Computational Accounting Theory: Addressing Breton's Epistemological Critique Through Artificial Intelligence and Blockchain Technology

Abstract

The scientific legitimacy of accounting has been contested for decades, with recent critiques questioning whether it possesses the characteristics of a scientific discipline. Breton's 2019 analysis argued that traditional accounting theory fails to meet fundamental scientific standards because of its normative orientation and its lack of predictive capability. This conceptual paper examines technological developments between 2020 and 2025 and proposes a theoretical framework, Computational Accounting Theory, that addresses these epistemological concerns. Through a systematic analysis of recent literature, we demonstrate that machine learning algorithms provide falsifiable predictions with measurable accuracy, while blockchain-based systems establish fundamentally different epistemological foundations compared to conventional accounting. Our framework identifies four distinguishing characteristics: descriptive-predictive orientation, empirical falsifiability, practical implementability, and paradigmatic structure. Evidence from implementations in major accounting firms and empirical studies supports the viability of this framework. However, challenges, including algorithmic bias, transparency deficits, and regulatory lag, remain significant. This research provides novel theoretical foundations for understanding accounting as a computational discipline, identifying implications for methodology, practice, and education.