Speed constant material perception via touch relies on natural material statistics.
Authors: Metzger, A., Toscani, M.
Journal: Perception
Publication Date: 30/03/2026
Pages: 3010066261429271
eISSN: 1468-4233
DOI: 10.1177/03010066261429271
Abstract:High-frequency vibrations from manual exploration of natural surfaces are crucial for differentiating materials. These signals depend not only on the spatial structure of materials' surfaces but also on exploration speed. How we achieve speed constant material perception, even in passive perception in which the texture is moved across the finger, is an open question. Here, we systematically varied exploration speed and recorded vibratory signals from human explorations of 74 material samples. We report that natural materials' power spectra can be described by the 1/fs function, with the exponent s differentiating between materials. Crucially, s is speed constant and can explain human correct and mistaken material classifications. Furthermore, s correlates with the highest layer of a speed constant neural network trained to classify natural materials and can be computed from the ratio of tactile afferents' (RA to PC) activation. We propose that s is a biologically plausible solution of speed constancy.
Source: PubMed
Speed constant material perception via touch relies on natural material statistics.
Authors: Metzger, A., Toscani, M.
Journal: Perception
Publication Date: 03/2026
Pages: 3010066261429271
eISSN: 1468-4233
ISSN: 0301-0066
DOI: 10.1177/03010066261429271
Abstract:High-frequency vibrations from manual exploration of natural surfaces are crucial for differentiating materials. These signals depend not only on the spatial structure of materials' surfaces but also on exploration speed. How we achieve speed constant material perception, even in passive perception in which the texture is moved across the finger, is an open question. Here, we systematically varied exploration speed and recorded vibratory signals from human explorations of 74 material samples. We report that natural materials' power spectra can be described by the 1/fs function, with the exponent s differentiating between materials. Crucially, s is speed constant and can explain human correct and mistaken material classifications. Furthermore, s correlates with the highest layer of a speed constant neural network trained to classify natural materials and can be computed from the ratio of tactile afferents' (RA to PC) activation. We propose that s is a biologically plausible solution of speed constancy.
Source: Europe PubMed Central