The relationship between artificial intelligence (AI) and cognitive psychology has been reciprocal and transformative since both fields emerged in the 1950s cognitive revolution. AI draws on cognitive psychology for inspiration about intelligent processing, while cognitive psychology uses AI techniques to build computational models of the mind. Recent advances in deep learning and large language models have created new challenges and opportunities for understanding both artificial and human cognition.
From Cognitive Models to AI
Early AI was explicitly modeled on cognitive processes: production systems (Newell and Simon) formalized human problem-solving, semantic networks represented knowledge as cognitive psychologists theorized, and expert systems captured domain knowledge in rule-based form. Connectionism drew on neural network metaphors. Modern deep learning, while less directly inspired by cognitive architecture, incorporates attention mechanisms, memory systems, and hierarchical representations that echo cognitive principles.
Large language models (LLMs) have created new questions for cognitive psychology. They produce human-like language without embodiment, social interaction, or sensory experience — challenging theories that ground language in these factors. They show patterns of both success and failure that partially overlap with human cognitive patterns. They raise questions about whether statistical learning from text alone can produce genuine understanding, or whether it produces a sophisticated simulation that fails in ways that reveal what is missing from purely text-based learning.
AI as Cognitive Tool
AI models serve as cognitive theories that can be formally tested against human behavioral data. Computational models of perception (convolutional neural networks), language processing (transformer models), decision-making (reinforcement learning), and memory (complementary learning systems) generate specific, testable predictions about human cognition. Comparing where AI succeeds and fails relative to human cognition reveals the computational principles that underlie human intelligence.