Review: A Face-Specific N170?-The Controversy Goes On

Event Related Potential studies of face processing Event Related Potential (ERP) studies on face recognition have isolated an early occipito-temporal negative component (N1) which is often associated with visual processing and encoding (see Bentin et al, 1996; Carmel & Bentin, 2002; Rousselet et al 2004; Rousselet et al 2005). This component falls within the low alpha range (7-8Hz), and has an early pick of 150-170ms after stimulus onset in visual tasks involving face processing. Contrary to recognition of non-facial objects, face detection produces a negative deflection in the N1 (see figure 1) and an early pick of 150-170 ms in the occipito-temporal electrodes after stimulus onset (Rousselet et al, 2004). Thus, when studied in connection with face recognition the N1 has often been termed 'N170' and it is found to be largely distributed at the posterior- inferior region of the temporal lobe and greater in the right than in left cerebral hemisphere (Bentin et al, 1996; Jeffreys, 1996; Kanwisher et al, 1997; Carmel & Bentin, 2002). It is, therefore, widely accepted that N170 responds preferentially to human faces as opposed to other objects (see Carmel & Bentin, 2002; Rousselet et al 2004; Itier & Taylor, 2004; Rousselet et al 2005 Bentin et al, 2007). [FIGURE 1 OMITTED] Figure 1 show ERPs elicited in one of the experiments conducted by Bentin et al (1996). Evoked potentials for faces and cars differed in latency range at locations T5 and T6. ERPs for both faces, scrambled faces and scrambled cars also differed in the 250-500ms latency range. Although not reaching statistical significance (p=0.3), the N170 for faces was more negative over the right (-4.07 [micro]V) than the left hemisphere (-3.02 [micro]V). Figure 1 also shows that the amplitude of the N170 at the left (T5) and right (T6) hemispheres is significantly larger for faces than for scrambled faces. These results show that face sensitive ERPs can be recorded with scalp electrodes, and also provide strong supportive evidence for a face-specific N170. Controversies surrounding the notion of a face-specific N170 The N170 specificity to human faces has become paradigmatic for much debate, as some argued that this component is not exclusively sensitive to faces (Rousselet et al, 2004; Rousselet et al, 2005; Thierry et al, 2007a; Thierry et al, 2007b). A number of researchers asserted that this component is a product of complex and interrelated conceptual and perceptual factors. For instance, Thierry et al (2007a) proposed that, rather than faces, the N170 is sensitive to Interstimulus Perceptual Variance (ISPV). Their main thesis was that this component has not been controlled for variability across stimuli categories. Thierry et al asserted that in most experimental conditions faces are always presented in full front view and, therefore, they are less susceptible to perceptual variance. In both high and low ISPV conditions they claim to have found that the N170 did not occur within the usual 170ms after stimulus onset, but 70ms earlier. In this way, Thierry et al argued that differences found between experimental conditions do not reflect categorical inequalities, as these may arise as a consequence of ISPV. However, Thierry et al results attracted critical attention from Bentin et al (2007) who demonstrated that the amplitude of the N170 does not vary linearly with the effect of ISPV. They also asserted that although the ISPV for cars and human faces are often identical, the N170 tend to be larger for faces than for cars. Furthermore, Thierry et al assertions have also been criticised on the grounds of not providing any theoretical explanation as to why the N170 was larger in amplitude low ISPV than for high ISPV (for a detailed discussion please see Bentin et al, 2007). Additionally, the N170 insensitivity to human faces found by Thierry et al can be explained through the way in which data was averaged across the different sites of participants scalp. …