Abstract
According to Bernard et al.'s (2012) Sexualized Body Inversion Hypothesis (SBIH), female bodies are viewed as objects, and processed as such by the visual system, while male bodies are processed as social objects. Their hypothesis is supported by a greater inversion effect for male than for female body images in a discrimination task. However, there are physical differences between the male and female image sets that might account for the reported differences in the inversion effect. The current study investigated how much of this sex difference can be accounted for by discriminability of images, and whether the SBIH is still supported. We replicated Bernard et al.'s study presenting participants with a target image, followed by a blank screen, then the target image presented alongside its mirror-image as a 2AFC recognition task. We found a significant Orientation by Target Sex Interaction (F (1,47) = 9.29, p < 0.003). While participants recognized upright and inverted images of females equally well (t (46) = 1.82, p > 0.05), they performed better for upright than for inverted images of males (t (46) = 5.56, p < 0.0001), consistent with Bernard et al. We conducted an ideal observer analysis to quantify the discriminability of the images, obtaining 100 thresholds for each of the 48 images (12 upright males, 12 inverted males, 12 upright females, 12 inverted females). A simple linear regression showed that discriminability predicted human performance (t (1,46) = -2.818, p < 0.000, adj. R² = 0.1287), suggesting that physical characteristics of the images account for some of the differences in the inversion effect. However, after accounting for discriminability, there was still a reliable residual difference in the inversion effects across stimulus sets as shown by a significant Target Sex by Orientation Interaction (F(1,43) = 4.99, p < 0.031), supporting the SBIH.
Meeting abstract presented at VSS 2016