Domain Specificity for Faces versus Expertise: A Critical Look at the Evidence

This page will be updated soon, in the interm the following articles provides the best overview:

McKone, E., Kanwihser, N., & Duchaine. B. (in press, due Jan 07) Can generic expertise explain special processing for faces? Trends in Cognitive Science. pdf icon

Op de Beeck, J., et al. (in press) Discrimination Training Alters Object Representations in Human Extrastriate Cortex. Journal of Neuroscience. pdf icon

Moore, C., Cohen, M., Ranganath, C. (2006) Neural Mechanisms of Expert Skills in Visual Working Memory. Journal of Nerusoscience. 26(43)11187–96 pdf icon

Rachel Robbins, Elinor McKone. (in press) No face-like processing for objects-of-expertise in three behavioural tasks. Cognition pdf icon

According to the most widely discussed current version of the 'expertise hypothesis', the cognitive and neural mechanisms that have been claimed to be specific for faces per se are in fact more generally engaged in fine-grained discrimination of exemplars of any visual object class for which for which the subject has gained substantial visual expertise, and for which exemplars share the same first-order configuration. For papers describing and supporting this position, see the original paper that put forth this hypothesis (Diamond & Carey, 1986, Why faces are not special: An effect of expertise, JEP:General, 115, 107-117), and the many publications listed on the Perceptual Expertise network.

The Expertise Hypothesis has received considerable attention in textbooks and in the popular press. Yet in our view, the data supporting the hypothesis are weak, and are opposed by powerful counterevidence. Much of the evidence against the expertise hypothesis and favoring domain specificity of face processing mechanisms has been published relatively recently (this web information is dated March 2005), so for convenience we have collected here a list of these articles and their abstracts. We include some conference abstracts for which papers are not yet available.

1. McKone, E. & Kanwisher, N. (2005). Does the human brain process objects of expertise like faces? A review of the evidence. In: Dehaene S, Duhamel JR, Hauser M, and Rizzolatti, editors: From Monkey Brain to Human Brain. MIT Press. pdf icon

This chapter provides a review of the evidence for and against the expertise hypothesis. Particularly important in this review is a critical analysis of the behavioral studies that have been claimed to support the expertise hypothesis; we present the actual data from those studies and show that despite the strong claims made in the abstracts the actual evidence for the expertise hypothesis is weak at best. This chapter was written about a year ago so does not include a discussion of many of the articles below.

2. Duchaine BC, Dingle K, Butterworth E, Nakayama K. (2004). Normal greeble learning in a severe case of developmental prosopagnosia, Neuron. 2004, 43(4):469-73. pdf icon (requires subscription to Neuron)

A central question in cognitive neuroscience is whether mechanisms exist that are specialized for particular domains. One of the most commonly cited examples of a domain-specific competence is the human ability to recognize upright faces. However, according to a widely discussed alternative hypothesis, face recognition is instead performed by mechanisms specialized for processing any object class for which an individual has expertise. Faces, according to this domain-general hypothesis, are just one example of an expert class. Nonface object expertise has been intensively investigated using a training procedure involving an artificial stimulus class known as greebles. A key prediction of this hypothesis is that individuals with face recognition impairments will also have impairments with other categories that control subjects have expertise with. Our results show that a man with severe prosopagnosia performed normally throughout the standard greeble training procedure. These findings indicate that face recognition and greeble recognition rely on separate mechanisms.

[Note: further evidence that face recognition is doubly dissociable from object expertise comes from two classic studies of neurological patients: one who lost visual expertise for previously expert stimulus categories despite retaining normal face recognition abilities (Moscovitch, M., Winocur, G., & Behrmann, M., 1997. What is special about face recognition? Nineteen experiments on a person with visual object agnosia and dyslexia but normal face recognition. Journal of Cognitive Neuroscience, 9, 555-604.), and another who became severely prosopagnosic but retained expertise for cars (Sergent, J. & Signoret, J.L., 1992. Varieties of functional deficits in prosopagnosia. Cerebral Cortex, 2, 375-388).]

3. Xu, Y., Liu, J., & Kanwisher, N. (2005). The M170 is Selective for faces, not for Expertise. Neuropsychologia, 43, 588-597. pdf icon

Are the mechanisms for face perception selectively involved in processing faces per se, or do they also participate in the processing of any class of visual stimuli that share the same basic configuration and for which the observer has gained substantial visual expertise? Here we tested the effects of visual expertise on the face-selective "M170", a magnetoencephalography (MEG) response component that occurs 170ms after stimulus onset and is involved in the identification of individual faces. In Experiment 1, cars did not elicit a higher M170 response (relative to control objects) in car experts compared to controls subjects. In Experiment 2, the M170 amplitude was correlated with successful face identification, but not with successful car identification in car experts. These results indicate that the early face processing mechanisms marked by the M170 are involved in the identification of faces in particular, not in the identification of any objects of expertise.

4. Grill-Spector, K. Knouf, N., & Kanwisher, N. (2004). The fusiform face area subserves face detection and face identification, not generic within-category identification. Nature Neuroscience, 7, 555-62. pdf icon

The function of the fusiform face area (FFA), a face-selective region in human extrastriate cortex, is a matter of active debate. Here we measured the correlation between FFA activity measured by functional magnetic resonance imaging (fMRI) and behavioral outcomes in perceptual tasks to determine the role of the FFA in the detection and within-category identification of faces and objects. Our data show that FFA activation is correlated on a trial-by-trial basis with both detecting the presence of faces and identifying specific faces. However, for most non-face objects (including cars seen by car experts), within-category identification performance was correlated with activation in other regions of the ventral occipitotemporal cortex, not the FFA. These results indicate that the FFA is involved in both detection and identification of faces, but that it has little involvement in within-category identification of non-face objects (including objects of expertise).

[Note that Xu (2005, Cerebral Cortex; see also Rhodes et al., 2004, JOCN) replicated Gauthier's finding of a significant but very small increase in the FFA response to cars in car experts. The fact that the study above did not find this result with masked stimuli suggests that this effect, when it happens, may be due to top-down feedback, not bottom-up perceptual processing of the stimulus, a hypothesis consistent with the lack of expertise effects in face-selective MEG responses described in Xu et al., 2005. In any event the FFA response to objects of expertise is not correlated with identification of those objects.]

5. Robbins R, McKone E. (2003). Can holistic processing be learned for inverted faces? Cognition, 88(1):79-107 pdf icon (requires subscription to Cognition)

The origin of "special" processing for upright faces has been a matter of ongoing debate. If it is due to generic expertise, as opposed to having some innate component, holistic processing should be learnable for stimuli other than upright faces. Here we assess inverted faces. We trained subjects to discriminate identical twins using up to 1100 exposures to each twin in different poses and images. In the upright orientation, twin discrimination was supported by holistic processing. Removal of a single face feature had no effect on performance, and a composite effect (Young, A. W., Hellawell, D., & Hay, D.C. (1987). Configurational information in face perception. Perception 16 (6), 747-759) was obtained. In the inverted orientation, however, above chance identification ability relied on (a) image specific learning, or (b) tiny local feature differences not noticed in the upright faces. The failure to learn holistic processing for inverted faces indicates that, in contrast to the situation for objects (Tarr, M.J., & Pinker, S. (1989). Mental rotation and orientation-dependence in shape recognition. Cognitive Psychology 21 (2), 233-282), orientation specificity of face processing is highly stable against practice.

6. Elinor McKone & Rachel Robbins (2005). No face-like processing for objects-of-expertise in three behavioral tasks. CNS 2005 pdf icon

Abstract: The domain-general expertise hypothesis predicts that face-like processing should emerge for objects-of-expertise. We test Labrador Retriever experts (mean experience = 23 yrs) on three hehavioral tasks involving individual-level identification of faces and Labradors. Dog experts showed only a small inversion effect on memory for dogs, that was much less than for faces, and no greater than for dogs in novices. There was no contrast reversal effect for dogs in dog experts. There was also no composite effect, a direct test of face-like holistic/configural processing. Both tasks produced the standard signature phenomena for faces. In no case was there any correlation between size of effects for dogs and number of years' experience. A literature review shows these results are not as surprising as they might seem: Diamond & Carey's (1986) original large inversion effect in dog experts has never been replicated, and other tests of holistic/configural processing generally show negative results.'

7. Hans P Op de Beeck, Chris I Baker, Sandra Rindler, Nancy Kanwisher (2005). An Increased BOLD Response for Trained Objects in Object-selective Regions of Human Visual Cortex. Vision Sciences Society 2005

Abstract: Little is known about how intensive experience in discriminating exemplars from an object class changes the processing and representation of these objects in human visual cortex. Physiological studies in macaque visual cortex have reported effects of discrimination training on both the strength and the selectivity of neural responses. We investigated the effect of object discrimination training on the BOLD response in several regions of the human visual cortex. We trained 9 subjects during 10 daily sessions on the discrimination of shaded 3D objects from one of three novel object classes: smoothies spikies and cubies All subjects showed an improvement of discrimination ability over training. Subjects were scanned using an 8-channel coil in a 3T scanner before and after training. During scanning, subjects performed an orthogonal task, detecting changes in object color. A comparison of the pre- and post-training scan sessions revealed that training increased the BOLD response (p < .001) for the trained object class relative to the response for the untrained classes in object-selective voxels (LOC). No consistent training effects were seen in other visual regions such as the right fusiform face area and early visual cortex. The training effect was restricted to a subset of the voxels in LOC. The localization of the training effects seemed to depend on the selectivity of voxels before training and on subjectsÕstrategies. In conclusion, object discrimination training increases the BOLD response in object-selective regions of human visual cortex.

[This is the first object training study that does not restrict analyses to face-selective regions, and it provides a much more accurate and complete picture of how training changes the profile of response across visual cortex than any previous study.]

8. Erin M Harley, Whitney B Pope, Pablo Villablanca, Stephen A Engel (2005). Neural bases of perceptual expertise in radiologists. Vision Sciences Society 2005

Abstract: Previous work has established that visual processing in expert radiologists differs from processing in non-experts. To better understand the mechanisms of this expertise we used functional magnetic resonance imaging (fMRI) to compare neural activity in expert thoracic radiologists and beginning radiology residents as they detected abnormalities in chest radiographs. Subjects viewed intact and scrambled chest radiographs and indicated whether a lung nodule was present in a cued region of the image while BOLD fMRI data were acquired using a rapid event-related design. For each subject, regions of interest—V1, V2, V3, lateral occipital complex (LOC), fusiform face area (FFA), and radiograph-selective regions—were defined using separate localizer scans. Average event-related responses were computed using ordinary least squares from the fMRI time course averaged within each region. In experts regions in lateral occipital cortex showed higher fMRI response amplitudes for intact compared to scrambled radiographs. These radiograph-selective regions were not coincident with the FFA and were weaker or absent in residents.

9. Bradley C. Duchaine, Galit Yovel, Edward J. Butterworth, Ken Nakayama. (2006) Prosopagnosia as an impairment to face-specific mechanisms: Elimination of the alternative hypotheses in a developmental case. Cognitive Neuropsychology. pdf icon

New in press article

10. Rachel Robbins, Elinor McKone. (in press) No face-like processing for objects-of-expertise in three behavioural tasks. pdf icon

New in press article


Nancy Kanwisher

Elinor McKone

Brad Duchaine