Humans use a local code for tactile perception

Humans are classically thought to use either spectral decomposition or averaging to identify vibrotactile signals. These are general purpose ‘global’ codes that require integration of the signal over long stretches of time. Natural vibrotactile signals, however, likely contain short signature events that can be detected and used for inference of textures, instantaneously, with minimal integration, suggesting a hitherto ignored ‘local code’. Here, by employing pulsatile stimuli and a change detection psychophysical task, we studied whether humans make use of local cues. We compared three local cues based on instantaneous skin position and its derivatives, as well as six global cues, calculated as summed powers (with exponents 1,2, and 3) of velocity and acceleration. Deliberate manipulation of pulse width and amplitude (local+global) as well as pulse frequency (global) allowed us to disentangle local from global codes. The results singled out maximum velocity, an instantaneous code, as a likely and dominant coding variable that humans rely on to perform the task. Comparing stimuli containing versus lacking local cues, demonstrated that performances exclusively using global cues are rather poor compared to situations where local ones are available as well. Our results are in line with the notion that humans not only do use local cues but that local cues may even play a dominant role in perception. Our results parallel previous results in rodents, pointing to the possibility that quite similar coding strategies evolved in whisker and finger tactile systems. Significance statement The brain is believed to select coding symbols in sensory signals that would most efficiently convey functionally relevant information about the world. For instance, the visual system is widely believed to use spatially local features, like edge orientation, to delineate a visual scene. For the tactile system only global, general purpose coding schemes have been discussed so far. Based on the insight that moving contacts, characteristic for active touch, feature short-lived stick-slip events, frictional movements that transfer fair amounts of texture information, one should expect the brain to use a temporally local code, extracting and instantaneously analyzing short snippets of skin movement. Here, we provide the first analytical psychophysical evidence in humans that this indeed is the case.