This survey maps how six branches of psychology — cognitive, developmental, behavioral, social, personality, and psycholinguistics — are informing the design and training of large language models. Foundational ideas from Piaget to reinforcement learning are reinterpreted as curriculum strategies, feedback signals, and scaffolding frameworks. Across more than 175 studies, the authors trace how psychology shapes data collection, training dynamics, and evaluation. They also highlight pitfalls of anthropomorphic language: transformer “attention” is a mathematical weighting, not human selective attention; “memory” and “understanding” can mislead if unqualified. The paper argues that psychology and AI are converging fields. Psychology can make AI more interpretable and ethical, while AI offers new tools to test psychological theory. Produced by Cognivault — insight, intelligence, and innovation made clear.