MARCS Research Seminar

Dr Lars T. Boenke from Leibniz Institute for Neurobiology, Magdeburg, Germany will be presenting "Perceptual uncertainty and information integration – A new theoretical framework to understand bias and interindividual differences in audiovisual simultaneity perception?"

Abstract: Studies of audiovisual simultaneity perception are characterized by mainly three puzzling things: Overall inconclusive results, high interindividual differences, and results depending on the task employed. One example for the former are studies assuming that the brain is capable of actively compensating for the different propagation time of sound and light (cf. Kopinska & Harris 2004).

Given that it is also widely stated that temporal coincidence in the brain is the most important cue to link crossmodal signals to an external multimodal event, such a mechanism seems to be an elegant solution how the brain “maintains” simultaneity between the auditory and visual modality (overview Keetels & Vroomen 2012). However, one reason why the results are inconclusive might be the fact that the assumptions are invalid. One necessity to actively compensate for different travel speed is to “know” at least one of the two tokens that the other token can be compensated to accordingly. Whereas this might be given in a typical lab situation, our brain is in the natural environment continuously bombarded by signals yielding to a temporal overlap of neural activation of many different signals. This temporal overlap creates for the perceptual system an ambiguous situation where it initially has no clues which signals belong to the compound external event (and need to be compensated for). This problem of underdetermined solutions is akin to what was formulated by Helmholtz already in 1853 in the context of electromagnetic dipoles and is known in vision science as the inverse optics problem (percept of a 3-dimensional object based on a 2-dimensional retina image).

In my talk I will present published and ongoing psychophysical and electrophysiological work which supports a perceptual uncertainty hypothesis instead of any active compensation mechanism. I will argue that above all the perceptual system is dealing with an inverse problem when linking signals out of a salmagundi of plenty of other signals within a temporal window of uncertainty competing for further processing for an action (including a consciousness percept). In such a situation the perceptual system has basically only two alternatives to deal with this uncertainty: First, the output (report, consciousness percept etc.) is random over trials. Second, the system integrates more information compared to the first option to overcome uncertainty (cf. past experience/prior information). I will propose an interpretation where the observed interindividual differences can be partly explained by the amount of information integration with some participants integrating more information than others and with an overall increased probability of integration proportional to task difficulty. Further I will demonstrate data which allow an alternative explanation why some studies found evidence for an active compensation mechanism despite the theoretical framework seems questionable.