Customer Decision-Making Processes Revisited: Insights from an Eye Tracking and ECG Study Using a Hidden Markov Model

Good timing is key for many activities in business and society. In the context of adaptive user assistance, it can work as door opener to further engage with the user. This paper presents a virtual commerce study which combines eye tracking, electrocardiography, and virtual reality with the goal to detect decision phases in two different purchase scenarios. We therefore collect objective sensor data in combination with subjective decision phase annotations. Shifts between decision phases are determined subjectively by the participants via retrospective video analysis. For decision phase recognition, we demonstrate how to use the neurophysiological sensor data to train a Hidden Markov Model with multivariate mixed Gaussian emission distributions and how to use it for inference. A main benefit of our approach is its lightweight character regarding both training and inference.