Abstract
Most behavioral and functional MRI studies of object processing have focused on object categorization and labeling. Behavioral studies have explored the similarity between objects based on their semantic labels, while neural studies have compared the behavioral similarities to the representations in the visual cortex. However, categorization is not the only purpose of visual object processing. For example, when grasping an object, some aspects of its shape is relevant for planning the proper movement. To explore the grasp-based similarity of objects, we 3D printed a set of 58 natural objects and asked participants to grasp and hold the objects. We recorded the movements of the participants’ fingers and explored the similarity between objects based on these movements. We used multi-dimensional scaling (MDS) to visualize the dimensions that explain most of the variance. The first two dimensions were jointly related to the orientation and size of the grasp-relevant part of the object. Changing the subjects’ initial position from sitting to lying down and hiding the object when the grasp was performed did not systematically change the similarity matrices. Next, in an odd-one-out behavioral experiment performed online, we showed pictures of the same objects to participants in sets of three and asked them to pick the most distinct item. From these judgements we constructed a visual/semantic similarity matrix. Here, an MDS analysis showed that the first two dimensions were jointly related to the visual orientation and semantic labels of the object. The visual/semantic similarity was not significantly correlated with the grasp-based similarity suggesting separate features influence the two behaviors. These results could inform future fMRI studies exploring the brain regions involved in processing objects for the purpose of grasping.