Mental Effort, Workload, Time on Task, and Certainty: Beyond Linear Models

Self-rated mental effort has been and continues to be the most widely used measure of cognitive load. This single-item measure is often used as a predictor variable in linear models for predicting performance or some other response variable. While an advantage of linear models is that they are fairly easy to understand, they fall short when the relations between variables of interest are clearly non-linear. The current study focused on a reanalysis of data of four recently published studies in which self-rated mental effort was measured repeatedly along with either workload, time on taskviewing time in one study, response time in another studyor self-rated certainty of correct task performance. A core outcome of the reanalysis is a preference towards mental effort as a non-linear instead of linear predictor in three of the four studies: quadratic (workload), quadratic (response time), and cubic (certainty); only in the case of viewing time, mental effort as a linear predictor was found to be the preferred solution. Implications of this finding for the interpretation of self-rated mental effort, indicators of cognitive load involving time on task, and metacognition-related response variables such as self-rated certainty are discussed.