Canada

The following studies investigated the perceptual processes that are the basis of the face inversion effect (FIE). We evaluated the effects of disrupting holistic information conveyed by the face contour/outline. In Experiment 1 (n=144) we blurred the contour of the faces and using an old/new recognition task we found that a robust inversion effect similar to that for normal faces remains for these new no-contour faces. However, a significant reduction in overall performance was found for no-contour vs normal faces. In Experiment 2 (n=74) instead of blurring we inserted a novel face contour to replace the normal one and found the same pattern of results as in Experiment 1. Our results suggest that the holistic information provided by the face contour does not on its own influence the FIE, however it plays a role in face recognition more generally.

CogSci 2022

Manipulating the face contour affects face recognition performance leaving the Face Inversion Effect unaltered

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. 





Quantifying the meaning of a word is a complex challenge. Humans can encode semantic information along a large and diverse range of semantic dimensions for any given word. Whilst a number of studies have applied a range of techniques to quantify word meaning along specific dimensions, little work has focussed on the socio-semantic dimensions of meaning. Here, we present data that quantifies the socio-semantic representations of 2,700 Czech words along the dimensions of gender, location, political, valence and age. We also demonstrate the utility of the data set by calculating an estimate of socio-semantic similarity between all words, which can be used to identify words that are either proximally close or distant in socio-semantic space.

Quantifying the Socio-semantic Representations of Words

In virtually every activity we engage in — from analyzing economic trends, to predicting which of two football teams is more likely to win a game — our minds are tasked with separating signal from noise. Such computations benefit from the fact that our minds are highly attuned to the statistical structure of the world. But how quickly do we detect statistical structure — and to what extent is our sensitivity to structure rooted in perceptual processes? To address this, we asked observers to judge whether briefly presented visual stimuli were generated randomly or non-randomly. In as little as a tenth of a second, people exhibited the same stable biases of statistical perception that they exhibit in classic cognitive tasks (i.e., without time constraints). These results suggest that certain biases of subjective probability may arise not from how we think about randomness, but from how we perceive statistical information in the first place.

The speed of statistical perception

English is the dominant language in computer science. In addition to English-based academic papers, English is frequently the only language provided in introduction sections and manuals of command and software libraries, which are essential aspects of computer programming. Hence, English-as-a-second-language (ESL) learners may have difficulty studying computer science because they must learn this field while also learning English. Despite this problem, few studies have assessed the difficulty level of computer science texts for ESL learners.
Ideally, the difficulty levels of texts are assessed by having groups of ESL learners read them. However, owing to the excessive time and financial costs involved, such practices can be impractical.
Hence, using two highly accurate automatic readability assessors based on natural language processing (NLP) techniques, we assessed the readability of various computer-science-related texts for ESL learners. The first assessor is based on state-of-the-art deep transfer learning, and the second is based on classical machine learning and applied linguistics. For training the assessors, we used a standard corpus employed in NLP, which was annotated by professional English teachers to evaluate the readability of the texts for ESL learners. To conduct the experiments, we built a collection of computer science texts ranging from academic papers to software manuals (READMEs) crawled from a source-code hosting website, namely GitHub. The experimental results showed that intermediate ESL learners were able to read most of the computer science related texts.

Neural Language Model-based Readability Assessment of Computer Science Introductory Texts for English-as-a-Second Language Learners

This study examines the relations between parent–educator communication and preschoolers’ numeracy, literacy, and executive function (EF), and explores if these relations work through the home numeracy environment (HNE) and the home literacy environment (HLE). Children (51.33% female, M = 57.52 months, SD = 3.62) were enrolled in a state-funded prekindergarten program (n = 211) or in another community-based preschool program (n = 52). Analyses revealed a significant relation only between parent–educator communication and numeracy skills in the spring of preschool (β = -.14, p = .002). Unexpectedly, higher parent–educator communication scores in the spring of preschool were related to lower numeracy skills at that same time. Furthermore, the indirect effects for the HNE and HLE were not significant for all outcomes. The findings suggest that it may be beneficial for educators to engage with families early in the academic year to support children’s numeracy skill development.

The relations between parent–educator communication, the home environment, and children’s outcomes in preschool

The present study assessed both implicit and explicit math self-concept, math-gender stereotype, math achievement, and math anxiety in 260 Swiss children from 1st, 3rd, and 5th grade. In line with prior research, explicit math self-concepts were found to decrease as a function of participant age cohort. Furthermore, already children in the first grade were found to implicitly endorse the math-gender stereotype. In line with these findings, math anxiety was found to be higher in older cohorts than in younger, and it was non-trivial in all age groups. Lastly, as expected, math achievement was higher for older children. Taken together, the present study found patterns of stereotypes increasing as a function of age, with simultaneous increases in math anxiety, and simultaneous decreases in math self-concept. These findings underline the importance of studying math-related constructs in children. Understanding these patterns will contribute to sustaining STEM interests in young children and adolescents.


Math Self-Concept, Stereotypes, Achievement and Anxiety – A Cross-Sectional Study

Math anxiety (MA) is negatively related to math performance. One proposed intervention with potential to disrupt the MA-math performance link is expressive writing. The current study aimed to conceptually replicate Park and colleagues (2014). In that study, the authors concluded that expressive writing effectively boosted math anxious students’ performance. In our current sample of 168 college students, participants randomly assigned to the expressive writing condition were no more accurate at posttest than were other participants assigned to a math self-concept intervention, active control, or passive control. Additionally, participants in the math self-concept and active control conditions reported lower state MA immediately following the intervention; participants in the expressive writing and passive control conditions reported no differences between pretest and posttest state MA. The current study provides boundary conditions for the effectiveness of expressive writing interventions in ameliorating MA during difficult math tasks and illuminates potential mechanisms underlying MA.

Does Expressive Writing Blunt the Effects of Math Anxiety on Math Performance? A Conceptual Replication and Extension of Park et al. (2014)

Learning to write is characterized by bottom-up mimicking of characters and top-down writing from memory. We introduce a CNN-RNN model that implements both pathways: It can (i) directly write a letter by generating a motion trajectory given an image, (ii) first classify the character in the image and then determine its motion trajectory `from memory', or (iii) use a combination of both pathways. The results show that, in one-shot and few-shot learning, the model profits from different combinations of the pathways: The generation of different character variants works best when the top-down is supported by the bottom-up pathway. Refilling occluded images of efficiently learned characters works best when using the top-down pathway alone. Overall, the architecture implies that a weighted merge of bottom-up and top-down information into a latent, generative code fosters the development of compositional encodings, which can be reused in efficient learning tasks.

Efficient learning through compositionality in a CNN-RNN model consisting of a bottom-up and a top-down pathway

Intelligent agents are capable of transfer and generalization. This flexibility in adapting to new tasks and environments often relies on representation learning and replay. Among these algorithms, successor representation learning and memory replay offer biologically plausible solutions. However, replay prioritization algorithms remain largely limited to value prediction errors. Here we propose PARSR, Priority-Adjusted Replay for Successor Representations, to address this caveat. Decoupling learning of the environment dynamics and rewards, PARSR can use prediction errors from either representation learning or values to prioritize memory replay. We compare PARSR to SR-Dyna, Dyna-Q, and a number of state of the art algorithms using replay and successor representations in cognitive neuroscience. We find that PARSR is able to reproduce human behavior in a number of revaluation tasks while also representing a performance improvement over SR-Dyna, its closest counterpart.

Priority-Adjusted Replay for Successor Representations

This paper describes an initial cognitive framework that captures the reasoning involved in scientific data analyses, drawing from close collaborations with scientists in different domains over many years. The framework aims to automate data analysis for science. In doing so, existing large repositories of data could be continuously and systematically analyzed by machines, updating findings and potentially making new discoveries as new data becomes available. The framework consists of a cycle with six phases: formulating an investigation, initiating the investigation, getting data, analyzing data, aggregating results, and integrating findings. The paper also describes our implementation of this framework and illustrates it with examples from different science domains.

Towards Capturing Scientific Reasoning to Automate Data Analysis

The introduction of deep learning and CNNs to image recognition problems has led to state-of-the-art classification accuracy. However, CNNs exacerbate the issue of algorithm explainability due to deep learning’s black box nature. Numerous explainable AI (XAI) algorithms have been developed that provide developers insight into the operations of deep learning. We aim to make XAI explanations more user-centric by introducing modifications to existing XAI algorithms based on cognitive theory. The goal of this research is to yield intuitive XAI explanations that more closely resemble explanations given by experts in the domain of bird watching. Using an existing base XAI algorithm, we conducted two user studies with expert bird watchers and found that our novel averaged and contrasting XAI algorithms are significantly preferred over the base XAI algorithm for bird identification.

Premium content

Manipulating the face contour affects face recognition performance leaving the Face Inversion Effect unaltered

Downloads

Next from CogSci 2022

Quantifying the Socio-semantic Representations of Words

Stay up to date with the latest Underline news!

PRESENTATIONS

CONFERENCES

COMPANY

RESOURCES