United States

CogSci 2025

Do we know enough to know what language models know?

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.

symposium

### 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.

This study investigates how adults in the third (60–79 years) and fourth age (80+ years) integrate words across sentence boundaries during reading. Participants read two-sentence passages involving direct lexical repetition or bridging inferences. Both age groups exhibited longer reading times when bridging was required, showing that inference-making is still present despite potential cognitive declines. However, while third-age adults showed immediate sensitivity to inference demands, fourth-age participants demonstrated a delayed response, suggesting compensatory strategies. These findings highlight a key role for semantic knowledge in sustaining reading comprehension in old age. Future research with more diverse samples and longitudinal methods should clarify how age-related changes interact with linguistic resources. Interventions may target processing speeds to support reading comprehension in late adulthood.

Word Integration Across Sentence Boundaries in Third and Fourth Age Adults: Evidence from Eye-Tracking during Reading

Speakers often communicate imprecisely, using expressions that are strictly speaking false yet felicitous. The degree to which imprecision is tolerated in discourse is governed by the Standard of Precision (SoP). While it is known that contextual factors can modulate the SoP, less is known about the discourse dynamics of imprecision. Previous accounts have claimed that implicit negotiations of the SoP are unidirectional: they only work upward, not downward. Here, we investigated whether parallel asymmetries arise when comprehenders adapt their SoPs in response to an interlocutor’s precision preferences. Results show bidirectional SoP adaptation effects: exposure to lower standards increased tolerance for imprecision, while exposure to higher standards reinforced stricter thresholds. These updates persisted beyond the dialogue, suggesting that exposure to (im)precise speakers modulates not only interpretations within a discourse, but also beyond the conversation. More broadly, our study provides a novel framework for studying real-time dynamic meaning negotiations during conversation.

Meaning adaptation in the discourse dynamics of imprecision

Modal cognition is foundational to human reasoning, enabling
us to construct and narrow a rich space of hypotheses about the
world. While this faculty has often been considered uniquely
human, its evolutionary roots remained uncertain. Nonhuman
primates, lacking a natural language modal lexicon, provide
a crucial test case for whether modal reasoning depends
on linguistic expression. Prior research has cast doubt on
primates’ ability to contrast mutually exclusive possibilities,
with subjects failing tasks that require reasoning about
uncertain outcomes. Across four experiments, we show
that rhesus macaques can reliably distinguish between certain
and possible rewards, exhibiting unprecedented success on
a 3-cup task. Primates’ previous failures may stem not
from an inability to reason modally but from the burden of
representing absence as a possible alternative. By alleviating
this burden, we uncover an early evolutionary footprint of
modal reasoning that extends beyond humans and reveals an
ancient, language-independent logical framework.

Nonhuman Primates Think About the Possible

Postural control strategies for upright stance adapt when balance is threatened. We investigate behavioral indicators for control strategy change in Virtual Reality (VR). Previous VR research has shown increased postural sway during virtual height exposure, but most studies focus on outdoor-like environments with extensive visual cues that may influence balance. In contrast to these outdoor studies, our indoor VR results indicate that virtual height exposure increases the mean power frequency (MPF) of sway while reducing anterior-posterior (AP) sway range. We also find an anterior shift of the Center of Pressure (CoP) when there are vertical drops both on the front and back. These findings suggest a strong context-dependence of the strategy humans employ to counteract perceived threat and heightened neuromotor control for balance stabilization.

Staring Down the Elevator Shaft: Postural Responses to Virtual Heights in an Indoor Environment

Many theories conceptualize thinking as search through a space of hypotheses. But what if your initial space is wrong? What cognitive skills support abandoning an ineffective hypothesis space and re-construing a problem with the correct hypothesis space? We examined the development of such abilities in n=172 children ages 3-7 years using Stumper riddles, which challenge respondents to explain seemingly impossible situations. 
We found evidence that children both learned the relevant hypothesis space for different riddle categories and generalized the cognitive strategy across riddle categories. Although older children showed greater overall accuracy, these effects of learning and meta-learning were found even for the youngest 3-5-year-olds. These results suggest a promising method for probing both flexible hypothesis search and meta-cognitive skills. We discuss ongoing plans to characterize individual differences as a way to uncover the underlying mechanisms of creative problem-solving.

Stumped! Learning to think outside the box in 3-7 year old children

While temporal contiguity and order effects have been shown to be highly robust in the word list memory literature, little is known about how the presence of boundaries signaling higher-order linguistic units affects temporal order memory for items within those units. Here, we present the results of two sentence memory experiments that show that linguistic boundaries block contiguity effects through a temporal context mechanism, whereby words encoded across distinct linguistic environments are also encoded as contextually dissimilar. Two consequences of this encoding mechanism are that (i) retrieval of an item results in reactivation of that item’s linguistic context alone due to co-activation of contextually similar content and (ii) significant linguistic boundaries reduce encoding interference between proximal, semantically similar items. We take these results to suggest that linguistic groupings map onto encoding contexts, which constrain the effect of item-to-item associations in sentence memory.

How linguistic boundaries form encoding contexts in memory: Evidence from temporal order effects

How are comprehenders able to extract meaning from utterances in the presence of production errors? The noisy-channel theory provides an account grounded in Bayesian inference: comprehenders may interpret utterances non-literally in favor of an alternative with higher prior probability that is close under some error model. However, we lack implemented computational models of prior expectation and error likelihood capable of predicting human processing of arbitrary utterances. Here, we model sentence processing for ``noisy" utterances as incremental and approximate probabilistic inference over intended sentences and production errors. We demonstrate that the model reproduces patterns in human behavior for anomalous sentences in three separate case studies from the noisy-channel literature. 
Our results offer a step towards an algorithmic account of inference during real-world language comprehension. Our model code, implemented in Gen, is available at https://github.com/thomashikaru/noisy_channel_model.

A Model of Approximate and Incremental Noisy-Channel Language Processing

Preceding talks in the symposium highlight an important tension between models of cognitive control vs semantics: whereas control is often thought to operate by potentiating localized and pre-specified representations or pathways, semantic representations are often thought to be highly distributed, graded, and learned. Giallanza & Cohen offer a way of thinking about how control can operate on, and may be penetrated by learned and distributed semantic structure. Frisby & Cox introduce methods for finding highly distributed structure in functional imaging data. Jackson & Lambon Ralph suggest an anatomical organization for semantic and control structures that, through maturational and learning process, promote learning of useful semantic representations. I will describe new work connecting these ideas by applying neural decoding methods to brain imaging data collected as participants make different kinds of semantic judgments about visual images of people and scenes. Behavioral results show that, across tasks, people prioritize different kinds of semantic information, most of which cannot be canonically localized to discrete brain areas (e.g. age of pictured person/object, time of day photo was taken, fine sub-categories such as type of building). Application of whole-brain image decoding then allows us to assess which neural systems show analogous changes to encoded representational structure, both within and across domains often thought to be modular and non-overlapping (faces and places). 

Downloads

Next from CogSci 2025

Word Integration Across Sentence Boundaries in Third and Fourth Age Adults: Evidence from Eye-Tracking during Reading

Similar lecture

How Can We Know When Language Models Know? On the Calibration of Language Models for Question Answering

.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

Word Integration Across Sentence Boundaries in Third and Fourth Age Adults: Evidence from Eye-Tracking during Reading

Similar lecture

How Can We Know When Language Models Know? On the Calibration of Language Models for Question Answering

Downloads