From Pixels to Preferences: Visual Salience Shapes Economic Choices Beyond Gaze Alone

Abstract Visually salient information often sways economic choice, an effect typically attributed to gaze bias. By combining eye-tracking with computer vision, we show that visual salience alters risky choice largely independent of gaze. People were more likely to accept a 50/50 monetary gamble when its potential gain, rather than its potential loss, was made more visually salient. Salience induced a pronounced gaze bias toward the salient attribute. Yet multiple complementary analyses converged to show that gaze accounted for less than 20% of the salience effect on choice, with most of the effect operating independently of gaze. To formalize these pathways, we introduce an image-computable attentional Visual Accumulator Model (aVAM): a convolutional neural network converts stimulus images into value representations that feed a gaze-modulated Linear Ballistic Accumulator. This dual-pathway framework reproduces the salience effect on gain-loss trade-offs and highlights the often-overlooked gaze-independent influence of visual salience on economic choice.