The interference of space on time and vice versa are common in daily life. The spatiotemporal interference effect is a phenomenon in which temporal perception is disturbed by spatial information or spatial perception is disturbed by temporal information. Some studies suggest that spatiotemporal interference is asymmetric, and the spatial interference on time is always greater. Other studies indicate that the strength of mutual interference between time and space is influenced by experimental factors. In general, spatial interference on time is greater, but time can also produce the same degree or even greater interference on space. Results of previous studies on the direction and strength of spatiotemporal interference effects have been controversial, and there are phenomena in which space interferes with time to a greater degree, time interferes with space to a greater degree, and time and space interfere with each other to an equal degree. A rational theory is required to explain the phenomena and mechanisms of the spatiotemporal interference effect. Some reviews have introduced related research, but they all explain the spatiotemporal interference effect from the perspective of metaphor theory and a theory of magnitude (ATOM). In recent years, researchers have applied Bayesian models to the field of spatiotemporal interference effects and have achieved rapid developments in this area, but no systematic review of these studies has been conducted. Therefore, it is necessary to summarize and discuss the research that has used Bayesian models to explain spatiotemporal interference effects.

First, this paper introduces the recent studies related to the spatiotemporal interference effect and reviews the main viewpoints of metaphor theory and ATOM to explain the spatiotemporal interference effect. Metaphor theory suggests that people’s tendency to use spatial metaphors to think about time extends into the perceptual domain, producing asymmetrical spatial interference on time, and ATOM proposes that temporal and spatial information are processed in a common magnitude system in the parietal cortex, which allows time and space to influence each other comparably but not necessarily symmetrically. Then we highlight four types of Bayesian models in the field of spatiotemporal interference effect research: the constant velocity Bayesian model, the slowness Bayesian model, the dimensional covariance Bayesian model, and the ATOM Based Bayesian model. Next, based on Bayesian models, we propose a new interpretation of the generation mechanism of spatiotemporal interference. The Bayesian model that explains the effect of temporal perception interfered by spatial information can serve as an example. The observer’s perceptual time (posterior) is an integration of the time in experience (prior, a belief that longer distances take longer to arrive) and the time of sensory input (likelihood). The more unreliable the sensory input (e.g., low salience of temporal stimuli), the more the observer will rely on the prior and be interfered by the spatial information from the prior, resulting in a spatiotemporal interference effect. The Bayesian models assume that the spatiotemporal interference effect is not always symmetrical or asymmetrical, and the direction and strength of the spatiotemporal interference effect is modulated by the relative noise of spatial and temporal information in working memory (i.e., the relative variance of spatial and temporal likelihood distributions). If they are of equal size, there will be symmetrical interference between space and time; if they are of different sizes, there will be asymmetrical interference from the dimension with less memory noise (i.e., smaller variance of the likelihood distribution) to the other dimension. Thus, the symmetry of the spatiotemporal interference effect can be affected by experimental factors such as stimulus saliency and perceptual acuity. Finally, we discuss the relationship between metaphor theory, ATOM, and Bayesian models. All three accounts propose that people expect space and time to co-vary in the same direction, which suggests that we can assimilate reasonable views of spatiotemporal relations in metaphor theory and ATOM to the prior hypotheses of Bayesian models and then construct an optimal model in the future by optimizing the prior and revealing the neural basis of the inference and decision-making processes. Specifically, three issues should be addressed in further studies: (a) expanding the scope of the Bayesian model to explain the spatiotemporal interference effect, (b) exploring the neural mechanism of spatiotemporal interference based on Bayesian inference, and (c) seeking regulation methods for spatiotemporal interference, which laid a foundation for unraveling the cognitive and neural mechanisms of spatiotemporal interference in the future.