Cognitive Load Assessment Study Based On Eye-Tracking Data Analysis

Cognitive load refers to the amount of cognitive resources expended by the brain during the processing of information and decision-making in a specific task.Due to its significant impact on an individual’s task performance or skill acquisition efficiency,the development of scientifically valid methods for measuring and evaluating cognitive load is essential in fields such as education,healthcare,aerospace,and other related domains.The existing methods for measuring cognitive load mainly suffer from low generalization,poor timeliness,and susceptibility to subjective influences.Additionally,some physiological data measurement methods involve invasive sampling equipment,which may negatively impact the completion of relevant tasks by experimental subjects.In contrast,eyetracking devices offer the advantage of non-contact data collection and hold a higher practical value.Therefore,this study has chosen eye movement data as the physiological data source for quantifying cognitive load and conducted research in the following areas:Theoretical Research section provides a comprehensive review of relevant literature,focusing on two main areas of research:(1)eye movement indicators research,which integrates existing evaluation methods and conclusions regarding eye movement indicators related to cognitive load;(2)cognitive load modeling techniques,which involve research on using eye movement data to model cognitive load,and systematically summarize the experimental designs,feature indicators,and modeling techniques of existing studies.Data Processing and Eye-movement Feature Set section primarily involves two areas of research:(1)processing the collected eye movement data,including missing value imputation,blink recognition,and calculation methods for traditional eye movement indicators;(2)expanding the eye movement indicator set,including Markov transition matrix-related indicators,time-frequency domain-related indicators,and head movement indicators.The new expanded indicators are combined with traditional eye movement indicators to construct an indicator set,and the effectiveness of the indicators is tested using statistical analysis methods.Establishment of Cognitive Load Models section mainly involves three areas of research:(1)using selected eye movement features to classify cognitive load levels using multi-class models,and comparing and analyzing the classification performance of different multi-class models to determine the best model.The results show that the integrated classifier achieved a classification accuracy of over 80%for all experimental tasks.(2)Due to the advantage of unsupervised learning,which does not require data labeling,this study attempted to identify high cognitive load levels using unsupervised models and achieved a relatively good classification accuracy on the test set,particularly for reading comprehension tasks,which achieved an accuracy of 81.7%.(3)Using the distance between data points and the hyper-sphere center in the SVDD model as a quantified cognitive load indicator,the effectiveness of the distance indicator was verified.Application ofModels section mainly involves two aspects of research:(1)To sort out the cognitive load analysis process based on eye movement data,conduct a needs analysis,and identify pain points that can be optimized in the evaluators’ evaluation process.(2)To study the eye movement data processing and analysis process,as well as the related cognitive load classification and recognition model,by designing a minimal prototype of the eye movement cognitive load assessment system to simplify the evaluation process for evaluators.This study designed experimental tasks based on cognitive load theory and collected eye movement data by processing the data and extracting features.By addressing the issues of cognitive load recognition and measurement,an effective measurement model was ultimately established,proving the feasibility and effectiveness of using eye movement data to expand features and measure cognitive load levels.It also designed a core functional prototype to assist evaluators in conducting evaluations.The study’s process and conclusions have certain reference and application value for the field of cognitive load recognition and measurement research.