Asking a naive question about the McGurk effect: Why does audio [b] give more [d] percepts with visual [g] than with visual [d]?

Why does audio [b] give more [d] percepts with visual [g] than with visual [d], as in the present classical McGurk experiment ? An explanation given for this asymmetry could be a language bias towards [d]. As concerns the visual information, and contrary to what is sometimes taken as granted in part of the lipreading literature, visual [g] does not give more [d] than [g] responses. In fact [d] and [g] are neither visemically, nor auditorily equivalent. They are fully distinguishable in audio as well as in vision, where 80% correct identifications are current in laboratory condition, as in the present experiment. We show here that in spite of these highly differenciating scores, FLMP modelling can quite surprisingly account for such an asymmetry by tuning very small remaining values; which is highly unsatisfactory. We suggest another explanation for this asymmetry which could be grounded on brain mechanisms dedicated to the computation of auditory and visual mouth opening movements, i.e. audio movement and visual velocity detectors and integrators, dedicated to the bimodal integration of place targets in speech.