LM Psychology Primer: Why Understand Large Language Models as Minds
Start here
On October 21, 2026, LessWrong published an article arguing for understanding large language models from a psychological perspective. The author holds that pretraining a model on large amounts of human text forces it to acquire latent representations of human mental states, and that attributes such as memory and a sense of self may converge across intelligent systems. As to whether a model can introspect, whether it has feelings or a sense of self, and whether post-training changes its personality, the author says the answer to all is "in some meaningful sense yes, but different from humans, and the mechanism is unclear," and holds that psychological-level description is closer to the native level at which humans understand the world.
Generated by AI from reports · Updated 23 days ago
Progress timeline
- LessWrongLM Psychology Primer: Why Treat Large Language Models as Minds to Understand
The article argues for understanding large language models from a psychological perspective: pretraining on large amounts of human text forces the model to acquire latent representations of human mental states, and attributes such as memory and a sense of self may converge in intelligent systems. The author states that whether LMs can introspect, whether they have feelings or a sense of self, and whether post-training changes their personality, the current answers are all "in some meaningful sense yes, but different from humans, and the mechanism is unclear." The author believes that psychological-level description is closer to the native level at which humans understand the world.