China

Humans can quickly learn a new word from a few illustrative examples, and then systematically and flexibly use it in novel contexts. Yet the abilities of current language models for few-shot word learning, and methods for improving these abilities, are underexplored. In this study, we introduce a novel method, Minnow. This method trains language models to generate new examples of a word&#39;s usage given a few in-context examples, using a special placeholder token to represent the new word. This training is repeated on many new words to develop a general word-learning ability. We find that training models from scratch with Minnow on human-scale child-directed language enables strong few-shot word learning, comparable to a large language model (LLM) pre-trained on orders of magnitude more data. Furthermore, through discriminative and generative evaluations, we demonstrate that finetuning pre-trained LLMs with Minnow improves their ability to discriminate between new words, identify syntactic categories of new words, and generate reasonable new usages and definitions for new words, based on one or a few in-context examples. These findings highlight the data efficiency of Minnow and its potential to improve language model performance in word learning tasks.

EMNLP 2025

Rapid Word Learning Through Meta In-Context Learning

in-context learning

language acquisition

meta-learning

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** 
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** 
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) 
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) 
1415: [Invited Symposium: Dynamics between minds and the environment](https://underline.io/events/465/sessions?eventSessionId=18016&searchGroup=lecture) 
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) 

**Friday, July 26** 
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) 
1030: [Rumelhart Symposium: Childhood as exploration](https://underline.io/events/465/sessions/18038/lecture/100346-childhood-as-exploration) 
1415: [Elman Prize Symposium](https://underline.io/events/465/sessions?eventSessionId=18053&searchGroup=lecture) 
1600: [Invited Symposium: Dynamics between minds](https://underline.io/events/465/sessions?eventSessionId=18065&searchGroup=lecture) 
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) 

**Saturday, July 27** 
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) 
1000: [Glushko Talks](https://underline.io/events/465/sessions?eventSessionId=18086&searchGroup=lecture) 
1500: [Invited Symposium: Dynamics within minds](https://underline.io/events/465/sessions?eventSessionId=18100&searchGroup=lecture) 


#### **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** 
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** 
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.

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.

Humans can quickly learn a new word from a few illustrative examples, and then systematically and flexibly use it in novel contexts. Yet the abilities of current language models for few-shot word learning, and methods for improving these abilities, are underexplored. In this study, we introduce a novel method, Minnow. This method trains language models to generate new examples of a word's usage given a few in-context examples, using a special placeholder token to represent the new word. This training is repeated on many new words to develop a general word-learning ability. We find that training models from scratch with Minnow on human-scale child-directed language enables strong few-shot word learning, comparable to a large language model (LLM) pre-trained on orders of magnitude more data. Furthermore, through discriminative and generative evaluations, we demonstrate that finetuning pre-trained LLMs with Minnow improves their ability to discriminate between new words, identify syntactic categories of new words, and generate reasonable new usages and definitions for new words, based on one or a few in-context examples. These findings highlight the data efficiency of Minnow and its potential to improve language model performance in word learning tasks.

poster

## Welcome!
"I am excited to welcome you to this year’s edition of the Conference on Empirical Methods in Natural Language Processing! Importantly, it marks the 30th edition of EMNLP. With over 8,000 submissions, more than 3,000 accepted papers, and thousands of attendees, we have come a long way from that first
workshop, which had 14 accepted papers. As the field looks ahead, Suzhou is the fitting location for celebrating this milestone: rooted in a long literary tradition, yet modern and forward-looking, and home to a large share of the NLP community." 

*Message from the General Chair, Dirk Hovy*

Celebrate 30 Years of EMNLP! 
EMNLP 2025 will be held in Suzhou, China from November 5th to November 9th, 2025.

Large language models (LLMs) are frequently fine-tuned or unlearned to adapt to new tasks or eliminate undesirable behaviors. While existing evaluation methods assess performance after such interventions, there remains no general approach for detecting unintended side effects—such as unlearning biology content degrading performance on chemistry tasks, particularly when these effects are unpredictable or emergent. To address this issue, we introduce MNEME, textitModel diffiNg for Evaluating Mechanistic Effects, a lightweight framework for identifying these side effects using sparse model diffing. MNEME compares base and fine-tuned models on task-agnostic data (e.g., The Pile, LMSYS-Chat-1M), without access to fine-tuning data, to isolate behavioral shifts. Applied to five LLMs across three scenarios, WMDP knowledge unlearning, emergent misalignment, and benign fine-tuning—MNEME achieves up to 95\% accuracy in predicting side effects, aligning with known benchmarks and requiring no custom heuristics. Furthermore, we show that retraining on high-activation samples can partially reverse these effects. Our results demonstrate that sparse probing and diffing offer a scalable and automated lens into fine-tuning-induced model changes, providing practical tools for understanding and managing LLM behavior.

REVIVING YOUR MNEME: Predicting The Side Effects of LLM Unlearning and Fine-Tuning via Sparse Model Diffing

Most existing Theory of Mind (ToM) benchmarks for foundation models rely on variations of the Sally-Anne test, offering only a very limited perspective on ToM and neglecting the complexity of human social interactions. To address this gap, we propose ToM-SSI: a new benchmark specifically designed to test ToM capabilities in environments rich with social interactions and spatial dynamics. While current ToM benchmarks are limited to text-only or dyadic interactions, ToM-SSI is multimodal and includes group interactions of up to four agents that communicate and move in situated environments. This unique design allows us to study, for the first time, mixed cooperative-obstructive settings and reasoning about multiple agents' mental state in parallel, thus capturing a wider range of social cognition than existing benchmarks. Our evaluations reveal that current models' performance is still severely limited, especially in these new tasks, highlighting critical gaps for future research.

ToM-SSI: Evaluating Theory of Mind in Situated Social Interactions

Despite Large Language Models (LLMs) demonstrating superior translation performance and long-context capabilities, evaluation methodologies remain constrained to sentence-level assessment due to dataset limitations, token number restrictions in metrics, and rigid sentence boundary requirements. We introduce an evaluation scheme that extends existing automatic metrics to long-document translation by treating documents as continuous text and applying sentence segmentation and alignment methods. Our approach enables previously unattainable document-level evaluation, handling translations of arbitrary length generated with document-level prompts while accounting for under-/over-translations and varied sentence boundaries. Experiments show our scheme significantly outperforms existing long-form document evaluation schemes, while comparable to evaluations performed with groundtruth sentence alignments. Additionally, we apply our scheme to book-length texts and newly demonstrate that many open-weight LLMs fail to effectively translate documents at their reported maximum context lengths.

Extending Automatic Machine Translation Evaluation to Book-Length Documents

We present NN-Rank, an algorithm for ranking source languages for cross-lingual transfer, which leverages hidden representations from multilingual models and unlabeled target-language data. We experiment with two pretrained multilingual models and two tasks: part-of-speech tagging (POS) and named entity recognition (NER). We consider 51 source languages and evaluate on 56 and 72 target languages for POS and NER, respectively. When using in-domain data, NN-Rank beats state-of-the-art baselines that leverage lexical and linguistic features, with average improvements of up to 35.56 NDCG for POS and 18.14 NDCG for NER. As prior approaches can fall back to language-level features if target language data is not available, we show that NN-Rank remains competitive using only the Bible, an out-of-domain corpus available for a large number of languages. Ablations on the amount of unlabeled target data show that, for subsets consisting of as few as 25 examples, NN-Rank produces high-quality rankings which achieve 92.8% of the NDCG achieved using all available target data for ranking.

Model-Based Ranking of Source Languages for Zero-Shot Cross-Lingual Transfer

To mitigate the hallucination problem in large language models, DoLa exploits early exit logits from the same model as a contrastive prior. However, we found that these early exit logits tend to be flat, low in magnitude, and fail to reflect meaningful contrasts. To address this, we propose PruneCD, a novel contrastive decoding method that constructs the amateur model via layer pruning rather than early exit. This design leads to more informative and well-aligned logits, enabling more effective contrastive decoding. Through qualitative and quantitative analyses, we demonstrate that PruneCD consistently improves factuality with minimal inference overhead, offering a robust and practical approach to mitigating hallucinations in LLMs.

PruneCD: Contrasting Pruned Self Model to Improve Decoding Factuality

Cognitive Restructuring (CR) uses multi-turn dialogue to identify and restructure one's negative thoughts, arising from mental health issues, into more helpful and positive ones. Clinician shortage and stigma urge the development of human-LLM interactive psychotherapy for CR. Yet, effectively implementing CR is hindered by entrenched cognitive distortions, emotional resistance, and individual differences, which existing works have not overcome. To bridge this gap, we propose CRDial, a novel framework that structures CR as theory-grounded multi-stage multi-turn dialogue, integrating multi-aspect supportive strategies for emotional management and a multi-channel loop mechanism to account for diverse individual distortions. With CRDial, we distill Crisp, a large-scale and high-quality bilingual dialogue dataset, from LLM. We then train Crispers, Crisp-based conversational LLMs for CR, at 7B and 14B scales. Extensive human studies show the superiority of Crispers in pointwise, pairwise, and intervention evaluations.

Crisp: Cognitive Restructuring of Negative Thoughts through Multi-turn Supportive Dialogues

Neural text-based models for detecting machine-translated texts can rely on named entities (NEs) as classification shortcuts. While masking NEs encourages learning genuine translationese signals, it degrades the classification performance. Incorporating speech features compensates for this loss, but their interaction with NE reliance requires careful investigation. Through systematic attribution analysis across modalities, we find that bimodal integration leads to more balanced feature utilization, reducing the reliance on NEs in text while moderating overemphasis attribution patterns in speech features. All the code and data will be publicly released.

From Shortcuts to Balance: Attribution Analysis of Speech-Text Feature Utilization in Distinguishing Original from Machine-Translated Texts

Large language models (LLMs) have shown remarkable performance in general translation tasks. However, the increasing demand for high-quality translations that are not only adequate but also fluent and elegant. To assess the extent to which current LLMs can meet these demands, we introduce a suitable benchmark (PoetMT) for translating classical Chinese poetry into English. This task requires not only adequacy in translating culturally and historically significant content but also a strict adherence to linguistic fluency and poetic elegance. Our study reveals that existing LLMs fall short of this task. To address these issues, we propose RAT, a Retrieval-Augmented machine Translation method that enhances the translation process by incorporating knowledge related to classical poetry. Additionally, we propose an automatic evaluation metric based on GPT-4, which better assesses translation quality in terms of adequacy, fluency, and elegance, overcoming the limitations of traditional metrics. Our dataset and code will be made available. Our dataset and code will be available upon acceptance.

Benchmarking LLMs for Translating Classical Chinese Poetry: Evaluating Adequacy, Fluency, and Elegance

Large language model (LLM) role-playing has gained widespread attention. Authentic character knowledge is crucial for constructing realistic LLM role-playing agents. However, existing works usually overlook the exploration of LLMs' ability to detect characters' known knowledge errors (KKE) and unknown knowledge errors (UKE) while playing roles, which would lead to low-quality automatic construction of character trainable corpus. In this paper, we propose RoleKE-Bench to evaluate LLMs' ability to detect errors in KKE and UKE. The results indicate that even the latest LLMs struggle to detect these two types of errors effectively, especially when it comes to familiar knowledge. We experimented with various reasoning strategies and propose an agent-based reasoning method, Self-Recollection and Self-Doubt (S²RD), to explore further the potential for improving error detection capabilities.

Revealing and Mitigating the Challenge of Detecting Character Knowledge Errors in LLM Role-Playing

Minimum Bayes risk (MBR) decoding is a decision rule of text generation, which selects the hypothesis that maximizes the expected utility, and robustly generates higher-quality texts than maximum a posteriori (MAP) decoding. However, it depends on sample texts drawn from the text generation model; thus, it is hard to choose the hypothesis that reflects the knowledge or information of out-of-domain. To tackle this issue, we propose case-based decision-theoretic (CBDT) decoding, another approach to the expected utility estimation using examples of domain data. CBDT decoding not only generates higher-quality texts than MAP decoding, but also the combination of MBR and CBDT decoding outperformed MBR decoding in seven domain GermanrightarrowEnglish and JapaneseleftrightarrowEnglish translation tasks and image captioning tasks on MSCOCO and nocaps datasets.

Downloads

Next from EMNLP 2025

REVIVING YOUR MNEME: Predicting The Side Effects of LLM Unlearning and Fine-Tuning via Sparse Model Diffing

Similar lecture

Dynamic Processes of Learning Words from Context