Cognitive Study Of Facial Expression Perceptual Learning

Perceptual learning is a phenomenon that intensive training on a perceptual task may lead to a large improvement on the task performance. Previous studies found that quite a lot perceptual abilities are susceptible to training, including discriminating contrast,orientation,motion direction,texture and so on。Perceptual learning could not only improve elementary feature discrimination, but also improve the recognition of complex stimuli such as object and face。For example,previous studies have depicted the characteristics of perceptual learning of face identity recognition and face view discrimination. However,the characteristics of perceptual learning of facial expression have not been investigated. It is a critical living ability to recognize the facial expression of others in the modern society. People seem to be very sensitive to facial expressions. it is quite unclear whether we could be trained to attain an extraordinary sensibility to facial expression detection and discrimination. In the current study, we used four experiment to solve this questions.Experiment 1. Each subject underwent eight daily training sessions to discriminate facial expressions around the fully happy or fully sad expression. Half of the subjects were trained to discriminate happy faces. The other half were trained with sad faces. Subjects were asked to make a two-alternative-forced-choice（2-AFC） judgment of the expression strength of the second face relative to the first one（stronger or weaker）. Before, immediately after, and one month after the 8-day training, we tested subjects’ discrimination performance. With eight days of training, subjects’ discrimination performance improved significantly and this improvement generalized to faces with the same expression but different gender to the trained face.Experiment 2. we first matched the expression strengths of male and female faces for each subject, and then trained and tested them in the same way as Experiment 1. Specifically, before training started, each subject participated in a study of matching facial expressions with the method of constant stimuli. Subjects were instructed to report which face was happier. Then, we calculate the point of subject equality（PSE） for either the male reference face or the female reference face. Our results in this experiment revealed that, with facial expression intensity carefully matched between female and male faces, the transfer of learning effect became nearly complete. This result further implicated that perceptual learning of facial expression was almost unaffected by factors unrelated to facial expression.Experiment 3. In this experiment, we aimed to study whether the learning effect could transfer from the trained strength of expression to a higher or lower strength of expression. Half of the subjects were trained with expression strength of happy₁00 face and the other half were trained with happy₅0 face. All the training procedure was the same as Experiment 1. Our results indicated that the learning effect of facial expression could partially transfer to a face with lower expression intensity but not a face with higher intensity.Experiment 4. In this experiment, we further investigated whether the learning effect of a happy face could transfer to a neutral face. We trained subjects with female happy₅0 face and tested them with female happy₁00 face, female happy₅0 face and female neutral face. All the procedure was similar to Experiment 1 except the neutral face trials. In a neutral face trial, a female neutral face and a female sad_θ face were presented. In order to keep the task and response consistent across different conditions, subjects were still instructed to indicate which face was happier. In this experiment, we further demonstrated that the learning effect of facial expression discrimination could transfer to a neutral face. We could therefore conclude that the transfer of facial expression learning was asymmetry between high intensity and low intensity expressions to a large extent.In all the experiments, we not only measured the performance immediately after training, but also the performance one month after training. Results showed that all the learning effect and transfer could persist at least one month, which implied that these findings revealed long-term mechanisms of training. In a series of experiments, we systematically studied the characteristics of perceptual learning of facial expression discrimination. Our results demonstrated that this learning was specific to the learned expression and could persist up to one month after training.