First Implementation of Causal Perception Framework for Fairness

A recent publication on arXiv (2608.03917) details the inaugural application of the causal perception framework established by Álvarez and Ruggieri in 2025. Causal perception arises when agents with differing Structural Causal Models (SCMs) for the same system deduce varying probability distributions, including those that are hypothetical under specific interventions. The authors have successfully operationalized this theoretical framework by creating algorithms to calculate interventional and counterfactual distributions, alongside proposing distance metrics to measure the extent of disagreement among agents. They differentiate between structural causal perception, where agents disagree on the causal graph, and parametrical causal perception, where they concur on the graph but differ in weights. The German Credit dataset is utilized to demonstrate the impact of causal perception on accuracy and fairness. This paper is classified as a new announcement and can be accessed via the provided URL.