Cross-category Face Adaptation Of Feature Association

Previous studies considered that there is face adaptation only when adaptor isface. When adaptor and test face are not belong to the same category (for example,adaptor is house,hand,animal), there is no face adaptation effect. That means faceadaptation is Category-selective. These studies suggested that the adaptation occurredin the level of sensory processing. That means face adaptation aftereffect issharp-selective adaptation. We want to prove that high-level perceptual adaptation canoccur when the property or features being adapted are automatically inferred ratherthan perceived in the adaptor. The Requirement of generating cross-categoryadaptation is adaptor and test face has correlation properties on the level of task. Faceadaptation is not only shape-selective, adaptation can also occur in the level meaningprocessing which have associated properties on the level of task.This study attempts to verify the existence of cross-category face adaptation,occurrence conditions of cross-category face adaptation, the perceptual results ofcross-category face adaptation and the inner mechanism of cross-category faceadaptation.Experiment1explored different duration of cross-category face adaptation, usingthe pictures of the objects which have gender feature as adaptors, While behavioraland electrophysiological aftereffect as the indicators of cross-category face adaptation.The difference between baseline and50ms was not significant; there is no adaptationeffect for50ms. The results showed that when duration of adaptor is among100ms to3200ms, the magnitude of behavioral and electrophysiological aftereffect wasinverted U-shaped; the magnitude of adaptation effect was greatest for400ms.The Speculate reason of cross-category face adaptation is that adaptation and teststimulus have associations on "gender" feature. With the sexual orientation of theobject makes the observer adapted to the gender of the general, abstract representation,and then make the subsequent presentation of faces produced a perception of bias. Todemonstrate a potential mechanism of cross-category face adaptation, in Experiment2, we take pictures with the sexual orientation of items, corresponding name of thepictures, the text features (male, female) as a stimulus to adapt, to explore whether theadaptation of the "gender" feature can Generally produce cross-category faceadaptation. Adaptation effects of picture are most significant, followed by the featurename, and finally the picture name.Experiment3validated the feature correlation strength impacts on cross-categoryadaptation. By controlling the gender correlation strength between adaptor and teststimulus (high, low, no, baseline), we explored the influence of correlation strength.The results showed that the magnitude of adaptation effect was greater for highcorrelation than low correlation. The difference between baseline and no correlationcondition was not significant.Experiment4, by varying attention load degrees on adaptor (high load, lowload, no load, baseline), we explore whether awareness of the adaptor impactscross-category face adaptation. These showed that the magnitude of adaptation effectwas greater for no attention load than low and high attention load. The differencebetween baseline and high attention load condition was not significant.We performed Experiment5to investigate if face gender adaptation could affectgender discrimination. The result revealed that adapting to a male/female face couldselectively enhance discrimination for contrast face. When the gender of adaptor andcontrast face is accordance, Cross-category adaptation could enhance discriminationfor contrast face. These results provide further evidence suggesting that, similar tolow-level vision, adaptation in high-level vision could calibrate the visual system tocurrent inputs of complex shapes(i.e. face)and improve discrimination at the adaptedcharacteristic.In summary, this study has proved high-level perceptual adaptation can occurwhen the property or features being adapted are automatically inferred rather thanperceived in the adaptor. The Requirement of generating cross-category adaptation isadaptor and test face has correlation properties (in this study, it was gender) on thelevel of task. Cross-category adaptation was affected by the form of adaptor,correlation intensity between adaptor and test stimulus, the attention load of adaptor, and so on. Cross-category adaptation could selectively enhance discrimination formale/female faces. When the gender of adaptor and contrast face is accordance,Cross-category adaptation could enhance discrimination for contrast face.