp force could therefore be extracted separately from grip type. These findings suggest an important role of AIP and F5 in grasp force control in addition to M1. F5 is likely also involved in planning grasp force, while the role of AIP and M1 are likely smaller in this process. Finally, since grip type and force information could be extracted separately, these results show that grasp force is possibly coded independently from hand shape in the cortical grasping network.