# Do language models have coherent mental models of everyday things?

arXiv: https://arxiv.org/abs/2212.10029 · HTML: https://mlsanity.com/paper/2212.10029/

## Key facts

- Cited by 13 papers in the map (data through 14 Oct 2026).
- Builds on 17 of its 39 references that are in the map.
- First posted on arXiv 20 Dec 2022.

## Builds on (most cited first, 17 of 17)

- [Language Models are Few-Shot Learners](https://mlsanity.com/paper/2005.14165/) (2020)
- [PIQA: Reasoning about Physical Commonsense in Natural Language](https://mlsanity.com/paper/1911.11641/) (2019)
- [CommonsenseQA: A Question Answering Challenge Targeting Commonsense Knowledge](https://mlsanity.com/paper/1811.00937/) (2018)
- [World Models](https://mlsanity.com/paper/1803.10122/) (2018)
- [Talking About Large Language Models](https://mlsanity.com/paper/2212.03551/) (2022)
- [Language Models as Agent Models](https://mlsanity.com/paper/2212.01681/) (2022)
- [Improving Coherence and Consistency in Neural Sequence Models with Dual-System, Neuro-Symbolic Reasoning](https://mlsanity.com/paper/2107.02794/) (2021)
- [A Logic-Driven Framework for Consistency of Neural Models](https://mlsanity.com/paper/1909.00126/) (2019)
- [BeliefBank: Adding Memory to a Pre-Trained Language Model for a Systematic Notion of Belief](https://mlsanity.com/paper/2109.14723/) (2021)
- [Abstract Visual Reasoning with Tangram Shapes](https://mlsanity.com/paper/2211.16492/) (2022)
- [General-Purpose Question-Answering with Macaw](https://mlsanity.com/paper/2109.02593/) (2021)
- [PTR: A Benchmark for Part-based Conceptual, Relational, and Physical Reasoning](https://mlsanity.com/paper/2112.05136/) (2021)
- [HAKE: Human Activity Knowledge Engine](https://mlsanity.com/paper/1904.06539/) (2019)
- [DREAM: Improving Situational QA by First Elaborating the Situation](https://mlsanity.com/paper/2112.08656/) (2021)
- [Unpacking Large Language Models with Conceptual Consistency](https://mlsanity.com/paper/2209.15093/) (2022)
- [Just-DREAM-about-it: Figurative Language Understanding with DREAM-FLUTE](https://mlsanity.com/paper/2210.16407/) (2022)
- [Understanding Spatial Relations through Multiple Modalities](https://mlsanity.com/paper/2007.09551/) (2020)

## Built on by (most cited first, 13 of 13)

- [ChatGPT is a Knowledgeable but Inexperienced Solver: An Investigation of Commonsense Problem in Large Language Models](https://mlsanity.com/paper/2303.16421/) (2023)
- [SimpleToM: Exposing the Gap between Explicit ToM Inference and Implicit ToM Application in LLMs](https://mlsanity.com/paper/2410.13648/) (2024)
- [A Vision Check-up for Language Models](https://mlsanity.com/paper/2401.01862/) (2024)
- [Language Models with Rationality](https://mlsanity.com/paper/2305.14250/) (2023)
- [Concept Space Alignment in Multilingual LLMs](https://mlsanity.com/paper/2410.01079/) (2024)
- [EpiK-Eval: Evaluation for Language Models as Epistemic Models](https://mlsanity.com/paper/2310.15372/) (2023)
- [Towards Concept-Aware Large Language Models](https://mlsanity.com/paper/2311.01866/) (2023)
- [WinoViz: Probing Visual Properties of Objects Under Different States](https://mlsanity.com/paper/2402.13584/) (2024)
- [The quasi-semantic competence of LLMs: a case study on the part-whole relation](https://mlsanity.com/paper/2504.02395/) (2025)
- [How Do Large Language Models Learn Concepts During Continual Pre-Training?](https://mlsanity.com/paper/2601.03570/) (2026)
- [Aligned at the Start: Conceptual Groupings in LLM Embeddings](https://mlsanity.com/paper/2406.05315/) (2024)
- [Large Language Models as a Tool for Mining Object Knowledge](https://mlsanity.com/paper/2410.12959/) (2024)
- [Beyond Pattern Recognition: Probing Mental Representations of LMs](https://mlsanity.com/paper/2502.16717/) (2025)

## Abstract (arXiv)

When people think of everyday things like an egg, they typically have a mental image associated with it. This allows them to correctly judge, for example, that "the yolk surrounds the shell" is a false statement. Do language models similarly have a coherent picture of such everyday things? To investigate this, we propose a benchmark dataset consisting of 100 everyday things, their parts, and the relationships between these parts, expressed as 11,720 "X relation Y?" true/false questions. Using these questions as probes, we observe that state-of-the-art pre-trained language models (LMs) like GPT-3 and Macaw have fragments of knowledge about these everyday things, but do not have fully coherent "parts mental models" (54-59% accurate, 19-43% conditional constraint violation). We propose an extension where we add a constraint satisfaction layer on top of the LM's raw predictions to apply commonsense constraints. As well as removing inconsistencies, we find that this also significantly improves accuracy (by 16-20%), suggesting how the incoherence of the LM's pictures of everyday things can be significantly reduced.