United States

Facial expressions are essential to human interaction, providing emotional cues that influence how we perceive and respond to others. Advances in artificial intelligence (AI) now enable real-time manipulation of these expressions, raising ethical concerns about AI-mediated communication (AI-MC). This project investigates the effects of AI-based facial expression alterations on prosocial behavior and empathy using a choice-exposure counterfactual paradigm. It addresses two key questions: (1) Do facial AI filters influence empathy and prosocial behavior? (2) Do individuals strategically use AI filters to justify self-serving actions? Participants chose between authentic and AI-filtered videos before making donation decisions after exposure. In an initial study, neutralizing AI filters showed no significant effects on empathy or donations. Building on these findings, we will explore stronger filters and more emotionally expressive portrayals to better understand their impact. This research aims to advance knowledge of the psychological and ethical implications of AI-MC.

CogSci 2025

Filtering out Empathy: The Impact of Other-Alteration through AI-Filters on Prosocial Behavior

empathy

human-computer interaction

psychology

artificial intelligence

emotion

EMNLP 2021 is planned to be a hybrid event in Punta Cana, Dominican Republic, with both on-site and fully virtual participation possible. The experience for on-site participants would closely approximate a normal pre-COVID *ACL conference, with 5-6 thematically organized parallel sessions and live Q/A and interactive discussion immediately after the talks. Presentations by virtual participants will be equitably interleaved with those of on-site participants, projected on the auditorium screens as if on-site, and also followed immediately by live Q/A and interactive discussion at a time during reasonable waking hours for the virtual presenter. For all participants, on-site and virtual, who are unable to attend a session due to either time-zone issues or because they are participating in another session live, talk recordings and slides will be available online at a minimum after the live presentation (and in many cases before as well), and questions may be submitted in advance on session-specific discussion boards and answered live in session with the usual visual aids if desired.

<iframe style="width:700px;height:400px" src="https://online.fliphtml5.com/ebtyf/ceby/"  seamless="seamless" scrolling="no" frameborder="0" allowtransparency="true" allowfullscreen="true" ></iframe>

Please Note: The EMNLP registration system is not currently connected to the underline site as we are still in the process of building out EMNLP 2021. You will receive access instructions from underline the week of November 1st. 

Access is given only to EMNLP upon registration, if you have not registered please do so [here](https://2021.emnlp.org/registration).

Registered attendees will receive access the week of November 1st.

EMNLP 2021

EMNLP 2021 is planned to be a hybrid event in Punta Cana, Dominican Republic, with both on-site and fully virtual participation possible.

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.

Recent research has compared representation models of word meaning (Brown et al., 2023, Cognitive Science 47:e13291), however, less research has compared representation models of words’ perceptual features. Thus, we compared vector-space word representation models that can be used to quantify words’ phonological similarity. The main structure of the model was adapted from Cox et al.’s orthographic representation model (Behavior Research Methods 43:602-15, 2011). Variations of the model included phonetic mapping scheme, encoding scheme, the inclusion of lexical stress, and the combination of orthographic and phonological representations. We tested the model variants against human-rated phonological similarity and both phonological and orthographic Damerau-Levenshtein distance. Open n-gram encoding (1 ≤ n ≤ 2) performed better overall than terminal relative encoding across all phonological similarity metrics.  Concatenated orthographic-phonological vectors improved the prediction of human ratings with terminal-relative encoding only. Using more fine-grained phonetic mapping or including lexical stress had minimal effects.

Exploring Vector Representations for Phonological Similarity

Interactions between agents are supported through a continuous process of detecting and responding to behaviors that are contingent upon the other agent's behavior. Here, we explore the temporal dependence of these mechanisms, focusing on the role of timescale compatibility in inter-agent interactions. Using continuous-time recurrent neural networks (CTRNNs) to control embodied agents in a minimal social interaction task, we demonstrate that effective interactions require agents to operate on compatible timescales. Our results indicate that time scale mismatches disrupt agents’ ability to distinguish other agents from non-social entities, revealing a timescale threshold beyond which agents begin misclassifying slower agents as static objects and faster agents as non-social animate objects.

The Effect of Timescale Dependence on Dyadic Interactions

Visuo-spatial abilities contribute to successful athletic performance (Millard et al., 2021), and the results of visio-spatial tests have a predictive and diagnostic value (Moreau et al., 2012). In our study, we test participants from grassroots sports (n=186) and adapted sports (n=30) with tests covering allocentric and egocentric cognition (Mental rotation (Shepard & Metzler, 1988), Perspective taking (Kozhevnikov & Hegarty, 2001), Santa Barbara Sense of Direction Scale (SBSOD) (Hegarty et al., 2002)) using a web application. We compare differences in spatial skills in both groups depending on demography (age, gender) and sports training characteristics (type of sport, training regularity, experience). Preliminary results suggest that (1) mental rotation performance is similar across groups; (2) participants from the grassroots sports perform better in the perspective-taking test; (3) SBSOD measurements are slightly better in the adapted sports.

Spatial skills in grassroots athletes and athletes with functional impairments: A screening study in Latvia

This study examines how instructional framing influences probability distortions in decision-making scenarios. With 136 participants, we explored four instruction conditions: direct calculation guidance, estimative averaging, and narrative framing from first- and third-person perspectives. Our findings indicate that skewness, rather than variance, significantly impacts estimation errors in expected value tasks. The estimative instruction condition notably reduced probability neglect, whereas direct calculation instructions unexpectedly introduced bias. Both narrative conditions amplified probability neglect, with no significant difference between perspectives. These outcomes challenge traditional assumptions in decision-making models, emphasizing the central role of skewness and the substantial effect of instructional framing on probability distortions. The results suggest that employing estimative instructions could effectively minimize biases in contexts where accurate evaluation of expected value is crucial. This research underscores the importance of instructional design in decision-making tasks and provides insights into minimizing probability neglect.

What Do People Expect from Expected Value?

Prosody affects how people produce and understand language, yet studies of how it does so have been hindered by the lack of efficient tools for analyzing prosodic stress. We fine-tune OpenAI Whisper large-v2, a state-of-the-art speech recognition model, to recognize phrasal, lexical, and contrastive stress using a small, carefully annotated dataset. Our results show that Whisper can learn distinct, gender-specific stress patterns to achieve near-human and super-human accuracy in stress classification and transfer its learning from one type of stress to another, surpassing traditional machine learning models. Furthermore, we explore how acoustic context influences its performance and propose a novel black-box evaluation method for characterizing the decision boundaries used by Whisper for prosodic stress interpretation. These findings open new avenues for large-scale, automated prosody research with implications for linguistic theory and speech processing.

Prosody in the Age of AI: Insights from Large Speech Models

Research has shown that infants prefer prosocial characters over antisocial ones, suggesting that sociomoral evaluation is early-emerging. However, some have argued that infants’ preferential responses stem from low-level perceptual processes rather than true social understanding. Using electroencephalography (EEG), past work has suggested that motivational and social, but not attentional, processes are implicated in infants’ responses to prosocial versus antisocial acts and individuals, however, the majority of past work utilized a single type of prosocial/antisocial interactions: helping a character to climb a hill. To test the generalizability of past neural findings from the hill paradigm, here we examined infants' responses in a distinct helping/hindering scenario in which a character tries but fails to open a box and is alternatively helped or hindered. Largely replicating past work, infants showed greater activity in social (indexed by the P400) but not attentional (indexed by the Nc) ERP components when seeing hinderers versus helpers, consistent with claims that infants’ responses to prosocial and antisocial agents are social. No evidence of differential approach/avoidance motivation during prosocial/antisocial events was found. These findings support the role of social processes in infants’ sociomoral evaluations.

Examining the Robustness of Neural Correlates of Infants’ Sociomoral Evaluations

Theoretical understanding of neurodevelopmental conditions (NCs) has shifted from a categorical approach to a dimensional one, characterized by an acceptance of comorbidity and heterogeneity. Previous computational modelling of NCs has tended only to accommodate categorical views. The current work presents a mechanistic simulation framework that fits with the dimensional view, using artificial neural networks to model populations of learners, with underlying causes of variation in developmental outcomes viewed as continuous, polygenic, and in part environmental. We show how the dimensional and categorical approaches can be linked using latent profile analysis and outlier methods, recovering profiles and specific deficits from dimensional variation. We show how altering the distribution of hyper-parameters shifts the population composition of developmental profiles and frequencies of deficit patterns, and we test their robustness to stochastic factors.

Categories from dimensions: Population-level computational modelling of neurodevelopmental conditions 

Dynamic Causal Modeling is a widely-used method for examining brain connectivity. Most commonly, it is applied to brain regions showing strong responses to experimental tasks, comparing different network configurations based on the temporal dynamics of the neural signals. It can further be applied to models employing a theory-driven selection of brain regions, showing a weaker experimental effect. However, it is unclear if these effects provide sufficient temporal information for Dynamic Causal Modeling to reliably identify the best-fitting model. This study investigated the regional predictive fit in a theory-driven model which has been found to consistently outperform alternatives using Dynamic Causal Modeling. Results revealed issues with the fit of some regions and subjects, raising concerns regarding the reliability of model comparisons using Dynamic Causal Modeling with regions selected based on theory instead of a strong experimental effect.

Less Than the Sum of its Parts: Complex Models of Cognition Struggle to Capture Regional Activity within Otherwise Well-Fitting Model Structures

Humans often interpret pointing as referring to an object, however, it can also indicate a direction or relevant spatial location. We investigated which one of these interpretations can explain 14-month-olds responses in a two-alternative choice task. We conducted three experiments, in which an experimenter pointed at one of the two lateral objects, swapped their positions in full view of the infant, and then allowed the infant to choose. Pointing was either produced in an Ostensive Addressing (Experiment 1), Nonostensive Addressing (Experiment 2), or Ostensive Labelling context (Experiment 3). In the Ostensive Addressing and Ostensive Labelling experiments infants chose the non-indicated object in the indicated direction significantly more often than predicted by chance. In contrast, in the Nonostenive Addressing experiment, infants’ performance was on chance. These findings suggest that infants follow the direction of pointing rather than interpreting it as indicating a specific object in a communicative context.

What or where? Infants Interpret Pointing as Referring to a Location Rather Than to a Specific Object

revious studies investigated infants’ ability to recognize turn-taking exchanges of signals that can serve communicative information transfer and draw pragmatic inferences from them. Here we investigate 13-month-olds’ expectations about the distal effects of communicative versus non-communicative actions and explore their understanding of the epistemic causal mechanisms through which communicative signals modify their addressees’ consequent intentional actions. In four looking time experiments (Ntotal = 80), we found that infants understand that communicative signals cannot bring about non-intentional state changes in other entities and expect their distal effects to be limited to inducing intentional behavioral reactions in recipient agents. These results indicate that human infants possess cognitive mechanisms to understand the unique causal affordances of ostensive communicative actions. Coupled with their evolved pragmatic inferential capacities and communicative mindreading skills, these abilities form a specialized cognitive system for interpreting ostensive communicative information exchange between communicating social partners.

Downloads

Next from CogSci 2025

Exploring Vector Representations for Phonological Similarity

Similar lecture

Distilling Knowledge for Empathy Detection