Eye Movement Characteristics Of Conscious Process In The Involved Self-report Scale Of Depression Patients

Background:Effectively preventing patients with mental disorders from entering the army and keeping the quality of soldiers at the source is an important measure to enhance the combat effectiveness of the army.At present,the international screening methods for mental disorders mainly rely on self-report scales guided by psychological test theory.However,due to inherent shortcomings such as subjectivity and score sensitivity of self-report scales,the personnel psychological selection in this way has gradually entered the bottleneck.With the improvement of measurement quality,technical changes to traditional psychological tests have become the actual needs and inevitable trends of military personnel selection.The conscious process of people answer the scale provides extremely rich and valuable information,but it has not been valued for a long time.So far,the scores of the self-report scale are still the only source of information for psychological tests.This research group proposed the theory named Neuro-Stimulus Fusion To Consciousness(NSFC)in 2014,which aims to use multiple cognitive neural technologies to perform fusion recognition on the conscious process when people answer the questionnaire,so as to achieving an objective and refined evaluation of subjective consciousness activities.The concise and abstract language description of traditional self-report scales is not conducive to inducing differences in conscious process among different groups of people.Therefore,exploring more optimized forms of scale stimulus is the focus of NSFC research.Depression,as one of the most common mental disorders,seriously affects the combat effectiveness of troops.This study base on the NSFC,taking depression as an example.We use the natural corpus of patients to innovate the form of language stimulation of traditional scales,and proceed from eye movement technology to explore a more specific implementation method for identifying mental disorders based on the self-report scale conscious process.This research provides theoretical and practical basis for the development of the next generation of recruiting psychological selection system.Objects and methods:The subjects included the depression patients group and the normal control group,which were divided into four parts.The first part of the study is the analysis of the difference between the conscious process of the depression and normal people in the involved scale.It aims to innovate the traditional form of stimulus and analyze the difference in the overall conscious process of the two groups.The study constructed a corpus of depression patients through a web crawler.The traditional scale was adapted using the patient’s language to form the involved scale.The study collected eye movement data of depression patients and normal people when answering the scale.A linear mixed model was used to compare the differences in conscious process between the two groups in the traditional scale and the involved scale.The second part of the study is the construction of depression patients’ recognition model in the conscious process of the involved scale,aiming to identify depression patients through the overall conscious process of the scale response.The study extracted35 eye movement features from three types of indicators: gaze,saccade,and pupil.Through four algorithms,which is logistic regression,naive Bayes,support vector machine,and random forest,to build a recognition model for depression patients.The model performance of the two scales was compared.Using the single response time feature,the recognition models of depression patients were constructed for the two scales respectively.The differences of recognition accuracy between the eye movement model and the response time model were compared.The third part of the study is the classification model construction and prediction analysis of eye movement response and the involved scale options,aiming to analyze the qualitative and quantitative correlation between eye movement response and response options.Research has developed a visualization tool—behavioral spatial spectrum,which fuses stimulus content with cognitive neural indicators,and analyzes the general relationship between different response options and eye movement response.Using the extracted eye movement features,a four-category model was constructed for the involved scale by logistic regression algorithm,and the classification accuracy rate of eye movements for the actual answer options of the subjects was analyzed.The fourth part of the study is the influence of involvement on eye movement response under negative language stimulation in depressed patients,aiming to exclude the influence of reading factors in the conscious process of involved scale.The study set two different negative materials,which were depressive materials and non-depressive materials.The eye movement data of two materials for depression patients and normal people were collected,and a linear mixed model was used to compare the differences between the two groups.Results:1.The scores of depression patients and normal people on the involved scale were significantly different(t=13.04,P<0.001).The spearman correlation coefficient between the scores of the traditional scale and the involved scale was 0.911(P<0.001).2.For the number of fixation in statement area and the number of regressive saccade in the statement area,the interaction between group and stimulus was significant(F=38.90,P<0.001;F=7.55,P<0.01).Compared with normal people,depression patients had more number of fixation in statement area(t=4.32,P<0.001)and number of regressive saccade in the statement area(t=2.84,P<0.01).There were no significant differences in the number of fixation(t=0.83,P=0.41)and the number of regressive saccade(t=1.68,P=0.10)on the traditional scale.3.For the relative number of regressive saccade in option area and the number of interactive saccades between statement and options,the interaction between groups and stimuli was significant(F=5.17,P<0.05;F=7.49,P<0.01).Compared to traditional scale,normal people have fewer relative number of regressive saccade in option area(t=3.59,P<0.001)and number of interactive saccades between statement and options(t=2.23,P<0.005).There were no significant differences in the relative number of regressive saccade(t= 0.34,P=0.73)and the number of interactive saccades(t=-1.64,P=0.11)for patients in the two scales.4.For the pupil diameter,there were no significant differences in the group main effect(F=0.40,P=0.53),stimulation main effect(F=1.12,P=0.29),and interaction between groups and stimuli(F=0.37,P=0.54).On the change of pupil diameter of the two scales,the depression patients were significantly larger than the normal people(t=2.39,P<0.05).5.Based on eye movement data,the recognition accuracy of the involved scale in four classifiers of logistic regression,naive Bayes,support vector machine and random forest was 91.80%,80.33%,81.97% and 80.33%,and the traditional scale was 75.41%,77.05%,67.21% and 77.05%.The highest prediction accuracy of the involved scale was14.75% higher than that of the traditional scale,and the average prediction accuracy was9.43% higher.6.Based on the single feature of response time,the recognition accuracy of the involved scale in the four classifiers of logistic regression,naive Bayes,support vector machine and random forest was 59.02%,67.21%,49.18% and 60.66%,and the traditional scale was 77.05%,73.77%,65.57% and 75.41%.The average prediction accuracy of the traditional scale of eye movement model was about 1.2% higher than that of the response time model,and the average prediction accuracy of the involved scale of eye movement model was about 25% higher than that of the response time model.7.Qualitative analysis of behavioral spatial spectrum shows that there are two regular connections between the eye movement response and the response options of the subjects.One is that with the change of response options from D to A,the intensity of the eye movement response gradually increases and the range of the eye movement response gradually increases.The other is that the intensity and range of eye movement response in response to B and C are greater than that in response to A and D.8.Based on the characteristics of eye movements,the predictive accuracy of the four types of answering options of A,B,C,and D under the logistic regression classifier were64%,37%,33%,and 27%.The average prediction accuracy of the model is 40.25%,which is 15.25% higher than the random prediction accuracy.9.During the reading task,the fixation duration(t=2.60,P<0.05),the number of fixation(t=2.87,P<0.01),number of regressive saccade(t=2.87,P<0.01)and saccadic amplitude(t=3.41,P<0.01)of depression patients on the non-depressive materials were greater than that of normal people.The fixation duration(t=1.29,P=0.20),number of fixation(t=1.58,P=0.12),number of regressive saccade(t=1.44,P=0.16),and saccade distance(t=1.70,P=0.10)on depressive materials had no significant difference from normal people.Conclusion:1.The difference between depression patients and normal people not only reflected in the results of the answers,but also in the conscious process.Compared with the traditional scale,the involved scale can better induce the difference in the conscious process between depressed patients and normal people.Specifically,it triggers the deeper level of cognitive process and greater degree of emotional response of depressive patients,while reducing the hesitation and thinking degree of normal people when answering questions.2.Based on the eye movement characteristics of the overall conscious process during the scale response,the involved scale can better identify patients with depression,and the model performance is better than the traditional scale.It has initially proved the possibility of introducing eye movement technology into the psychological test.3.The subject’s cognitive and emotional responses induced by the question can be represented through eye movement,and there is a general qualitative connection between the subject’s eye movement responses and the actual answer options.Based on eye movement behavior,the respondents’ response options can be predicted at a certain level.4.The eye movement difference between depression and normal people in text reading is not the main source of the difference of two groups in the involved scale.Depression patients showed more attention under the negative language stimulation of low involvement than normal control,but there was no significant difference under the negative language stimulation of high involvement.