United States

The Aspect Hypothesis suggests children’s early verb production choices to use perfect or progressive verb form (-ed vs. -ing respectively) can depend on event semantics (Shirai &amp; Andersen 1995). Children tend to use perfective constructions with verbs denoting completion, while present constructions mostly occur with verbs denoting ongoing actions. Li and Shirai (2000) suggest that children may stray from this pattern as caregiver input changes throughout development.

However, these changes are underexplored, and previous findings supporting the Aspect Hypothesis emerge from limited corpora. Our study used NLP on all English corpora in the CHILDES corpus to extract the main verbs from the utterances of children and caregivers. We confirmed children and caregivers’ general adherence to the predicted pattern. We also found preliminary support for caregivers’ shifting in their inflection of multiple verb types. Our findings support the Aspect Hypothesis and provide insight into how children come to broaden their inflections.

CogSci 2025

Testing the Aspect Hypothesis: Relating child and caregiver verb inflection across development

developmental analysis

corpus studies

language production

psychology

language acquisition

morphology

linguistics

The Aspect Hypothesis suggests children’s early verb production choices to use perfect or progressive verb form (-ed vs. -ing respectively) can depend on event semantics (Shirai & Andersen 1995). Children tend to use perfective constructions with verbs denoting completion, while present constructions mostly occur with verbs denoting ongoing actions. Li and Shirai (2000) suggest that children may stray from this pattern as caregiver input changes throughout development.

However, these changes are underexplored, and previous findings supporting the Aspect Hypothesis emerge from limited corpora. Our study used NLP on all English corpora in the CHILDES corpus to extract the main verbs from the utterances of children and caregivers. We confirmed children and caregivers’ general adherence to the predicted pattern. We also found preliminary support for caregivers’ shifting in their inflection of multiple verb types. Our findings support the Aspect Hypothesis and provide insight into how children come to broaden their inflections.

poster

### 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.

Learning new information without forgetting prior knowledge is central to human intelligence. In contrast, neural network models suffer from catastrophic forgetting: a significant degradation in performance on previously learned tasks when acquiring new information. The Complementary Learning Systems (CLS) theory offers an explanation for this human ability, proposing that the brain has distinct systems for pattern separation (encoding distinct memories) and pattern completion (retrieving complete memories from partial cues). To capture these complementary functions, we leverage the representational generalization capabilities of variational autoencoders (VAEs) and the robust memory storage properties of Modern Hopfield networks (MHNs), combining them into a neurally plausible continual learning model.  We evaluate this model on the Split-MNIST task, a popular continual learning benchmark, and achieve close to state-of-the-art accuracy ~90%, substantially reducing forgetting. Representational analyses empirically confirm the functional dissociation: the VAE underwrites pattern completion, while the MHN drives pattern separation. By capturing pattern separation and completion in scalable architectures, our work provides a functional template for modeling memory consolidation, generalization, and continual learning in both biological and artificial systems.

A Neural Network Model of Complementary Learning Systems: Pattern Separation and Completion for Continual Learning

Recent studies use semantic structural priming to show that various cases of linguistic strengthening happen through a common mechanism: generation of implicatures through alternative-based (scalar) reasoning. In this paper, we used priming to investigate another group of cases, where strengthening is postulated to follow from the tendency to systematically neglect structures that verify a sentence by virtue of an empty configuration (neglect-zero): empty-set quantifiers ('at most/fewer than') and disjunction under a universal quantifier. We report data indicating semantic priming between these two structures, but not between them and scalar 'some'. We propose that 1. there is a common mechanism in use for strengthening constructions postulated to follow from the neglect-zero tendency, and that 2. this mechanism is different from the one involved in alternative-based reasoning.

Neglect zero: evidence from priming across constructions

Recent advancements in Large Language Models (LLMs) have brought them closer to matching human cognition across a variety of tasks. How well do these models align with human performance in detecting and mapping analogies? Prior research has shown that LLMs can extract similarities from analogy problems but lack robust human-like reasoning. Building on Webb, Holyoak, and Lu (2023), the current study focused on a story-based analogical mapping task and conducted a fine-grained evaluation of LLM reasoning abilities compared to human performance. First, it explored the semantic representation of analogies in LLMs, using sentence embeddings to assess whether they capture the similarity between the source and target texts of an analogy, and the dissimilarity between the source and distractor texts. Second, it investigated the effectiveness of explicitly prompting LLMs to explain analogies. Throughout, we examine whether LLMs exhibit similar performance profiles to those observed in humans by evaluating their reasoning at the level of individual analogies, and not just at the level of overall accuracy (as prior studies have done). Our experiments include evaluating the impact of model size (8B
vs. 70B parameters) and performance variation across state-of-the- art model architectures such as GPT-4 and LLaMA3. This work advances our understanding of the analogical reasoning abilities of LLMs and their potential as models of human reasoning.

Modeling Understanding of Story-Based Analogies Using Large Language Models

Recent legal rulings have denied copyright protection to artworks derived from AI-generated sources, because AI is assumed to be incompatible with qualities that define human authorship. We empirically test lay intuitions related to these assumptions in two studies (N = 235, N = 119) by investigating how creator attribution of initial source material (AI- vs. human-generated), effort investment in generating source material, and modification level of a derivative work influence perceptions of transformativeness, essence change, and creativity in derivative artworks. Modification level exerted the strongest influence across all measures, with dramatic modifications rated significantly higher than slight or no modifications. Effort investment in generating source material only influenced creativity ratings, with less effort sometimes perceived as more creative. Most notably, creator attribution for source material had minimal impact. These results challenge current copyright doctrine by demonstrating that lay human observers prioritize degree of transformation over both effort and creator attribution for source material. Our findings suggest that legal frameworks should recognize that  AI assistance in generating artworks does not preclude a genuine human contribution that merits copyright protection.

Is AI-assisted Creativity an “Original Sin”?: Lay Judgments of Qualities Justifying Copyright Protection for Artworks Derived from AI- vs. Human-generated Sources

Automatic depression detection is gaining traction at the intersection of psychology and machine learning, but concerns over data privacy and scarcity persist. We propose a pipeline using Large Language Models (LLMs) to generate synthetic data that enhances depression prediction while addressing ethical concerns. Starting from recorded clinical transcripts, our chain-of-thought prompting involves two steps: the generation of a synopsis and sentiment analysis from the original transcript, and the generation of synthetic data based on these summaries and a new depression score. The resulting synthetic data not only performs well in terms of utility and fidelity, but also balances the severity distribution in training datasets, improving prediction of depression intensity. Our method offers a practical solution to augmenting limited, sensitive data while preserving statistical integrity. This framework provides a robust framework for advancing mental health research and applications without compromising patient privacy.

Synthetic Data Generation with Large Language Models for Improved Depression Prediction

Large Language Models like ChatGPT are becoming every-day writing partners in the workplace. This study asked: how does simply knowing an email was “edited by ChatGPT” affect its persuasiveness and the perceived cred-ibility of the sender? We collected data from 308 profes-sionals using experimental vignettes that simulated realis-tic workplace emails. Some emails were described as en-tirely human-written, while others were labeled as AI-edited, with variations in the sender's reliability (who is sending the message) and strength of the argument (how well the content is constructed). A Bayesian Model of Ar-gumentation provided normative predictions for how reli-ability and argument quality should influence persuasion. We found that when an email was labeled as “edited by ChatGPT,” receivers saw it as less persuasive overall. However, AI-mediation did not diminish the relative in-fluence of source reliability and argument quality. In other words, while the AI-edited label reduced overall persua-siveness, it didn’t change how recipients inherently evalu-ated credibility. They still adjusted their beliefs primarily based on who sent the message and how strong the argu-ment was. To our knowledge, this is the first study to ap-ply a Bayesian framework to understanding how people process AI-mediated communication.

“Can you tell I used ChatGPT?” How Perceived AI-Mediation Affects Workplace Email Persuasiveness— A Bayesian Approach

Increased cognitive load is linked to decreased fixation counts and longer dwell times during visual search. However, past work studied eye movements in 2D tasks, whereas in 3D environments the head and body help the eyes gather information. Thus, we examined how cognitive load affects the eyes, head, and body in 3D visual search. Cognitive load may create difficulty planning the eye, head, and body movements needed to search a 3D space. We used an eye-tracker to record gaze and inertial sensors to measure the motion of the head and body during a dual task paradigm: Participants searched a set of 27 images distributed in a 270º space for stimuli of a specified criteria. Cognitive load was manipulated by having participants count backwards by set intervals (1, 3, 5, 7, and no counting). Results showed a decrease in total fixations and an increase in stimulus dwell times when under load.

The Effects of Cognitive Load on Full-Body Gaze Control During 3D Visual Search

We commonly use the language of physics to describe social interactions, even those that do not involve physical contact at all (e.g., pushing, pressuring, blocking). Is there a common conceptual basis for perceiving causal interactions? This study asks whether there is a shared neural code for physical and social interactions. For instance, when a gust of wind knocks a sailboat off its trajectory, is that interaction processed similarly to when an employee is kept from going home by their boss? To investigate domain-invariant representations of causal interactions involving enabling vs. preventing, we presented fMRI participants with vignettes depicting physical interactions between objects or purely social interactions (agents interacting socially but not physically). Multivoxel pattern analyses revealed that brain regions classically associated with physical reasoning and social reasoning contained domain-invariant representations of enabling vs. preventing. These results suggest a shared conceptual structure for making sense of physical and social interactions.

Pushing people: the neural basis of social interaction perception

Short-term phonetic accommodation is a fundamental driver behind accent change, but how does real-time input from another speaker's voice shape the speech planning representations of an interlocutor? We advance a computational model of change in speech planning representations during phonetic accommodation, grounded in dynamic neural field equations for movement planning and memory dynamics. A dual-layer planning/memory field predicts that convergence to a model talker on one trial can trigger divergence on subsequent trials, due to a delayed inhibitory effect in the more slowly evolving memory field. The model's predictions are compared with empirical patterns of accommodation from an experimental pilot study. We show that observed empirical phenomena may correspond to variation in the magnitude of inhibitory memory dynamics, which could reflect resistance to accommodation due to phonological and/or sociolinguistic pressures. We discuss the implications of these results for the relations between short-term phonetic accommodation and sound change.

Phonetic accommodation and inhibition in a dynamic neural field model

Children must learn individual word meanings and the semantic connections among words. Prior work has probed children’s sensitivity to different semantic relations, but little is known about how this knowledge develops. This study examined whether parents and children jointly created semantic regularities in naturalistic everyday interactions. Forty-four parents and their toddlers (M = 20.4 months, range: 13.2-31 months) participated in a 10-minute free play session with 27 toys from three categories (food, animals, vehicles) while wearing head-mounted eye-trackers. Children’s looking from one object to another was categorized as a same category (e.g., dog-cat) or different category (e.g., dog-car) transition, revealing temporal sequences of semantically related toy play: Children were more likely to transition between objects from the same category and heard more unique words from the same category during these toy transitions. Together, these findings demonstrate that parents and children create semantic regularities for learning through shared attention and action.

Downloads

Next from CogSci 2025

A Neural Network Model of Complementary Learning Systems: Pattern Separation and Completion for Continual Learning

.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

A Neural Network Model of Complementary Learning Systems: Pattern Separation and Completion for Continual Learning

Downloads