The paper ‘Causal Perception’ by Jose M. Alvarez and Salvatore Ruggieri, published in January 2024, explores the concept of perception in the context of automated decision-making (ADM) systems. The authors argue that perception, influenced by individual experiences and interpretations, can significantly affect the fairness of ADM systems. They formalize perception under causal reasoning, defining it as the act of interpretation by an individual. The study also introduces the concept of ‘loaded attributes’ which are prone to evoke perception, such as gender and race. The authors propose two types of causal perception, ‘unfaithful’ and ‘inconsistent’, and illustrate their framework through decision-making examples. The goal of the study is to highlight the role of perception in ADM and position it as a parameter of interest for further research.
Publication date: 24 Jan 2024
Project Page: https://arxiv.org/abs/2401.13408v1
Paper: https://arxiv.org/pdf/2401.13408
