The results of the present study are broadly consistent with previous work on multimodal perception that provided evidence for the automatic, pre-attentive nature of multisensory integration, mostly in the auditory–visual domain (e.g., Bertelson et al.,
2000; Driver,
1996; Vroomen et al.,
2001a). Evidence in support of the view that multisensory integration is an early, pre-attentive process comes also from neurophysiological (ERP, fMRI) studies. Recent event-related potential (ERP) studies revealed that auditory mismatch negativity (MMN) can be evoked (Stekelenburg, Vroomen, & de Gelder,
2004) or eliminated (Colin, Radeau, Soquet, Dachy, & Deltenre,
2002) by illusory sound shifts induced by ventriloquism. MMN is typically evoked by an occasional deviant in a homogenous sequence of auditory stimuli and is thought to reflect pre-attentive processes (for a review, see Näätänen,
1992). Therefore, it can be inferred that ventriloquism, i.e., audiovisual integration, occurs at an early, pre-attentive processing stage. Similarly, MMN can be elicited by illusory McGurk percepts (Colin, Radeau, Soquet, Demolin, et al.,
2002), indicating that audiovisual integration mechanisms in speech occur early during the perceptual processes. Furthermore, in accordance with the notion that multisensory integration is an early process, a large number of functional magnetic resonance imaging studies (fMRI) observed interactions of inputs from different sensory modalities in primary sensory areas typically thought of as being strictly modality specific (e.g., Calvert et al.,
1997; Kayser, Petkov, Augath, & Logothetis,
2005; Macaluso, Frith, & Driver,
2000; Pekkola et al.,
2005; van Atteveldt, Formisano, Goebel, & Blomert,
2004).