Work

Reciprocal haptic cross-entropy links Gibsonian specification and Shannon uncertainty in human gait

Year: 2026
Type: article
Abstract: Ecological psychology treats information as a relation that specifies an environmental source and its affordances, whereas Shannon theory quantifies uncertainty among alternatives. We propose a limited operational bridge between these views for haptic locomotion. Knee posture is represented by a finite helical axis reconstructed from consecutive thigh and shank marker configurations, and touch is represented by the ground reaction force line. Their canonical Klein product defines a signed reciprocal separation for every posture–touch pairing. Because the initial position of the knee axis family in the laboratory frame was unknown, one unconstrained three-dimensional translation was estimated from all synchronized pairs and interpreted as a coordinate boundary condition rather than an anatomical displacement. The verified analysis used 120 pairs (first knee transition index 29), yielding a translation of [33.585, −1296.928, −112.445] mm and a median post-registration line separation of 5.555 mm. Gaussian compatibility scores over the complete pairwise separation matrix produced a posture-to-touch cross-entropy of 2.993 nats. We correct an important mathematical point: with one-hot synchronized targets, this cross-entropy equals a Kullback–Leibler divergence and is distinct from the entropy of the predicted conditional distribution. Cross-entropy measures relative temporal specificity, whereas a separate Gaussian residual term measures absolute reciprocity. A circular shift audit that refitted translation and scale for every shift gave 4.623 ± 0.379 nats for the 119 nonzero shifts ( p = 0.0083). These single-trial, in-sample results support a testable framework in which reciprocal geometry supplies ecological compatibility and information theory quantifies its specificity and breakdown.
Language: English
Open Access status: gold
Estimated APC paid: $3,801
OpenAlex ID: W7234089017
FWCI:
Cites: 9
Cited by:
Related to: 10