Understanding the Recognition of Facial Identity and Facial Expression
Total Page:16
File Type:pdf, Size:1020Kb
REVIEWS UNDERSTANDING THE RECOGNITION OF FACIAL IDENTITY AND FACIAL EXPRESSION Andrew J. Calder* and Andrew W. Young‡ Abstract | Faces convey a wealth of social signals. A dominant view in face-perception research has been that the recognition of facial identity and facial expression involves separable visual pathways at the functional and neural levels, and data from experimental, neuropsychological, functional imaging and cell-recording studies are commonly interpreted within this framework. However, the existing evidence supports this model less strongly than is often assumed. Alongside this two-pathway framework, other possible models of facial identity and expression recognition, including one that has emerged from principal component analysis techniques, should be considered. 1 FUNCTIONAL ACCOUNT Nearly 20 years ago, Bruce and Young presented a of distinct pathways for the visual analysis of facial A theoretical framework that is model of face recognition that posited separate func- identity and expression, but differ in terms of whether based on the information- tional routes for the recognition of facial identity and the perceptual coding of expression is carried out by a processing characteristics of a facial expression BOX 1. This framework has remained dedicated system for expressions1 or by a system that set of cognitive subsystems rather than their underlying the dominant account of face perception; few papers codes expression alongside other changeable facial 2 neural mechanisms have challenged it and none has offered a widely characteristics . accepted alternative. Here we discuss the relevant evi- At the heart of both models is the idea that facial dence, and show why a different conception from that identity and expression are recognized by function- offered by Bruce and Young1 should be considered. ally and — by implication for Bruce and Young1, and As a FUNCTIONAL ACCOUNT, the Bruce and Young1 explicitly for Haxby and colleagues2 — neurologically model did not incorporate a neural topography of its independent systems. This idea is supported by many separate components. However, the recent neurologi- psychological studies. For example, the familiarity of cal account of face perception proposed by Haxby and a face does not affect the ability of a healthy partici- colleagues2 BOX 1 is compatible with the general con- pant to identify its expression and vice versa3–6. Brain *Medical Research Council ception offered by Bruce and Young. The core system injury in humans can produce selective impairments Cognition and Brain of Haxby and colleagues’ model2 contains two func- in the recognition of facial identity or facial expres- Sciences Unit, 15 Chaucer tionally and neurologically distinct pathways for the sion7–11, in nonhuman primates, different cell popula- Road, Cambridge CB2 2EF, visual analysis of faces BOX 1: one codes changeable tions respond to facial identity and expression12, and UK. ‡Department of Psychology and York facial properties (such as expression, lipspeech and functional imaging studies show that the perception of Neuroimaging Centre, eye gaze) and involves the inferior occipital gyri and facial identity and facial expression have different neu- University of York, superior temporal sulcus (STS), whereas the other ral correlates13–15. Therefore, the central idea of some York YO10 5DD, UK. codes invariant facial properties (such as identity) form of dissociation between these two facial cues is Correspondence to A.J.C. e-mail: andy.calder@mrc- and involves the inferior occipital gyri and lateral not at issue. Rather, we focus on how this dissociation cbu.cam.ac.uk fusiform gyrus. The models proposed by Haxby and should be interpreted. In particular, we ask whether the doi:10.1038/nrn1724 colleagues2, and Bruce and Young1, share the idea concept of distinct parallel visual routes that process NATURE REVIEWS | NEUROSCIENCE VOLUME 6 | AUGUST 2005 | 641 © 2005 Nature Publishing Group REVIEWS This review is structured around four questions Box 1 | Cognitive models of face perception that we regard as central to interpreting dissociations a Expression between facial identity and expression. The first asks at analysis what level of analysis the facial identity route bifurcates Facial speech from the facial expression route. The remaining three analysis questions relate specifically to the recognition of facial Directed Cognitive expressions: are all expressions processed by a single visual system system, do the mechanisms for recognizing expression processing incorporate a multimodal level of analysis, and does the facial expression system deal with anything other than expression? View-centred Expression- Face Person Name descriptions independent recognition identity retrieval Where do the two routes separate? descriptions units nodes Although each face is a single object, it conveys many Structural encoding socially important characteristics (such as identity, b age, sex, expression, lipspeech and gaze), at least some of which (for example, identity and expression) show Intraparietal sulcus considerable functional independence. Face processing Spatially directed attention Superior temporal therefore requires a different conceptual framework sulcus Changeable aspects Auditory cortex from object recognition, and any plausible model of faces — perception Prelexical speech perception requires a ‘front-end’ system that can both extract and of eye gaze, expression and lip movement Amygdala, insula, limbic system separate different facial cues. Inferior occipital gyri Emotion 1 2 Early perception of Both Bruce and Young and Haxby and colleagues facial features propose that functional (and neural) separation of the Lateral fusiform Anterior temporal facial identity and expression routes occurs immediately gyrus Personal identity, name and Invariant aspects of biographical information after the front-end component, which is involved in the faces — perception initial structural and visual analysis of faces, with each of unique identity Extended system Further processing with other route incorporating distinct visuoperceptual represent- Core system: visual analysis neural systems ations of the relevant facial characteristic. So, what evi- dence is there that facial identity and facial expression 1 Bruce and Young’s functional model of face processing (panel a) contains separate are coded in distinct visual representational systems? parallel routes for recognizing facial identity, facial expression and lipspeech. The In cognitive neuropsychology, support for this route labelled ‘directed visual processing’ is involved in the direction of attention to a framework requires the identification of patients with particular face or facial feature. It is generally considered that the idea of separate impairments in the visual recognition of facial identity routes for the recognition of facial identity and expression is supported by studies in or facial expression alone. Cases of PROSOPAGNOSIA with- cognitive psychology4,5, cognitive neuropsychology7–10, single-cell recording in out impaired facial expression recognition would sup- nonhuman primates12 and functional imaging13,14. Panel a modified, with permission, port the independence of identity processing; however, from REF. 1 © (1986) British Psychological Society. The Bruce and Young1 framework is compatible with the distributed human neural remarkably few prosopagnosics show well-preserved system for face perception proposed by Haxby and colleagues2 (panel b). This facial expression recognition. In fact, the idea that pro- identifies the neural structures that are involved in recognizing two types of facial sopagnosics can recognize facial expression is usually information: changeable (dynamic) characteristics, such as expression, gaze and supported only by anecdote; on formal testing, most such lipspeech; and invariant (relatively nonchangeable) characteristics, such as identity. patients show impairments of facial expression recogni- Panel b reproduced, with permission, from REF. 2 © (2000) Elsevier Science. tion. These difficulties are usually less severe than the This model is divided into a core system for the visual analysis of faces, which problems with recognizing facial identity, although this comprises three occipital/temporal regions, and an extended system that includes might reflect the many procedural differences between neural systems that are involved in other cognitive functions. Visuoperceptual standard tests of facial identity and expression. Instead, representations of changeable characteristics are associated with the superior much of the evidence for impaired facial identity with temporal sulcus (STS), whereas invariant characteristics are coded by the lateral intact facial expression recognition comes from studies fusiform gyrus (including the fusiform face area or FFA). The extended systems act in which the cause of the identity impairments has not together with the core system to facilitate the recognition of different facial cues. been established7,8,16,17. Such data can provide evidence of a dissociation between the recognition of identity and expression, but they do not prove that this dissociation PROSOPAGNOSIA facial identity and facial expression offers the best fit to has a visuoperceptual origin. A visual agnosia that is largely current findings. This view is endorsed by both mod- So, a crucial issue that is often overlooked is that restricted to face recognition, 1,2 but leaves intact recognition of els of face perception