Key points are not available for this paper at this time.
Theory of Mind (TOM) – the cognitive capacity to attribute mental states such as beliefs, desires, and emotions to oneself and others – is increasingly relevant for AI systems. This scoping review addresses three questions: whether AI should incorporate TOM capabilities, whether such integration can preserve data-driven analytical fidelity, and which computational architectures support implementation. Following Joanna Briggs Institute (JBI) scoping review methodology, 45 sources across psychology, cognitive science, neuroscience, philosophy, and AI ethics were synthesised. Searches were conducted from October 2024 to November 2025. Adopting a functionalist emergentist position, the review evaluates whether hybrid neuro-symbolic architectures – combining neural pattern recognition with symbolic reasoning – can achieve functional mental-state attribution. TOM-enabled systems demonstrate moderate effectiveness in mental health applications and high emotion-recognition accuracy on benchmarks. However, fidelity challenges persist: error rates rise substantially for underrepresented groups, and accuracy drops significantly from laboratory datasets to real-world conditions. Cross-cultural perspectives from Buddhist, Confucian, and Ubuntu traditions highlight relational approaches to mental-state inference. The evidence supports affirmative answers to all three questions. AI-TOM systems can enhance human interaction through hybrid neuro-symbolic architectures while preserving data-driven strengths. Realising these benefits requires robust governance frameworks for ethical deployment.
Daoud Matta (Wed,) studied this question.