United States

In daily life, humans perform tasks in teams, in which the labor division is decided by one individual in the team and not jointly by the team members (e.g., if an employer delegates a task to an employee). In the present study, we tested how the labor is divided if one team member decides the labor division in a joint multiple object tracking (MOT) task. We found that humans preferred to equally split the number of tracked targets in the joint MOT task. When comparing the data with our previous study, in which participants performed the same task with a computer partner, we found that this preference for an equal labor division is specific to interactions with a human but not with a computer partner. Moreover, participants also tended to take into account the tracking difficulty of the delegated targets more with a human compared to a computer partner.

CogSci 2025

Play fair: Humans prefer an equal division of labor in a joint multiple object tracking task

distributed cognition

group behaviour

social cognition

human-computer interaction

### 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.

poster

Parents' beliefs about children's abilities shape their parenting practices. But how accurate are parents at estimating what children are truly capable of? Here, we test the hypothesis that U.S. parents underestimate young children's abilities to complete challenging, multi-step tasks, and in turn, intervene beyond children's developmental needs (a behavior known as "overparenting"). In Studies 1A and 1B, parents (N = 130) of preschool-aged children underestimated their children's abilities, especially on practical (vs. academic) and novel (vs. familiar) multi-step tasks. In Studies 2A and 2B, we found that parents' (N = 109) underestimation has potential negative consequences: Parents who believed their child was less capable were more likely to take over tasks and provided less encouragement for independent actions. These findings suggest that parents underestimate young children's abilities, which may hinder the development of children's learning and autonomy.

Parents underestimate young children's abilities which may undermine their parenting practices

Advances in deep multi-agent reinforcement learning learning (MARL) enable sequential decision making for a range of exciting multi-agent applications.
The black-box characteristic of MARL restricts the safe and scalable application of decision models in practical deployment.
However, existing interpretability methods for deep reinforcement learning models are are not suitable for addressing challenges posed by multi-agent environments and often inadequate in generating logical sequential decisions.
We present an innovative framework called BT4MARL, which introduces the behavior tree structure to explainable MARL.
The proposed method clusters state space by aggregating temporally related states and divides agents into several groups in the new state.
Based on these clustered states and agents, we constructs behavior tree structures.
In this way, we use an exploration technique based on pairing a combined behavior tree with the target model.
We empirically show that our framework is effective in four benchmark MARL domains. Moreover, the results of a user study show that the generated explanations significantly improve performance and satisfaction.
This work represents a significant stride towards addressing the challenges of explainability and performance in MARL applications.

 A Framework for Modeling Cognitive Processes in Intelligent Agents Using Behavior Trees

With the rapid improvement of machine learning (ML) models, cognitive scientists are increasingly asking about their alignment with how humans think. Here, we ask this question for computer vision models and human sensitivity to geometric and topological (GT) concepts. Under the core knowledge account, these concepts are innate and supported by dedicated neural circuitry.
In this work, we investigate an alternative explanation, that GT concepts are learned ``for free'' through everyday interaction with the environment. We do so using computer visions models, which are trained on large image datasets. We build on prior studies to investigate the overall performance and human alignment of three classes of models -- convolutional neural networks (CNNs), transformer-based models, and vision-language models -- on an odd-one-out task testing 43 GT concepts spanning seven classes. Transformer-based models achieve the highest overall accuracy, surpassing that of young children. They also show strong alignment with children's performance, finding the same classes of concepts easy vs. difficult. By contrast, vision-language models underperform their vision-only counterparts and deviate further from human profiles, indicating that naïve multimodality might compromise abstract geometric sensitivity. These findings support the use of computer vision models to evaluate the sufficiency of the learning account for explaining human sensitivity to GT concepts, while also suggesting that integrating linguistic and visual representations might have unpredicted deleterious consequences.

Computer Vision Models Show Human-Like Sensitivity to Geometric and Topological Concepts

The rising prevalence of adolescent depression is a major public health concern. Current diagnostic methods, often burdensome and lacking objective biomarkers, hinder early detection. This paper proposes a novel multi-task framework integrating attention and resting EEG signals using Biosignal Learning and Agent Transformer (BLAT) for depression detection. EEG data is segmented, channeled, and positionally encoded, followed by feature extraction and classification via Agent Transformer. Experiments with 50 depressed adolescents and 50 controls achieved 85% accuracy using BLAT, offering a promising method for early adolescent depression screening.

Advancing Adolescent Depression Detection through Multi-Task EEG Signals and Biosignal Learning

Learning abstract concepts is a core component of human cognition, yet remains challenging for artificial intelligent. We present a computational model that investigates whether abstract categories can be acquired through shared perceptual features, using a Label-Free Concept Bottleneck Model (CBM) trained to induce basic-level concepts using shared features. Concepts are represented through intermediate concepts layer, enabling the model to form grounded representations of basic-level categories. To evaluate conceptual robustness beyond surface-level accuracy, we conduct a series of generalization and ablation experiments. These assess whether the model forms robust conceptual representations rather than merely mapping inputs to labels. Our results show that the CBM achieves high accuracy on a dataset comprising four basic-level classes and twelve subordinate Image-Net subclasses, while also yielding interpretable intermediate representations. This framework demonstrates that abstract categorization can emerge through feature based induction, and suggests a pathway for cognitive models of concept learning.

Can Abstract Categories Be Represented by Shared Features in Concept Bottleneck Models?

The “thinking for speaking” hypothesis proposes that our language can influence cognition during language produc- tion or interpretation, directing our attention to the grammat- ical and/or semantic categories readily codable in our lan- guage. Beyond the codability of grammatical and semantic categories, the role of syntactic hierarchy–a core feature of human language—has not been studied so far. This study addresses this gap by investigating the effect of learning dif- ferent complex noun phrase (NP) structures on English na- tive participants’ similarity judgments of objects which dif- fered in color or number. In Experiment 1, as the training proceeded, participants who learned to describe novel objects following an unusual NP structure that highlights the dimen- sion of number over color were more likely to judge objects matching on number; by contrast the judgments of partici- pants who learned a more typical NP structure that highlights color over number did not change significantly over time. The training-specific effect observed in Experiment 1 failed to emerge in Experiment 2 where online language involvement was reduced. These results extend the scope of “thinking for speaking”, suggesting that hierarchical structure in syntax may also influence cognition during language use. They also shed light on the potential for cognitive flexibility in representations of the NP.

Thinking through syntax: Expanding the scope of "thinking for speaking"

The human mind can segment continuous streams of activity in the world into meaningful, discrete units known as events. However, not all events are created equal. We draw a distinction between bounded events (e.g., folding a handkerchief) that have a predictable structure that develops in distinct stages (i.e., a beginning, middle, and end) and a well-defined endpoint, and unbounded events (e.g., waving a handkerchief) that lack such a well-defined structure and endpoint. We predict that event boundedness (bounded vs. unbounded) will affect attention allocation patterns over the course of the event. Here, we tested this prediction using a dwell time paradigm by measuring the time participants spent on each still frame of an activity. We found that event endpoints attracted increased attention compared to midpoints; importantly, this increase was significantly greater when people viewed bounded events, compared to unbounded events. In addition, event endpoints attracted increased attention compared to event beginnings, but this pattern also interacted with event boundedness. We conclude that abstract internal event structure (specifically, event boundedness) affects attention allocation during online event apprehension.

Event Boundedness Affects Attention Allocation during Online Event Processing

Synchronized development of possibility reasoning and theory-of-mind: Evidence from a cross-cultural study

Multi-Agent Systems (MAS) are increasingly used to simulate social interactions, but most of the frameworks miss the underlying cognitive complexity of human behavior. In this paper, we introduce Trans-ACT (Transactional Analysis Cognitive Toolkit), an approach embedding Transactional Analysis (TA) principles into MAS to generate agents with realistic psychological dynamics. Trans-ACT integrates the Parent, Adult, and Child ego states into an agent’s cognitive architecture. Each ego state retrieves context-specific memories and uses them to shape response to new situations. The final answer is chosen according to the underlying life script of the agent. Our experimental simulation, which reproduces the Stupid game scenario, demonstrates that agents grounded in cognitive and TA principles produce deeper and context-aware interactions. Looking ahead, our research opens a new way for a variety of applications, including conflict resolution, educational support, and advanced social psychology studies.

Games Agents Play: Towards Transactional Analysis in LLM-based Multi-Agent Systems

Many causal relations can be represented by continuous functions that map inputs to outputs. Can young children learn continuous causal functions and generalize them from observed data to new scenarios? We found that 4- and 5-year-olds can represent continuous functions with different abstract forms. After observing a few input-output pairs, children can accurately infer positive linear and step functions by predicting the outputs of novel input values. They also have emerging knowledge of negative linear and triangular functions. While children do not yet make consistently accurate predictions for these functions, they can distinguish these functions from the positive linear function and show inferential patterns that are consistent with the respective functions. Like adults and older children, preschoolers show an inductive bias towards positive linear functions. Their understanding of negative linear functions--which strongly requires inhibiting this inductive bias--improves with age.

Downloads

Next from CogSci 2025

Parents underestimate young children's abilities which may undermine their parenting practices

Similar lecture

The Role of Feedback in Cognitive Offloading during Human-Computer Collaboration

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES