Abstract
Finger proprioception is essential for haptic perception, particularly in flat versus curved surface recognition, but is often overlooked in designing and interacting with haptic proxies. When multiple fingers simultaneously contact an object, perception emerges as an integrated sensory experience rather than independent processing of each finger's input. It is interesting to know whether such multi-finger integration on curved shape perception is predicable. We measured how 2-, 3-, and 4-finger combinations perceive curved surfaces and modeled the integration of proprioceptive signals. Results show that integration is not a simple sum of subsets; finger pairings closer to the index finger contribute more due to lower sensory uncertainty. Nonlinear Scaled Maximum-Likelihood Estimation (The NL-ScaledMLE Model) and Linear models best describe this process. In a follow-up study, we used models to render curved surfaces, considering finger spacing, and validated perception accuracy with different methods. Findings indicate the models effectively guide haptic rendering of curved surfaces and have potential for optimizing haptic devices with limited finger-lift actuation. This work advances the understanding of multi-finger proprioceptive integration and provides proprioceptive rendering strategies through empirical data and mathematical modeling.

This publication has 65 references indexed in Scilit: