Changing faces: Direction is important

PLOS ONE, 2018

It can be difficult to recognise new instances of an unfamiliar face. Recognition errors in this particular situation appear to be viewpoint dependent with error rates increasing with the angular distance between the face views. Studies using front views for comparison have shown that recognising faces rotated in yaw can be difficult and that recognition of faces rotated in pitch is more challenging still. Here we investigate the extent to which viewpoint dependent face recognition depends on the comparison view. Participants were assigned to one of four different comparison view groups: front, ¾ yaw (right), ¾ pitch-up (above) or ¾ pitch-down (below). On each trial, participants matched their particular comparison view to a range of yaw or pitch rotated test views. Results showed that groups with a front or ¾ yaw comparison view had superior overall performance and more successful generalisation to a broader range of both pitch and yaw test views compared to groups with pitch-up or pitchdown comparison views, both of which had a very restricted generalisation range. Regression analyses revealed the importance of image similarity between views for generalisation, with a lesser role for 3D face depth. These findings are consistent with a view interpolation solution to view generalisation of face recognition, with front and ¾ yaw views being most informative.

1995

In two experiments we examined the ability of human observers to recognize faces from novel viewpoints. Previous work has indicated that there are marked declines in recognition performance when observers learn a particular view of a face and are asked to recognize the face from a novel viewpoint. We replicate these ndings and extend them in several ways. First, we replicate the well-known 3/4 view advantage for recognition and extend it to show that this advantage is stronger than would be expected simply due to the 3/4 view being the center of the learned views. In the second experiment, we found little evidence for advantageous transfer to a symmetric view of the other side of the face, in all cases, observers were much better at recognizing a face from the side learned. Third, we extended past results to explore the consistency of face recognizability for individual faces across di erent views and view transfer conditions. We found only a modest relationship between the recognizability of individual faces in the di erent view conditions. These data give insight into the organization of memory for faces and its stability across changes in viewpoint. Many thanks are due to Niko Troje and Isabelle B ultho for the stimulus creation and processing. Alice O'Toole gratefully acknowledges support by the Alexander von Humboldt Stiftung and the hospitality of the Max-Planck Institut f ur biologische Kybernetik.

Vision research, 2007

Swiss Journal of Psychology, 2006

When faces are viewed from different angles the appearance of facial features undergoes dramatic changes. We investigated two types of 3D-head models in frontal and threequarter views, varying either in componential information such as different eyes, mouths and noses, or in relational information. Variations of the latter can only be investigated

Proceedings of the Royal Society B: Biological Sciences, 2004

Understanding how the human visual system recognizes objects is one of the key challenges in neuroscience. Inspired by a large body of physiological evidence, a general class of recognition models has emerged, which is based on a hierarchical organization of visual processing, with succeeding stages being sensitive to image features of increasing complexity. However, these models appear to be incompatible with some well-known psychophysical results. Prominent among these are experiments investigating recognition impairments caused by vertical inversion of images, especially those of faces. It has been reported that faces that differ 'featurally' are much easier to distinguish when inverted than those that differ 'conf igurally'; a f inding that is difficult to reconcile with the physiological models. Here, we show that after controlling for subjects' expectations, there is no difference between 'featurally' and 'conf igurally' transformed faces in terms of inversion effect. This result reinforces the plausibility of simple hierarchical models of object representation and recognition in the cortex.

Perception, 2007

2019

Viewing an unfamiliar face at an orientation in depth that differs only moderately from its previously viewed orientation incurs sizeable costs in the reaction time (RT) and accuracy for its recognition. There has been no general account of these costs. We assessed the effects of orientation disparity in a minimal match-to-sample paradigm of a display of three faces, with one of the two test faces physically identical to the sample, the other being a foil. Both test faces had the same orientation in depth at an angle that differed by 0° to 20° from the sample. The dissimilarities of the face images were scaled by a model based on V1 simple cell tuning that correlates almost perfectly with psychophysical similarity. Performance on this task could be modeled with two linear factors that produced additive effects on RTs: 1. The orientation disparity between the matching face and the sample, and 2. the similarity of the foil to the matching face. An orientation disparity between sample ...

Journal of Vision, 2010

Although perceptual learning of simple visual features has been studied extensively and intensively for many years, we still know little about the mechanisms of perceptual learning of complex object recognition. In a series of seven experiments, human perceptual learning in discrimination of in-depth orientation of face view was studied using psychophysical methods. We trained subjects to discriminate face orientations around a face view (i.e., 30°) over eight daily sessions, which resulted in a significant improvement in sensitivity to the face view orientation. This improved sensitivity was highly specific to the trained orientation and persisted up to 6 mo. Different from perceptual learning of simple visual features, this orientation-specific learning effect could completely transfer across changes in face size, visual field, and face identity. A complete transfer also occurred between two partial face images that were mutually exclusive but constituted a complete face. However, the transfer of the learning effect between upright and inverted faces and between a face and a paperclip object was very weak. These results shed light on the mechanisms of the perceptual learning of face view discrimination. They suggest that the visual system had learned how to compute face orientation from face configural information more accurately and that a large amount of plastic changes took place at a level of higher visual processing where size-, location-, and identity-invariant face views are represented.

2009

The aim of this study was to separately analyze the role of featural and configural face representations. Stimuli containing only featural information were created by cutting the faces into their parts and scrambling them. Stimuli only containing configural information were created by blurring the faces. Employing an old-new recognition task, the aim of Experiments 1 and 2 was to investigate whether unfamiliar faces (Exp. 1) or familiar faces (Exp. 2) can be recognized if only featural or configural information is provided. Both scrambled and blurred faces could be recognized above chance level. A further aim of Experiments 1 and 2 was to investigate whether our method of creating configural and featural stimuli is valid. Pre-activation of one form of representation did not facilitate recognition of the other, neither for unfamiliar faces (Exp. 1) nor for familiar faces (Exp. 2). This indicates a high internal validity of our method for creating configural and featural face stimuli. Experiment 3 examined whether features placed in their correct categorical relational position but with distorted metrical distances facilitated recognition of unfamiliar faces. These faces were recognized no better than the scrambled faces in Experiment 1, providing further evidence that facial features are stored independently of configural information. From these results we conclude that both featural and configural information are important to recognize a face and argue for a dual-mode hypothesis of face processing. Using the psychophysical results as motivation, we propose a computational framework that implements featural and configural processing routes using an appearance-based representation based on local features and their spatial relations. In three computational experiments (Experiments 4-6) using the same sets of stimuli, we show how this framework is able to model the psychophysical data.