Thinking and Consciousness as Reflection and Response An Eight Factor Framework for Artificial Intelligence Consciousness
Study at a glance
AI-extracted from the abstract| Characteristics | Theoretical or philosophical paper |
|---|---|
| Key points | Argues that thinking (first-order reflection) should be distinguished from consciousness (subject response), and that affective consciousness is grounded in embodied, life-maintaining activity while rational consciousness is thinking's response to thinking. On this embodied and autopoietic account, current AI systems can functionally simulate aspects of rational monitoring and self-correction but do not provide sufficient grounds for attributing affective experience or complete moral subjectivity to them. |
Abstract
This paper develops a reflection-response framework for distinguishing thinking from consciousness and applies the framework to debates about artificial intelligence. Thinking is defined as first-order reflection: the representation, retrieval, recombination, and inference of object-states. Consciousness is defined as subject response: awareness, evaluation, regulation, and feedback in relation to represented content, bodily states, and a system's own cognitive activity. The paper further distinguishes affective consciousness from rational consciousness. Affective consciousness concerns embodied and interoceptive responses grounded in life-maintaining activity, whereas rational consciousness concerns the reflective evaluation and regulation of thinking, or thinking's response to thinking. Drawing on debates concerning phenomenal and access consciousness, global workspace theory, higher-order theories, cognitive phenomenology, embodied cognition, and the phenomenology of life, the paper defends a qualified position between unrestricted functionalism and strict biologism. Current artificial intelligence systems possess powerful representational and inferential capacities and can functionally simulate aspects of rational monitoring and self-correction. Under the embodied and autopoietic account defended here, however, current systems do not provide sufficient grounds for attributing affective experience or complete moral subjectivity to them. The framework supplies a layered vocabulary for evaluating AI cognition, consciousness, and responsibility.