The AI editor for research papers

Write your paper where your papers live.

LaTeX editing, your library, and agents that actually read the literature — in one window. This is the real interface: click around.

▾PERSPECTIVE-TAKING-PAPER

›figs2

▾BROWSE

▾My Projects56

›My first project21

›hi4

›hey2

›SIMPLe_ICML__Copy__1_20

›Git Bridge Demo0

›Overleaf import test0

›test20

›Overleaf import0

›Difference_Masking__Copy_0

›Self-Play: from Games to Language0

›Formatting Instructions For NeurIPS 20260

›perspective-taking-paper26

›to-read6

›qa-night-editor6

›qa-night-vc0

›qa-night-files0

›qa-night-papers7

›qa-night-settings0

›hi1

›hey3

›new5

Literati database

▾TAGS

● to-read● baseline● benchmark

Filter Tags

File Tree

figs

fig1.png

fig2.png

literati.bib

main.tex

neurips_2026.sty

File Outline

§Introduction8

§Background14

Ask anything…

chat (shift+tab to cycle)esc to interrupt

Aaab.*

Type to search main.tex

Changes

Message (⌘↵ to commit)

Staged Changes1

results.texsections/A

Working Changes2

main.texM

abstract.texsections/M

History

›Draft intro + method2hRevert

›Initial skeleton1dRevert

main.tex×

Editing ⌄

8\section{Introduction}

10Large language models increasingly exhibit theory-of-mind

11(ToM) capabilities, yet their accuracy on false-belief

12tasks remains inconsistent across benchmarks.

14\section{Background}

All fields ⌄ 6 of 6 Filters1× Best match

2023Think Twice: Perspective-Taking Improves Large Language Models’ Theory-of-Mind CapabilitiesWilfwilf2023think● baseline2

2023Minding Language Models’ (Lack of) Theory of Mind: A Plug-and-Play Multi-Character Belief TrackerSclarsclar2023minding1

2023How FaR Are Large Language Models From Agents with Theory-of-Mind?Zhouzhou2023far1

2023Evaluating Large Language Models in Theory of Mind TasksKosinskikosinski2023theory● to-read—

2017Attention Is All You NeedVaswanivaswani2017attentionneed—

2020Language Models are Few-Shot LearnersBrownbrown2020languagemodelsfewshotlearners—

1/ 1

Perspective-Taking Improves LLM Theory-of-Mind

Anonymous Author(s)
Affiliation
Address
email

Abstract

1 We study theory-of-mind in language models and propose a method.

2 Across five benchmarks it yields consistent gains.

1 Introduction

3 Large language models increasingly exhibit theory-of-mind (ToM)

4 capabilities, yet their accuracy on false-belief tasks remains

5 inconsistent across benchmarks.

Submitted to 40th Conference on Neural Information Processing Systems (NeurIPS 2026). Do not distribute.

All logs Errors Warnings Info

click anything — Compile, the logs icon, the files (figs, literati.bib, the .sty), your papers, source control, find & replace

10Large language models increasingly exhibit theory-of-mind

11(ToM) capabilities, yet their accuracy on false-belief

12-tasks remains inconsistent.

12+tasks remains inconsistent across benchmarks. We show

13+that a simple perspective-taking prompt closes much of

14+this gap \citep{wilf2023think}.

37We evaluate on the ToMI benchmark.

38-Our method improves accuracy.

38+Our method improves accuracy by 47\% on ToMI

39+and 14\% on BigToM over the GPT-3.5 baseline

40+\citep{wilf2023think}.

41We report mean over 5 seeds.

42-% TODO: add ablation table

42+Ablations in Table 2 isolate each component.

1We study theory-of-mind in language models and

2-propose a method.

2+propose Think-Twice, a perspective-taking prompt that

3+reframes each question from the target agent's view.

4Across five benchmarks it yields consistent gains.

Think Twice: Perspective-Taking Improves Large Language Models’ Theory-of-Mind Capabilities

Alex Wilf · Sihyun Shawn Lee · Paul Pu Liang · Louis-Philippe Morency

Carnegie Mellon University · arXiv:2311.10227

AbstractHuman interactions are deeply rooted in the interplay of thoughts, beliefs, and desires made possible by Theory of Mind (ToM). Although ToM may come naturally to us, emulating it presents a challenge to even the most advanced Large Language Models. We turn to the cognitive science theory "Simulation Theory" and introduce SimToM, a two-stage prompting framework based on perspective-taking: first filter context based on what the character in question knows, then answer a question about their mental state. Requiring no additional training and minimal prompt-tuning, SimToM shows substantial improvement over existing methods.

1 Introduction

Large language models increasingly succeed on tasks once thought to require social reasoning, yet their performance on false-belief tasks remains brittle. We hypothesize that much of this brittleness stems from a failure to anchor reasoning to a specific agent's knowledge state, rather than from a lack of underlying capability.

We test this hypothesis with a lightweight intervention that asks the model to restate what each character can and cannot observe before answering. The result is a large, consistent improvement that requires no additional parameters.

41Base

67CoT

88Ours

95Human

Figure 1. Accuracy on ToMI (%). SimToM (Ours) closes most of the gap to human performance.

2 Method

Given a scenario and a question about agent a, we prompt the model to enumerate the events a witnessed, mark events hidden from a, and only then resolve the query conditioned on a's partial observation. This decomposition mirrors the sub-steps humans perform implicitly.

[1] Sclar et al. Minding Language Models' (Lack of) Theory of Mind. 2023.

[2] Zhou et al. How FaR Are Large Language Models From Agents with Theory-of-Mind? 2023.

[3] Kosinski. Evaluating Large Language Models in Theory of Mind Tasks. 2023.

Minding Language Models' (Lack of) Theory of Mind: A Plug-and-Play Multi-Character Belief Tracker

Melanie Sclar · Sachin Kumar · Peter West · Alane Suhr · Yejin Choi · Yulia Tsvetkov

University of Washington · Allen Institute for AI · arXiv:2306.00924

AbstractTheory of mind — tracking what others believe — is a hallmark of social intelligence, but language models struggle to maintain consistent beliefs for multiple characters. We present SymbolicToM, a plug-and-play inference-time tracker that maintains an explicit belief graph per character. It substantially improves robustness on ToMI, especially under adversarial reorderings that break end-to-end models.

1 Motivation

End-to-end prompting conflates the beliefs of different characters, causing errors that compound as stories grow longer. We argue that an explicit, symbolic representation of "who knows what" is both interpretable and more robust than relying on the model's latent state.

49Vanilla

58CoT

93SymbolicToM

Figure 1. Robust accuracy under story reordering. The symbolic tracker degrades far less.

2 Belief Graphs

For each character we instantiate a graph whose nodes are propositions and whose edges encode observation events. Queries are answered by reading off the relevant character's sub-graph, sidestepping the interference that plagues monolithic prompting.

[1] Nematzadeh et al. Evaluating Theory of Mind in Question Answering. 2018.

[2] Le et al. Revisiting the Evaluation of Theory of Mind through Question Answering. 2019.

How FaR Are Large Language Models From Agents with Theory-of-Mind?

Pei Zhou · Aman Madaan · Srividya Pranavi Potharaju · Aditya Gupta · Kevin R. McKee · Ari Holtzman · Jay Pujara · Xiang Ren · Swaroop Mishra · Aida Nematzadeh · Shyam Upadhyay · Manaal Faruqui

University of Southern California · Google DeepMind · arXiv:2310.03051

Abstract"Thinking is for Doing." Humans infer other people’s mental states and act pragmatically on those inferences. Benchmarks such as ToMi ask models about characters’ beliefs, but do not test whether models can use those inferences to guide their actions. We propose Thinking for Doing (T4D), which requires connecting inferences about others’ mental states to actions in social scenarios, and introduce Foresee and Reflect (FaR), a zero-shot prompting framework that encourages LLMs to anticipate future challenges and reason about potential actions.

1 Thinking is for Doing

LLMs such as GPT-4 seemingly excel at tracking characters’ beliefs in stories, yet struggle to translate this capability into strategic action. The core challenge lies in identifying the implicit inferences about mental states — without being explicitly asked, as in ToMi — that lead to choosing the correct action.

50GPT-4

71+ FaR

Figure 1. Accuracy on T4D (%). FaR boosts GPT-4 from 50% to 71%, outperforming Chain-of-Thought and Self-Ask.

2 Foresee and Reflect

The foresee step prompts the model to anticipate likely future events and the challenges facing each character; the reflect step reasons over candidate actions against those futures before committing. FaR generalizes to diverse out-of-distribution story structures and scenarios.

[1] Yao et al. ReAct: Synergizing Reasoning and Acting in Language Models. 2022.

[2] Shinn et al. Reflexion: Language Agents with Verbal Reinforcement Learning. 2023.

FANToM: A Benchmark for Stress-testing Machine Theory of Mind in Interactions

Hyunwoo Kim · Melanie Sclar · Xuhui Zhou · Ronan Le Bras · Gunhee Kim · Yejin Choi · Maarten Sap

Allen Institute for AI · University of Washington · arXiv:2310.15421

AbstractTheory of mind evaluations have largely used passive narratives. We introduce FANToM, a benchmark that stress-tests theory of mind within information-asymmetric conversations, requiring models to track what each participant heard and did not hear. FANToM is designed to be adversarial to shortcuts: models must answer the same belief question in multiple forms to be scored correct. State-of-the-art LLMs, including those using chain-of-thought and fine-tuning, perform substantially worse than humans.

1 Interactions, not narratives

Existing benchmarks present a story and ask a question about a character. Real theory of mind is exercised while conversing: participants join and leave, so beliefs diverge continuously. FANToM generates multiparty conversations in which a participant steps away and misses a span of dialogue.

24GPT-3.5

41GPT-4

48+ CoT

92Human

Figure 1. All-question accuracy (%). The gap to humans is far larger than on narrative benchmarks.

2 Scoring

Each belief is probed with free-response, multiple-choice, and list variants; a model scores only when all forms agree. This penalizes the illusory competence that single-format evaluations reward.

[1] Sclar et al. Minding Language Models' (Lack of) Theory of Mind. 2023.

[2] Kosinski. Evaluating Large Language Models in Theory of Mind Tasks. 2023.

1% literati.bib — generated by Literati from your library

2% papers are added via the arXiv API · do not edit by hand

4@article{wilf2023think,

5 title = {Think Twice: Perspective-Taking Improves Large Language

6 Models' Theory-of-Mind Capabilities},

7 author = {Wilf, Alex and Lee, Sihyun Shawn and Liang, Paul Pu

8 and Morency, Louis-Philippe},

9 year = {2023},

10 eprint = {2311.10227},

11}

13@article{sclar2023minding,

14 title = {Minding Language Models' (Lack of) Theory of Mind:

15 A Plug-and-Play Multi-Character Belief Tracker},

16 author = {Sclar, Melanie and Kumar, Sachin and West, Peter and

17 Suhr, Alane and Choi, Yejin and Tsvetkov, Yulia},

18 year = {2023},

19 eprint = {2306.00924},

20}

22@article{zhou2023far,

23 title = {How FaR Are Large Language Models From Agents

24 with Theory-of-Mind?},

25 author = {Zhou, Pei and Madaan, Aman and Potharaju, Srividya Pranavi

26 and Gupta, Aditya and McKee, Kevin R. and Holtzman, Ari

27 and Pujara, Jay and Ren, Xiang and Mishra, Swaroop

28 and Nematzadeh, Aida and Upadhyay, Shyam and Faruqui, Manaal},

29 year = {2023},

30 eprint = {2310.03051},

31}

33@article{kosinski2023theory,

34 title = {Evaluating Large Language Models in Theory of Mind Tasks},

35 author = {Kosinski, Michal},

36 year = {2023},

37 eprint = {2302.02083},

38}

1% neurips_2026.sty — NeurIPS 2026 official style

2% do not modify — conference formatting

3\NeedsTeXFormat{LaTeX2e}

4\ProvidesPackage{neurips_2026}[2026/01/15 v1.0]

5\RequirePackage{lineno}

6\RequirePackage[colorlinks=true]{hyperref}

8\newcommand{\@noticestring}{%

9 Submitted to 40th Conference on Neural

10 Information Processing Systems (NeurIPS 2026).

11 Do not distribute.}

13\setlength{\textwidth}{5.5in}

14\setlength{\textheight}{9in}

15\widowpenalty=10000

16\clubpenalty=10000

17\flushbottom

18\sloppy

Perspective-Taking Improves LLM Theory-of-Mind

Anonymous Author(s)
Affiliation
Address
email

Abstract

1 We study theory-of-mind in language models and propose a method.

2 Across five benchmarks it yields consistent gains.

1 Introduction

3 Large language models increasingly exhibit theory-of-mind (ToM)

4 capabilities, yet their accuracy on false-belief tasks remains

5 inconsistent across benchmarks.

2 Method

ℒ = −Σ_i y_i log ŷ_i (1)

Submitted to 40th Conference on Neural Information Processing Systems (NeurIPS 2026). Do not distribute.

Evaluating Large Language Models in Theory of Mind Tasks

Michal Kosinski

Stanford University · arXiv:2302.02083

AbstractEleven LLMs were assessed using a custom-made battery of false-belief tasks — a gold standard in testing Theory of Mind in humans. Smaller and older models solved no tasks; GPT-3.5-turbo solved 20%, and GPT-4 solved 75% of the tasks, matching the performance of six-year-old children observed in past studies. We discuss whether this reflects genuine mentalizing or the statistics of language, and argue the distinction is increasingly hard to draw.

1 Tasks

We use the Sally–Anne unexpected-transfer task and the Smarties unexpected-contents task, each with carefully constructed controls to rule out shallow lexical shortcuts. Items are presented one sentence at a time to probe incremental belief updates.

0Older

20davinci-003

20GPT-3.5

75GPT-4

Figure 1. Share of false-belief tasks solved (%). GPT-4 matches six-year-old children.

2 Interpretation

Emergence of ToM-like behavior may be an unintended by-product of training on human-generated text. We caution against both over-attribution of mental states and premature dismissal, and call for adversarial benchmarks that separate the two.

[1] Wimmer & Perner. Beliefs about beliefs. 1983.

[2] Ullman. Large Language Models Fail on Trivial Alterations to Theory-of-Mind Tasks. 2023.

Gemini finds. Claude builds. Your library keeps them honest.