The effect of inserting an inter-stimulus interval in face–voice matching tasks

Abstract

Based on the results of three experiments, Smith, Dunn, Baguley, and Stacey (2016b) concluded that source identity information is shared by voices and faces regardless of whether the faces are static or dynamic (i.e., articulating but muted). The balance of evidence suggests that voices and static faces do provide sufficient concordant identity information (Smith et al., 2016a) so that it is possible to accurately match an unfamiliar face to a voice (Smith et al., 2016b). All previous tests of face-voice matching have presented faces and voices close together in time, with a maximum 1-s inter-stimulus interval (ISI;

Matching Unfamiliar Voices to Static and Dynamic Faces: No Evidence for a Dynamic Face Advantage in a Simultaneous Presentation Paradigm

“…Taken together, the most likely explanation for the lack of a significant face-voice matching effect in the static group using the first trial per participant only is simply a lack statistical power for this particular analysis, especially as the direction of the effect in the static group pointed into the expected direction. Finally, the assumption of significant static face-voice matching ability is also confirmed by previous studies demonstrating corresponding effects (Krauss et al, 2002; Mavica and Barenholtz, 2013, Experiment 2; Smith et al, 2016b, Experiment 3), and some studies suggested that the absence of strong memory demands (as realized in the present design) is especially beneficial for yielding above-chance static face-voice matching ability (Smith et al, 2016a, 2018; Stevenage et al, 2017). Taken together, there is no substantial reason to doubt the presence of face-voice matching abilities in both groups of participants.…”

“…This paradigm also revealed above-chance matching performance. Following up on this, Smith et al (2018) confirmed that any delay between the presentation of the voice and the static face yields matching performance at chance level only. Recently, Stevenage et al (2017) used a simultaneous same/different matching task and showed above-chance performance for matching voices to static faces.…”

“…To conclude, it can be considered unlikely that cross-modal interference based on a mismatch between head voice and visual face substantially increased task difficulty selectively in the dynamic condition, especially since the static condition also involved a mismatch (i.e., the lips of the static faces do not move despite the presentation of spoken text). Conversely, I firmly believe that comparing static and dynamic visual faces in a simultaneous presentation paradigm for the first time allows for a fair comparison of static and dynamic conditions as participants do not have to rely on memory, which is already well-known to substantially compromise any interpretation of face-voice matching performance (Smith et al, 2016a, 2018; Stevenage et al, 2017).…”

Explaining face-voice matching decisions: The contribution of mouth movements, stimulus effects and response biases

“…Distinct profiles of response behaviour were apparent in our dataa significant interaction of trial type (same vs. different) by condition indicated that participants were showing systematic response biases that varied by condition. Such response biases have been reported in previous face-voice matching studies employing a samedifferent procedure (Smith et al, 2016a(Smith et al, , 2016cStevenage et al, 2017), where it has been observed that participants have an overall tendency to accept face-voice pairings as belonging to the same identity. Here, we therefore asked a second research question, addressed in an additional set of exploratory analyses: Beyond mouth movements, how are participants' face-voice matching responses for static and dynamic faces affected by experimental design choices?…”

“…Trial-wise accuracy was the outcome variable, condition was entered as a fixed factor. We now also included trial type (same or different identity) and an interaction of trial type and condition fixed effects based on previous studies showing differences in accuracy varying along these two factors (Smith et al, 2016a(Smith et al, , 2016cStevenage et al, 2017). The random-effects structure was the same as described above for the exploratory analysis, with only trial type having been moved into the fixed effects as it now became an effect of interest.…”