Algorithm 1 describes the MCMC sampling for posterior inference.

The human ability to navigate through familiar environments, such as one's residence, even under low light conditions, is underpinned by a sophisticated cognitive mechanism referred to as spatial navigation Chen et al. [2023a], Wilkins [2011]. Humans use spatial navigation as a complex cognitive process that is important in finding their way around the environment by utilizing different senses and areas of the brain McNamara and Chen [2022], Chen et al. [2023b], Garg et al. [2024], Verghese and Blumen [2022]. It involves cues such as landmarks and information on self-motion to determine positions and achieve goals Roth et al. [2020]. A thorough understanding of spatial navigation is essential for improving destination efficiency and reducing anxiety in unfamiliar settings. Assessing spatial navigation is crucial for evaluating cognitive health, especially in neurological and neurodegenerative diseases Roth et al. [2020]. Spatial navigation tasks can detect structural changes in subcortical brain areas related to cognitive decline risk Chen et al. [2023b]. Different neurodegenerative conditions see impaired spatial navigation as a symptom at the onset; thus, it can be a valuable predictor of dementia in subjective cognitive decline patients or those with mild cognitive impairment Tangen et al. [2022]. These deficits worsen with aging, highlighting the urgent need for efficient assessment tools such as the Virtual Environments Navigation Assessment (VIENNA), which evaluates spatial navigation abilities Rekers and Finke [2024a]. This research is critical for detecting cognitive impairments and guiding clinical decisions.

In this example we are going to look at a classification analysis of eye movement patterns. Although complex preprocessing steps can be performed to extract higher-order features from the raw coordinate timeseries provided by an eye-tracker, we are keeping it simple. It contains coordinate timeseries of 144 trials (recorded with 350 Hz), where subjects either looked at upright or inverted images of human faces. Each timeseries snippet covers 3 seconds. This data has been pre-processed to remove eyeblink artefacts.

Recent studies in the field of human vision science suggest that the human responses to the stimuli on a visual display are non-deterministic. People may attend to different locations on the same visual input at the same time. Based on this knowledge, we propose a new stochastic model of visual attention by introducing a dynamic Bayesian network to predict the likelihood of where humans typically focus on a video scene. The proposed model is composed of a dynamic Bayesian network with 4 layers. Our model provides a framework that simulates and combines the visual saliency response and the cognitive state of a person to estimate the most probable attended regions. Sample-based inference with Markov chain Monte-Carlo based particle filter and stream processing with multi-core processors enable us to estimate human visual attention in near real time. Experimental results have demonstrated that our model performs significantly better in predicting human visual attention compared to the previous deterministic models.