United States

Marvin Minsky, Claude Shannon, Allen Newell, and Herbert Simon were four of eleven participants at the 1956 Dartmouth Conference, where the field of Artificial Intelligence was named. Each of these pioneering researchers helped lay the foundation for future AI research. Four of their seminal ideas are: 1) Society of Mind (Minsky), 2) Information Theory (Shannon), 3) Problem-Solving Theory (Newell &amp; Simon), and 4) Bounded Rationality (Simon). Society of Mind contains a hidden blueprint for developing safe, SuperIntelligent AGI. Information Theory helps us identify the datasets that can best catalyze the growth of intelligent systems. Problem-Solving Theory explains how AI agents can communicate effectively and how we can increase the safety of AGI. Finally, Bounded Rationality illuminates the type of SuperIntelligence we might expect in the future and how humans might remain relevant when SuperIntelligent AI becomes vastly more intelligent than us.

CogSci 2025

Four Gifts from AI’s Founders

behavioral science

group behaviour

quantitative behavior

cognitive neuroscience

intelligent agents

consciousness

theory of mind

problem solving

qualitative analysis

artificial intelligence

machine learning

reasoning

logic

The AAAI 2025 presidential panel on the future of AI research aims to help all AI stakeholders navigate the recent significant transformations in AI capabilities, as well as AI research methodologies, environments, and communities. It includes 17 chapters, each covering one topic related to AI research, and sketching its history, current trends and open challenges. The study has been conducted by 25 AI researchers and supported by 15 additional contributors and 475 respondents to a community survey.

The Future of AI Research - Webinar Series

Marvin Minsky, Claude Shannon, Allen Newell, and Herbert Simon were four of eleven participants at the 1956 Dartmouth Conference, where the field of Artificial Intelligence was named. Each of these pioneering researchers helped lay the foundation for future AI research. Four of their seminal ideas are: 1) Society of Mind (Minsky), 2) Information Theory (Shannon), 3) Problem-Solving Theory (Newell & Simon), and 4) Bounded Rationality (Simon). Society of Mind contains a hidden blueprint for developing safe, SuperIntelligent AGI. Information Theory helps us identify the datasets that can best catalyze the growth of intelligent systems. Problem-Solving Theory explains how AI agents can communicate effectively and how we can increase the safety of AGI. Finally, Bounded Rationality illuminates the type of SuperIntelligence we might expect in the future and how humans might remain relevant when SuperIntelligent AI becomes vastly more intelligent than us.

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.

Each person develops a unique emotional landscape shaped by their experiences and linguistic-cultural contexts, partly personal and partly shared with others. This enables personally unique emotional experiences while maintaining shared understanding. This work aims to advance a framework for investigating what's shared and distinct across individuals, beginning with linguistic communities as an essential level of analysis, using English and Korean speakers as our case study. We examined how emotion concept representations differ between English and Korean speakers using representational similarity analysis and network analysis. English and Korean speakers’ judgments of pairwise similarity between 57 emotion concepts evidenced both substantial shared structure and language-specific patterns (Spearman's ρ = 0.72, indicating 48% unshared variance). While valence emerged as a key organizing dimension in both languages, network analyses with strength centrality showed distinct patterns for each language. First, the Korean emotion concept network demonstrated higher strength centrality across all emotion concepts than the English network, indicating higher interconnectedness between concepts. Second, high-centrality emotions were predominantly negative in both languages but formed language-specific local networks with different sets of neighboring concepts. The statistics of language usage encode a substantial part of the conceptual structure of emotion, enabling large language models to capture aspects of human emotion. Despite their advanced multilingual capabilities, GPT4-o and Claude-3.5 showed stronger alignment with English speakers' representations, regardless of prompt language. These findings demonstrate that while languages reflect common principles in emotion representation, they shape distinct patterns, with implications for cross-linguistic/cultural emotion understanding and AI system development.

Cross-Cultural Emotion Concept Representation: A Comparison of English, Korean, and Large Language Model Representations

Iconicity has recently been shown to be widespread in language and to play a particularly important role in bootstrapping new referring expressions or even getting whole new languages off the ground. The basis of this role has long been assumed to depend primarily on transparency for the receiver of the iconic signal, but might there also be producer-side advantages that play a significant role? We investigated this using an experimental referential communication game in which dyads communicated fruit and vegetables. We manipulated whether the sender could generate iconic signals and whether the receiver saw them. Results suggested that iconicity gave dyads a head-start, via stability in production, even if the receiver did not perceive the iconicity. However, this benefit declined over time, most likely due to memory constraints on the receiver.

The benefits of one-sided iconicity

Adult participants were asked to accept or reject third-party distributions that were fair or unfair in ways that created advantageous or disadvantageous inequities while their manual responses were tracked. Reach tracking measures how participants resolve conflict between alternate options and the timing of that resolution. Participants were less likely to accept disadvantageous inequities than fair distributions or advantageous inequities. Participants showed greater deviance in their curvature towards the alternative option when they rejected the distribution than when they accepted it. However, they resolved their deviance towards the alternate option faster when they rejected a distribution. This suggests a new way of considering the role of inhibitory control in economic decision making: Participants more quickly detect conflict when rejecting distributions, but the act of costing oneself resources requires more inhibitory processes related to conflict resolution.

Distinct inhibitory control systems underlie individual differences in dynamic responses to an ultimatum game: A preliminary investigation

Instruction tuning (IT) has been a fruitful technique for aligning Large Language Models with human preferences. However, the linguistic implications of IT remain unclear. In two experiments on coreference and coherence biases in the context of Implicit Causality, we investigate how IT modulates these discourse biases in relation to model size. Our results show that IT interacts with model size -- instruction-tuned models display enhanced coherence biases and more human-like coreference patterns, sometimes exceeding human performance. However, this effect appears size-dependent, suggesting that IT causes some linguistic patterns to emerge that are dormant in the respective foundation models.

Instruction tuning modulates discourse biases in language models

Infants’ attention to faces is a critical component of social development, yet little is known about how face looking unfolds during real-world interactions. Using head-mounted eye tracking and computer vision, we investigated the availability of faces in infants’ field of view and their face-looking behaviors while playing with caregivers. Faces were visible only 23% of the time, and even then, infants looked at them only 19% of the time. Time series analyses revealed that face visibility is unrelated to face looking; Infants orient their head and body to bring a face into view when they intend to look at it, rather than looking because a face is present. Face looking is related to the face’s visual properties and is only slightly influenced by parents’ face looking. These findings highlight the active nature of infant face looking and suggest that it is shaped by both visual features and infants’ own interest.

Using Head-Mounted Eye Tracking to Examine Infant Face Looking During Naturalistic Freeplay

A large body of research suggests that approach-related emotional states are lateralized to the left prefrontal cortex (Harmon-Jones et & Gable, 2018). However, because affective motivation and valence have often been entangled in experimental designs, it is unclear which construct drives this laterality. In one fMRI study designed to dissociate motivation and valence, Berkman and Lieberman (2010) found that approach motivation was more left-lateralized than avoidance motivation in the dorsolateral prefrontal cortex (DLPFC), controlling for valence.  Our study did not replicate this key finding from Berkman and Lieberman (2010).  Furthermore, whereas Berkman and Lieberman (2010) found that individuals' trait approach motivation predicted the laterality of approach-related DLPFC activity, we found that trait approach motivation predicted the laterality of positive valence, controlling for motivation.  Overall, our results do not provide any clear support for the 'textbook' model of affective motivation in the frontal lobes.

Replication of Prefrontal Asymmetry in Approach-Avoidance Motivation in fMRI

Vague temporal adverbials, such as ”recently," "just" and ”long time ago,” describe the temporal distance between a past event and the utterance time, but leave the exact duration underspecified. 
In this paper, we introduce a factorized model that captures the semantics of these adverbials as probabilistic distributions. These distributions are composed with event-specific distributions to yield a contextualized meaning for an adverbial applied to a specific event. 
We fit the model's parameters using existing data capturing judgements of native speakers regarding the applicability of these vague temporal adverbials to events that took place a given time ago. Comparing our approach to a non-factorized model based on a single Gaussian distribution for each pair of event and temporal adverbial, we find out that, while both models have similar predictive power, our model is preferable in terms of Occam’s razor, as it is simpler and has a better extendability.

A Factorized Probabilistic Model of the Semantics of Vague Temporal Adverbials Relative to Different Events

The Tritone Paradox is an auditory illusion in which a sequence of two complex tones is perceived as either ascending or descending, depending on the individual. It presents an interesting phenomenon for investigating pitch perception in contexts. however, no neurophysiological study has been conducted. This study identified event-related potential (ERP) correlates of pitch judgments under different pitch contexts. Twenty-seven participants judged whether the tritone pair was perceived as ascending or descending after listening to a sequence of ascending or descending tone pairs. Cortical auditory evoked responses to the second tone of the tritone pair were compared across contexts. In the Rise context, standard stimuli evoked larger responses at Fp1; in the Fall context, deviant stimuli elicited stronger responses across all sites. These results suggest that frontal and central brain regions are involved in processing ambiguous pitch stimuli, and that ERP responses reflect the interaction between stimulus context and perceptual.

Neural responses of Interval Judgment in the Tritone Paradox 

To learn about the world, people must gather information from external sources (e.g., books, the internet, other people). People prefer sources that have been reliable in the past—but how do we seek information when uncertain about a source’s quality? We test the hypothesis that these decisions are governed by an explore-exploit tradeoff. Prior work has researched this in reward search. Here, we adapt the tradeoff to information search. We propose that people weigh exploring an information source that is more uncertain with exploiting an information source that is more informationally promising. In one experiment (N = 171), we found evidence for both random and directed exploration in people’s choices between information sources. However, people explored uncertain information sources even without the possibility of future exploitation (differing from typical reward search). People may be broadly interested in learning about sources’ quality, and they explore to gain this information.

Learning how to learn: Evidence for an explore-exploit tradeoff in information search 

Cognitive functioning across development has predominantly been assessed through task performance. However, the role of cognitive effort infants and young children exert has been largely neglected in understanding cognitive functioning. In a large longitudinal sample (YOUth cohort, N = 2241) of infants and young children aged 5, 10 and 36 months, we extracted dynamic baseline-corrected pupil responses to measure cognitive effort during a gap-overlap eye-tracking task. Results revealed a shift from predominantly reactive effort when infants were younger to more preparatory effort in older children, especially for the more demanding condition. Moreover, preparatory effort in infancy predicted cognitive effort in childhood. These findings underscore the importance of measuring cognitive effort in addition to task performance to capture a more complete picture of early cognitive development.

Downloads

Next from CogSci 2025

Cross-Cultural Emotion Concept Representation: A Comparison of English, Korean, and Large Language Model Representations

Similar lecture

Artificial General Intelligence (AGI)

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES