关键词: 交互进化计算, 社交焦虑, 情绪识别, 心理健康测量

Abstract: Previous literatures demonstrate that clinical anxious patients are very sensitive to threatening stimuli, and further, it is easier for them to be engaged in threatening stimuli, including the negative pictures, images and faces. In the present article, we use the faces as the experiment material, which are showed in ways different to the previous researches. This study tries to explore the applicability of this method to anxiety measurement.
In our research, two groups of subjects are involved. The first group includes members that are requested to choose and assess the CG pictures showed on the computer. And the whole process of assessment is carried out based on a new computer technology IEC, (interactive evolutionary computation). IEC is a method for optimizing target systems based on human evaluation criteria, which is broadly used in other fields including facial image generation, speech processing, etc. However, the application of IEC in psychology has not been widely used.
And we have five social anxiety subjects with high scores in IAS and also four ones with low scores in the first group of subjects. IAS is broadly used in measuring the level of social anxiety, which is of good validity and credulity. The social anxiety subjects with high scores in IAS are regarded as experimental subjects and the latter ones are regarded as the control subjects. Other 70 subjects (the second group) in the same university are chosen to be the second group to evaluate the CG images that have been chosen by the first group with Scheff’s method of paired comparison. That is how we have obtained the final psychological scale.
The first step of this experiment has two dimensions: “happy” and “fear”. The subjects of the first group choose and assess the CG images showed on the computer by IEC using five-level rating scale. And then the second group is required to evaluate all the images that have been chosen by the first group. They also used the five-level rating scale to estimate the degree of comparability of all the images. And then we analyze the variance of these estimate data. Statistical tests of the evaluations show: (1) the range of emotional impressions perceived by the five experimental subjects that have high social anxiety between happy-fear is significantly narrower than the control ones (p<0.001). (2) IEC has the potential for the measurement of mental health, especially in the emotional recognition.
It can be concluded that IEC is a new and advanced technology that can be used in the research of psychology. Times’ changing, traditional scales with paper and pen cannot satisfy the need of so many people’s desire for knowing the level of mental health of themselves. This provides an opportunity to apply this computer technology to psychology.

Key words: interactive evolutionary computation, social anxiety, emotional recognition, mental health measurement