Exploring Cognitive Load in the Secretary Problem Using EEG Signals

Human beings often face decision problems involving sequential selections with incomplete information and a trade-off between exploring different options and committing to the current best option. These problems have been called “The Secretary Problem” in Economics, and while their research yielded optimal mathematical solutions, bounded rational agents such as human beings behave sub-optimally in such scenarios. In this work, we seek to explore the problem through the lenses of its neurophysiological markers and contribute new insights to understanding the decision-making process. Specifically, we used an EEG to record the participants’ brain patterns while engaged in a “Secretary Problem” task and analyze their brain signals, such as Theta and Beta bands. Our analysis yields two exciting results: (1) the cognitive load associated with the decision-making process varies during the task. In the early and later stages of the selection process, more cognitive resources are being invested than in the middle stages of the experiment. (2) There is a consistent increase in the Theta band power as the number of offers increases, which hints at an increasing need to invest more cognitive resources as the task progresses. These insights have practical implications for the construction of strategic automated agents.