Adaptive Sampling Policies Imply Biased Beliefs: A Generalization of the Hot Stove Effect
This work addresses a fundamental bias in adaptive learning algorithms, which is incremental as it extends existing theory to more general sampling scenarios.
The authors generalized the Hot Stove Effect theory to settings where adaptive sampling policies lead to biased beliefs, showing that negativity bias persists even when learners do not entirely avoid alternatives with negative estimates but sample them less, and they demonstrated that most Bayesian learners underestimate expected values.
The Hot Stove Effect is a negativity bias resulting from the adaptive character of learning. The mechanism is that learning algorithms that pursue alternatives with positive estimated values, but avoid alternatives with negative estimated values, will correct errors of overestimation but fail to correct errors of underestimation. Here, we generalize the theory behind the Hot Stove Effect to settings in which negative estimates do not necessarily lead to avoidance but to a smaller sample size (i.e., a learner selects fewer of alternative B if B is believed to be inferior but does not entirely avoid B). We formally demonstrate that the negativity bias remains in this set-up. We also show there is a negativity bias for Bayesian learners in the sense that most such learners underestimate the expected value of an alternative.