Attention plays a key role in how the brain integrates information across the different senses for multisensory perception. Humans integrate information if they infer that the signals arose from a common cause, but they segregate information from independent causes. Still, how spatial attention influences these causal inference processes remains unclear. To address this, we employed a variant of the classical spatial-cueing paradigm (Posner, 1980) to investigate how exogenous and endogenous attention modulate audiovisual perception. Participants (n = 20) completed three sessions (~2.5 h each), comprising 24 runs of audiovisual and unisensory trials. Exogenous attention was manipulated via briefly flashed cues (50% validity), and endogenous attention via arrows (75% validity). Subsequent audiovisual spatial target stimuli were presented independently at six horizontal locations, and participants selectively localized either auditory or visual targets.

Behavioral analyses focus on crossmodal bias (CMB) as a function of cue validity, spatial disparity, and task relevance, separately for exogenous and endogenous attention. Preliminary results suggest that CMB shows the key pattern of causal inference, that is, stronger integration for audiovisual signals of small as compared to large disparity. This key pattern is most pronounced for auditory targets, which show particularly strong visual biases. The validity of exogenous and endogenous cues further modulates this effect.

Our results suggest that exogenous and endogenous attention interact with causal inference processes. The study provides a comprehensive characterization of how spatial attention shapes multisensory integration, paving the way for neuroimaging studies linking attention-driven causal inference to its neural dynamics.