Can holistic processing be learned for inverted faces?

The researchers employed a rigorous training design to evaluate perceptual learning. Subjects participated in a long-term discrimination task involving identical twins. The approach required up to 1100 exposures for each individual twin. These exposures featured diverse images and varying head poses to ensure robust learning. The team assessed performance by comparing outcomes between upright and inverted orientations. They utilized feature removal techniques to test for integrated versus local processing. The study also applied the composite effect as a metric for holistic integration. This methodology allowed for a direct comparison of processing strategies across different orientations.

The strongest finding indicates that holistic processing cannot be learned for inverted faces despite extensive practice. In the upright orientation, participants successfully utilized holistic processing to distinguish between identical twins. The study observed that removing a single face feature did not affect upright performance. Furthermore, the composite effect was clearly obtained for upright faces. Conversely, inverted face identification relied on image-specific learning or tiny local feature differences. The researchers noted that performance improvements for inverted faces did not mirror the integrated strategies seen in upright conditions. This stability suggests that orientation specificity is resistant to training. The results demonstrate a clear divergence between face recognition and object recognition strategies.

The authors propose that holistic processing for faces remains strictly tied to upright orientations despite extensive training. This finding suggests that face perception mechanisms are highly stable against practice. The researchers conclude that learning to identify inverted faces relies on different strategies than those used for upright faces. Participants utilized image-specific memorization or local feature cues rather than integrated processing. This contrasts with object recognition, which often demonstrates greater flexibility regarding orientation. The study indicates that the specialized nature of face processing is not merely a product of generic expertise. These results highlight a fundamental distinction between face and object recognition systems. The authors maintain that orientation specificity is a robust feature of human visual cognition.