Estimating Present Bias using Behavior History

抄録

Present bias, the cognitive bias that prioritizes immediate rewards over future ones, is considered one of the factors that can hinder goal achievement. Estimation of present bias enables the development of effective intervention strategies for behavioral change. This paper proposes a novel method using behavior history, captured by wearable devices for estimating the present bias. We employ Transformer due to its proficiency in learning relationships within sequential data like behavioral history, including continuous (e.g., heart rate) and event data (e.g., sleep onset). To allow Transformer to capture behavior patterns affected by present bias from behavior history, we introduce two novel architectures for effectively processing continuous and event data timestamp information in behavioral history: temporal and event encoders (TE and EE). TE discerns the periodic characteristics of continuous data, while EE examines temporal interdependencies in the event data. These encoders enable our proposed model to capture temporally (ir)regular behavioral patterns associated with present bias. Our experiments using the behavior history logs of 257 subjects collected over 28 days demonstrated that our method estimates the subjects' present bias accurately.