Visual Fixation Does Not Equal Perceptual Salience: Eye-Tracking-Based Kansei Evaluation of Ming-Style Chair Form

Abstract

This study addresses a limitation of the assumption in Kansei Engineering (KE) that morphological features contribute equally to affective evaluation by developing a KE framework based on the Feature Integration Theory (FIT). Specifically, eye-tracking data based on Areas of Interest (AOI) are used to calculate the coefficient of variation (CV) of fixation duration between stimuli, which is then applied to rescale morphological features before modeling. In practice, ten Ming-style chairs were encoded with 30 morphological features, and Kansei ratings were collected from 389 participants, along with eye-tracking data from 30 participants. Accordingly, six Partial Least Squares (PLS) models constructed based on the rescaled features demonstrated robust performance (Q2 = 0.544–0.885), supported by LOOCV, bootstrap, and permutation tests, with external validation further confirming their generalizability. Notably, the results revealed a dissociation between fixation distribution and dispersion-based scaling, indicating that stable gaze concentration does not translate into higher predictive contributions in Kansei responses. This framework provides a repeatable method for incorporating visual attention variability into feature-based KE for product evaluation.