Model Selection in AVSP: Some old and not so old news.
Dominic W. Massaro · 2003
We reiterate a paradigm of inquiry in auditory visual speech processing, focusing on appropriate experimental procedures and methods of model selection. Several methods of model selection find evidence in support of the fuzzy logical model of perception (FLMP). We caution investigators to not limit themselves to simply testing the classic McGurk effect because its outcomes cannot distinguish among alternative interpretations. 1. A Paradigm for Inquiry The study of speech perception by ear and eye has been and continues to be a powerful paradigm for uncovering fundamental properties of the information sources in speech and how speech is perceived and understood. Our general framework documents the value of a combined experimental/theoretical approach. The research has contributed to our understanding of the characteristics used in speech perception, how speech is perceived and recognized, and the fundamental psychological processes that occur in speech perception and pattern recognition in a variety of other domains. We believe that our empirical work would be inadequate and perhaps invalid without the corresponding theoretical framework. Thus, the work continues to address both empirical and theoretical issues. At the empirical level, experiments have been carried out to determine how visible speech is used alone and with auditory speech for a broad range of individuals and across a wide variation of situational domains. At the theoretical level, the assumptions and predictions of several models have been analyzed, contrasted, and tested. In addition, a general framework for inquiry and a universal principle of behavior has been proposed.