The Effects of Cue Reliability on Crossmodal Recalibration in Adults and Children

Reliability-based cue combination is a hallmark of multisensory integration, while the role of cue reliability for crossmodal recalibration is less understood. The present study investigated whether visual cue reliability affects audiovisual recalibration in adults and children. Participants had to localize sounds, which were presented either alone or in combination with a spatially discrepant high- or low-reliability visual stimulus. In a previous study we had shown that the ventriloquist effect (indicating multisensory integration) was overall larger in the children groups and that the shift in sound localization toward the spatially discrepant visual stimulus decreased with visual cue reliability in all groups. The present study replicated the onset of the immediate ventriloquist aftereffect (a shift in unimodal sound localization following a single exposure of a spatially discrepant audiovisual stimulus) at the age of 6–7 years. In adults the immediate ventriloquist aftereffect depended on visual cue reliability, whereas the cumulative ventriloquist aftereffect (reflecting the audiovisual spatial discrepancies over the complete experiment) did not. In 6–7-year-olds the immediate ventriloquist aftereffect was independent of visual cue reliability. The present results are compatible with the idea of immediate and cumulative crossmodal recalibrations being dissociable processes and that the immediate ventriloquist aftereffect is more closely related to genuine multisensory integration.

Semantic and spatial congruency mould audiovisual integration depending on perceptual awareness

Information integration is considered a hallmark of human consciousness. Recent research has challenged this tenet by showing multisensory interactions in the absence of awareness. This psychophysics study assessed the impact of spatial and semantic correspondences on audiovisual binding in the presence and absence of visual awareness by combining forward–backward masking with spatial ventriloquism. Observers were presented with object pictures and synchronous sounds that were spatially and/or semantically congruent or incongruent. On each trial observers located the sound, identified the picture and rated the picture’s visibility. We observed a robust ventriloquist effect for subjectively visible and invisible pictures indicating that pictures that evade our perceptual awareness influence where we perceive sounds. Critically, semantic congruency enhanced these visual biases on perceived sound location only when the picture entered observers’ awareness. Our results demonstrate that crossmodal influences operating from vision to audition and vice versa are interactively controlled by spatial and semantic congruency in the presence of awareness. However, when visual processing is disrupted by masking procedures audiovisual interactions no longer depend on semantic correspondences.

High-Level Perception and Multimodal Perception

What is the correct procedure for determining the contents of perception? Philosophers tackling this question increasingly rely on empirically oriented procedures. This chapter argues that this strategy constitutes an improvement over the armchair methodology of phenomenal contrast arguments, but that there is a respect in which current empirical procedures remain limited: they are unimodal in nature, wrongly treating the senses as isolatable. The chapter thus has two aims: first, to motivate a reorientation of the admissible contents debate into a multimodal framework. The second is to explore whether experimental studies of multimodal perception support a so-called Liberal account of perception’s admissible contents. The chapter concludes that the McGurk effect and the ventriloquist effect are both explicable without the postulation of high-level content, but that at least one multimodal experimental paradigm may necessitate such content: the rubber hand illusion. One upshot is that Conservatives who claim that the Liberal view intolerably broadens the scope of perceptual illusions, particularly from the perspective of perceptual psychology, should pursue other arguments against that view.

The Ventriloquist Effect is not Consistently Affected by Stimulus Realism

Despite more than 60 years of research, it has remained uncertain if and how realism affects the ventriloquist effect. Here, a sound localization experiment was run using spatially disparate audio-visual stimuli. The visual stimuli were presented using virtual reality, allowing for easy manipulation of the degree of realism of the stimuli. Starting from stimuli commonly used in ventriloquist experiments, i.e., a light flash and noise burst, a new factor was added or changed in each condition to investigate the effect of movement and realism without confounding the effects of an increased temporal correlation of the audio-visual stimuli. First, a distractor task was introduced to ensure that participants fixated their eye gaze during the experiment. Next, movement was added to the visual stimuli while maintaining a similar temporal correlation between the stimuli. Finally, by changing the stimuli from the flash and noise stimuli to the visuals of a bouncing ball that made a matching impact sound, the effect of realism was assessed. No evidence for an effect of realism and movement of the stimuli was found, suggesting that, in simple scenarios, the ventriloquist effect might not be affected by stimulus realism.

The Ventriloquist Effect is not Consistently Affected by Stimulus Realism †

Despite more than 60 years of research, it has remained uncertain if and how realism affects the ventriloquist effect. Here, a sound localization experiment was run using spatially disparate audio-visual stimuli. The visual stimuli were presented using virtual reality, allowing for easy manipulation of the degree of realism of the stimuli. Starting from stimuli commonly used in ventriloquist experiments, i.e., a light flash and noise burst, a new factor was added or changed in each condition to investigate the effect of movement and realism without confounding the effects of an increased temporal correlation of the audio-visual stimuli. First, a distractor task was introduced to ensure that participants fixated their eye gaze during the experiment. Next, movement was added to the visual stimuli while maintaining a similar temporal correlation between the stimuli. Finally, by changing the stimuli from the flash and noise stimuli to the visuals of a bouncing ball that made a matching impact sound, the effect of realism was assessed. No evidence for an effect of realism and movement of the stimuli was found, suggesting that, in simple scenarios, the ventriloquist effect might not be affected by stimulus realism.

Transfer of Audio-Visual Temporal Training to Temporal and Spatial Audio-Visual Tasks

Temporal and spatial characteristics of sensory inputs are fundamental to multisensory integration because they provide probabilistic information as to whether or not multiple sensory inputs belong to the same event. The multisensory temporal binding window defines the time range within which two stimuli of different sensory modalities are merged into one percept and has been shown to depend on training. The aim of the present study was to evaluate the role of the training procedure for improving multisensory temporal discrimination and to test for a possible transfer of training to other multisensory tasks. Participants were trained over five sessions in a two-alternative forced-choice simultaneity judgment task. The task difficulty of each trial was either at each participant’s threshold (adaptive group) or randomly chosen (control group). A possible transfer of improved multisensory temporal discrimination on multisensory binding was tested with a redundant signal paradigm in which the temporal alignment of auditory and visual stimuli was systematically varied. Moreover, the size of the spatial audio-visual ventriloquist effect was assessed. Adaptive training resulted in faster improvements compared to the control condition. Transfer effects were found for both tasks: The processing speed of auditory inputs and the size of the ventriloquist effect increased in the adaptive group following the training. We suggest that the relative precision of the temporal and spatial features of a cross-modal stimulus is weighted during multisensory integration. Thus, changes in the precision of temporal processing are expected to enhance the likelihood of multisensory integration for temporally aligned cross-modal stimuli.