A new methodology for consciousness science treats computational models as providing negative data—information about what consciousness is not—rather than positive evidence. This approach avoids metaphysical commitments while supporting quantitative research. It combines computational modeling as an integrative framework across cognitive sciences, echoing Alan Newell's call for computer science concepts as a common language, with a validation method that uses models to constrain theories by ruling out alternatives. The methodology addresses the challenge that consciousness is inherently subjective while scientific data are intersubjective, enabling empirical investigation without resolving philosophical debates.
For artificial agents to participate in human social and cultural life, their behavior should be explainable using folk psychological concepts like beliefs, desires, obligations, and especially intentions. Drawing on philosophy and psychology, the authors argue that attention is critical for intentional action and agency. They contend that existing AI and cognitive systems research has not adequately developed computational architectures that incorporate these insights. To address this gap, they present the ARCADIA attention-driven cognitive system as an initial step toward an architecture capable of supporting the kind of agency required for rich human–machine interaction.