Canada

One of the key features of natural languages is that they exhibit long-distance filler-gap dependencies (FGDs): In the sentence `What do you think the pilot sent __?&#39; the wh-filler what is interpreted as the object of the verb sent across multiple words. The ability to establish FGDs is thought to require hierarchical syntactic structure. However, recent research suggests that recurrent neural networks (RNNs) without specific hierarchical bias can learn complex generalizations about wh-questions in English from raw text data (Wilcox et al. 2018; 2019). Across two experiments, we probe the generality of this result by testing whether a long short-term memory (LSTM) RNN model can learn basic generalizations about FGDs in Norwegian. Testing Norwegian allows us to assess whether previous results were due to distributional statistics of the English input or whether models can extract similar generalizations in languages with different syntactic distributions. We also test the model&#39;s performance on two different FGDs: wh-questions and relative clauses, allowing us to determine if the model learns abstract generalizations about FGDs that extend beyond a single construction type. Results from Experiment 1 suggest that the model expects fillers to be paired with gaps and that this expectation generalizes across different syntactic positions. Results from Experiment 2 suggest that the model&#39;s expectations are largely unaffected by the increased linear distance between the filler and the gap. Our findings provide support for the conclusion that LSTM RNN&#39;s ability to learn basic generalizations about FGDs is robust across dependency type and language.

CogSci 2022

LSTMs Can Learn Basic Wh- and Relative Clause Dependencies in Norwegian

## Welcome to EMNLP 2024! 
We are excited to welcome you to one of the most prominent conferences in the field of Natural Language Processing. This year, EMNLP 2024 is being held in a hybrid format,
offering both virtual and in-person participation in beautiful Miami. Due to a record-breaking number of submissions, we've expanded the total number of accepted papers to accommodate more cutting-edge research from around the globe.
### [Conference Handbook](https://drive.google.com/file/d/1WPROgxjLAC96AJL7Ugy0tEnYm7dkrbHt/view?usp=sharing)

You are required to register for this event. **Please register [here](https://2024.emnlp.org/registration/).** The EMNLP 2024 event page on Underline will be open to public one week prior to the event.

Please register!

EMNLP 2024

EMNLP 2024 will take place in Miami, Florida from Nov 12th to Nov 16th, 2024, at the Hyatt Regency Miami Hote and on Underline for remote participants.

One of the key features of natural languages is that they exhibit long-distance filler-gap dependencies (FGDs): In the sentence `What do you think the pilot sent __?' the wh-filler what is interpreted as the object of the verb sent across multiple words. The ability to establish FGDs is thought to require hierarchical syntactic structure. However, recent research suggests that recurrent neural networks (RNNs) without specific hierarchical bias can learn complex generalizations about wh-questions in English from raw text data (Wilcox et al. 2018; 2019). Across two experiments, we probe the generality of this result by testing whether a long short-term memory (LSTM) RNN model can learn basic generalizations about FGDs in Norwegian. Testing Norwegian allows us to assess whether previous results were due to distributional statistics of the English input or whether models can extract similar generalizations in languages with different syntactic distributions. We also test the model's performance on two different FGDs: wh-questions and relative clauses, allowing us to determine if the model learns abstract generalizations about FGDs that extend beyond a single construction type. Results from Experiment 1 suggest that the model expects fillers to be paired with gaps and that this expectation generalizes across different syntactic positions. Results from Experiment 2 suggest that the model's expectations are largely unaffected by the increased linear distance between the filler and the gap. Our findings provide support for the conclusion that LSTM RNN's ability to learn basic generalizations about FGDs is robust across dependency type and language.

poster

Join us for CogSci2022! Our 44th Annual Conference will be centered around the theme of diversity, with an accompanying range of keynotes and invited speakers who themselves highlight the diversity of our field. Our three symposia are Cognition across Diverse Populations, Cognition within Diverse Societies, and Global Challenges for Cognitive Science. The conference is being offered in a fully hybrid environment for onsite and online attendees.
 
 [**The Program at a Glance can be found here**](https://cognitivesciencesociety.org/wp-content/uploads/2022/07/Cogsci-2022-PAAG-Detailed-July-12final-1.pdf)
 [**The Detailed Program can be found here**](https://cognitivesciencesociety.org/wp-content/uploads/2022/08/DeArmond-CogSci2022-program-v113-final.pdf)

This conference requires a registration to attend. Please use the link below to secure your ticket. The website will be available to all registered attendees on July 25th. 

[Register to Attend](https://cognitivesciencesociety.org/registration/)

Registration Is Required

The 44th Annual Meeting of the Cognitive Science Society highlighted research on the theme Cognitive Diversity in addition to the full breadth and diversity of research topics offered by the society’s membership. 





We study pension planning and financial wealth of natives and immigrants (N=1177) in the Netherlands, in relation to their temporal values (past/future-focused), financial knowledge, IQ, and other individual characteristics. We find that, compared to natives, immigrants are less financially literate and rely more on the government for their retirement income, but are more future-focused and think more about their retirement. Second, controlling for financial knowledge, IQ, saving intention, self-control and demographic factors, temporal values help to predict many aspects of pension planning: how much people think about retirement, their desired retirement age, whether they develop a plan to save for retirement, perceived saving adequacy, and home ownership. Furthermore, temporal values predict savings, risky assets and financial wealth in 2016 and 2020, even after controlling for the financial situation in 2016. Our results have strong implications for policies related to pension communication and contribute to the theory on relationships between economic decisions, time and cognition.

Beyond financial knowledge and IQ: The effect of temporal values on pension planning and financial wealth of natives and immigrants in the Netherlands

The representation of time depends heavily on spatial skills.
Saj et al. (2014) demonstrated that left-hemispatial neglect
patients, who lost the ability to detect objects in their left visual
field, have a selective deficit in remembering items
corresponding to the past, i.e., the left side of their mental
timeline. The current study used the same memory task but
tested neurotypical individuals (N = 76) to examine whether
individual differences in spatial ability as well as learning order
(chronological vs. random) predict how well participants
remember items and associations between the item and time
(past or future). Our results indicate that higher spatial ability
and chronological learning both lead to better memory. This
study is among the first to demonstrate how individual
differences may impact time representation and memory that
relies on a mental timeline.

Locating past and future: The Influence of Spatial Ability on Time Representation

We study the effect of people’s temporal values (habits of attending to past or future events) on their health, labour market performance and happiness. Participants’(N=1177) data were initially collected in 2016 and followed in 2020-2021. We find that habitually more attending to the future negatively correlates to diseases (heart attack; high cholesterol; diabetes; high-blood pressure), but positively associates with health-related behaviour (eating vegetables and fruit; less smoking), health status (e.g., healthy weight; long life expectancy), income, hourly wages, financial satisfaction and happiness. Furthermore, such temporal values predict participants’ future situation of these aspects in 2020-2021, even after controlling for the 2016 baseline situation, IQ, self-control, patience, risk aversion and demographic information. We propose a temporal values and well-being hypothesis, suggesting that individuals’ temporal values can predict their concurrent and longitudinal all-around well-being. Our findings have strong implications for theories of time perception, and for a better understanding of factors that influence people’s health, income, and happiness.

Time to get attention: The effect of temporal values on health, income and happiness

Children have intuitive theories of several conceptual domains, but it is unknown if adults’ common sense beliefs about time reflect an intuitive theory. Here, in an online survey, 165 3- to 6-year-old children judged whether 13 time-related phenomena (e.g., the future, going back in time) were real or not real and provided confidence ratings for their judgments. Beforehand, parents provided their own responses to the same items and predicted their children’s responses. As early as age 3, children’s responses to most items resembled those of adults. Children’s responses to past-related items (e.g. changing the past) were more similar to adults’ than were their responses to future-related items (e.g., changing the future). Parents predicted their children’s responses with high accuracy. These results suggest that many, but not all, adult beliefs about time emerge early in development, and may be part of an intuitive theory.

Do humans have intuitive theories of time?

Temporal perspectives have been studied as a part of spatial thinking of time. They allow us to place ourselves and temporal events on a timeline, making it easier to spatialize time. This study investigates how we adopt temporal perspectives in temporal gestures. We asked participants to retell temporal scenarios written in the Moving-Ego (ME), Moving-Time (MT), and Time-Reference-Point (Time-RP) perspectives in spontaneous and elicited gesture conditions. Participants adopted temporal perspectives similarly regardless of the gesture condition, with few differences. Our results showed that participants’ temporal gestures resonated better with the Ego-Reference-Point versus Time-Reference-Point distinction. Participants produced more ME and Time-RP gestures for the corresponding scenarios and speech, however the MT perspective was not adopted more than the others in any condition. Our findings show that we incorporate temporal perspectives into our temporal gestures to a considerable extent, however, the classical ME and MT classification may not hold for temporal gestures.

Temporal Gestures in Different Temporal Perspectives

Cumulative culture is believed to be a uniquely human form of social learning, and is therefore believed to be key in understanding how humans evolved such complex social organization and technologies. It is reasonable to believe that our capacity for cumulative culture is the result of a series of interconnected, complex evolutionary processes, but humans abilities for teaching, imitation and communication are believed to play a central role. We plan on combining brain imaging techniques and a transmissions chain design with transmission of evolutionary relevant tasks (knot tying and symbol-production) in order to look at the mechanisms of cumulative culture in a new way. By looking at the brain activation involved in acquiring and transferring these skills we aim to offer new insight on the cognitive and behavioral demands of these technologies and their effect on cultural evolution.

The cognitive neuroscience of social learning in human evolution

When researching social insects, interpretations drawing parallels between behaviours and social systems of humans and animals are particularly prevalent. While these comparisons have value for illustrative purposes, failing to recognize them as metaphors can result in erroneous interpretations of data. Previous studies have found this anthropomorphic bias to be connected to extraversion and sociability. 
Autistic individuals are overrepresented in the life sciences, may be less prone to cognitive bias, yet still face significant stigma in the workplace. In this interdisciplinary study, we therefore surveyed an international sample of social insect scientists to explore the impact of autistic traits on their research.
We herein present the first results of this study, and discuss them from the standpoints of cognitive science, philosophy of psychiatry, and entomology.
We thereby hope to elucidate the role of neurodiversity and atypical modes of perception in science, and contribute to creating a more inclusive academic landscape.

Neurodiversity and anthropomorphism in social insect research

This study explored how sociocultural pressures and internalized beauty ideals play a role in how women and men perceive the attractiveness of different body types of the same and opposite gender. Results showed that when judging the attractiveness of bodies of the same gender, internalized beauty ideals have different effects on women and men. Women’s judgments of the attractiveness of female bodies are predicted by the pressure exerted by a thin beauty ideal, while men’s judgments of the attractiveness of male bodies are predicted instead by a muscular beauty ideal. Attractiveness judgments for bodies of the opposite gender are influenced by the pressure to be thin and the perceived influence of significant others. Sociocultural pressures also have a stronger effect on women than men. These findings offer an initial window into the distinct factors that shape body image construction for the digital generation of women and men.

Internalized Beauty Ideals and Sociocultural Pressures Shape How Young Women and Men Perceive Body Attractiveness

How do individuals select which of multiple sources of information to attend to, and which events and entities in their environment to solicit more information about? This study
aims at understanding whether adults actively solicit information that they are missing to fill gaps in their knowledge of recently learned novel word-object associations. In other
words, we ask whether adults actively solicit the labels of objects they are not confident about. Furthermore, given the role of confidence on the influence of feedback on word learning, we ask whether the beneficial effects of feedback on errors vary as a result of the confidence learners have in their knowledge of newly learned novel word-object associations. We will also compare the findings of this study to the results of a study with the same design that was conducted with preschoolers.

Active learning, feedback and hypercorrection effect in word learning

According to Rozenblit & Keil (2002), attempting a mechanistic explanation of a concrete phenomenon results in reduced self-reported knowledge, suggesting that the latter is normally overestimated (Illusion of Explanatory Depth or IoED). Fernbach and Sloman (2013) demonstrated the effect in sociopolitical issues, finding that causal explanation results in short-term moderation of belief extremity. While the effect of explanation on knowledge self-estimation has been robustly replicated, debates about the impact on extremity continue. Here, we replicate both effects in an online sample of Iranians (N=100 and p<.05). Given IoED’s reliance on mechanistic framing, we further hypothesized that the more concrete construal afforded national hot-button issues would produce a more robust effect compared to international ones. Our experiment indeed found stronger IoED effects on both knowledge estimation (p<.005) and belief extremity (p<.003) across national issues. We discuss how this moderation intervention can be helpful in extremism context.

Premium content

Downloads

Next from CogSci 2022

Beyond financial knowledge and IQ: The effect of temporal values on pension planning and financial wealth of natives and immigrants in the Netherlands

Similar lecture

Generalizations across filler-gap dependencies in neural language models