Purchase this article with an account.
Mayu Nishimura, Samidha Joglekar, Daphne Maurer; The effect of training on the recognition of faces across changes in viewpoint. Journal of Vision 2008;8(6):160. doi: 10.1167/8.6.160.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Recognizing a face from a novel viewpoint requires processing the structural properties of the face that are reliable cues to identity and view-invariant. One such property may be second-order relations (e.g., spacing between eyes and mouth). In Experiment 1, we investigated whether 10-year-old children's and adults' recognition of faces across changes in viewpoint could be improved through training, and whether training results were correlated with sensitivity to second-order relations. Over two one-hour sessions 10-year-olds and adults (n = 10) were trained to make same/different judgments about facial identity between faces seen from different viewpoints. Consistent with previous studies (e.g. Mondloch et al., 2003), 10-year-olds were worse overall than adults. However, both groups improved at a similar rate during training, with 10-year-olds' final accuracy being comparable to adults' accuracy prior to training. There was no correlation between performance on the viewpoint training task and sensitivity to second-order relations either before or after training in either age group, perhaps because observers may have learned to match specific views of the training faces and not a general skill. In Experiment 2, we investigated whether training adults (n = 12) would be more effective if novel faces were introduced as training progressed over the two-day period. Improvement in matching faces across changes in viewpoint transferred from the first 7 facial identities to the next 7 identities, a result suggesting that training improved a general skill in view-invariant recognition. However, improvement failed to transfer to the third set of 7 identities and was not correlated with sensitivity to second-order relations, results suggesting that the learning also involved the linking of view-specific exemplars. Collectively, the results indicate that improvements in recognizing faces across changes in viewpoint involve both view-specific and view-independent processes, and are not directly related to sensitivity to second-order relations.
This PDF is available to Subscribers Only