Hierarchical processing of face viewpoint in human visual cortex

Abstract

The ability to recognize objects across different viewpoints (view invariance) is a remarkable property of the primate visual system. According to a prominent theory, view information is represented by view-selective mechanisms at early stages of visual processing and gradually becomes view invariant in high-level visual areas. Single-cell recording studies have also reported an intermediate step of partial view invariance for mirror-symmetric face views. Nevertheless, similar evidence for this type of hierarchical processing for face view has not been reported yet in the human visual cortex. The present functional magnetic resonance imaging study used state-of-the-art multivariate pattern analysis to explore face-view tuning in the human visual cortex. Our results revealed that consistent with a view-selective representation, face view can be successfully decoded in face and object-selective regions as well as in early visual cortex. Critically, similar neural representations for mirror-symmetric views were found in high-level but not in low-level visual areas. Our results support the notion of gradual emergence of view-invariant representation with invariance for mirror-symmetric images as an intermediate step and propose putative neural correlates of mirror-image confusion in the human brain.

Figure 1.

Activations and correct classification for the five face views in face-selective areas (FFA, OFA, and STS) in Experiment 1. A, Images of the five face views used in the experiment. B, Percent signal change for each of the five face views. C, Correct classification rates averaged across the five face views. The black line indicates chance level of 0.2. All error bars indicate the SEM.

Figure 2.

Correct classification and confusion rates in face-selective areas in Experiment 1. A, Correct classification and confusion rates for frontal view faces. B, Correct classification and confusion rates for profile view faces. C, Pairwise classification of the profile view with each of the other four views in face-selective areas. The black line indicates a chance level of 0.5. All error bars indicate the SEM.

Figure 3.

Activations, correct classification, and confusion rates in the LO object area in Experiment 1. A, Average percent signal change for the five face views. B, Correct classification and confusion rates for the profile view. C, Pairwise classification of the profile view with each of the other views. The black line indicates a chance level of 0.5. All error bars indicate the SEM.

Figure 4.

Laterality bias for face views in the one-back identity task in Experiment 1. A, To perform the one-back identity task subjects had to look at the facial features (indicated by the arrow) and therefore right and left profiles (and partly also half profiles) were located in different visual fields. B, Average percent signal change for the five face views in EVC indicates strong laterality bias for the different face views. The error bars indicate the SEM.

Figure 5.

Activations, correct classification, and confusion rates in EVC in Experiment 2. A, The five face views used in Experiment 2. B, Average percent signal change for the five face views. C, Correct classification rates and confusion errors for profile view. D, Pairwise classification of the profile view with each of the other views. The black line indicates a chance level of 0.5. All error bars indicate the SEM.

Figure 6.

Activations and correct classification to the five face views in face-selective areas in Experiment 2. A, Average percent signal change for the 5 face views. B, Correct classification rates and confusion errors for profile view. C, Pairwise classification of the profile view with each of the other views. The black line indicates a chance level of 0.5. All error bars indicate the SEM.

Figure 7.

Classification rates in the LO object area in Experiment 2. A, Correct classification rates and confusion errors for profile view. B, Pairwise classification of the profile view with each of the other views. The black line indicates a chance level of 0.5. All error bars indicate the SEM.