United States

Recent advances in generative language models, such as ChatGPT have demonstrated an uncanny ability to produce texts that appear to be comparable to those produced by humans. Nevertheless, machine generated texts differ from those produced by humans in important aspects, such as routinely including references to nonexistent sources. In this paper, we use both psycholinguistic measurements and participant responses to compare texts generated by machine with equivalent texts generated by humans. Our analysis demonstrates some of the ways in which machine-generated texts differ from human-generated ones in both style (e.g., increased use of positive connectives) and content (e.g., increased confidence). We also note multiple ways in which texts generated by these models are similar to those generated by humans (e.g., their use of emotion words). We believe this research provides insights that can be useful to understanding how language is generated by both humans and machines.

CogSci 2025

Man or Machine: Evaluations of Human and Machine-Generated Movie Reviews

language understanding

psychology

artificial intelligence

discourse

natural language processing

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.

Animated films can provide a context for caregivers and children to discuss death, potentially furthering children’s understanding of this concept (Bridgewater et al., 2021). However, white caregivers in the United States tend to shield their children from this topic (Rosengren et al., 2014). The purpose of the study was to delve into the dynamics of child-caregiver discussions about death. We recruited 29 children (ages 4-6) and their caregivers and observed their discussion about a death depicted in an animated film. We found that most families (92.6%) discussed death, and often mentioned affective topics (e.g., sadness; 81.5%), but some mentioned biological (33.3%) or spiritual topics (11.1%). Children's age or caregivers' reports of shielding were not linked to the content or frequency of these discussions. This study highlights how media can serve as a context for the development of spiritual and biological concepts.

Once Upon a Goodbye: Exploring How Animated Films Spark Child-Caregiver Conversations About Death

We feel uncanniness for human-looking robots (uncanny valley). Regarding the mechanism, we hypothesized that the visual system automatically tries but misses to categorize a registered object, and therefore negatively evaluates it. However, it is unclear whether the automatic categorization induced the negative evaluation since all previous studies directed participants to evaluate the object in terms of a previously categorized dimension. The present study examined whether the uncanny valley is automatically triggered by the categorization failure in a task-irrelevant dimension. Participants categorized a morphed figure, the shape of which has been known to trigger the uncanny valley, in terms of color, and evaluated the likeability of it. The uncanny valley occurred based on the task-irrelevant shape dimension of the objects, even though the preceding color categorization was successful. These findings suggest that the visual system evaluates the likeability of the registered objects in response to the automatic categorization and its failure.

The uncanny valley phenomenon triggered by a task-irrelevant dimension of objects

Conversations are intricately structured forms of social interaction in which talkers move through interconnected topics with nested levels of semantic specificity. What principles govern how conversational partners jointly navigate an expansive topic space? To characterize these dynamics, we introduce a new dataset of annotated topic shifts from N=1,505 annotators on 200 distinct video call conversations between strangers (Reece et al., 2023). Conversational dyads made stochastic but systematic transitions between topics, and within individual topics, we find that dyads begin concentrated in semantic space before dispersing to more idiosyncratic regions as topics progress. The same dispersion pattern also holds over entire conversations, providing quantitative evidence for nested levels of increasing specificity over conversations. Overall, our findings suggest that strangers get to know one another through systematic exploration of topic space, revealing hierarchical structure in idle talk.

Dynamics of topic exploration in conversation

How well students learn depends on their ability to sustain attention. However, it is currently unclear how to measure sustained attention in the classroom and relate those underlying attentional dynamics to academic engagement and performance. Here we leverage a suite of sustained attention instruments to explore how individual differences in sustained attention account for differences in learning outcomes in a university STEM course (N=248). We found that a student's ability to sustain attention predicted their subsequent academic achievement in the course. Sustaining attention was also associated with STEM-related stress, anxiety, and students' confidence in their ability to learn the course material. We are additionally exploring interaction logs from the digital textbook students used to investigate the mechanisms linking sustained attention to subsequent achievement. Together, these findings highlight the promise of studying attention and learning across timescales to advance mechanistic understanding of human cognition in real-world environments.

Measuring sustained attention across timescales to predict learning in real-world environments

Lewis signaling game (LSG) and similar coordination 
games have been used to model the emergence and
evolution of language. However both Nash
equilibria and learning or evolutionary
dynamics often result in suboptimal signaling systems.  
We present a sequential reinforcement  
learning (SRL) model based on a novel sequential binary decision
process. SRL has low cognitive demands and parameter
count and exhibits lateral inhibition without additional
assumptions. 
We prove all scenarios converge to an optimal signaling
system in all N state, N signal LSGs with arbitrary
state probabilities and further
explore its properties with numerical simulations. 
Next, we model a signaling game with 
agents who both speak and hear while using 
one state of learning (instead of two,  as is common).  
Agents have a probability
distribution for meanings in a given context.
Speaking agents use the distribution to 
choose a meaning and use SRL model to 
choose a signal. 
Hearing agents use Bayes to combine their state of learning
with their  meaning distribution to guess a meaning.
An agent's state of learning is reinforced
from habit of speaking and guessing a meaning.
Numerical simulations indicate both agents
converge to the same optimal system
without external reinforcement
as happens in language acquisition.

Coordination Games with Sequential Stochastic Learning and Language Emergence

This study investigates Hubei dialect speakers’ ability to distinguish and identify Mandarin and English words with [l]-[n] onsets through AX discrimination and identification tasks, along with their production in Hubei dialect, Mandarin, and English. Results showed lower accuracy in discriminating [l] and [n] in English and Mandarin, especially with stimuli containing a high vowel, nasal coda, or diphthong beginning with a high or front vowel. Participants with more dialect use and later Mandarin acquisition were less sensitive to the contrast. Directionality of merger varied across participants. No significant effects of linguistic or social factors were found in identification, except a positive effect of initial high vowels in Mandarin diphthongs. Acoustic measures showed no significant differences, suggesting an [l]-[n] merger. These findings provide insights into non-native speech perception and production and suggest implications for pronunciation training.

An Investigation of [l]-[n] Merger among Hubei Dialect Speakers: Cross-Linguistic Evidence from Perception and Production

This study examined the motivational effects of incorporating instructor images into educational materials in the early reading stages. Participants viewed a page of disaster prevention materials for two seconds and rated it on a five-point scale for motivation to read ("Did the page motivate you to read?") and understandability ("Did the page look easy to understand?"). The materials were three types: those without any instructor photographs or illustrations, those with real instructor photographs, and those with instructor illustrations generated from the photographs. Results showed that mean scores were highest for the condition with real photographs, followed by the condition with illustrations, and lowest for the condition without images. These findings suggest that instructor images enhance motivation to read, with real photographs having a stronger effect.

Motivational Effects of Instructor Images in Educational Materials on Early Reading Stages

Research has begun exploring the performance of large vision language models (LVLMs) in recognizing illusions. However, studies often have not distinguished actual and apparent features, leading to ambiguous assessments of machine cognition.  
We introduce a visual question answering (VQA) dataset, categorized into genuine and fake illusions. Genuine illusions present discrepancies between actual and apparent features, whereas fake illusions have the same actual and apparent features even though they look illusory. We evaluate the performance of LVLMs for genuine and fake illusion VQA tasks and investigate whether the models discern actual and apparent features. Our findings indicate that although LVLMs may appear to recognize illusions by correctly answering questions about both feature types, they predict the same answers for both Genuine Illusion and Fake Illusion VQA questions. This suggests that their responses might be based on prior knowledge of illusions rather than genuine visual understanding.

Do Large Vision-Language Models Distinguish between the Actual and Apparent Features of Illusions?

We explore how time pressure affects accuracy at various stages of learning in a complex, dynamic task: the game of Tetris. We emulate human decision-making processes under time pressure in several reinforcement learning models by training them under time pressures present on humans. Subsequently, we compare the performance and the behavior of human players against the ones demonstrated by AI players of equivalent skill. At the surface level, the AI models are able to achieve human-like performance levels at different stages of expertise. However, when probed at lower levels, we find that their behavior and strategies are considerably different from the ones employed by human experts. Examining why and how the models differ from humans highlights the promise of using AI models to study the nuances of human decision-making in dynamic tasks, along with the need to explain both human and AI performance at multiple performance levels for accurate understanding.

The Need for Speed? Exploring the Contribution of Motor Speed to Expertise in a Complex, Dynamic Task

Humans who use a left-to-right writing system often associate smaller numbers with the left side of space and larger numbers with the right. Whether this left-to-right number-space mapping is innate or culturally learned is unclear. Here, we test whether monkeys who lack human cultural practices show a left-to-right number-space mapping. Previous work in monkeys has found mixed evidence on whether monkeys show a left-to-right bias in their number-space mappings. Replicating the methods of Drucker and Brannon (2014), monkeys were trained to touch the fourth circle from the bottom in a vertical array of five circles. Then, they were tested with a horizontal array of five circles. Overall, monkeys showed no preference for the fourth circle from the left compared to the fourth from the right. This suggests monkeys may not have a directionality bias for number-space mappings. Therefore, the left-to-right bias in humans may be due to specific cultural practices.

Downloads

Next from CogSci 2025

Once Upon a Goodbye: Exploring How Animated Films Spark Child-Caregiver Conversations About Death

.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

Once Upon a Goodbye: Exploring How Animated Films Spark Child-Caregiver Conversations About Death

Downloads