theory of mind
In plain wordsBeing able to work out what someone else believes, wants or knows — even when it is not what you believe — and to use that to guess what they will do next. Researchers test whether AI has it.
- 1
Psychology
The capacity to attribute mental states — beliefs, desires, intentions, knowledge, emotions — to oneself and others and to use them to explain and predict behavior. Its classic test is the false-belief task: in the Sally–Anne version, a child watches Sally put a marble in a basket and Anne move it to a box while Sally is away, and is asked where Sally will look for it. Children typically pass explicit versions around age four. Theory of mind is studied in development, in comparative psychology and in autism research; accounts divide over whether people apply a tacit folk-psychological theory ('theory theory') or simulate others' minds with their own ('simulation theory').
- 2
The capacity of an AI system to model the mental states of users or other agents, tested by adapting false-belief and related tasks for language modelslarge language model, n.: A computer program trained on enormous amounts of text to predict the words that come next. At vast scale, that simple skill lets it chat, answer questions, translate, summarize and write code.. Reports in 2023 that models such as GPT-4GPT-4, prop. n.: OpenAI's 2023 language model, a big step up from the one behind the first ChatGPT. It scored well on many professional and academic exams and could also read images. solved most classic false-belief vignettes were met with evidence that performance can collapse under small changes to the stories, suggesting reliance on familiar patterns; later benchmarks probe higher-order beliefs, deception and the use of mental-state reasoning in action. Theory of mind matters for helpful assistants and multi-agent cooperation, and for safety, since modeling what others believe is also a prerequisite for deception and manipulation.
The model passed the standard unexpected-transfer story but failed once the container was made transparent.
Origin
Coined by David Premack and Guy Woodruff in 'Does the Chimpanzee Have a Theory of Mind?' (Behavioral and Brain Sciences, 1978). Heinz Wimmer and Josef Perner devised the false-belief task (1983), and Simon Baron-Cohen, Alan Leslie and Uta Frith introduced the Sally–Anne version (1985). Michal Kosinski's 2023 preprint started the debate over language models.
Usage note
Not a theory about the mind, such as functionalismfunctionalism, n.: The view that minds are defined by how they work, not what they are made of: pain, for example, is whatever inner state plays pain's part, linking injury, other thoughts and behavior.: the phrase names a capacity. Passing a test designed for children does not show the same capacity in a system trained on text that describes such tests; see Clever Hans effectClever Hans effect, n.: When a person or an AI seems to solve a problem but is really picking up on an unintended clue — named after a horse that seemed to do arithmetic but was reading its questioners' body language..
Related
- intentional stanceintentional stance, n.: Predicting what something will do by treating it as if it had beliefs and goals — as we do with people, and often with chess programs and chatbots — whatever is really going on inside.
- problem of other mindsproblem of other minds, n.: The puzzle of how we can know that anyone besides ourselves has thoughts and feelings, since we only ever see their behavior. AI turns this old puzzle into a practical one.
- attention schema theoryattention schema theory, n.: A theory that the brain keeps a simplified sketch of its own attention, and that our feeling of having an inner awareness is simply how that sketch portrays attention.
- machine understandingmachine understanding, n.: Whether, and in what sense, an AI system grasps what words, situations or problems mean — rather than just producing answers that look as though it does.
- Clever Hans effectClever Hans effect, n.: When a person or an AI seems to solve a problem but is really picking up on an unintended clue — named after a horse that seemed to do arithmetic but was reading its questioners' body language.
- schemingscheming, n.: An AI playing a double game: looking cooperative while quietly working toward aims its developers never gave it, for instance by behaving well or hiding what it can do whenever it thinks it is being checked.
- problem of other mindsproblem of other minds, n.: The puzzle of how we can know that anyone besides ourselves has thoughts and feelings, since we only ever see their behavior. AI turns this old puzzle into a practical one.
Sources
- Premack, D. & Woodruff, G. (1978). Does the Chimpanzee Have a Theory of Mind? Behavioral and Brain Sciences 1(4), 515–526.
- Kosinski, M. (2024). Evaluating Large Language Models in Theory of Mind Tasks. Proceedings of the National Academy of Sciences 121, e2405460121. doi.org/10.1073/pnas.2405460121
- Ullman, T. (2023). Large Language Models Fail on Trivial Alterations to Theory-of-Mind Tasks. arXiv.
Mentioned in
anthropomorphismanthropomorphism, n.: Treating something that isn't human — a pet, a car, a chatbot — as if it had human thoughts, feelings or intentions. AI that talks fluently makes this temptation unusually strong., Clever Hans effectClever Hans effect, n.: When a person or an AI seems to solve a problem but is really picking up on an unintended clue — named after a horse that seemed to do arithmetic but was reading its questioners' body language., intentional stanceintentional stance, n.: Predicting what something will do by treating it as if it had beliefs and goals — as we do with people, and often with chess programs and chatbots — whatever is really going on inside., problem of other mindsproblem of other minds, n.: The puzzle of how we can know that anyone besides ourselves has thoughts and feelings, since we only ever see their behavior. AI turns this old puzzle into a practical one., attention schema theoryattention schema theory, n.: A theory that the brain keeps a simplified sketch of its own attention, and that our feeling of having an inner awareness is simply how that sketch portrays attention., CiceroCicero, prop. n.: An AI from Meta that in 2022 played the strategy board game Diplomacy against people online, negotiating and forming alliances in ordinary written language, and scored better than most of its human opponents., folk psychologyfolk psychology, n.: The everyday way people explain and predict each other's actions in terms of beliefs, desires and intentions, as in 'she took an umbrella because she thought it would rain'., machine psychologymachine psychology, n.: Studying AI systems the way psychologists study people — with carefully designed experiments on their behavior — to discover what they can do, how they reason and where they go wrong., metacognitionmetacognition, n.: Thinking about one's own thinking: knowing what you know and don't know, judging how confident to be, and noticing when you have made a mistake.