Search

It takes children considerable learning and development to accurately predict whether an object is safely balanced or will fall -- something that happens if its center of mass is not supported from below. In the meantime, children go through a characteristic set of mistaken beliefs. Here we use an adapted version of the classical balance task to evaluate whether different models go through the same stages. Preliminary results show that convolutional neural networks (CNNs) do learn the task but do not necessarily go through the same stages. We are also testing several simulation-based accounts. We anticipate completing this work in time for the conference. The findings will help clarify the space of possible accounts of children's acquisition of intuitions about gravity and balance.

Humans are born with an intuitive representation of the physics world. How accurate is intuitive physics? Researchers from education focus on the failures, students' errors and misconceptions while cognitive psychologists argue humans anticipate and manipulate physical environments in ways betraying veridical knowledge of classical mechanics. One solution is to hypothesize there are distinct systems of “cognitive physics” with different limitations and deployment in the tasks favored by the two literatures. The goal of current study is to gather evidence from psychometric studies by estimating how many distinct factors explain performance on intuitive physics assessments. We build an augmented concept inventory including several previously-validated concept inventories, around 120 items. The pilot study indicated that participants recruited online from Prolific displayed expected accuracy on the tasks. We are now collecting around 1,000 participants and applying multidimensional item response theory (MIRT) analyzes to identify the latent structure.

Since the Workshop on Scaling Cognitive Science was held at CogSci 2020, the methodological landscape of the field has undergone a significant shift toward online experiments. The COVID-19 pandemic forced greater numbers of cognitive scientists to collect data over the internet and to explore online alternatives to methods which they had previously used only in the lab. For example, since this time, eye-tracking capabilities have been introduced to a number of software packages for online experiments (Ba ́nki, de Eccher, Falschlehner, Hoehl, & Markova, 2022; Slim & Hartsuiker, 2022; Yang & Krajbich, 2021).

Bilingualism varies greatly in terms of language background, age of acquisition, and learning environment, making it complex to study the relationship between language and brain structure. Age of acquisition and linguistic proficiency are both crucial factors in language acquisition research and are closely related. To better understand the relationship between bilingualism and brain structure, we created a dataset of 2,000+ brain 3D images from bilinguals and monolinguals and trained a neural network (Contrastive Variational Autoencoder, CVAE) to distinguish bilingual-specific neuroanatomical variations from those shared by monolingual groups. The initial results reveal that the common variation (e.g. scanner and sex) in brain structure shared by two groups can be identified by CVAE. More importantly, CVAE disentangled linguistic proficiency from the common variation but not age of acquisition. It suggests that compared to age of acquisition, linguistic proficiency is more correlated with brain’s structure.

The impact of an individual's first language (L1) on their acquisition of a second language (L2) is widely recognized in the field of psycholinguistics. However, previous research has focused on limited L1-L2 pairs, leaving questions about how different linguistic structures, such as genetic relationship, phonology, or syntax, can either facilitate or impede the learning of a new language. In our current study, we aimed to address these gaps by analyzing standardized English assessment scores of English Language Learners (ELLs) from 46 international schools across 30 countries. We compared linguistic similarities between more than 40 L1s and examined how different linguistic structures of the L1 influenced the acquisition of L2, specifically in terms of genetic, phonological, and syntactic similarities. Our findings indicate that older ELLs learn English at a faster rate than younger ELLs, and among older ELLs, those who speak an L1 that shares linguistic similarities with English acquire the language faster than ELLs whose L1 is less closely related to English. These results highlight the transfer of linguistic knowledge from L1 to L2 and emphasize the importance of age of acquisition in L2 learning. Our study has important theoretical and pedagogical implications for research on second language acquisition.

Humans, even from infancy, are capable of unsupervised (“sta- tistical”) learning of linguistic information. However, it re- mains unclear which of the myriad algorithms for unsuper- vised learning captures human abilities. This matters because unsupervised learning algorithms vary greatly in how much can be learned how quickly. Thus, which algorithm(s) humans use may place a strong bound on how much of language can ac- tually be learned in an unsupervised fashion. As a step towards more precisely characterizing human unsupervised learning capabilities, we quantitatively synthesize the literature on adult unsupervised (“statistical”) word segmentation. Unfortunately, most confidence intervals were very large, and few moderators were found to be significant. These findings are consistent with prior work suggesting low power and precision in the litera- ture. Constraining theory will require more, higher-powered studies.

The Simon, Stroop, and flanker tasks are commonly used to investigate cognitive control. However, it remains unclear whether these three tasks in fact measure the same cognitive abilities and in the same proportion. We take a developmental approach to this question: if the tasks all roughly measure the same capacity, they should show similar patterns of age-related change. We present data from two massive online studies: Study 1 included 9,642 participants 10 to 80 years of age who completed the Simon and Stroop tasks, and Study 2 included 13,448 participants 10 to 79 years of age who completed the flanker task. The results revealed markedly different developmental trajectories among the tasks, with only the flanker task following an inverted U-shaped trajectory. These findings caution against using standard congruency tasks to draw general conclusions about the development of cognitive control and underscore the importance of developing more psychometrically rigorous measures.

Second-language (L2) learning is characterized by both positive and negative transfer from the first language (L1). However, psycholinguistic studies focus on a few syntactic phenomena and L1-L2 pairs at a time, resulting in an incomplete picture. We apply machine learning to seven learner corpora in English and Spanish with 39 language pairs, showing that statistical models combined with simple $n$-grams of part-of-speech tags and syntactic dependency relations achieve good performance in recovering the L1, indicating structural transfer from L1 to L2. Further machine learning using a rich hand-curated linguistic feature set allowed us to identify aspects of L2 linguistic structure particularly influenced by L1 (verbal morphology, average dependency tree parse depth, and headedness of clausal structures) as well as those with minimal influence (distributions of dependency relations, basic word orders, or non-projective dependencies).

Replication studies in psychological science sometimes fail to reproduce prior findings. If these studies use methods that are unfaithful to the original study or ineffective in eliciting the phenomenon of interest, then a failure to replicate may be a failure of the protocol rather than a challenge to the original finding. Formal pre-data-collection peer review by experts may address shortcomings and increase replicability rates. We selected 10 replication studies from the Reproducibility Project: Psychology (RP:P; Open Science Collaboration, 2015) for which the original authors had expressed concerns about the replication designs before data collection; only one of these studies had yielded a statistically significant effect (p

The proliferation of web-connected devices has presentedsignificant opportunities and challenges to cognitive sci-ence — opportunities in that cognitive scientists can collectdata relevant to human cognition orders of magnitude fasterthan before, addressing questions that were otherwise impos-sible to address; and challenges in that cognitive scientistsrequire new infrastructure to collect these data and new meth-ods to analyze them once collected.This workshop brings together cognitive scientists who areat the forefront of these opportunities and challenges in scal-ing cognitive science, along with cognitive scientists whowould like to be, to engage in a day of interactive exchangeand development of ideas related to scaling cognitive science.The workshop is a full day. Each presentation addresses adifferent opportunity or challenge. One set of presentationshighlights opportunities. A second set highlights challengesof statistical analysis and data collection. In a break-out ses-sion, attendees address these opportunities and challengeshead on.

Language is frequently ambiguous, with the same sentence having several possibleinterpretations. One prevalent exampleis third-person pronouns. Hartshorne, Gerstenberg, & Tenenbaum (2014) HGT2014 model pronoun interpretation asan inference over a generative model of the speaker. An advantage of the generative intuitive theory approach is thatit incorporates a flexible, quantitative model of world knowledge rather than a list of facts and heuristics. The authorsformalized this world knowledge as inference over a generative model of the world. We directly test this flexibilityby changing the rules of the world (e.g., through scenarios that reverse the normal relationship between strength andprobability of winning tug-of-war), which according to HGT2014 should directly affect pronoun interpretation. We findthat model predictions and participant judgments align well in such scenarios, supporting HGT2014 and challenging othertheories of pronoun resolution. We discuss this work in the context of recent work on intuitive theories.

Ambridge argues that the existence of exemplar models for individual phenomena (words, inflection rules, etc.) suggests the feasibility of a unified, exemplars-everywhere model that eschews abstraction. The argument would be strengthened by a description of such a model. However, none is provided. I show that any attempt to do so would immediately run into significant difficulties -- difficulties that illustrate the utility of abstractions. I conclude with a brief review of modern symbolic approaches that address the concerns Ambridge raises about abstractions. [For Ben Ambridge's "Against Stored Abstractions: A Radical Exemplar Model of Language Acquisition," see EJ1269951.]

This full-day workshop focuses on Massive OnlineExperiments (MOEs). MOEs have transformative potential,as they effectively allow researchers to run hundreds ofexperiments simultaneously (cf. Hartshorne & Germine,2015; Reinecke & Gajos, 2014). The goal of the workshopis to help a broad cross-section of cognitive scientists beginto incorporate MOEs into their research.

Moral judgment often involves pinning causation for harm toa particular person. Since it reveals “who one sides with”, ex-pression of moral judgment can be a costly social act that peo-ple may be motivated to conceal. Here, we demonstrate thata simple, well-studied psycholinguistic task (implicit causal-ity) can be leveraged as a novel implicit measure of morallyrelevant causal attributions. Participants decided whether tocontinue sentences like “Amy killed Bob because...” with ei-ther the pronoun he or she. We found that (1) implicit causal-ity selections predicted explicit causal judgments, (2) select-ing the object (victim) for harm/force events (e.g., kill, rape)predicted endorsement of moral values previously linked tovictim-blame, and (3) higher hostile sexism predicted select-ing the female as the cause in male-on-female harm/force. Theimplicit causality task is a new measure of morally motivatedcausal attribution that may circumvent social desirability con-cerns.

The relationship between a verb and the syntactic frames in which it can appear has been closely studied by psychol-ogists and linguists. Research suggests that the semantics of a verb and its arguments determine the verb’s syntactic frames,but various theories (Levin & Hovav, 2005) disagree on the nature and complexity of these relationships, in part because mostinvestigations have focused on a small subset of verbs that may not generalize. Investigating the semantic and syntactic rela-tionships present in larger sets of verbs would provide more substantial evidence for evaluating and selecting theories of verbargument structure. We report on initial analyses of the 6000+ verbs and 280+ syntactic frames of VerbNet (Kipper et al., 2008),the largest English verb syntax resource available, using nonparametric Bayesian methods (e.g. Shafto et al., 2006) for clusteranalysis and dimensionality reduction.

Recent reports have suggested that many published results are unreliable. To increase the reliability and accuracy of published papers, multiple changes have been proposed, such as changes in statistical methods. We support such reforms. However, we believe that the incentive structure of scientific publishing must change for such reforms to be successful. Under the current system, the quality of individual scientists is judged on the basis of their number of publications and citations, with journals similarly judged via numbers of citations. Neither of these measures takes into account the replicability of the published findings, as false or controversial results are often particularly widely cited. We propose tracking replications as a means of post-publication evaluation, both to help researchers identify reliable findings and to incentivize the publication of reliable results. Tracking replications requires a database linking published studies that replicate one another. As any such database is limited by the number of replication attempts published, we propose establishing an open-access journal dedicated to publishing replication attempts. Data quality of both the database and the affiliated journal would be ensured through a combination of crowd-sourcing and peer review. As reports in the database are aggregated, ultimately it will be possible to calculate replicability scores, which may be used alongside citation counts to evaluate the quality of work published in individual journals. In this paper, we lay out a detailed description of how this system could be implemented, including mechanisms for compiling the information, ensuring data quality, and incentivizing the research community to participate.

Widespread failures of replication and generalization are, ironically, a scientific triumph, in that they confirm the fundamental metascientific theory that underlies our field. Generalizable and replicable findings require testing large numbers of subjects from a wide range of demographics with a large, randomly‐sampled stimulus set, and using a variety of experimental parameters. Because few studies accomplish any of this, meta‐scientists predict that findings will frequently fail to replicate or generalize. We argue that to be more robust and replicable, developmental psychology needs to find a mechanism for collecting data at a greater scale and from more diverse populations. Luckily, this mechanism already exists as follows: Citizen science, in which large numbers of uncompensated volunteers provide data. While best‐known for its contributions to astronomy and ecology, citizen science has also produced major findings in neuroscience and psychology, and increasingly in developmental psychology. We provide examples, address practical challenges, discuss limitations, and compare to other methods of obtaining large datasets. Ultimately, we argue that the range of studies where it makes sense *not* to use citizen science is steadily dwindling. [ABSTRACT FROM AUTHOR]

Famously, "dog bites man" is trivia whereas "man bites dog" is news. This illustrates not just a fact about the world but about language: to know who did what to whom, we must correctly identify the mapping between semantic role and syntactic position. These mappings are typically predictable, and previous work demonstrates that young children are sensitive to these patterns and so could use them in acquisition. However, there is only limited and mixed evidence that children do use this information to guide acquisition outside of the laboratory. We find that children understand emotion verbs which follow the canonical CAUSE-VERB-PATIENT pattern ("Mary frightened/delighted John") earlier than those which do not ("Mary feared/liked John"), despite the latter's higher frequency, suggesting children's generalization of the mapping between causativity and transitivity is broad and active in acquisition.