| Face recognition is an important cognitive activity. But its mechanisms and processes are still not clear. To clarify the mechanisms of face recognition is not only important for theoretical psychology and physiology, but also necessary for understanding the pathogenesis of diseases such as prosopagnosia and name anomia, as well as developing feasible therapies for them.In this study, we combine behavioral performance, event-related potential (ERP), ERP topography and dipole localization method (DLM) to investigate the spatial-temporal characteristics of face perception and recognition, and the underlying mechanisms of them as well. Our target problems include perceptive encoding of face, face specificity, mechanisms of the face inversion effect (FIE), retrieval of face and relative information and implicit processing of face,and so on.Methods and results:1. In order to identify the differences in face and other-object recognition, behavioral and ERP studies were performed with the "learn-test" paradigm. During the "learn" stage, subjects were required to study the upright pictures of human faces, dogs and mobile phones. And then subjects were required to recognize the learned pictures of human faces, dogs and mobile phones whether upright or inverted in the "test" stage. The results showed that recognition of inverted human faces had lower accuracy and longer RT than those of other objects. The P170 and N170 of faces were the larger compared to those of non-faces. And the inverted human face significantly increased P100, P170 and N170 amplitudes and delayed the N170 and P100 latencies. The old-new effect of the ERPs was least in human faces, even disappeared in inversion. Moreover, P170 and N170 of faces were generated from different sources with those of non-faces. 2. To understand the contribution of different facial features to FIE, the faces were modified to reflect the local, relative or holistic features of faces. Then the accumulative values of the degrees of the modification were evaluated quantitatively for both upright and inverted faces. It was found that the inversion effects was the largest for background features, the second grade relation features and local-exchanged features, but the lowest for local features, extra features and the first grade relation features. Besides, the ERPs of normal faces were not affected by physiognomy. 3. The faces were divided into 15 groups based on the local, relative and holistic features to compare the encoding and retrieval processes of individual facial features. The results showed that ERP components, e.g. N100, P100, P170 and N170 were sensitive to the complicity of facial features. Besides, the accuracy for retrieval of faces with external features was the highest, followed by those with internal features, second grade relation features and local facial features, while it was the lowest for those with contour facial features. At parietal-occipital regions, the LPC amplitudes for the faces with external features were also the largest. 4. The spatial-temporal characteristics of the ERPs evoked by famous faces and by "learned" (familiar) faces were compared. It turned out that the stimulus categories significantly influenced LPC, with that elicited by famous faces with correct name retrieval the biggest, that by "learned" familiar faces and famous faces without retrieval of names smaller, and that by new faces the smallest. The topographies showed that LPC amplitudes of the learned familiar faces were more positive in left centroparietal and prefrontal; the identified famous faces without name retrieval more positive at right centrofrontal regions, and the named famous faces more positive at right centroparietal and prefrontal regions.5. When the stimuli were presented promptly (150ms) and attention was diverted away to search the target "f " in a serial letters below the stimuli, the N170 and P170 of faces were enlarged and forwarded compared to those of non-faces. ERP components were not affected by the orientations and races of f... |
PDF Full Text Request