United States

Pre record assistants have become fixtures in everyday settings, but most research focuses on their development rather than their use following deployment. To facilitate study of their use in office settings, we introduce OfficeDial, a multimodal dataset containing audio recordings, transcriptions, eye tracking data, and screen recordings from conversations between humans and Pre record assistants in office environments. Conversations are paired with physical and behavioral measures of cognitive load. We study the associations between verbal behavior and noise level and reveal key relationships between verbal redundancy, disfluency, and noise level. We make our new dataset available to interested researchers to inspire further exploration.


CogSci 2023

Linguistic Cognitive Load Analysis on Dialogues with an Intelligent Pre record Assistant

technical paper

The 45th Annual Meeting of the Cognitive Science Society will take place in Sydney, Australia as a modified hybrid conference. Presentations from the main-session room will be live-streamed every day for all events. This will include all invited talks and award sessions. Recordings of all CogSci 2023 sessions will be made available for on-demand viewing after the conference.<br>
The in-person component will be held at the International Convention Centre Sydney at Darling Harbour. The Conference will take place Wednesday July 26th – Saturday July 29th, 2023 (Wednesday, July 26th is the Workshops Day).

Join us!

<br>[**The Program at a Glance can be found here**](https://cognitivesciencesociety.org/wp-content/uploads/2023/07/COgSci-PAAG-V2.pdf)
<br>[**The Detailed Program can be found here**](https://online.fliphtml5.com/ebtyf/mswz/)
<br><br>

#### **CogSci 2023 Program Book**
<iframe style="width:700px;height:400px" src="https://online.fliphtml5.com/ebtyf/tzgk/"  seamless="seamless" scrolling="no" frameborder="0" allowtransparency="true" allowfullscreen="true" ></iframe>
<br>

The 45th Annual Meeting of the Cognitive Science Society presents the latest research across cognitive science and highlights the theme of Cognition in Context.

The best-performing artificial intelligence systems for object recognition are deep neural networks (DNNs). Engineers and neuroscientists have claimed that DNNs show human-level performance in object recognition. However, Barbu et al. (2019) reported accuracies of around 30% for state-of-the-art object recognition systems when testing on their better-controlled image set – ObjectNet. How do humans perform on ObjectNet? In our experiment, we tested 25 undergraduates on their ability to classify the ten categories of objects in ObjectNet that the Deep Convolutional Neural Networks (DCNNs) found easiest, moderate and hardest. Although humans and DCNNs had similar levels of overall accuracy, there were some everyday, basic-level categories for which machine performance was much lower than humans. The pattern of errors generated by the DCNNs were about as similar to human error patterns as the individual human error patterns were to each other. Implications of these results and plans for future work are discussed.

A Comparison of Human and Machine Performance in Object Recognition Using The ObjectNet Image Set

Collaboration can make collective judgments more accurate than individual judgments, but it also comes with costs in time, effort, and social cohesion. But how do we estimate these costs? In two experiments, we introduce children and adults to two teams in which the teammates disagree about the optimal solution to a novel problem, and ask which team would need more time to reach a consensus decision. We find that all ages expect slower decisions from teams with more people or factions, and expect the number of factions to matter more than the number of people. But only adults expect decisions initially endorsed by a stronger faction to be faster than those endorsed by a weaker faction. Results are discussed in context of children’s reasoning about dominance, and models of time-rational collective decision-making.


Herding cats: children’s intuitive theories of persuasion predict slower collective decisions in larger and more diverse groups, but disregard factional power

When you encounter a new claim, how do you decide whether to trust it or to seek more evidence first? In two studies, we show that children use environmental cues about informational reliability to rationally adapt their information seeking. In Study 1, children exposed to an environment containing some misinformation (as opposed to all true information) sampled more evidence before verifying a test claim in a novel domain. In Study 2, children showed a graded sensitivity to environmental reliability: information seeking increased linearly with the proportion of false statements heard during exposure. Additionally, these statements were presented as online search results, which demonstrated that children make sophisticated inferences about their informational environments, beyond speaker-specific cues, to adjust their skepticism toward new information. These results further emphasize the importance of considering the environmental context of learning when developing interventions to promote healthy skepticism and lifelong learning.


Children flexibly adapt their evidentiary standards to their informational environment

How can we quantify the degree to which a piece of evidence affects a person's belief? Philosophers investigating theories of Bayesian Confirmation have identified a plurality of potential measures, each with their own virtues and shortcomings. Psychologists meanwhile have largely neglected this question, which has limited their ability to understand differential belief updating, cases where certain individuals or groups respond to the same evidence in different ways. In this study, we examine how competing Bayesian confirmation measures track commonsense notions of evidential strength. We demonstrate how these measures can be computed from participants' belief reports, and identify cases where the measures come apart in their characterization of participants' belief updating. In so doing, this project seeks to build a connections between investigations of psychological belief updating processes and formal epistemic theories of confirmation.


Bayesian confirmation and commonsense notions of evidential strength

In deductive domains, three metacognitive knowledge types in ascending order are declarative, procedural, and conditional learning. This work leverages Deep Reinforcement Learning (DRL) in providing adaptive metacognitive interventions to bridge the gap between the three knowledge types and prepare students for future learning across Intelligent Tutoring Systems (ITSs). Students received these interventions that taught how and when to use a backward-chaining (BC) strategy on a logic tutor that supports a default forward-chaining strategy. Six weeks later, we trained students on a probability tutor that only supports BC without interventions. Our results show that on both ITSs, DRL bridged the metacognitive knowledge gap between students and significantly improved their learning performance over their control peers. Furthermore, the DRL policy adapted to the metacognitive development on the logic tutor across declarative, procedural, and conditional students, causing their strategic decisions to be more autonomous.


Bridging Declarative, Procedural, and Conditional Knowledge Gap Using Deep Reinforcement Learning

We present a computational-level model of language production in terms of a combination of information theory and control theory, in which words are chosen incrementally in order to maximize communicative value subject to an information-theoretic capacity constraint. The theory generally predicts a tradeoff between ease of production and communicative accuracy. We apply the theory to two cases of apparent availability effects in language production, in which words are selected on the basis of their accessibility to a speaker who has not yet perfectly planned the rest of the utterance. Using corpus data on English relative clause complementizer dropping from Levy & Jaeger (2007) and experimental data on Mandarin noun classifier choice from Zhan & Levy (2019), we show that the theory reproduces the observed phenomena and makes successful new predictions, providing an alternative account to Uniform Information Density (UID) and a promising general model of language production which is tightly linked to emerging theo- ries in computational neuroscience.


An information-theoretic account of availability effects in language production

The world’s languages differ in how they order adjectives and nouns relative to each other. We ask whether cross-linguistic variation and systematicity in adjective-noun order can be explained by the iterated pressure for pragmatic referential communication. To this end, we formalize cooperative pressures on referential communication in the Rational Speech Act framework in an iterated learning setting. First, we show that the less informative adjectives are relative to nouns, the more likely they are to occur post-nominally. This is the case when informativeness is manipulated via the composition of the lexical space (i.e., changing the relative number of adjectives vs. nouns that are available for reference), and via the inherent referential utility of adjectives vs. nouns. Secondly, we show that under the assumption that nouns are on average more informative than adjectives, the model predicts a cross-linguistic distribution of ordering preferences that qualitatively resembles the empirical one, with these biases becoming further entrenched with iterated language use. Taken together, these results suggest a possible pathway for syntactic preferences to be calcified over time as the result of pragmatic communicative pressures on language.


The cross-linguistic order of adjectives and nouns may be the result of iterated pragmatic pressures on referential communication

Why are some words more frequent than others? Some reasons are self-evident. We have far more reasons to use a word like ”eat” than a word like ”diagonalize”. But robust differences in frequency are also observed for words with seemingly equal communicative usefulness. For example, hot and cold seem equally important for communicating temperature yet hot in English is more frequent than cold. We focused on antonym pairs such as these and sought to predict differences in frequency from the connection patterns of these words in a semantic network while controlling for predictors like the number of word senses. Two network properties predicted word frequency especially well: the number of connections the word and its surrounding words have, and the ability of the
word to connect less interconnected words. These two network properties not only predicted present word frequency, but also predicted future frequency changes suggesting a potential causality relationship between network properties and word frequency. Overall, this study offers new insights into the underlying causes of differences in word frequency and highlights the importance of considering a network perspective when examining how word frequency evolves.


Understanding the Frequency of a Word by its Associates: A Network Perspective

Sampling assumptions --- the assumptions people make about how an example of a category or concept has been chosen --- help us learn from examples efficiently. One context where sampling assumptions are particularly important is in social contexts, where a learner needs to infer the knowledge and intentions of the information provider and vice-versa. The pedagogical sampling assumptions model describes a Bayesian account of how learners and providers should behave given different assumptions they have about the other (e.g., is the provider trying to deceive or help me? Does the learner trust me?). In this study, we tested how well this model could describe learning behaviour in the rectangle game, where a fictional information provider revealed clues about the structure of a rectangle that the learner (a participant) needed to guess. Participants received clues from either a helpful information provider, a provider who was randomly sampling clues, or one of two kinds of unhelpful providers (who could mislead but could not lie). We found that people learned efficiently and in line with model predictions when the provider was helpful and that this was the case even when no cover story was provided. However, although participants could identify that unhelpful providers were not being helpful, they struggled to learn the strategy those providers were using.


Premium content

Downloads

Next from CogSci 2023

A Comparison of Human and Machine Performance in Object Recognition Using The ObjectNet Image Set