Purchase this article with an account.
Dobkins Karen, Sampath Vanitha; Face inversion effects in infants are driven more by high, than low, spatial frequencies. Journal of Vision 2008;8(6):193. doi: https://doi.org/10.1167/8.6.193.
Download citation file:
© ARVO (1962-2015); The Authors (2016-present)
Background: To investigate the mechanisms underlying development of face processing in infants, we used a contrast threshold paradigm to measure face inversion effects (FIEs) for “low” vs. “high” spatial frequency (SF)-filtered faces, with the notion that “low” and “high” SFs tap different visual mechanisms. Methods: Using forced-choice preferential looking, we obtained contrast thresholds for “low” and “high” SF-filtered faces, for both “upright” and “inverted” images (4 stimulus types per subject). For each age tested, the “low” and “high” SF cutoffs were determined based on the published peaks in contrast sensitivity functions. (For example, for four-month-olds, whose peak is near 0.4 c/deg, “low” was 0.5 c/deg.) Face stimuli were 10.8 degrees wide, presented 13.3 degrees to the left or right of monitor center, at one of five contrasts (6.13% – 95%). For each subject, contrast thresholds were obtained for each of the four stimulus types, and a threshold ratio was calculated: Thr-inverted/Thr-upright, separately for “low” and “high” SF-filtered faces. Values greater than 1.0 indicate a face inversion effect (FIE), i.e., lower thresholds (greater sensitivity) for “upright” faces. Results: Data from four-month-olds show a significant FIE for “high” SF-filtered faces (ratio=1.91, p=0.013), but not “low” SF-filtered faces (ratio=1.10, p=NS). Conclusions: The mechanisms underlying FIEs in infants (i.e., greater sensitivity to upright faces) appear to be selective for high spatial frequencies. The discrepancy with previous findings suggesting that infant face processing relies more on low spatial frequencies (de Heering et al. 2007) may be explained by proposing that: 1) previous studies did not control for detectability of “low” vs. “high” SFs (i.e., the low SF faces may have simply been more detectable), or 2) reliance on “low” vs. “high” SFs may depend on the nature of the face processing measure, which differs between studies.
This PDF is available to Subscribers Only