A human parietal face area contains alignedhead-centered visual and tactile mapsMartin I Sereno1& Ruey-Song Huang1,2Visually guided eating, biting and kissing, and avoiding objects moving toward the face and toward which the face moves requireprompt, coordinated processing of spatial visual and somatosensory information in order to protect the face and the brain. Single-cell recordings in parietal cortex have identified multisensory neurons with spatially restricted, aligned visual and somatosensoryreceptive fields, but so far, there has been no evidence for a topographic map in this area. Here we mapped the organization of amultisensory parietal face area in humans by acquiring functional magnetic resonance images while varying the polar angle offacial air puffs and close-up visual stimuli. We found aligned maps of tactile and near-face visual stimuli at the highest level ofhuman association cortex—namely, in the superior part of the postcentral sulcus. We show that this area may code the location ofvisual stimuli with respect to the face, not with respect to the retina.In macaque monkeys, neurons in a multisensory parietal area at theborder between visual cortex and somatosensory cortex—the ventralintraparietal area, VIP—have spatially restricted visual and somato-sensory receptive fields that are aligned with each other1. For example, aVIP neuron with a visual receptive field in the upper right part of thevisual field will also typically have a somatosensory receptive fieldlocated on the upper right part of the forehead. VIP neurons respondselectively to optical flow stimuli2,3and have connections with visualmotion areas, somatosensory areas and motor areas controlling faceand eye movements4,5. Electrical stimulation of VIP results in defensivemovements including flinching6.The mobility of the eyes with respect to the face and head, however,could potentially misalign somatosensory and visual inputs7,8.Thereisevidence that some VIP neurons remap their visual receptive fields tocancel the effects of eye movements1,9, by using information about eyeposition. Thus, if the monkey moves its eyes downward so that thevisual stimulus moves farther into the upper visual field with respect tothe eye, the visual receptive field will nevertheless maintain its align-ment with the upper right forehead rather than moving downwardwith the eye. These VIP neurons thus represent both visual andsomatosensory information in a somatosensory coordinate system.This is in contrast to the nearby lateral intraparietal area (LIP), whichrepresents and updates potential visual targets—including those initi-ally detected by way of another modality—in a visual (retinotopic)coordinate system10,11.Previous functional magnetic resonance imaging (fMRI) studies ofmultisensory processing in humans12,13have shown using conjunctionanalysis that there is a small region in parietal cortex that responds tosomatosensory, visual and auditory stimuli just as monkey VIP does. Inboth humans and monkeys, however, there has been no evidence thatthis region contains retinotopic or somatotopic maps of near-facespace. Earlier studies of LIP in humans, which is situated just posteriorto the multisensory focus, showed that there, potential visual targets arein fact represented in a retinotopic cortical map14–16. This study set outto determine if human VIP, too, contains a topographic map—but inthis case, a multisensory map in which visual space is superimposedand aligned with a somatosensory map.RESULTSSubjects (n ¼ 12) were posed in the scanner with their heads tiltedslightly forward so that they could directly view (without a mirror) awide-field visual stimulus projected onto a translucent back-projectionscreen located very close to the face. The tilt also made it possible todeliver gentle, computer-controlled air puffs to 12 approximatelyequally spaced locations around the face, through adjustable plasticnozzles (see Fig. 1; the near-face back-projection screen has beenremoved in Fig. 1g). Subjects wore ear plugs, and white noise wasdelivered via magnetic resonance (MR)-compatible headphones inorder to completely mask the sound of the air puffs. Data were analyzedusing surface-based Fourier methods17.Identification of a new parietal face areaFirst, we used four different block-design stimulus protocols(Fig. 1a–d) to identify multisensory areas of interest (Fig. 2a)andtomake a connection with previous studies. The first of these, ‘Whole faceair puffs versus OFF’ (Fig. 2b, data shown on dorsolateral view ofinflated surface), compared 16-s periods in which air puffs weredelivered to random locations on the face to 16-s periods of nothing.The experiment was conducted in the dark and subjects kept their eyesclosed. As expected, this strongly activated face primary somatosensory© 2006 Nature Publishing Group http://www.nature.com/natureneuroscienceReceived 11 August; accepted 1 September; published online 24 September 2006; doi:10.1038/nn17771Department of Cognitive Science, and2Swartz Center for Computational Neuroscience, Institute for Neural Computation, University of California San Diego, La Jolla,California 92093, USA. Correspondence should be addressed to M.I.S. ([email protected]).NATURE NEUROSCIENCE ADVANCE ONLINE PUBLICATION 1ARTICLEScortex (areas 3b and 1) on the posterior bank of the central sulcus as wellas secondary somatosensory cortex (areas S-II and PV, not visible in thisview) on the upper bank of the lateral sulcus. Another area in superiorparietal cortex (dotted red circles in Fig. 2b) was bilaterally activated asstrongly (3–5% peak-to-peak signal amplitude) as primary somatosen-sory cortex. This area was located at the confluence of the postcentraland intraparietal sulci, near ‘‘region 1’’ in ref. 12. Very similar resultswere obtained when subjects maintained central fixation on an other-wise blank screen instead of keeping their eyes closed (data not shown).In a second block-design experiment, we compared air puffsdelivered to random locations on the right versus the left half of theface (again in the dark). We found similarly strong activations inprimary and secondary somatosensory cortex, but also in the superiorparietal focus. A number of regions were significantly more stronglyactivated by stimulation of the contralateral face than the ipsilateral face(Fig. 2c), and no region was more strongly activated by stimulation ofthe ipsilateral face. This result is expected in