Answers are grounded in this paper and cite their sources.
Standard models of reading propose that reading is driven by lexical efficiency and word‐level variables (length, frequency, predictability), treating it as a purely computational act while overlooking the physiological “feeling” of words. This study demonstrates the influence of a word's affective properties (valence, arousal, dominance) on eye‐movements during naturalistic reading alongside standard lexical constraints. Using Linear Mixed‐Effects modeling on Affective Provo Corpus, a triangulated dataset constructed by integrating the Provo Corpus, SUBTLEX‐US, and Warriner et al.'s affective norms, we reveal a distinct temporal dissociation in affective processing. Early processing (First Fixation/Gaze Duration) is influenced by arousal, acting as an automatic somatic “brake” independent of valence. Intermediate processing (Regression Path) is influenced by a valence × arousal interaction, triggering regressions to verify potent information. Late processing (Total Dwell Time) is regulated by a dominance × arousal interaction, where feelings of agency buffer arousal's cognitive cost. Based on these findings, we propose the Alert‐Check‐Regulate (ACR) Model to map affective reading mechanics, and the GUPTA Theory (Graded Unfolding and Physiological Theory of Affect) to explain its homeostatic function. The oculomotor system acts as a homeostatic regulator to negotiate between information intake and evaluation‐confirmation for physiological survival preparation, prioritizing somatic alerting and information confirmation over processing speed.
No takes yet. Share an insight, caveat, or question.
Gupta et al. (2026) studied this question.
Synapse has enriched 10 related papers. Consider them for comparative context: