Cognitive Frameworks for Mitigating Antiblack Bias: Advancing Ethical AI Design and Development

Abstract

This paper explores the utilization of cognitive modeling to address the influence of antiblackness and racism on the design and development of AI systems. Through the lens of the ACT-R/Φ cognitive architecture and ConceptNet, an existing knowledge graph system, we investigate this issue from cognitive, sociocultural, and physiological perspectives. We propose an approach that not only examines how antiblackness may permeate AI system design and development, particularly within the realm of software engineering, but also establishes links between antiblackness, human cognition, and computational cognitive modeling. We contend that overlooking sociocultural factors in cognitive architectures perpetuates a colorblind approach to modeling, obscuring the inherent sociocultural context that shapes human behavior and cognitive processes.