Automatic Detection Of Configural Processing Of Faces Under Non-attentional Condition

Faces are socially very important visual objects and the detection of a change in facesis an essential evolutionary skill. Indeed, face recognition is a well documented domain ofhuman perceptual expertise and adults are experts at recognizing faces. It has been shownthat adults expertise in recognizing faces is attributed to configural processing (i.e.,processing not just the shapes of individual features but also the relations among them),whereas the object identification is mainly based on featural processing. The faceinversion effect (FIE), indeed, has been taken as diagnostic of configural processing. Alarge body of evidence of face specificity in visual processing initially comes from studiesbased on the face inversion effect (FIE). Stimulus rotation by180°in the picture planehas been observed to impair the processing of faces to a larger extent than other non-facialstimuli suggesting face perception mainly depends on configural and holistic processingstrongly affected by picture-plane inversion. However, non-facial object perceptiontypically involves feature-based analytic processing. Therefore, it is not surprising thatconsiderable efforts have been invested to investigate how the brain processes faceperception, which the studies were based on behavioral level or neural mechanisms underattention condition, however, it is that the lack of the consideration how the brainprocesses face perception under non-attentional condition. Recently, several studies examined the influence of attention on configural processing of faces, however, therelationship between attention and configural processing failed to produce consistentresults. Actually, it is unclear whether facial configural processing is dependent onattentional resources. In other words, whether the facial configural processing could occurat the pre-attentive stage under non-attentional condition is still open. Convergingevidence revealed that the MMN is a reliable indicator for evaluating the automaticchange detection ability of human beings, to address the question of whether the humanbrain can automatically detect the configual changes of faces under non-attentionalcondition, in the present study, we will investigate whether configural computation offaces versus non-face objects automatically occurs under non-attentional condition byrecoding and analyzing the mismatch negativity (MMN) of event-related potentials(ERPs). The difference between face-orientation and object-orientation MMN would helpus to understand the mechanism of configural processing effects on non-attentionalcondition, which could provide the electrophysiological evidence. The present study wasdivided into five parts. The first part, we investigated the visual MMNs by elicited bydeviant orientation for schematic faces and schematic non-face objects(arrows, houses)in Oddball paradigm to answer the processing mechanism between schematic face andnon-face under non-attentional condition. The second part, we further focused on how thebrain process the face and non-face objects automatically under non-attentional conditionby recording deviant-standard-reverse Oddball paradigm visual MMNs elicited by deviantorientation for real faces and houses, respectively. The third part, the aim of present studywas to how process the brain the spacing of the features about face’s configuralinformation automatically under non-attentional condition by recording Oddball paradigmvisual MMNs elicited by deviant orientation schematic faces and non-face object(scrambled faces and face contour), respectively. In the fourth part, the aim of presentstudy was to use real face and non-face (houses) at three different orientations (90°,180°and270°) rather than contrasting only upright and inverted faces as in previous studies,which was to employ the equal probable paradigm to verify memory-comparison-basedexplanations of MMN elicited by deviant orientation real face and non-face (houses). The difference between real face-orientation and object-orientation MMN would help us tounderstand that the inversion affects face processing qualitatively or quantitatively onnon-attentional condition. In the fifth part, the aim of the present study was to collectempirical evidence on the differential processing of own-race faces and other-race faces atthe perceptual level and investigate whether the second order relation configuralcomputation of own-race face and other-race faces automatically occurs undernon-attentional condition. To achieve this objective, electrophysiological data vMMNmodulation were recorded in Chinese participants performing task with Caucasian andChinese faces, which configural changes were made by simultaneously altering (reducing)the spacing between the eyes.The main results of present study were drawn as follows:First, in the present study, visual mismatch negativity (vMMN) elicited by deviantorientation (90°/270°vs.0°) for schematic faces and schematic non-face objects (arrows,houses), respectively, were investigated. We found compared to standard stimuli (0°orientation), the orientation changes elicited posterior visual MMNs for all deviant stimuli.vMMN amplitudes were similar between houses and arrows except the significant delayedpeak latency for houses. Importantly, compared with houses and arrows, vMMN elicitedby the orientation change of faces was significant decreased and delayed. However, thepresent source analysis supports the view that face-MMN reflects differential neutralmechanism compared to non-face-MMN.Second, as predicted, the visual mismatch negativity (vMMN) elicited by deviantorientation (90°/270°vs.0°) for real faces and houses, respectively, were recorded usingthe deviant-standard-reversed paradigm. We found that compared to standard stimuli, theorientation changes elicited posterior visual MMNs for deviant stimuli. Compared withvMMNs in response to houses, the vMMN elicited by the orientation change of faces wassignificant enhanced with shorter latency. The face MMN was larger for the orientatedthan upright conditions but the orientation did not modulate the amplitudes of houseMMN.Third，the visual mismatch negativity (vMMN) elicited by deviant orientation forschematic faces and non-face object (scrambled faces and face contour), respectively, were recorded using the Oddball paradigm. Our ERP findings showed that compared withvMMNs in response to non-face (scrambled faces and face contour), the vMMN elicitedby the orientation change of faces was significant enhanced with shorter latency, and theamplitudes of face contour and scrambled faces were no significant differences.Furthermore, using Control-MMN paradigm, ERPs were recorded in response to realfaces and houses presented at three different orientations from90°,180°and270°ratherthan contrasting only upright and inverted faces as in previous studies. We found thatcompare with the vMMN elicited by180°orientation, the vMMN elicited by theorientation change (90°and270°) whether faces or houses were significant enhanced withshorter latency. Compared with vMMNs in response to houses, the vMMN elicited by theorientation change of faces was significant enhanced with shorter latency. Mostimportantly, our ERP findings showed that multiple face orientations affected the N170and vMMN component by modulating its latency and amplitude in a non-linear fashion,which the vMMN using Control-MMN ERPs elicited by the90°orientation changewhether faces or houses were significant enhanced with shorter latency. We also found thatface MMN was larger on the right than left occipital-temporal regions.Finally, we investigated whether the second order relation configural computation ofown-race face and other-race faces automatically occurs under non-attentional condition.Our ERP findings showed that the N170component by modulating its latency andamplitude are in the same pattern irrespective of original face or faces with configuralinformation changes. Besides, compare with original faces, faces with configuralinformation changes elicited larger P200amplitudes by Caucasian faces. It is worthnoticing that own-race faces elicited larger vMMN amplitudes than other-race faces.The main conclusions of this study are as follows:The present main findings indicated that the mismatch coding of face individuationwas automatically engaged under unattended condition and they based on the differentialneutral mechanism underlying face MMN versus objects MMN, the differences are mainlyoriginated from the adults are experts at subordinate-level face recognition with aperceptual strategy of configural processing. Then, the brain is more sensitive to the first-order configural information by manipulated shape of the stimuli with non-facestimuli (scrambled faces and face contour) versus faces. Besides, it is worth noticing thatusing the equal probable paradigm was to verify memory-comparison-based explanationsof MMN elicited by deviant orientation real face and non-face (houses), which90°orientation change on face the most serious interfere with configural information encoding.And the inversion affects face processing qualitatively on non-attentional condition.Finally，the second-order relation configural computation of own-race face and other-racefaces automatically occurs under non-attentional condition, which second-order relationcan be independent on attention. Importantly, the lager vMMN elicited by own-race facesindicated configural processing is reduced for faces of other racial groups.In summary, the detection of a change in faces is an essential evolutionary skill，as acentral source of information in social interactions, faces are considered special amongvisual categories, therefore, automatic detection of configural processing of faces undernon-attentional condition deem particularly important. In the present study, we investigatewhether configural computation of faces automatically occurs under non-attentionalcondition by recoding and analyzing the mismatch negativity (MMN) of event-relatedpotentials (ERPs) and here we directly assessed adults sensitivity to configural componentchanges in own-race and other-race faces. The mismatch coding of face was automaticallyengaged under unattended condition and they based on the differential neutral mechanismunderlying face MMN versus objects MMN, and second-order relation can be independenton attention, in addition configural processing is reduced for faces of other racial groups.These data provided electrophysiological evidence for automatic detection of configuralchanges of faces under unattended conditions.