Eye movements during decision making show systematic patterns such as increased fixations to the chosen option (i.e.gaze cascades) and multiple gaze transitions between fixated options. Existing formalisms, such as multivariate decisionfield theory, only provide limited scope for describing multiple reversals in the attentional focus and it is therefore unclearhow they can be applied to the underlying attentional dynamics. Here, we present an open systems dynamical model fromquantum theory to describe gaze transitions between choice options and the gaze cascade effect. Our model was tested ona decision task, in which participants repeatedly decided among two complex options (i.e. that lacked easily quantifiable,matched characteristics). The model can describe the gaze patterns on the individual trial level. It reveals structure inthe gaze dynamics that is predictive for choice behavior. The explanatory value of this account for studying attentionaldynamics during decision making will be discussed.

People judge an object's criterion value by relying on its similarity to previously experienced objects, the so-called exemplars. This work investigates exemplar-based processes in quantitative judgments by applying cognitive modeling to data from an eye-tracking experiment. Participants (N = 49) first learned the criterion value and location on the screen of each of four exemplars. Then, they assessed the criterion value of briefly presented test stimuli, and eye-tracking measured the gaze proportion to the now blank exemplar locations (looking-at-nothing). Participants who showed more looking-at-nothing also relied more on exemplars according to cognitive modeling of the test phase responses in the RulEx-J framework. Furthermore, looking-at-nothing was directed in particular at locations of exemplars similar to the test stimulus. Our multi-method approach thus suggests tight links between eye-tracking and cognitive modeling. The insights from process-tracing methods might be particularly valuable, when cognitive modeling cannot distinguish between alternative processes to perform quantitative judgments.