Background
Cognitive load describes the amount of working memory resources used at a given time and is a measure of how hard the brain is working on a current task. Cognitive overload occurs when your working memory is overloaded, which negatively affects learning, planning, problem solving, and decision making. As a result, tools to assess cognitive load are desirable for detecting instances of cognitive overload. Psychophysiological measurements such as eye tracking and electroencephalography (EEG) have been used to objectively assess cognitive load; however, current implementations of these methods are largely constrained with respect to the task space in which they are applied. Existing assessments cannot be used to estimate cognitive performance or to assess brain health in a meaningful way. This PDIR addresses this gap by capturing psychophysiological measurements of subjects during cognitive loading tasks and analyzing the data to identify the best measure or multi-modal measure of cognitive load and overall decision-making capability. The identified measure(s) are functionalized to a fieldable implementation, which enables us to gather objective measures of load in relevant settings.
Approach
Prior to administering cognitive loading tasks, a subject’s cognitive baseline was established using psychophysiological measurements. Psychophysiological measurements included electroencephalography (EEG), electrocardiogram (ECG), galvanic skin response (GSR), eye-tracking, and speech. Psychophysiological measurements were gathered as subjects were loaded using three cognitive loading tasks. These tasks examined progressively higher-level aspects of cognition that play a role in decision making. Overall task performance was used as ground truth measures of cognitive load. Psychophysiological measurements gathered during cognitive loading tasks were correlated to our ground truth measures of load. From this analysis, the psychophysiological measurement that most highly and consistently correlated with our ground truth estimates was identified. Multi-modal measurements were also considered.
Accomplishments
We developed a data acquisition software (DAQ) that interfaces with data streams acquired from the various measurement instruments. This development included the implementation of the cognitive tasks in MATLAB, including design of the user interface for the presentation of stimuli and input of subject responses. We then used various preprocessing techniques to clean the data from the five modalities. These techniques were necessary to eliminate measurement noise inherent to the data collection instruments and to eliminate noise from the environment (e.g., movement artifacts or background noise). Finally, we examined the change in physiological signals due to increasing levels of cognitive load. We correlated the psychophysiological measurements to ground truth measures of cognitive load (task difficulty and task performance). In future work, modalities that correlate with our ground truth measures will be used to create a fieldable cognitive assessment system.