3D Hand-Motion Tracking and Bottom-Up Classification Sheds Light on the Physical Properties of Gentle Stroking.

Despite much research on the perception of gentle stroking, its motion characteristics and modulation by social intent remain largely unknown. Here we addressed this situation by asking volunteers to gently stroke a dog, the arm of their romantic partner, their own arm, or a foam arm in a pleasant manner, as if to provide comfort, or as fast/slow as possible. Stroking motion was tracked with a sensor attached to the back of the hand and processed using MPAL, a new 3D motion analysis tool. Statistical testing was both hypothesis-driven and exploratory. Hypothesis-driven tests revealed comparable stroking velocities for social (dog, partner) and non-social touch targets, but an overall slower velocity for pleasant and slow as compared with fast stroking. Additionally, stroking a social target or with a pleasant intent entailed less motion along the target's front/back axis, increased motion along the left/right axis and increased temporal variability between main strokes. An exploratory linear discriminant analysis on 26 motion features revealed that stroking a social target was more distinct than stroking in a pleasant manner and that the former, and to a lesser extent the latter, were strongly associated with features indexing spatio-temporal variability. Thus, touchers socially tune their stroking motion by reducing its predictability, which may make the touchee's experiences more pleasurable by facilitating the differentiation between self- and other touch. Together, our results offer useful directions for future research on gentle stroking and emphasize the need to consider the natural physical properties of touch.