Understanding the relationship between eye movements and salience arising from one or two visual dimensions

Poster Presentation 53.404: Tuesday, May 19, 2026, 8:30 am – 12:30 pm, Pavilion
Session: Attention: Capture 1

Alejandro Lleras1 (), Zoe J. Xu2, Jun-Ming Yu1, Howard J. H. Tan1, Simona Buetti1; 1University of Illinois, 2University of Washington

The impact of salience on eye movements has been studied mostly via manipulations of luminance, a relatively easy to manipulate feature dimension. Humans as well as monkeys show an inverse relation where the stronger the luminance is, the smaller the saccadic initiation time. This finding has been modeled in terms of the time it takes for signals to reach the superior colliculus. Here, we explore the relationship between salience and the eye movement system along three different visual dimensions: color, shape, and orientation. We measured the time taken to initiate and land on a unique element. The unique element was randomly presented in a circle around fixation, embedded in a 12 x 12 grid of identical objects. In Experiment 1, the unique element was defined by a difference along a single dimension, with four different levels of salience, and all three feature dimensions, randomly varied along trials. We replicated the finding that saccadic initiation times are inversely related to salience in all three dimensions. In Experiments 2A-C, the unique element was defined along two dimensions (color-orientation, color-shape, and shape-orientation). Saccadic initiation times were not associated with the precision of the saccades in neither unidimensional (R2=0.24) or bidimensional conditions (R2=0.31). However, saccadic initiation times were strongly and linearly predicted by the strength of the guidance signal in both unidimensional (R2=0.93) and bidimensional conditions (R2=0.85), with the slope of the bidimensional condition being twice that of the unidimensional condition. Importantly, in Experiment 1, all data points fell along the same function, suggesting all visual dimensions produced signals that have equivalent impacts on the eye movement system. This was also observed in Experiment 2: the bidimensional data fell along the same function, irrespective of which two dimensions defined the unique element. Surprisingly, saccadic initiation times were faster in unidimensional conditions.