Revisiting the Evaluation of Theory of Mind through Question Answering

By Matt Le, Y-Lan Boureau, Maximilian Nickel. 2019

189 citations recorded by the source corpus.

Theory of mind, i.e., the ability to reason about intents and beliefs of agents is an important task in artificial intelligence and central to resolving ambiguous references in natural language dialogue. In this work, we revisit the evaluation of theory of mind through question answering. We show that current evaluation methods are flawed and that existing benchmark tasks can be solved without theory of mind due to dataset biases. Based on prior work, we propose an improved evaluation protocol and dataset in which we explicitly control for data regularities via a careful examination of the answer space. We show that state-of-the-art methods which are successful on existing benchmarks fail to solve theory-of-mind tasks in our proposed approach.

Subfields: Artificial Intelligence, Computer Vision and Pattern Recognition, Developmental and Educational Psychology, Natural Language Processing, Evaluation & Benchmarks

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Artificial Intelligence

In Computer Vision and Pattern Recognition

In Developmental and Educational Psychology

In Natural Language Processing

In Evaluation & Benchmarks

Citations in the corpus

Experience Grounds Language (2020, background): “Using traditional-style datasets can be problematic due to the risk of embedding spurious patterns (Le et al., 2019).”
Anticipating Safety Issues in E2E Conversational AI: Framework and Tooling (2021, background): “…a benchmark for theory of mind in conversational AI systems being solvable without extracting any information about agents (extensively discussed in Le et al. (2019)), or achieving state-of-the-art results on visual dialog without the need to consider dialog history and thus rendering it as VQA…”
Neural Theory-of-Mind? On the Limits of Social Intelligence in Large LMs (2022, methodology): “1; bottom; Le et al., 2019), which was inspired by the classic Sally-Ann False Belief Theory of Mind test (Baron-Cohen et al., 1985).”
Theory of Mind and Preference Learning at the Interface of Cognitive Science, Neuroscience, and AI: A Review (2022, methodology): “Inspired by false-belief ToM evaluation protocols in cognitive science, Nematzadeh et al. (2018) and Le et al. (2019) have proposed ﬁrst-order and second-order belief evaluation tasks for language models (also see Section Dysfunction of Theory of Mind for discussion about the interaction of ToM and…”
Playing repeated games with large language models (2023, unknown): “Experiments were performed in accordance with the relevant guidelines and regulations approved by the ethics committee of the University of Tübingen (protocol nr. 701/2020BO).”
SOTOPIA: Interactive Evaluation for Social Intelligence in Language Agents (2023, background): “On one hand, many of the social intelligence benchmarks are not interactive (Sap et al., 2019; Le et al., 2019; Zadeh et al., 2019b), which is sub-optimal for evaluating social intelligence ((Mehri et al., 2022; Hoppler et al., 2022; Lee et al., 2023)).”