United States

CogSci 2025

Do LLMs Comprehend the Meaning of Language?

## Welcome to EMNLP 2022!
I am delighted to welcome you to EMNLP 2022!  I believe this conference has been complicated beyond any precedent.  Over the past year, it’s been thrilling to see the organization team approach each new puzzle with creativity and enthusiasm.  We hope that those participating in Abu Dhabi as well as those joining remotely will leave the conference feeling newly inspired by the program and newly connected to our ever-growing community.  Following EMNLP 2021 and major NLP conferences since, EMNLP 2022 is “hybrid,” serving both virtual and in-person participants.

Our key innovations for EMNLP 2022 include:

* EMNLP 2022 is “hybrid” in a second sense, as well:  we allowed both direct and rolling review paper submissions, building on the pilot experiment of EMNLP 2021, which considered a small number of ARR submissions. 
* Familiar from NAACL but new to EMNLP, we’ve added an industry track.
* During the conference, “portals” will link virtual poster sessions to in-person conference participants during poster sessions each day.
* The first ACL-family conference in the United Arab Emirates.

<br>*Message from Noah A. Smith, University of Washington and Allen Institute for AI, Seattle, Washington, USA*<br>
***EMNLP 2022 General Chair***
<br><br><br>
[![](https://assets.underline.io/uploads/markdown_image/1/image/9eec7d4a287ee18c278b08229290aa83.png)](https://drive.google.com/file/d/1OlPv6QBeo62VVTughj2jkiLeyHd1WnUt/view)
<br><br>
[![](https://assets.underline.io/uploads/markdown_image/1/image/a3db7a768409f05192210d98601edb25.png)](https://emnlp2022.rocket.chat/)



To access this site you need to register. Please register [here](https://2022.emnlp.org/registration/).

Register here

EMNLP 2022

Welcome!
EMNLP 2022 will take place in Abu Dhabi from December 7th to December 11th, 2022. And it will be held in hybrid mode, both online and offline.

symposium

### Welcome to CogSci Conference 2025!

The 47th Annual Meeting of the Cognitive Science Society was a hybrid meeting held in San Francisco. 

<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/amvr/"  seamless="seamless" scrolling="no" frameborder="0" allowtransparency="true" allowfullscreen="true" ></iframe></div>

#### About

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.

#### Our History 

* **Society Creation**<br>
The Society was incorporated as a 501(c)(3) non-profit professional organization in Massachusetts in 1979. The organizing committee included Roger Schank, Allan Collins, Donald Norman, and a number of other scholars from psychology, linguistics, computer science, and philosophy. 
<br><br>
* **Conference Creation**<br>
The first conference on cognitive science was held at La Jolla, California in August, 1979, and has occurred annually since then. The proceedings of each conference are published, and those from most years are available through Lawrence Erlbaum Associates, Inc. The annual proceedings of the Cognitive Science Conference represent a major source of information on new work and new ideas in the scientific study of thinking. In 1990, the Society, with help from an anonymous donor, established the David Marr Prize for the best student paper at each annual meeting.
<br><br>
* **Journal Creation**<br>
The Journal, Cognitive Science, began publication in 1976, and is now published by Wiley-Blackwell. The Executive Editor is currently Richard P. Cooper of Birkbeck, University of London, and there are 18 Associate Editors and a 30-member editorial board. It serves as the premier outlet for research reports that intersect two or more disciplines. Copyrights for articles published in the journal are held by the Society. The Governing Board of the Cognitive Science Society voted in late 2006 to found a new journal, Topics in Cognitive Science (topiCS). The Editor in Chief is Wayne Gray, Cognitive Science Department, Rensselaer Polytechnic Institute. The journal seeks to fill a niche not occupied by Cognitive Science Journal or other cognitive science journals. Membership in the Society includes a subscription to Cognitive Science and TopiCS. Copyrights for articles published in the journal are held by the Society.
<br><br>

#### Code of Conduct

By attending the CogSci 2025 Conference, you are required to adhere to the society’s **[Code of Conduct](https://drive.google.com/file/d/1ChPuihLy6jE_BWqfO7J2KKgX35JW2zsM/view?usp=sharing)**.
<br><br>


You need to log in with the email address you registered with. 

Login credentials were sent to you from Underline -  subject line "Welcome to the CogSci 2025 Conference". Please be sure to check your spam/promotional inbox  if you do not see an email confirmation right away.





Please log in to join this event.

To access the site, please register [**here**](https://cognitivesciencesociety.org/registration/).

If you are registered and feel like you are seeing this message by mistake, please make sure you are logged in with the same email that you registered with. 

Please register!

The 47th Annual Meeting of the Cognitive Science Society presents the latest research across cognitive science and highlights the theme of Cognition in Context.

Deep reinforcement learning has achieved remarkable success in complex decision-making tasks, yet its black-box nature limits practical deployment in safety-critical domains. Current explainable reinforcement learning methods often fail to align with the hierarchical and temporal structure of human mental models, which are central to cognitive science theories of decision making. To bridge this gap, we propose Mental Model Alignment (MMA), a novel framework that constructs cognitive interfaces using behavior trees to harmonize AI decision-making with human-understandable reasoning. MMA introduces three innovations: (1) a mental model encoder that captures the hierarchical decomposition of tasks into subgoals, mirroring human cognitive processes; (2) a cognitive pruning algorithm that simplifies BTs while preserving decision-critical nodes aligned with human mental schemas; and (3) a mental effort metric to quantify the cognitive load required for users to interpret policies. Evaluated across six benchmark environments, MMA outperforms state-of-the-art methods in interpretability, policy fidelity, and computational efficiency. Our results demonstrate that aligning AI policies with human mental models significantly enhances trust and usability in real-world applications.

Mental Model Alignment: Building Cognitive Interfaces for Explainable Reinforcement Learning

We investigated gender differences in multimodal communication directed to children and adults. Eighty-two broadcasters (46 females and 36 males) participated in hosting adult-directed and child-directed broadcasting programmes respectively, and their lexical/syntactic features, prosody, and gestures were compared. Results revealed that broadcasters adapted their communication styles when addressing children. However, notable gender differences emerged: male broadcasters exhibited less diverse vocabulary, longer utterances, lower pitch but higher intensity, faster speaking rate with more pauses, and fewer referential gestures than their female counterparts. Furthermore, male broadcasters demonstrated larger adjustments in word frequency and vocal intensity but smaller adjustments in the use of questions and gestures than females. These findings highlight distinct patterns in how men and women adapt multimodal communication to children, offering valuable insights into gendered strategies in child-directed language production and recipient design. Moreover, it offers implications for developing tailored broadcast training.

The effect of gender on multimodal child-directed language: Evidence from analyses of broadcast programmes

Approximately half of the world’s languages are tonal, and how lexical tone is encoded in spoken word production is still unclear. In Mandarin word production, there are two contrasting views regarding the mechanisms of tonal encoding. The two-stage model assumes that the lexical tone is selected first at the early stage of production, and then integrated with the atonal syllable at the later stage, while other researchers proposed that the lexical tone is retrieved only at the later stage of production. In this study, we performed computational simulations on disyllabic words to uncover the mechanisms underlying the facilitation and interference effects on naming latencies observed in previous primed picture naming studies, which intended to verify the two theoretical accounts of tonal encoding in Mandarin spoken word production. The results supported the two-stage model of tonal encoding in disyllabic Mandarin word production. Increased inhibition between atonal syllables and decreased activation between the tonal frame and the syllable motor program, implying slower tone-to-syllable integration, appear to be the prerequisite for generating the interference effect of tonal overlap without shared syllabic information.

Computational Modeling of Tonal Encoding in Disyllabic Mandarin Word Production

Empathy has been linked to music preferences in controlled laboratory settings, but naturalistic settings like music streaming platforms remain unexplored. This study investigates how trait empathy influences music preferences and sequential listening behaviors in real-world settings. To this end, we collected and analyzed one-year Spotify listening histories of 290 Indian university students alongside their trait empathy scores, measured using the IRI scale. Our results reveal that individuals who score high on the IRI subscales of Empathic Concern, Fantasy, and Perspective Taking prefer sad music. Moreover, those scoring high on Empathic Concern or Perspective Taking were found to be more likely to transition from happy to sad music. These findings partially align with previous lab-based research, specifically for the subscales of Empathic Concern and Fantasy, while providing novel insights into the relations between the Perspective Taking subscale and music consumption. The study also provides novel insights into sequential listening behaviours, thus, strengthening the evidence that empathy shapes musical preferences and listening behaviors across diverse contexts.

Empathy and Music Preferences: Exploring Valence-Arousal Patterns, and Sequential Listening Behaviors in Naturalistic Settings

Social categorization processes often rely on observable, psychologically salient attributes, including ascribed characteristics (e.g., age, gender, nationality) and chosen affiliations (e.g., profession, ideology). These dimensions shape how individuals classify others into ingroups and outgroups, influencing perception, evaluation, and behavior. This study examined whether the strength of ingroup bias varies depending on whether group membership is endowed (inherent and unchosen) or agentic (self-selected). Using a within-subjects design, participants completed a perceptual matching task across two experiments: one comparing endowed groups (INDIA vs CHINA), the other comparing agentic groups (COGSCI vs GEOLOGY). Ingroup bias was measured through reaction times and accuracy. Results showed significantly stronger bias toward endowed groups, with a notable interaction between group type and task context. These findings suggest that the origin of group membership fundamentally shapes the intensity of social bias and offer new insights into the mechanisms underlying intergroup behavior and the dynamics of social categorization.

Chosen or Assigned: Exploring the Notion of Choice in the Manifestation of the Ingroup Bias

To advance our understanding of referential communication and common ground formation, this study presents a novel generative cognitive model that integrates deep neural networks for visual perception, image generation, and language captioning. Using the Tangram Naming Task (TNT), we simulate the sender–receiver interaction with modular processes replicating holistic cognitive strategies. Through controlled simulation experiments, we reveal that language generation plays a more crucial role than visual perception in establishing common ground, while intermediate image generation enhances linguistic diversity—a key aspect of natural communication. Our results bridge cognitive modeling and large generative models, demonstrating how internal cognitive dynamics can be visualized and quantitatively evaluated. This study contributes to the growing field of cognitive-inspired human–AI communication and provides a blueprint for grounding-rich simulations in collaborative tasks.

Common Ground Building through Generative Cognitive Modules: Examining the Roles of Initial Perception, Imaging and Captioning

Errors during sentence production have revealed crucial insights about the cognitive underpinnings of language processing. One such widely studied error is the agreement attraction error. Such errors occur when the subject-verb agreement, a crucial linguistic dependency, falters such that the verb shows the features of a ‘distractor’ noun rather than that of the target subject. Previous work on agreement attraction has established similar cross-linguistic patterns, such as the number mismatch asymmetry effect. Such research suggests that the underlying mechanism might be universal. Recent studies, however, indicate that Hindi employs a language-specific strategy during agreement processing that is not reported in other languages. This raises an important question: do the cross-linguistic patterns observed in agreement processing also manifest in Hindi? Our experiment addresses this gap by using a preamble repetition task to elicit errors. Based on the nature of mismatch asymmetry and the structure of Hindi nouns, we hypothesize that if number mismatch asymmetry occurs, it should be limited to feminine nouns in Hindi.

Our findings confirm the presence of mismatch asymmetry in Hindi but exclusively for feminine nouns. This suggests that while agreement mechanisms are indeed universal, they are influenced by language-specific configurations and strategies. Overall, our results can be interpreted better within a cue-based retrieval framework.

Interaction of language-specific and cross-linguistic strategies during agreement computation - Evidence from Hindi

Recently, AI-generated content (AIGC) technologies have made remarkable advancements, even achieving superhuman performance across various domains. However, few previous studies have investigated its impact on emotion-focused therapy with artistic content, e.g., music. In this paper, we conducted an EEG experiment to explore the effects of generative music on emotion-focused music therapy based on the ISO principle. This experiment compared AI-generated and human-created music regarding the changes in participants' valence and arousal following negative emotion induction with the ISO principle adherence and non-adherence. The results show that generative music, with its harmonic consistency and simple rhythm, is more effective in supporting positive emotions and improving temporal lobe activity. Besides, the therapeutic effectiveness of generative music adhering to the ISO principle has also been validated. This study highlights the distinct emotional and neural mechanisms of AI-generated music, offering valuable insights into future AI-powered emotion-focused therapy strategies.

How Generative Music Affects the ISO Principle-Based Emotion-Focused Therapy: An EEG Study

Empathy was mainly considered a stable trait, and few studies have investigated whether
it can vary across different situations. This investigation explores how contextual empathy
of study participants varies across different social group relationship and positive/negative
event valence. In this study, participants were divided into high- or low-empathy group by
their scores of Empathy Scale. The in/out group membership was manipulated through a
point estimation paradigm, and event valence was operationalized by the emotion status
of the character in the story event. As expected, results showed main effects across all
factors that participants demonstrated contextual empathy differently. More importantly, an
ingroup bias is significantly emerged, with participants exhibiting enhanced contextual
empathy toward ingroup than outgroup character. Furthermore, positive story events
elicited more contextual empathic responses than negative events. These findings
evidently provide an empirical support for the context-dependent nature of empathy,
challenging its traditional conceptualization as an invariant trait.

Contextual Malleability of Empathy: Effects of Trait Level, Group Membership, and Event Valence on Emotional Empathy

Access to higher education (HE) is heralded as a pathway to social mobility and equity but remains elusive for Indigenous populations in high-income countries like the United States and Australia. Systemic racial inequities, deeply rooted in colonial histories, perpetuate barriers to HE access and attainment for Native American and Aboriginal and Torres Strait Islander communities. This essay employs a comparative analysis based on statistics and Indigenous policy frameworks, using Critical Race Theory (CRT) and marketisation as analytical lenses to interrogate these challenges. It examines how “Whiteness” shapes educational discourses and institutional practices, reinforcing exclusion and inequality. Key disparities are analyzed, including lower enrollment, geographic isolation, socio-economic disadvantage, and financial barriers. Contrasting outcomes—declining Native American enrollment in the U.S. versus rising Indigenous completion rates in Australia—underscore the importance of community-led, equity-focused policies. The essay advocates for transformative reforms that prioritize Indigenous voices, dismantle systemic barriers, and address colonial and neoliberal legacies.

Downloads

Next from CogSci 2025

Mental Model Alignment: Building Cognitive Interfaces for Explainable Reinforcement Learning

Similar lecture

Meaning Representations for Natural Languages: Design, Models and Applications

.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 CogSci 2025

Mental Model Alignment: Building Cognitive Interfaces for Explainable Reinforcement Learning

Similar lecture

Meaning Representations for Natural Languages: Design, Models and Applications

Downloads