Speed constant material perception via touch relies on natural material statistics.

Area of Science:

• Neuroscience
• Robotics
• Materials Science

Background:

• High-frequency vibrations during manual exploration are key to material differentiation.
• Perception of material texture is influenced by surface properties and exploration speed.
• Achieving constant material perception across different speeds, even passively, remains an open question.

Purpose of the Study:

• To investigate the relationship between exploration speed, vibratory signals, and material perception.
• To identify a speed-invariant feature that explains accurate and inaccurate material classifications.
• To explore the neural and computational basis of speed-invariant tactile perception.

Main Methods:

• Systematic variation of exploration speed during manual exploration of 74 natural material samples.
• Recording and analysis of vibratory signals using power spectra.
• Development and training of a speed-invariant neural network for material classification.
• Correlation analysis between the derived exponent (s) and tactile afferent activation ratios (RA to PC).

Main Results:

• Natural materials' power spectra conform to a 1/f^s function, with the exponent 's' effectively differentiating materials.
• The exponent 's' was found to be speed-constant across different exploration speeds.
• 's' successfully explained both correct and mistaken human material classifications.
• The exponent 's' correlated with the output of a trained neural network and could be computed from tactile afferent ratios.

Conclusions:

• The exponent 's' is a speed-invariant feature crucial for material perception.
• This exponent offers a biologically plausible mechanism for achieving speed-constant tactile perception.
• The findings provide insights into how the human nervous system processes tactile information for material identification.