Many everyday decisions, such as choosing a new phone or finding the right hotel, require integrating information across multiple attributes. Our recently proposed theory of Multi-Attribute Search and Choice (MASC; Gluth et al., 2024) formalises this process within a hierarchical Bayesian framework, assuming that posterior beliefs about attribute values, and ultimately option values, are iteratively updated through information acquired via fixations. When one option is deemed sufficiently superior, a choice is made.

Using EEG and eye-tracking in a multi-attribute choice task (N=57), we explored whether the belief-updating process proposed by MASC is predictive of EEG activity evoked during fixations (fixation-evoked potentials, FEPs). By fitting MASC to the eye-tracking data, we derived predictions of the strength of belief updating at each fixation. Linear deconvolution modelling allowed us to assess the effect of these model-derived estimates, while controlling for overlapping activity from successive fixations and confounds such as saccade amplitude.

We demonstrate that MASC not only captures key behavioural patterns, such as the distribution of attention across attributes by importance, but also provides meaningful predictions at the neural level. Regression analyses and threshold-free cluster enhancement revealed two separable dynamics influenced by the predicted strength of belief updating: a P3-like modulation associated with attribute-level updating, followed by a later, centro-parietal positivity associated with option-level updating.

Our results provide converging behavioural and neural evidence that multi-attribute decision making involves dynamic, fixation-driven updating consistent with Bayesian inference as formalized by MASC. Furthermore, these model-predicted belief updates are reflected in FEPs with comparable timing and topology.