Work

Algorithmic Antidotes: Heuristic-Aligned Explanations for Debiasing Judgment

Type: article
Abstract: Cognitive biases routinely impair human judgment, yet existing debiasing interventions are often narrow in scope, resource-intensive, and difficult to scale. We advance a heuristic-centered framework for algorithmic debiasing in which algorithms augment human judgment through real-time advice and explanations aligned with the psychological mechanisms of representativeness, anchoring, and availability. Across three experiments using a bankruptcy estimation task, we test how distinct forms of algorithmic explanations mitigate errors associated with each heuristic. Example-based explanations most effectively reduce representativeness errors, whereas feature-based explanations attenuate availability-driven distortions. For anchoring, accurate algorithmic advice reduces errors independent of explanation type. These findings demonstrate that effective algorithmic debiasing depends on aligning intervention design with the cognitive mechanism producing the bias. By embedding mechanism-aligned interventions within the decision process, algorithmic decision-support offers a scalable approach to real-time debiasing and provides actionable guidance for designing systems that counter predictable failures in human judgment.
Language: English
Open Access status: green
OpenAlex ID: W7222989558
Cites: 43
Cited by:
Related to: 10