Cognitive Perspectives on Context-based Decisions and Explanations

Human explanations are fundamentally about context: The paper demonstrates that when humans explain their decisions, they rely on mental representations that are deeply tied to specific contexts and situations. This suggests that AI explanations will be more trustworthy and understandable to people when they similarly account for contextual factors rather than providing generic, one-size-fits-all reasoning.

AI explanations should mirror how humans actually think: Rather than using abstract mathematical justifications, the research shows that effective AI explanations should align with how human cognition naturally works—through action-oriented mental models that consider what matters in a particular situation. This bridges the gap between how machines make decisions and how humans intuitively understand decision-making.

The Contextual Importance and Utility (CIU) method offers a practical path forward: The paper identifies CIU as an explainable AI approach that successfully incorporates cognitive science principles, making it a validated tool for creating explanations that humans can both relate to and trust, rather than explanations that are merely technically accurate.

Responsible AI requires understanding human cognition: The research emphasizes that building AI systems people can trust isn't just an engineering problem—it requires deep insights from philosophy and cognitive science about how humans actually rationalize and explain their own decisions.