Authors:

Rotshtein, P., Mattout, J., Friston, K. & HENSON, R.

Reference:

Year of publication:

2007

CBU number:

6506

Abstract:

Background. Cognitive1 and neuroscience2 models predict separate pathways for processing the identity and the expression of a face, with the former engaging the fusiform gyrus, and the latter the superior temporal sulcus2. The aim of this study was to test the above predictions. Importantly we used a constant set of face stimuli, and tested for dissociable processing by manipulating which property of faces (identity or expression) was task-relevant. Methods. 16 subjects participated in the fMRI experiment (1.5T). Face stimuli included six different identities each depicting the six basic expressions, plus six houses. Faces and houses were presented in separate blocks. Subjects performed a one-back task in each face block on the basis of either expression or identity; in house blocks based on identity. Data were analysed using SPM5. Consistent effects across subjects were tested using t-tests.Results.Behaviourally, reaction times responses for facial expression (1535msec±299s.e.m) and identity (1425±381) did not differ, but were faster for houses (1279±308) than for faces (t15=5.3,P<0.001). In the fMRI data, we observed dissociable responses (Figure A) during processing of facial expression compared with processing of identity in bilateral superior temporal sulcus (STS), bilateral inferior frontal sulcus/gyrus (IFS), all Z’s>3.19, P’s<0.001. Processing of facial identity was associated with larger responses in left inferior temporal gyrus (ITG) and medial orbital frontal cortex (OFC), all Z’s>2.67, P’s<0.005. In contrast, bilateral fusiform gyrus (FFG) and right inferior occipital gyrus (IOG, Figure 2) both showed larger responses for faces compared to houses (all Z’s>3.19, P’s<0.001), but these responses did not differ reliably according to the task- relevant facial aspect (i.e, expression versus identity; P’s>0.05). Discussion.Our results suggest that face identity and expression are processed using dissociable and common brain regions. Face processing in FFG and IOG were not affected by the task manipulation which may suggest that both expression and identity were processed in the FFG and IOG, or that these regions are mostly affected by the nature of the stimuli rather than task specific requirements. In contrast, bilateral STS, IFS were involved processing expressions, while left ITG and OFC were engaged in processing facial identity. These results concur with previous neuroimaging findings that suggest processing of both expression and identity in the IOG and FFG, but only expression in the STS3-5. The involvement of the left ITG in processing identity independent of expression accords with previous reports that this region encodes faces independent of their view6. Unlike previous studies however, we demonstrate that dissociations between expression and identity processing are also seen in frontal cortex, with IFS more engaged when processing expressions and medial OFC more engaged when processing identity. References1.V.Bruce and A.Young, Br. J. Psychol. 77, (1986).2. J.V. Haxby, et al., Trends Cogn. Sci. 4, (2000).3. J.S.Winston, et al., J. Neurophysiol. 92 (2004).4. T.Ganel, et al., Neuropsychologia 43 (2005).5. S.L.Fairhall and A.Ishai, Cereb. Cortex (2006).6. E. Eger, et al., Neuroimage, 26 (2005).