1We rarely experience difficulty picking up objects, yet of all potential grasp points on an object's 2 surface, only a small proportion yield stable, comfortable grasps. Here, we present extensive 3 behavioral data alongside a computational model that correctly predicts human precision 4 grasping of unfamiliar 3D objects. We tracked participants' forefinger and thumb as they picked 5 up objects of 10 wood and brass cubes configured to tease apart effects of shape, weight, 6 orientation, and mass distribution. Grasps were highly systematic and consistent across 7 repetitions and participants. The model combines five cost functions related to force closure, 8 torque, natural grasp axis, grasp aperture, and visibility. Even without free parameters, we find 9 that the model predicts human grasps with striking fidelity: indeed, it predicts individual grasps 10 almost as well as different individuals predict one another's. Adding fittable weights to the model 11 reveals the relative importance of the different constraints: the combination of force closure, 12 hand posture, and grasp size explains most of human grasping behavior, while our participants 13 cared surprisingly little about minimizing torque and optimizing object visibility. Together, these 14 findings provide a unified account of how we derive effective grasps from objects' 3D shape and 15 material properties to interact with them successfully. 16 17
Significance Statement 18Working out how we pick up and interact with objects effectively is one of the most important 19 challenges in behavioral science. Of all the potential contact points on an object's surface, only 20 a small proportion yield effective grasps. Despite this, we rarely experience any difficulty 21 choosing where and how to pick objects up. Here, we present a computational model that 22unifies the varied and fragmented literature on human grasp selection. We find that the model 23 correctly predicts human grasps across a wide variety of conditions, taking into account the 24 object's 3D shape, material properties and orientation. 25
42Despite the extensive literature describing human grasping patterns and movement 43 kinematics(2-11), little is understood about the computational basis of human grasp selection. 44Few authors have attempted to study and model how humans select grasps (e.g. (12, 13)), and 45 even then, only for 2D shapes. This is because, even for two-digit precision grip, many factors 46 influence how we grasp objects. Object shape must be considered, since the surface normals at 47