United States

Evidence is accumulating that patterns of use for singular “they” are changing in English. The pronoun is becoming the preferred generic when the gender of the referent is unknown or backgrounded. This change reflects a shift in patterns of acceptability for uses of singular “they” which is in turn linked to the increased frequency of singular they. We predict that adaptation may be a cognitive mechanism underlying this change, and if so, we may see short-term adaptation within a lab session. In the present study, we use a between-subjects priming paradigm to test whether participants adapt to the frequency with which they encounter singular or plural senses of “they” in the local discourse. We find that selections of singular “they” are significantly more likely after participants have been exposed to unambiguously singular vs. plural uses of “they”. This finding implicates adaptation and suggests that adaptation may link changes in the frequency of linguistic forms to changes in their acceptability.

CogSci 2025

Singular “they” exposure increases singular interpretations in ambiguous pronouns

statistical learning

language comprehension

psychology

linguistics

discourse

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.

Despite extensive research on mentalization, few studies target the representations and the cognitive systems that underlie different mental state attributions. In two eye-tracking experiments with adults (n=32) and 19-month-old infants (n=24), we examined whether factive (knowledge, ignorance) and non-factive (false belief, true belief) mental state attributions belong to seperate representational systems, relying on the assumption that transfer within-system should occur faster than between-systems. Participants watched animated videos of an agent tracking a hidden ball that could hide in two locations, requiring mental state attribution updates from non-factive to either another non-factive or to a factive mental state. Saccadic reaction times (SRTs) to the ball’s reappearance were measured. Results showed that both adults and infants had faster SRTs when updates occurred between two non-factive mental states compared to updates between a non-factive and a factive mental state. This supports the existence of distinct systems for factive and non-factive mental state attribution.

Evidence for Distinct Factive and Non-Factive Mentalization Systems in Adults and Infants

A well-known empirical result in human linguistic processing finds that humans are quicker to correctly recognize a string of letters as word when they are first shown a word that is semantically related to the word they must recognize. This is known as the “semantic priming effect.” Since Collins and Loftus (1975), it has been widely theorized that this effect is due to graphical storage of words in memory and a “spreading activation” model of priming. On this theory, words are related to one another in human semantic memory via a graphical structure encoding semantic relationships between words, with participants more likely to quickly recognize a word when they are primed with one that is graphically nearby; the prime word “activates” the node of a participant’s semantic memory network representing the prime word and this activation “spreads” to words at nearby nodes. Today, large language models increasingly excel at generating structured data representations, like graphs, when prompted to do so (Ghanem & Cruz, 2024; Dagdelen et al., 2024). In the current paper we investigate whether a language model can be prompted to represent a set of words as a semantic graph, and whether human reaction times in a word recognition task are predicted by the minimum path length between words in such an LLM-generated semantic graph. Using two versions of the Gemini language model, we use a prompting strategy to generate semantic graphs relating all words used in a large semantic priming experiment conducted by Hutchinson et al. (2013), under a variety of different temperatures and settings for the number of maximum output tokens. While we find that all LLM-generated semantic graphs produced during our experiments are such that the minimum path length between two words predicts the reaction time in which a person primed by one word recognizes the other, this effect is most pronounced for graphs generated via a smaller version of the model. It is under these conditions, we find, that LLMs produce the dense graphs that are more predictive of human semantic priming effects in lexical decision tasks.

LLM-Generated Semantic Networks Predict Semantic Priming Effects on Human Reaction Times in a Word-Recognition Task

Sense-Making, Cultural Scripts, and the Inferential Basis of Meaningful Experience

Can LLMs model meaning in restorying interventions?

We asked whether the concreteness and specificity of the language used by conversation participants change depending upon the familiarity and the presence/absence of an object discussed. Additionally, we explored whether interlocutors engaged in distinct abstraction processes (analogical comparison; superordinate categorization) and whether they focused more on object’s features or on their own experience.  

We used the ECOLANG corpus (Gu et al., 2025), a semi-naturalistic dataset of interactions in which 31 knowledgeable “speakers” describe novel/known objects to an “addressee” when the object is physically present or absent. We analyzed 22,581 sentences produced by the “speaker” and measured the concreteness and specificity of 1,612 nouns used.   

Results showed that more concrete and specific nouns were used for novel objects suggesting a need for precise information.  Additionally, abstraction processes were more likely when the object was present and novel. Finally, when the object was present and known, interlocutors focused more on personal experience.

Be concrete and specific: how speakers introduce novel topics in naturalistic language

Creativity assessment in science and engineering is increasingly based on both human and AI judgment, but the cognitive processes and biases behind these evaluations remain poorly understood. We conducted two experiments examining how including example solutions with ratings impact creativity evaluation, using a finegrained annotation protocol where raters were tasked with explaining their originality scores and rating for the facets of remoteness (whether the response is ``far'' from everyday ideas), uncommonness (whether the response is rare), and cleverness. In Study 1, we analyzed creativity ratings from 72 experts with formal science or engineering training, comparing those who received example solutions with ratings (example) to those who did not (no example). Computational text analysis revealed that, compared to experts with examples, no-example experts used more comparative language (e.g., ``better/worse'') and emphasized solution uncommonness, suggesting they may have relied more on memory retrieval for comparisons. In Study 2, parallel analyses with state-of-the-art LLMs revealed that models prioritized uncommonness and remoteness of ideas when rating originality, suggesting an evaluative process rooted around the semantic similarity of ideas. In the example condition, while LLM accuracy in predicting the true originality scores improved, the correlations of remoteness, uncommonness, and cleverness with originality also increased substantially --- to upwards of $0.99$ --- suggesting a homogenization in the LLMs evaluation of the individual facets. These findings highlight important implications for how humans and AI reason about creativity and suggest diverging preferences for what different populations prioritize when rating.

How do Humans and Language Models Reason About Creativity? A Comparative Analysis

Work on human reference processing has shown that, in sentences like “Mary asked her daughter Sally if she understood the assignment”, readers overwhelmingly interpret “she” as co-referring with “Sally”. This reflects perspective inference, or reasoning about who possesses at-issue information, and is inconsistent with a statistically-learned bias toward subject antecedent selections. The flexibility of inferencing is evident from the effect of manipulating the object character description (“Mary asked her tutor…”), where readers now prefer Mary as the antecedent. Until recently, these patterns have been largely unaccounted for by large language models (LLMs). Leveraging advancements in LLM interpretability techniques, the present study systematically examines how LLMs fare in relation to human judgments. We determine which layer activations impact these inferences and perturb them to causally link activations to model performance. Finally, we examine performance across training iterations, analyzing the point where subjecthood biases become evident and when more nuanced inferencing emerges.

The emergence of flexible perspective reasoning in large language models

Classifier choice has been widely studied, with previous research highlighting the influence of semantic features such as shape and animacy. This study, however, demonstrates that classifier choice—specifically the selection between general and specific classifiers—is also influenced by taxonomic categorization, where nouns are divided into three levels based on specificity: basic (e.g., “apple”), superordinate (e.g., “fruit”), and subordinate (e.g., “golden apple”). A picture naming task was conducted and our findings reveal a tendency for individuals to favor specific classifiers when nouns are at the basic than at the subordinate level. This challenges the prevalent assumption that general classifiers are predominantly chosen. We attribute this tendency to a cognitive economy principle and propose a novel explanation for classifier choice based on the theory of Uniform Information Density, a perspective previously rejected in previous studies. Overall, this research suggests new directions for investigating the cognitive and linguistic factors influencing classifier choice.

The Impact of Taxonomic Levels on Classifier Choice in Mandarin Chinese

Computational modelling offers a powerful tool for formalising psychological theories, making them more transparent, testable, and applicable in digital contexts. Yet, the question often remains: how should one computationally model a theory? We provide a demonstration of how formalisms taken from artificial intelligence can offer a fertile starting point. Specifically, we focus on the "need for competence", postulated as a key basic psychological need within Self-Determination Theory (SDT)—arguably the most influential framework for intrinsic motivation (IM) in psychology. Recent research has identified multiple distinct facets of competence in key SDT texts: effectance, skill use, task performance, and capacity growth. We draw on the computational IM literature in reinforcement learning to suggest that different existing formalisms may be appropriate for modelling these different facets. Using these formalisms, we reveal underlying preconditions that SDT fails to make explicit, demonstrating how computational models can improve our understanding of IM. More generally, our work can support a cycle of theory development by inspiring new computational models, which can then be tested empirically to refine the theory. Thus, we provide a foundation for advancing competence-related theory in SDT and motivational psychology more broadly.

Towards a Formal Theory of the Need for Competence via Computational Intrinsic Motivation

How we update our beliefs when encountering new evidence is the basis of evidential reasoning. Often, this will involve weighing up multiple pieces of evidence communicated to us by several sources (i.e., testimony). However, the testimonies of multiple sources are rarely truly independent; they may have used the same data or evidence, have the same training or background, or simply be repeating the same story as another. The nature of these dependencies among our evidence items is normatively impactful on the conclusions we should draw. Here we investigate whether participants are sensitive to such complex, yet impactful influences on their reasoning. We find a general preference for source diversity that heuristically gels with normative assertions. To our knowledge, it is the first paradigm that integrates shared background, shared evidence, and corroboration in the same design. We discuss challenges with developing and testing the intricacies of this.

Downloads

Next from CogSci 2025

Evidence for Distinct Factive and Non-Factive Mentalization Systems in Adults and Infants

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES