This is the point at which the participant is at Selleckchem Osimertinib chance (50%) deciding whether the sound came first or second relative to the visual onset. The same software was used to find the slope of the function and to derive 95% confidence intervals for both PSS and slope estimates, via a bootstrapping

procedure. Finally, we estimated the additional auditory lag required for the participant to go from responding at chance to responding ‘voice second’ 75% of the time. The resulting value quantifies the lag that can produce a Just Noticeable Difference (JND) between subjectively synchronous and asynchronous stimuli. For the phoneme discrimination task we obtained the proportion of trials in which the reported phoneme was consistent with the lip-movements, averaged across incongruous conditions only. For example, a ‘ba’ response to /da/ + [ba] and a ‘da’ response to /ba/ + [ga] IOX1 were scored as ‘consistent’. This was plotted

as a psychometric function of auditory lag. The data from each of the two incongruent conditions, plus the average across them, were fit using an asymmetric double sigmoid function (ADS, following van Wassenhove et al., 2007), which results in a bell-shaped curve with adjustable height, width and asymmetry, using the following equation: y=12[tanh(x−c1w1)−tanh(x−c2w2)]withconstraintsw1>0andw2>0 The optimal auditory lag for maximum McGurk interference (tMcG) from vision was read off at the peak of each

of these interpolated functions and averaged, with 95% confidence intervals derived from fits of 1000 bootstrapped samples. For stream–bounce judgements, ADS functions were fitted to the proportion of ‘bounce’ responses. Across subjects, MTMR9 mean (and SD) of R2 values for goodness of fit of functions to the psychometric data were .89 (.13) for the TOJ task, and .75 (.18) for the phoneme discrimination task. All inferential statistics reported in the following are based on parametric statistics, as data did not deviate significantly from normality (Kolmogorov–Smirnov p > .05). PH’s TOJs corroborated his subjective report of voice leading lips. His PSS was shifted away from veridical to 210 msec auditory lag. This means that subjective synchrony could only be restored for PH by artificially lagging voices relative to lip-movements (by 210 msec, see Table 2), at which point temporal order became indistinguishable (Fig. 3a). Also very curiously, the optimal asynchrony for maximum McGurk (tMcG) showed almost exactly the opposite asynchrony (240 msec auditory lead was required for optimum McGurk). Thus voices effectively lagged lip-movements for the purposes of audiovisual speech integration (Fig. 3b). To investigate the generality of PH’s auditory lead we tested him on a variety of biological and artificial non-speech stimuli, using single-task TOJs.