Measuring cognitive load in augmented reality with physiological methods: A systematic review

BackgroundCognitive load during AR use has been measured conventionally by performance tests and subjective rating. With the growing interest in physiological measurement using non-invasive biometric sensors, unbiased real-time detection of cognitive load in AR is expected. However, a range of sensors and parameters are used in various subject fields, and reported results are fragmented.ObjectivesThe aim of this review is to analyse systematically how physiological methods have been used to measure cognitive load and what the implications are for the future research on AR-based tools.MethodsThis paper took the systematic review approach. Through screening with 10 exclusion criteria, 23 studies, that contain 3 key elements: AR-based intervention, cognitive state examination and physiological methods, were identified, analysed and synthesised.ResultsPhysiological methods in their current form require reference to provide meaningful interpretations and suggestions. Therefore, they are often combined with conventional methods. Many studies investigate the effect of wearable devices in comparison with non-AR stimuli, which has been controversial, but detection of different causes of cognitive load are on the horizon. Eye-tracking is the method most used and most consistent in the use of its parameters.ConclusionsA multi-method approach combining two or more evaluation instruments is essential for the validation of users' cognitive state. In addition to the AR stimuli in question, having another independent variable such as task difficulty in experiment design is useful. Statistical approaches with more data input could help establish a reliable scale. The future research should attempt to dissociate cognitive load caused by different effects such as device, instruction, and other AR techniques as well as intrinsic and extraneous aspects, in a better experimental setup with multiple parameters. What is already known about this topicImplications for practiceThe efficacy of AR-based instructional tools can be evaluated using the concept of cognitive load.Cognitive load has been conventionally measured by self-reporting and performance.There is a growing interest in physiological measurement of cognitive load with non-invasive biometric sensors.What this paper addsThis paper systematically analyses how physiological methods are used to measure cognitive load in AR.It synthesises the fragmented results of AR studies conducted with a range of physiological methods.Combining two or more evaluation instruments is essential for the validation of users' cognitive state.Eye-tracking is the method most used, and fixation duration is the most consistent parameter.A statistical approach with more biometric data input could help establish a reliable cognitive load scale.Future research should attempt to separate cognitive load caused by different effects and different cognitive aspects.