Netherlands

Value alignment is essential for building AI systems that can safely and reliably interact with people. However, what a person values---and is even capable of valuing---depends on the concepts that they are currently using to understand and evaluate what happens in the world. The dependence of values on concepts means that concept alignment is a prerequisite for value alignment---agents need to align their representation of a situation with that of humans in order to successfully align their values. Here, we formally analyze the concept alignment problem in the inverse reinforcement learning setting, show how neglecting concept alignment can lead to systematic value mis-alignment, and describe an approach that helps minimize such failure modes by jointly reasoning about a person&#39;s concepts and values. Additionally, we report experimental results with human participants showing that humans reason about the concepts used by an agent when acting intentionally, in line with our joint reasoning model.

**Authors:**

Sunayana Rane: Princeton University; Mark K Ho: Stevens Institute of Technology; Ilia Sucholutsky: Princeton University; Tom Griffiths: Princeton University 

CogSci 2024

Concept Alignment as a Prerequisite for Value Alignment

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

ACL 2026

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

Value alignment is essential for building AI systems that can safely and reliably interact with people. However, what a person values---and is even capable of valuing---depends on the concepts that they are currently using to understand and evaluate what happens in the world. The dependence of values on concepts means that concept alignment is a prerequisite for value alignment---agents need to align their representation of a situation with that of humans in order to successfully align their values. Here, we formally analyze the concept alignment problem in the inverse reinforcement learning setting, show how neglecting concept alignment can lead to systematic value mis-alignment, and describe an approach that helps minimize such failure modes by jointly reasoning about a person's concepts and values. Additionally, we report experimental results with human participants showing that humans reason about the concepts used by an agent when acting intentionally, in line with our joint reasoning model.

**Authors:**

Sunayana Rane: Princeton University; Mark K Ho: Stevens Institute of Technology; Ilia Sucholutsky: Princeton University; Tom Griffiths: Princeton University 

poster

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!

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

A hallmark of human vision is the ability to rapidly recognize objects in a complex naturalistic scene. However, the exact mechanisms behind the computational invariance of object recognition remain unknown. In this study, we investigate object constancy by estimating how the ventral visual stream processes shading, shadows, textures, and specularities. To accomplish this, we use object meshes from the Objaverse dataset to create distinct multiclass classification tasks. For every task, we render a dataset by excluding exactly one of the previously stated features at a time. Subsequently, we train a ResNet50 model on each dataset. The trained model is evaluated on Brain-Score; deviations in these metrics indicate the importance of a brain region in achieving invariance to a specific feature. A reduced score for a removed feature in a particular region implies its crucial role in processing that feature since the model classifies objects based on remaining scene intrinsics.

**Authors:**

Shreya Kapoor: Friedrich-Alexander-Universität Erlangen-Nürnberg; Bernhard Egger: Friedrich-Alexander-Universität Erlangen-Nürnberg

Processing of scene intrinsics in the ventral visual stream for object recognition

Understanding abstraction is a stepping stone towards understanding intelligence. We ask the question: How do abstract representations arise when learning sequences? From a normative perspective, we show that abstraction is necessary for an intelligent agent when the perceptual sequence contains objects of similar interaction properties appearing in identical contexts. A rational agent should identify categories of objects of similar properties as an abstract concept, enabling the discovery of higher-order sequential relations that span a longer part of the sequence. We propose a hierarchical variable learning model (HVM) that learns chunks and abstract concepts from sequential data in a cognitively plausible manner. HVM gradually discovers abstraction via a conjunction of variable discovery and chunking, resembling the process of concept discovery during development. In a sequence recall experiment that demands learning and transferring variables, we observe that the model’s sequence complexity can explain human behavior in a sequence memorization experiment.

**Authors:**

Shuchen Wu: Max Planck Institute for Biological Cybernetics; Mirko Thalmann: Max Planck Institute for Biological Cybernetics; Eric Schulz: Max Planck Institute for Biological Cybernetics

Learning abstractions from discrete sequences

How do children acquire language through unsupervised or noisy supervision and what happens in their brain? Most prior works primarily focused on tracking the co-occurrence between individual words and referents. We model cross-situational learning (CSL) with full sentence on a few training examples (1000 sentences) with reservoir computing (RC) and LSTMs, on three datasets including complex grounding robotics. We observe that reservoirs yield superior performance over LSTMs, indicating that more biological plausibility can also lead to better performance in sample efficient context. LSTMs with few hidden units exhibit good performance for small size vocabulary datasets, while more hidden units are needed to perform reasonably well on larger corpora. Importantly, reservoirs demonstrate robust generalization with increasingly large vocabularies while LSTMs do no show such behavior even if the number of hidden units is greatly increased. We analysed the evolution of internal representations during training and propose explanations why resservoirs generalize faster.

**Authors:**

Xavier Hinaut: Inria; SUBBA REDDY OOTA: Inria; Alexandre Variengien: Inria; Frederic Alexandre: Inria-Research

Modelling Cross-Situational Learning on Full Sentences in Few Shots with Simple RNNs

While humans have the remarkable ability to learn concepts from few examples, machine learning algorithms oftentimes require complex architectures that struggle to learn from minimal data. We introduce a simple computational framework for one-shot learning to encode higher-level representations of visual concepts using Gaussian Mixture Models (GMMs). Distinct topological subparts of concepts are represented as inferred Gaussian components, which can generate abstracted subparts to build robust prototypes for each concept. Our framework addresses both one-shot classification tasks through a similarity metric inspired by Tverksy’s (1977) contrast model, as well as one-shot generative tasks through a novel pipeline employing variational autoencoders (VAEs) to generate new class variants. Our approach yields impressive classification accuracy while also performing a breadth of conceptual tasks that most approaches do not even attempt. Results from human judges reveal that our generative pipeline produces novel classes of visual concepts broadly indistinguishable from those made by humans.

**Authors:**

Chelsea Zou: Binghamton University; Kenneth Kurtz: Binghamton University

Abstracted Gaussian Prototypes for One-Shot Concept Learning

In visual cognitive neuroscience, there are two main theories about the function of the ventral visual system. One suggests that it serves to classify objects (H1); the other suggests that it generates intermediate representations from which people can generate verbal descriptions, actions, and other kinds of information (H2). To adjudicate these, we trained two deep convolutional AlexNet models on 330,000 images belonging to 86 classes, representing the intersection of Ecoset images and the semantic norms collected by the Leuven group. One model was trained to produce category labels (H1) , the other to generate all of an item’s semantic features (H2). The two models learned very different representational geometries throughout the network. The representations acquired by the feature-generating model aligned better with human-perceived similarities amongst images, and better predicted human judgments in a triadic comparison task. The results thus support H2.

**Authors:**

Siddharth Suresh: University of Wisconsin-Madison; Wei-Chun Huang: University of Wisconsin-Madison; Kushin Mukherjee: University of Wisconsin-Madison; Timothy T Rogers: University of Wisconsin- Madison

Can deep convolutional networks explain the semantic structure that humans see in photographs?

In hierarchical perception-action (PA) learning, agents discover invariants between percepts and actions that are structured hierarchically, from very basic immediate links to higher-level, more abstract notions. In practice, existing work tends to either focus on the general theory at the expense of details of the proposed mechanisms, or specify a-priori the contents of some layers. Here, we introduce a framework that does without such constraints. We demonstrate the framework in a simple 2D environment using an agent that has minimal perceptual and action abilities. We vary the perceptual abilities of the agent to explore how the specifics of this aspect of the agent's body might affect PA learning and find unexpected consequences. The contribution of this paper is therefore twofold, (1) we add a novel framework to the literature on PA learning, using, in particular curiosity-based reinforcement learning (RL) to implement the necessary learning mechanisms, and (2) we demonstrate that even for very simple agents, the relation between the specifics of an agent's body and its cognitive abilities is not straightforward.

**Authors:**

Tara Carette: Radboud University; Serge Thill: Radboud University

A general framework for hierarchical perception-action learning

Compositionality is considered a key hallmark of human language. However the degree to which people utilize compositionality in practice has not been directly quantified and recent studies showing re-use of learned linguistic “chunks” suggest that compositionality may be employed less frequently than commonly believed. Here, we aim to assess the degree of compositionality in English, focusing on the case of adjective-noun combinations as a testbed for our approach. Using corpus analysis, large language models (LLMs), and newly collected human acceptability ratings, we find that: (1) English vastly underutilizes the compositional potential of adjective–noun combinations; and (2) LLMs struggle to predict human acceptability ratings of unattested adjective-noun pairs, exposing an important shortcoming of these state-of-the-art models. Taken together, our findings shed new light on the role of compositionality in language and suggest a new challenging benchmark for further improving LLMs.

**Authors:**

Aalok Sathe: Massachusetts Institute of Technology; Evelina Fedorenko: Massachusetts Institute of Technology; Noga Zaslavsky: UCI

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

The space of human goals is tremendously vast; and yet, from just a few moments of watching a scene or reading a story, we seem to spontaneously infer a range of plausible motivations for the people and characters involved. What explains this remarkable capacity for intuiting other agents' goals, despite the infinitude of ends they might pursue? And how does this cohere with our understanding of other people as approximately rational agents? In this paper, we introduce a sequential Monte Carlo model of open-ended goal inference, which combines top-down Bayesian inverse planning with bottom-up sampling based on the statistics of co-occurring subgoals. By proposing goal hypotheses related to the subgoals achieved by an agent, our model rapidly generates plausible goals without exhaustive search, then filters out goals that would be irrational given the actions taken so far.

We validate this model in a goal inference task called Block Words, where participants try to guess the word that someone is stacking out of lettered blocks. In comparison to both heuristic bottom-up guessing and exact Bayesian inference over hundreds of goals, our model better predicts the mean, variance, efficiency, and resource rationality of human goal inferences, achieving similar accuracy to the exact model at a fraction of the cognitive cost, while also explaining garden-path effects that arise from misleading bottom-up cues. Our experiments thus highlight the importance of uniting top-down and bottom-up models for explaining the speed, accuracy, and generality of human theory-of-mind.

**Authors:**

Tan Zhi-Xuan: Massachusetts Institute of Technology ; Gloria Kang: Massachusetts Institute of Technology; Vikash Mansinghka: MIT; Josh Tenenbaum: MIT

Infinite Ends from Finite Samples: Open-Ended Goal Inference as Top-Down Bayesian Filtering of Bottom-Up Proposals

The commonsense question answering (CQA) system is a type of conversational system that can generate answers to commonsense questions posed by human users. Most current research on commonsense question-answering has focused on proposing different techniques in natural language processing and text information retrieval. However, for humans, retrieving and organizing desired answers from text knowledge related to commonsense questions is far less intuitive and comprehensive than it is when using multi-modal knowledge, such as related images and videos. Motivated by this, our paper proposes a framework for trying the acquisition of diverse modal information from various perspectives, and embedding and integrating it into CQA tasks, further improving the performance and user experience. Specifically, this paper proposes the integration of multi-modal knowledge, including images retrieved with questions, image description statements, image scene graphs, and knowledge sub-graphs retrieved with questions, into a CQA system. It introduces a parallel embedding technique for this multi-modal knowledge and employs an alignment-interaction-fusion mechanism to facilitate the seamless integration of this multi-modal knowledge. Through extensive and rigorous experiments, the effectiveness and superiority of our proposed method are demonstrated, achieving a notable 4.0% and 3.4% improvement in accuracy on the OpenBookQA and CommonsenseQA datasets, respectively.

**Authors:**

Yongqiang Zhao: Peking University; Zhi Jin: Peking University; Feng Zhang: Academy of Military Sciences; Xinhai Xu: Academy of Military Sciences; Donghong Liu: Academy of Military Sciences

Bridging the Gap: Advancing Commonsense Question Answering with Integrated Multi-Modal Knowledge

People’s color-concept associations influence many processes underlying visual cognition from object recognition to information visualization interpretation. Thus, a key goal in cognitive science is developing efficient methods for estimating color-concept association distributions over color space to model these processes. Here, we investigated the ability of GPT-4, a multimodal large language model, to estimate human-like color-concept associations. We collected human association ratings between 70 concepts spanning abstractness and 71 colors spanning perceptual color space and compared these ratings to analogous ratings from GPT-4, when it was given concepts as words and colors as hexadecimal codes. GPT-4 ratings were correlated with human ratings, comparably to state-of-the-art image-based methods. Variation in human-GPT rating correlations across concepts was predicted by concept abstractness, but this effect was superseded by specificity (peakiness; inverse entropy) of color-concept association distributions. Our results highlight the viability of using model-generated color-concept association ratings to better understand human color semantics.

**Authors:**

Kushin Mukherjee: University of Wisconsin-Madison; Timothy Rogers: UW-Madison; Karen Schloss: University of Wisconsin - Madison

Premium content

Downloads

Next from CogSci 2024

Processing of scene intrinsics in the ventral visual stream for object recognition

Similar lecture

VALUE ALIGNMENT TAX: Measuring Value Trade-offs in LLM Alignment