Thinking Fast or slow? Understanding Answering Behavior Using Dual-Process Theory through Mouse Cursor Movements

Abstract

Users’ underlying cognitive states govern their behaviors online. For instance, an extreme cognitive burden during live system use would negatively influence important user behaviors such as using the system and purchasing a product. Thus, inferring the user's cognitive state has practical significance for the commercialized systems. We use Dual-Process Theory to explain how the mouse cursor movements can be an effective measure of cognitive load. In an experimental study with five hundred and thirty-four subjects, we induced cognitive burden then monitored mouse cursor movements when the participants answered questions in an online survey. We found that participants' mouse cursor movements slow down when they are engaged in cognitively demanding tasks. With the newly derived measures, we were able to infer the state of heightened cognitive load with an overall accuracy of 70.22%. The results enable researchers to measure users' cognitive load with more granularity and present a new, theoretically sound method to assess the user's cognitive state.