Purchase this article with an account.
Nicolas Dupuis-Roy, Laurent Caplette, Couet-Garand Alexandre, Valérie English, Maxime Fortin, Mélissa Talbot, Daniel Fiset, Frederic Gosselin; Contrasting the use of interattribute distances with that of all other face-gender discrimination cues. Journal of Vision 2013;13(9):420. doi: https://doi.org/10.1167/13.9.420.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
The use of relational cues such as the distances between the main internal features of a face (interattribute distances; IADs) is believed to be a distinctive and critical aspect of face processing (Maurer et al., 2002). Yet, little is known about the information contained in real-world relational cues compared to other facial cues, and the relative ability of humans to use these cues. Recently, Dupuis-Roy et al. (2012) found that observers performed rather poorly in a face-gender discrimination task when only real-world IADs were available. However, given the scarce information in these cues, this low performance amounted to a high efficiency. To date, no study has ever systematically compared the efficiency associated with real-world IADs to the efficiency associated with the remaining gender-discriminative cues. This was the aim of this follow-up study. Thirty observers (15 females) completed three face-gender discrimination tasks: Task 1 tested the performance when solely real-world IADs were available; Task 2 measured the performance when all cues except IADs were available; and Task 3 assessed the performance when all cues were available. Each task had 1000 trials (500 female faces). To assess the information available in each task, three regularized linear classifiers were first trained on 250 face stimuli and tested on 250 novel ones. The observers’ sensitivity measure was then combined with the respective ideal observer’s sensitivity to compute an efficiency index. Results revealed higher efficiencies in task 1 (IADs only) than in task 2 (all cues except IADs) and task 3 (all cues), suggesting that participants are better at using the available scarce IADs information than any other facial cues. A further analysis comparing the observed performance in task 3 and the one predicted by a linear combination of the other two tasks, suggests a suboptimal integration of the available information.
Meeting abstract presented at VSS 2013
This PDF is available to Subscribers Only