Untangling Cue Conflicts: Understanding Spatial Cue Interaction in Navigation through the Bayesian Causal Inference Approach

Spatial navigation is essential for survival, involving the integration of various cues. The current study applied the Bayesian causal inference (BCI) model for multisensory perception to investigate navigation behavior in cue conflict situations for the first time. By contrasting visual landmarks and optic-flow-based path integration cues, we examined how landmark instability influenced underlying cognitive processes of spatial navigation. In the main experiment, 141 participants were assigned to three groups: in the large-conflict-absent group (LC-absent), participants only experienced a small degree of cue conflict; in the large-conflict-present group (LC-present), participants experienced a much larger degree of cue conflict; in the large-conflict-informed group (LC-informed), participants experienced a much larger degree of cue conflict and were made explicitly aware of the landmark instability. Behavioral results showed that increased landmark instability alone did not elicit any noticeable changes in navigation behavior. However, adding explicit awareness to increased landmark instability changed participants’ behavior by reducing their reliance on landmarks. Additionally, combining increased landmark instability with explicit awareness decreased spatial localization precision when only landmarks were used for navigation. The BCI model provided a good account of the data. The modeling results showed that increased landmark instability plus the explicit awareness reduced participants’ reliance on unstable landmarks via two independent mechanisms: first, sensory noise level was increased for unstable landmarks, which naturally resulted in a lower weight assigned to landmarks in the common-cause judgment; second, a lower weight was assigned to unstable landmarks in the different-cause judgment. Surprisingly, increased landmark instability did not decrease the prior belief of a common cause, even when explicit awareness of landmark instability was imposed. Finally, while the cue-weighting strategy in the same-cause judgment is determined by objective cue relative reliability in a bottom-up manner, the cue-weighting strategy in the different-cause judgment correlated with participants’ subjective evaluation of relative cue quality, demonstrating top-down influences from the meta-cognitive process. Further experiments and modeling work verified key findings in the main experiment. Together, the current study offers valuable insights into navigators' reactions to spatial cue conflicts and their strategies in resolving conflicting spatial cues, highlighting the utility of the BCI model in dissecting mechanisms underlying cue interaction during spatial navigation.