United States

Compositionality is considered central to language abilities. As performant language systems, how do large language models (LLMs) do on compositional tasks? We evaluate adjective–noun compositionality in LLMs using two complementary setups: prompt-based functional assessment and a representational analysis of internal model states. Our results reveal a striking divergence between task performance and internal states. While LLMs reliably develop compositional representations, they fail to translate consistently into functional task success across model variants. Consequently, we highlight the importance of contrastive evaluation for obtaining a more complete understanding of model capabilities.

ACL 2026

Evaluating Adjective-Noun Compositionality in LLMs: Functional vs Representational Perspectives

representations

compositionality

reasoning

evaluation

The 46th Annual Meeting of the Cognitive Science Society was an in-person meeting held in Rotterdam, The Netherlands at the Postillion Hotel & Conference Centre.

**ON-DEMAND PROGRAM ACCESS AFTER THE CONFERENCE**<br>
Recordings of the invited program are now available. You can access them by clicking on the 'Schedule' icon on the left. Select view by week and navigate to the day and time slot of the recording you wish to view. Click on the time slot to access the recording.

Recordings available (click on each recording to view):

**Thursday, July 25**<br>
0900: [Keynote Speaker Morgan Barense on Enhancing real-world event memory](https://underline.io/events/465/sessions/17997/lecture/99084-dynamics-between-minds-and-the-environment)<br>
1000: [Gleitman Award Winner Isabelle Dautriche on Language Foundations: Insights from acquisition, communication, cognition, and more](https://underline.io/events/465/sessions/18000/lecture/99085-gleitman-talk)<br>
1415: [Invited Symposium: Dynamics between minds and the environment](https://underline.io/events/465/sessions?eventSessionId=18016&searchGroup=lecture)<br>
1700: [Rumelhart Prize Presentation Speaker Alison Gopnik on Exploit, explore, empower: Three ages and three intelligences](https://underline.io/events/465/sessions/18028/lecture/99161-rp1-rumelhart-prize-presentation)<br>

**Friday, July 26**<br>
0900: [C.L. de Carvalho-Heineken Prize Keynote Speaker Kia Nobre on Focusing in memory](https://underline.io/events/465/sessions/18035/lecture/99162-hp1-c-l-de-carvalho-heineken-prize-keynote-address)<br>
1030: [Rumelhart Symposium: Childhood as exploration](https://underline.io/events/465/sessions/18038/lecture/100346-childhood-as-exploration)<br>
1415: [Elman Prize Symposium](https://underline.io/events/465/sessions?eventSessionId=18053&searchGroup=lecture)<br>
1600: [Invited Symposium: Dynamics between minds](https://underline.io/events/465/sessions?eventSessionId=18065&searchGroup=lecture)<br>
1745: [Keynote Speaker Andrea E. Martin on Neural dynamics encode the structure and statistics of language](https://underline.io/events/465/sessions/18077/lecture/99271-neural-dynamics-encode-the-structure-and-statistics-of-language)<br>

**Saturday, July 27**<br>
0900: [Keynote Speaker Gregor Schöner on How higher cognition emerges from the dynamics of strongly interacting neural populations](https://underline.io/events/465/sessions/18083/lecture/99273-dynamics-within-the-mind)<br>
1000: [Glushko Talks](https://underline.io/events/465/sessions?eventSessionId=18086&searchGroup=lecture)<br>
1500: [Invited Symposium: Dynamics within minds](https://underline.io/events/465/sessions?eventSessionId=18100&searchGroup=lecture)<br>


#### **CogSci 2024 Program Book**
<div style="position:relative;padding-top:0;width:900px;height:500px;"><iframe style="position:absolute;border:none;width:100%;height:100%;left:0;top:0;" src="https://online.fliphtml5.com/ebtyf/cble/"  seamless="seamless" scrolling="no" frameborder="0" allowtransparency="true" allowfullscreen="true" ></iframe></div>


**The Program Book can be downloaded [here](https://drive.google.com/file/d/1yRVp1cBqRnAbdYi4GXWV_pZeyWNsRquv/view?usp=sharing).**

#### **Letter from the President**

Welcome to the 2024 Meeting of the Cognitive Science Society in Rotterdam! CogSci 2024 brings together a large community of cognitive scientists who have traveled here from around the world, as well as a smaller group of remote presenters. I would like to extend an especially warm welcome to our first-time CogSci meeting attendees, and hope that 
their experience inspires them to join our future meetings for many years to come.

I want to thank this year’s conference co-chairs, Larissa K Samuelson, Stefan Frank, Mariya Toneva, Allyson Mackey, and Eliot Hazeltine, for putting together an amazing series of invited talks and symposia around the theme of Dynamics of Cognition. The co-chairs also deserve credit for handling a record-breaking number of submissions and producing a truly exciting conference program 

Beyond the invited talks and symposia dedicated to this year’s conference theme, I would like to highlight the presentations and events honoring the 
Rumelhart, Elman, Gleitman and C .L . de Carvalho Heineken prize winners, including the Rumelhart reception on Thursday. As you may know, we are recording all of the invited talks and symposia, and all prize-winners’ talks, and will be making this content available after the conference. 

The Cognitive Science Society relies on the volunteer efforts of our Governing Board members who work on matters related to our membership; conference policies; diversity and inclusion, international, and outreach initiatives; prizes and much more. We welcome interest from Society members who would like to become more involved in what we do. The 
Society is deeply grateful to Rick Dale and Andrea Bender, who serve as editors of the two Society journals, Cognitive Science and Topics in Cognitive Science respectively. We especially acknowledge our editors’ generosity in donating their stipends each year to D&I and outreach efforts. 

This year’s Annual Meeting depends critically on the support of Marischal De Armond and his team at Podium Conference Specialists: Sharon Zwack, Cendrine De Vis, Sarah-Kate Burke, and Rachel Waller worked tirelessly to secure the venue and handled a multitude of logistical and organizational issues, from arranging coffee breaks and receptions to making sure that a huge number of posters fits comfortably into the available space. Finally, and importantly, I would like to acknowledge the Cognitive Science Society Executive Officer, Erica Wojcik, for managing the complex and ever expanding sphere of our Society’s activities.

I hope you enjoy this year’s meeting and the many different opportunities to engage with the vibrant community of cognitive scientists gathered here. I also hope you get a chance to explore the great city of Rotterdam with its many attractions, museums, restaurants and scenic views. If you’d like to talk more about our Society, or simply want to say hello, please stop by!

![](https://assets.underline.io/markdown_image/1/image/3c66d0e21fff30e204df1138ffd15173.jpeg)

**Anna Papafragou**  
President, Cognitive Science 
Society 2023-2024

**CODE OF CONDUCT**<br>
By attending the CogSci 2024 Conference, you are required to adhere to the society’s [Code of Conduct](https://drive.google.com/file/d/16-6KkptF0Gn3ZYGJDlpqpwTdImw45Ng0/view?usp=drive_link).

**ABOUT THE COGNITIVE SCIENCE SOCIETY**<br>
The Cognitive Science Society brings together researchers from around the world who hold a common goal: understanding the nature of the human mind. The mission of the Society is to promote Cognitive Science as a discipline, and to foster scientific interchange among researchers in various areas of study, including Artificial Intelligence, Linguistics, Anthropology, Psychology, Neuroscience, Philosophy, and Education.

The Society is a non-profit professional organization and its activities include sponsoring an annual conference and publishing the journals Cognitive Science and TopiCS.
<br><br>

You need to log in with the email address you registered with. Access credentials have been sent to your email. 

Please be sure to check your spam and other email folders if you do not see an email confirmation right away.

Please log in to explore this event.

It looks like you are not registered for this event. 

To access the site please register [**here**](https://cognitivesciencesociety.org/registration/). 

Please register!

CogSci 2024

The 46th Annual Meeting of the Cognitive Science Society presents the latest research across cognitive science and highlights the theme of Cognition in Context.

workshop paper

It is my great pleasure and honor to welcome you to the 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026), held in San Diego, California, USA, July 2–7, 2026. ACL 2026 continues our fields tradition of excellence in scholarship, innovation, and inclusivity, and I am deeply grateful to the many volunteers who have worked tirelessly to bring this event to life.

It is exciting to see the many developments in the field reflected at our conference, such as the advancement of large language models (born out of our community) into many application directions. This is not without challenges, partly due to unsolved problems, and partly due to misuse of the technology. All this is reflected well at ACL 2026. The conference also reflects that the field is not only about advancing technology, but also the science of fundamentally understanding language and how it is use by many different speaker communities. <br>
[Full Message from General Chair Philipp Koehn](https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing) <br><br><br>
<html><button style="display: inline-flex; align-items: center; justify-content: center; white-space: nowrap; border-radius: 9999px; font-weight: bold; background: #7c3aed; color: white; font-family: 'Space Grotesk', sans-serif; height: 40px; font-size: 16px; padding: 0 20px; border: none; cursor: pointer" onclick="window.open('https://drive.google.com/file/d/1QiJFOmMjvDgsDo3i1XzlQnImxe3RU_Cy/view?usp=sharing')">Conference Handbook</button></html>
<br><br><br><br>

To access this event page, you need to log in with the **email address you registered with**. <br>Access credentials will be sent to your email from Underline -  subject line "Welcome to ACL 2026". Please be sure to check your spam email folder if you do not see an email confirmation right away.

To access this event page, you are required to register.
Please complete your registration to continue.

We recommend reading [**the registration information**](https://2026.aclweb.org/registration/) first.

**Online Registration Form**: https://acl.swoogo.com/acl2026

Registration Required

The 64th Annual Meeting of the Association for Computational Linguistics (ACL 2026) will take place in San Diego, California, from July 2 - 7, 2026.

We evaluate GPT-4o's color naming across nine languages using both synthetic and human-derived stimuli. Using hue wheels, fixed basic categories, low-chroma hue lines, and dense binned CIELAB grids, we separate lexical availability of color terms from distributional agreement with human color naming. GPT-4o reliably names vivid, high-chroma colors and reproduces several known language-specific distinctions under constrained settings. However, its performance degrades sharply for low-chroma colors and for stimuli near human category boundaries. In these regions, model-human divergence remains high. 
Overall, GPT-4o shows strong cross-linguistic lexical knowledge but does not reliably match human color-naming distributions, especially in low-chroma and boundary regions.

Lexical Availability and Human Distributional Agreement in GPT-4o's Color Naming

Zero-shot Named Entity Recognition is critical for low-resource domains, yet existing approaches rely on opaque prompting of large language models or dense representations that suffer from polysemanticity. We propose an alternative approach that leverages monosemantic features of Sparse Autoencoders. We introduce $\textbf{SAE-NER}$, a training-free framework that maps monosemantic SAE feature activations to entity types through direct precision estimation, requiring no supervision or prompting. Experiments across general and biomedical domains show that SAE-NER consistently outperforms trained probing classifiers, with especially a large margin in the biomedical domain (up to +20 F1). Finally, we evaluate the utility of SAE-NER predictions as silver training data for downstream NER models. Using controlled perturbations of gold annotations to simulate realistic annotation noise, we show that false negatives are the primary bottleneck for silver-data quality, outweighing the impact of boundary imprecision and false positives.

From Latents to Labels: Zero-Shot Named Entity Recognition using Sparse Autoencoder Features

Large language models (LLMs) appear successful in emulating compositional language, yet it remains unclear what these results entail about their underlying compositional semantic representations. The probing classifier paradigm has emerged as a tool to remedy this. This paper proposes to critically review the findings of 24 probing studies targeting a wide range of linguistic and semantic phenomena. It proposes a taxonomy of probing tasks based on the linguistic primitives they presuppose, distinguishing four tiers: lexical semantics, the syntax–semantics interface, propositional semantics, and discourse and pragmatics. A gradient in representational evidence emerges: LLMs robustly encode lexical information, display less consistent sensitivity to structural relations within sentences, and obtain unsatisfactory results on tasks requiring propositional content, speech acts, or pragmatic inference. The review underscores the need for a clearer theoretical grounding of what probing tasks measure and reflects on how probing can illuminate the compositional pathways available within current language models.

Compositional Meaning Representations in LLMs: a Critical Review of Probing Studies

We present a systematic study on paraphrase detection in Tamil by constructing a unified dataset through translation and semantic verification of three English benchmarks QQP, PAWS, and MRPC. Unlike prior efforts that focus on individual sources or limited scales, our dataset combines multiple paraphrase detection paradigms and is evaluated using semantic similarity metrics, round-trip translation checks, and classifier agreement analysis. We fine-tune five multilingual transformer models (mBERT, XLM-R, IndicBERT, MuRIL, and DistilmBERT) and a Tamil-specific compact model, TLMR (Tamil Language Model - DeBERTa), pretrained on 525M Tamil tokens. Furthermore, we assess the representational quality of the sentence embeddings that are taken from these models using lightweight classifiers (SVM, XGBoost, and Logistic Regression). We formulate an efficiency-oriented metric that incorporates top-5 accuracy, vocabulary usage, and script fidelity in relation to perplexity in order to facilitate resource-aware evaluation. The experimental findings lay the groundwork for future Tamil semantic understanding tasks by highlighting differences in generalization and efficiency across models.

Under the Surface: Probing Tamil Paraphrase Intelligence

Social biases based on regional identity (or regional bias) are often observed in Indian contexts on major online social networks and require critical attention. However, due to large linguistic and cultural diversity, high annotation costs, and inherent human biases, very little annotated data exists on regional biases in the Indian context. Recently, Large Language Models (LLMs) have garnered attention for the automatic annotation of text. However, such annotation efforts are largely limited to English texts, and LLMs often perform poorly when applied to low-resource languages. Therefore, this paper focuses on understanding the capabilities and challenges of popular open-source LLMs in annotating Indian regional biases. We utilize the recently proposed IndRegBias dataset, which consists of Indian regionally biased social media comments in both English and code-mixed formats.
First, we assess the annotation capabilities of LLMs in a zero-shot setting and critically analyze their performance across different writing styles, including code-mixing, transliteration, and English. We find that the majority of LLMs exhibit low agreement with human annotations (measured using Cohen’s kappa). Consequently, we extend our study by fine-tuning the models using 50% of the data and evaluating them on the remaining 50%. We observe a significant improvement in annotation agreement (kappa) for all the LLMs. To further assess the capabilities of the fine-tuned models, we evaluate them on 500 newly collected social media comments discussing regional issues in India. The results show that most fine-tuned LLMs outperform their zero-shot counterparts when annotating these new comments.

Annotating Indian Regional Biases using Large Language Models: Evaluation and Analysis

Updating bilingual dictionary entries is a tedious, time-consuming, and highly subjective task, especially when a new sense in the source language requires identifying an appropriate translation equivalent. To date, there have been no attempts to automatize the discovery of new bilingual sense entries. Related tasks such as Word-level Bilingual Dictionary Induction and cross-lingual embedding alignment do not account for polysemy and are not applied to lexicographic data.
In contrast to their monolingual counterparts, bilingual dictionaries fall short in terms of completeness, detail with respect to examples and glosses, and diachronic information.
We introduce a novel NLP task, Sense-Level Bilingual Dictionary Induction (\textsc{SenseBDI}), at the intersection of lexical semantics, cross-lingual, and diachronic NLP.
We construct a dataset of time-stamped sense-level bilingual dictionary entries by aligning two bilingual dictionaries, two monolingual dictionaries, and the multilingual resource BabelNet, thereby enriching bilingual entries with monolingual source-language information.
We propose a baseline based on nearest-neighbor search over cross-lingual embeddings of glosses and usages. We find that usages contribute more strongly than glosses, with substantial variation across language pairs and discuss task-specific challenges with regards to target language polysemy and future directions such as transfer to real-world scenarios.

Towards Sense-level Bilingual Dictionary Induction

Suicide ideation detection models are typically evaluated using aggregate performance metrics, yet little is known about how they internally represent psychologically meaningful risk factors. In high-stakes mental health applications, understanding these internal representations is essential for safety, transparency, and responsible deployment. In this work, we move beyond accuracy and analyze how suicide detection models trained on original and topic-augmented datasets encode psychological risk factors in their internal representation space. Using visualization and geometric analysis, we examine the coherence and separability of topic-related features. Our results show that topic-aware augmentation increases the clarity and distinctness of underrepresented psychosocial risk factors such as immigration, family issues, and financial crisis. These findings suggest that augmentation not only improves model performance but also leads to more structured and interpretable internal representations.

Beyond Accuracy: Interpreting Topic Representation in Suicide Ideation Detection Models

Stance detection identifies whether a text expresses support, opposition, or neutrality toward a target and is central to applications such as political analysis and misinformation monitoring. With the shift toward large language models (LLMs), stance classification increasingly relies on prompting and lightweight adaptation. Yet the generalization behavior of open-source LLMs across new targets and domains remains uneven. We conduct a large-scale diagnostic study of four open-source LLMs (3B–24B parameters), examining how model size, prompting strategies, and Low-Rank Adaptation (LoRA) interact across in-target, cross-target, and cross-domain settings. Across 912 experiments, three patterns emerge: (1) larger models improve prompting-based in-target performance, but this advantage diminishes after fine-tuning; (2) LoRA boosts in-target accuracy yet often harms cross-context transfer; (3) optimal prompting depends on model size. These results reveal a consistent tension between specialization and generalization, offering practical guidance for configuring LLM-based stance detection under transfer.

Diagnosing Generalization in Open-Source LLMs for Stance Detection

News headlines and summaries shape how events are interpreted through selective emphasis and omission, a phenomenon commonly referred to as framing. Large language models are now routinely used to generate such content, yet existing evaluation frameworks largely overlook this dimension. We introduce Frame In, Frame Out (FIFO), the first large-scale benchmark for measuring framing presence in LLM-generated news summaries, grounded in the widely used XSum dataset. FIFO combines 15,499 jury-annotated examples with 320 expert-labeled instances ($\kappa = 0.61$) to validate and calibrate model-based annotations. Using FIFO, we analyze measured framing rates across 27 summarization models. We find that LLM-generated summaries often exhibit higher calibrated framing rates than human-written references, with substantial variation across topics and training regimes, including elevated rates in scientific and public health summaries. Our results establish framing as an underexplored and consequential dimension of summarization quality.

Frame In, Frame Out: Measuring Framing Bias in LLM-Generated News Summaries

We disentangle multilingual sentence embeddings into language-dependent and language-agnostic components, leveraging the latter to improve cross-lingual similarity estimation.
Previous studies on this approach have trained disentanglers by combining intra-component constraints, which either align or disalign language-dependent embeddings or language-agnostic embeddings, with inter-component constraints across both embeddings.
However, when and how these constraints are effective remains unclear.
Our experiments on sentence similarity estimation and machine translation quality estimation revealed that while intra-component constraints and the combination of both constraints are effective for encoder-based multilingual sentence embeddings, inter-component constraints are effective for decoder-based ones.
Furthermore, our detailed analysis revealed distinct roles: intra-component constraints improve uniformity within the embedding space, while inter-component constraints enhance cross-lingual alignment between parallel sentences.

Downloads

Next from ACL 2026

Lexical Availability and Human Distributional Agreement in GPT-4o's Color Naming

Similar lecture

Compositionality is underutilized in language: The case of English adjective-noun phrases in humans and large language models

.css-70qvj9{display:-webkit-box;display:-webkit-flex;display:-ms-flexbox;display:flex;-webkit-align-items:center;-webkit-box-align:center;-ms-flex-align:center;align-items:center;}Downloads

Next from ACL 2026

Lexical Availability and Human Distributional Agreement in GPT-4o's Color Naming

Similar lecture

Compositionality is underutilized in language: The case of English adjective-noun phrases in humans and large language models

Downloads