Search

Despite the importance of risk attitude on decision making, its role on belief updating has been overlooked. In this paper, we aim to answer this fundamental question. We show, using economic theory, that stronger risk aversion drives more conservative actions, decreases the instrumental value of information relative to the importance of belief-based utility. Thus, with utility for clarity, stronger risk attitude yields to more belief change; whereas with updating cost, stronger risk attitude leads to less belief change. We confirm the result experimentally in two settings where subjects update their belief about their IQ and a randomly drawn number, respectively.

Frequently in psychology, simple tasks that are designed to tap a particular feature of cognition are used without considering the other mechanisms that might be at play. For example, the delayed-match-to-sample (DMTS) task is often used to examine short-term memory; however, a number of cognitive mechanisms interact to produce the observed behaviour, such as decision-making and attention processes. As these simple tasks form the basis of more complex psychological experiments and theories, it is critical to understand what strategies might be producing the recorded behaviour. The current paper uses the GEMS methodology, a system that generates models of cognition using genetic programming, and applies it to differing DMTS experimental conditions. We investigate the strategies that participants might be using, while looking at similarities and differences in strategy depending on task variations; in this case, changes to the interval between study and recall affected the strategies used by the generated models.

Research in child development has shown that embodied experience handling physical objects contributes to many cognitive abilities, including visual learning. One characteristic of such experience is that the learner sees the same object from several different viewpoints. In this paper, we study how learning signals that equate different viewpoints---e.g., assigning similar representations to different views of a single object---can support robust visual learning. We use the Toybox dataset, which contains egocentric videos of humans manipulating different objects, and conduct experiments using a computer vision framework for self-supervised contrastive learning. We find that representations learned by equating different physical viewpoints of an object benefit downstream image classification accuracy. Further experiments show that this performance improvement is robust to variations in the gaps between viewpoints, and that the benefits transfer to several different image classification tasks.

Existing studies have suggested that there may be a link between perception and production of speech sounds. In this paper we examine this link using data collected on Singapore English-Mandarin bilingual adults. We used a 2AFC task to measure perception of the Singapore Mandarin [tsʰuə́n]-[ʈʂʰuə́n] alveolar-retroflex contrast, and measured production using centre of gravity (CoG) measurements on speech recordings of alveolar and retroflex words. We found that a perception-production link may exist in our participants, with participants who exhibit less retroflexion in speech showing less sensitivity to retroflexion in ambiguous speech sounds.

Forgiveness plays a significant role in our everyday social life, and, because of that, it has received an increasing amount of attention in academic research. However, philosophers and psychologists are equally worried by the fact that we still lack an empirically adequate characterization of forgiveness. In this paper, we present two preregistered studies in which we explore what ordinary people believe a speaker does when he or she performs the speech act of forgiving by uttering the phrase, “I forgive you”. Study 1 uses a vignette-based stimulus to examine what participants believe to change after the victim granted forgiveness to their wrong-doer. In Study 2, we apply a linguistic test, the cancellability test, to determine whether participants consider forgiving the wrong-doer but still blaming them compatible.

When faced with uncertainty, human observers maximize performance by integrating sensory information with learned task-relevant regularities. Does this behavior similarly occur in social settings? In this paper, we explore how reward-seeking behavior in an intergroup context is affected by readily available but task-irrelevant social information (in the form of group membership) when task-relevant reward information can be learned over time. Across two experiments, we show that participants learned and utilized task-relevant regularities to inform their choices. We also show that human observers are not universally biased towards utilizing social information in all settings––participants learned to disregard social information when not relevant to the task at hand. However, learning about the utility of social information (Experiment 2) had a long-term influence on observers’ ability to subsequently learn and utilize available sources of information. Real-world intergroup contexts typically encompass situations and stimuli that have been previously experienced by the observer. Our findings highlight the powerful influence of learning in such contexts.

Vaccines are among the best tools to limit the spread of preventable diseases. yet, recent years have seen a rise in anti-vaccination sentiments for vaccines against COVID-19, the MMR, and more. It is critical to understand the factors that influence whether a person will accept or reject a vaccine for a given disease. This paper tests a Bayesian model to predict attitudes toward vaccines. with five factors (subjective beliefs concerning danger of illness, safety of the vaccine, prevalence of the disease, perceived social norm, and governmental recommendation). To parameterize the model, Study 1 elicits the full conditional probability table while Study 2 tests model predictions by eliciting people’s priors for COVID-19 and the common flu. We find a good fit between predictions and observations, accounting for 53% and 44% of the variance. This suggests the usefulness of a formal model to capture people’s beliefs about vaccination.

Visual number adaption is a widely accepted phenomenon. This paper advances an alternative explanation for putative cases of the phenomenon. We propose that such cases may simply reflect observers adapting to the items in perceived displays, rather than their numerical quantity. Three experiments motivate consideration of this novel proposal and call into question the evidential basis for received formulations of the number adaptation hypothesis.

This paper presents transdisciplinary research on laryngectomy and a methodological stance to broaden research paradigms for the cognitive sciences. Studying daily experiences of people communicating without biological larynx in an interactive context, we put special emphasis on methodology combining engaged epistemology with ethnographic fieldwork. Our results made evident i) the role of anatomical and physiological adaptations in shaping communication and social relations, ii) the existence of multimodal and context-dependent alternative strategies of conversation, iii) the crucial role of participants’ agency. The dialogue between epistemologically engaged cognitive science and ethnographic fieldwork allowed us to remain open to novel interpretations of the communicative situations and led to unexpected observations. Results of this study point to the importance of integrating qualitative methodologies within research on cognition, and may prove useful for guiding therapeutic interventions and novel technological designs.

This paper presents a challenge to enactivist approaches to cognition (e.g. Ward, D., Silverman, D. & Villalobos, M. 2017) that is based on the theoretical commitments behind forms of looking time studies that have been extensively used to probe into the cognitive abilities of infants and nonhuman animals. I briefly summarize the Violation of Expectation (VoE) paradigm (Ginnobili & Olmos 2021) to illustrate why such methods might pose a problem for enactivists and their conception of cognition as a largely representation-free dynamic coupling between organism and environment. I argue that despite the lack of clarity regarding how the notion of expectation should be applied to the minds of neonates and nonhuman animals, there is an inherently representational aspect to expectation, given that it embodies satisfaction conditions. The challenge is, then: given that many forms of enactivism seem to reject the notion of representation as it is used in the VoE literature, how can enactivists make sense of data and results obtained using such research methods?

In Mandarin Chinese, abbreviation happens commonly to compound words across different syntactic categories. What is the motivation behind this shortening of words? This paper presents an investigation of this phenomenon from an information-theoretic point of view. A corpus study was con- ducted to measure the average amount of information contained in the full (long) form and the abbreviated (short) form of words given certain contexts. The amount of information was then compared between the long and short forms of a word, revealing that the short one usually contains less information, and therefore is more likely to be used in more predictive contexts. This result indicates that speakers of Chinese can choose to use shorter words when the context is more predictive, in accordance with considerations of efficiency.

Behavior Trees are widely used in many games, however, they are normally designed in a static way which is hard to deal with dynamic environments. One of the factors that restrict the performance of BTs is the optimization of resources used among subtasks, which has not been paid much attention. In this paper, we concentrate on the problem of dynamic allocation of agents for BTs, i.e., an important resource in games. We design a novel hierarchical learning framework, including the allocation learning module and the action learning module, to find strategies for agent allocation and action selection. Each agent is associated with a network for encoding observations, which is then assigned to a subtask by the allocation learning network. Each subtask uses a network to compute action for the corresponding agent. We design how the losses are computed for the two modules respectively and present the whole training procedure.

Modelling of early language acquisition aims to understand how infants bootstrap their language skills. The modelling encompasses properties of the input data used for training the models, the cognitive hypotheses and their algorithmic implementations being tested, and the evaluation methodologies to compare models to human data. Recent developments have enabled the use of more naturalistic training data for computational models. This also motivates development of more naturalistic tests of model behaviour. A crucial step towards such an aim is to develop representative speech datasets consisting of speech heard by infants in their natural environments. However, a major drawback of such recordings is that they are typically noisy, and it is currently unclear how the sound quality could affect analyses and modelling experiments conducted on such data. In this paper, we explore this aspect for the case of infant-directed speech (IDS) and adult-directed speech (ADS) analysis. First, we manually and automatically annotated audio quality of utterances extracted from two corpora of child-centred long-form recordings (in English and French). We then compared acoustic features of IDS and ADS in an in-lab dataset and across different audio quality subsets of naturalistic data. Finally, we assessed how the audio quality and recording environment may change the conclusions of a modelling analysis using a recent self-supervised learning model. Our results show that the use of modest and high audio quality naturalistic speech data result in largely similar conclusions on IDS and ADS in terms of acoustic analyses and modelling experiments. We also found that an automatic sound quality assessment tool can be used to screen out useful parts of long-form recordings for a closer analysis with comparable results to that of manual quality annotation.

We investigated the potential and limitations of generative artificial intelligence (AI) in reflecting the authors' cognitive processes through creative expression. The focus is on the AI-generated artwork's ability to understand human intent (alignment) and visually represent emotions based on criteria such as creativity, aesthetic, novelty, amusement, and depth. Results show a preference for images based on the descriptions of the authors' emotions over the main events. We also found that images that overrepresent specific elements or stereotypes negatively impact AI alignment. Our findings suggest that AI could facilitate creativity and the self-expression of emotions. Our research framework with generative AIs can help design AI-based interventions in related fields (e.g., mental health education, therapy, and counseling). The full version of the paper can be found at https://arxiv.org/abs/2304.13324.

Although there is a great deal of interest in conditional probabilities in Bayesian cognitive science, there is still little understanding of how well human agents can learn them. This paper addresses the issue by theoretical and experimental means. In the theoretical part, we distinguish between cases of vacuous learning, where learned probabilistic information is not new, and belief revision is unwarranted, from cases with truly new information. In the experimental part, we investigate how well participants can distinguish these cases and how well they respect the probabilistic norms, thus adding new insights to the long-standing question of the extent to which the human mind is adapted to probabilistic norms.

Previous work has suggested that abstraction should not be treated as a unitary construct, but it should be split in at least two components. Concreteness is based on the proportion of sensory information in a concept, while precision is based on the aggregation of information corresponding to the concept position in a semantic taxonomy. This line of work showed that while both of these components have been used as operationalizations of abstraction, they can have opposite effects on cognitive performance. In this paper we extend those findings by showing that precision and concreteness are also differently related to affective content. Our current findings are relevant to some inconsistencies and puzzles in the existing research on the link between abstraction and affective content, and further emphasize the multi-component structure of abstraction.

What is the difference between language that describes the world and language that evaluates it? It has been suggested that an essential, distinguishing feature of evaluative language is its potential to guide actions by providing us with reasons to act. Calling an action “cruel” not only evaluates it negatively, its cruelty also provides us with a reason to refrain from it. Descriptive language, in and by itself, is relatively inert in this respect. In this paper, we examine whether this undisputed assumption is empirically adequate. We present three preregistered studies that demonstrate that evaluative language provides reasons for action when an agent contemplates how she should act, and also in conversational contexts. However, we also demonstrate that the speaker can easily deny the intention to provide such reasons to act.

Dependency tree is efficient for relation extraction model to exploit relations between words.Recent approaches have achieved promising performance while still suffering from inherent limitations, such as the computation efficiency and flexible pruning strategies. In this paper, we propose a novel relation extraction framework called Progressive Graph Learning over pruned dependency tree (PGLNet). PGLNet constructs a set of graphs by progressively adapting to input sentence. Specially, we implement the model to construct progressive weighted adjacency matrices by learning the relations among graph nodes with multi-head self-attention mechanism.Then, the model takes the learned weights as reference to prune dependency tree in order to preserve useful relevant sub-structures for the relation extraction while removing irrelevant words. Next, progressive convolution module is designed to encode the relations of entities and followed by relations classification.We evaluate our proposed model using public real-world datasets, experimental results demonstrate that the proposed model achieves state-of-the-art performance with consistency in all datasets. We conclude that the ability of PGLNet to progressively adapt to input data and enable the model with robustness.

On-line decision problems – in which a decision is made based on a sequence of past events without knowledge of the future – have been extensively studied in theoretical computer science. A famous example is the Prediction from Expert Advice problem, in which an agent has to make a decision informed by the predictions of a set of experts. An optimal solution to this problem is the Multiplicative Weights Update Method (MWUM). In this paper, we investigate how humans behave in a Prediction from Expert Advice task. We compare MWUM and several other algorithms proposed in the computer science literature against human behavior. We find that MWUM provides the best fit to people’s choices.

Decades of research have revealed that spatial language is the result of a complex interplay between language-independent, conceptual factors and language-specific forces (e.g., Bowerman, 1996; Johnston & Slobin, 1979; Landau & Jackendoff, 1993; Levinson & Meira 2003; Levinson & Wilkins, 2006). However, currently, a growing body of research acknowledges the importance of a further, much less explored but highly important factor having to do with pragmatic pressures – the general communicative need to convey informative meanings with appropriate levels of required effort (Grice, 1975; Zipf, 1949). This symposium aims to present research documenting the effect of communicative/pragmatic pressures on how spatial language systems are organized, used by speakers of different languages and acquired during development. The contributed papers explore a variety of spatial language phenomena, across many different languages, using a variety of empirical methods and diverse populations of participants.

Computational models of perceptual decision-making depend heavily on empirical goodness-of-fit measures for model selection. However, it is not possible to improve models' fit to data indefinitely, particularly when the data in question are variable across multiple elicitations. The completeness of a model or a theory assesses the extent to which it can predict observations in comparison with an ideal model. We measure the completeness of contemporary race models on a paradigmatic perceptual decision-making task - random dot motion discrimination - and show that the simple drift diffusion model is already close to complete in describing random dot motion discrimination data, with more complex models being in fact over-fit to datasets. Thus, in this paper, we quantitatively demonstrate limits to the ability of conventional choice fraction and response time data to disambiguate complex models of perceptual decision-making.

Solving problems with others not only reduces the time required to complete a challenge but may also enable the discovery of novel strategies that qualitatively change how a problem is approached. At the dyadic level, the laboratory-based ‘shepherding task’ demonstrated that, when tasked to contain evasive agents to a centralized location, some participants discover a non-obvious but optimal strategy to solve the task. This paper quantified the interactions between participants engaged in the task using Multidimensional Cross-Recurrence Quantification Analysis (MdCRQA), applied to each participant’s gaze and hand movements. The results demonstrated that strategy discoverers exhibited greater amounts of incidental coupling than non-discoverers prior to discovery. Once discovered, the strategy reduced the strength of coupling between participants, indicating that the strategy also reduced coordination demands. Future work will investigate whether differences in problem-solving can be attributable to differences in the perceptual features participants use which scaffold the discovery of task-optimal solutions.

It is rather uncontroversial that gender should have no influence on treating others as equal epistemic agents. However, is this view reflected in practice? This paper aims to test whether the gender of the testifier and the accused of assault is related to the perception of a testimony’s reliability and the guilt of the potential perpetrator. Two experiments were conducted, in which the subjects (n = 361, 47% females, 53% males) assessed the reliability of the testifier and the guilt of the potential perpetrator in four scenarios of assault accusation, in which the only difference was the gender of the people presented. During the study, we have observed dependencies of gender and ascription of reliability, and, to a lesser extent, of guilt. The results of our research may constitute an argument for the existence of different epistemic status endowed on people depending on their gender and existing gender stereotypes.

Humans can readily generalize their learning to new visual concepts, and infer their associated meanings. How do people align the different conceptual systems learned from different modalities? In the present paper, we examine emojis— pictographs uniquely situated between visual and linguistic modalities—to explore the role of alignment and multimodality in visual and linguistic semantics. Simulation experiments show that relational structures of emojis captured in visual and linguistic conceptual systems can be aligned, and that the ease of alignment increases as the number of emojis increases. We also found that emojis with subjective impressions of high popularity are easier to align between their visual and linguistic representations. A behavioral experiment was conducted to measure similarity patterns between 48 emojis, and to compare human similarity judgments with three models based on visual, semantic and multimodal-joint representations of emojis. We found that the model trained with multimodal data by aligning visual and semantic spaces best accounts for human judgments.

Recent studies suggest that errors facilitate learning in certain conditions. Despite this, reinforcement paradigms dominate learning methods, subscribing to the narrative that errorless learning is the foundation of an ideal learning environment. If we continue to view learning from this restrictive perspective, we may fail to capture and apply the benefits of errors. In this paper, we investigate two potential mechanisms of post-error learning. Participants (N = 61) learned word pairs in either a study or error trial before taking a final test. Supporting past error learning literature, errors before a study opportunity led to better performance on a final test. Differences in reaction times between conditions support the theory that errors increase learning by acting as a mediator, or secondary cue, to the correct answer on subsequent tests.

Recent years have seen a surge of interest in probabilistic measures of argument strength. Such measures have been used to elucidate longstanding questions of both theoretical and practical interest around fallacies of argumentation, scientific argument, legal argument, or argument and evidence aggregation. The measures typically used have drawn, in one form or other, on conditional probability as a central component for measuring argument quality, strength or confirmation. Work in this area has also consistently highlighted the value of Bayesian Belief Networks for computing argument strength in multi-argument (variable) contexts. However, to date, there has been no consideration of whether the key structural property underlying such networks – the notion of conditional independence- has a bearing on argument quality. In this paper, we consider the potential impact of causal structure on the utility of arguments.

Deep neural network representations of entities can serve as inputs to computational models of human mental representations to predict people's behavioral and physiological responses to those entities. Though increasingly successful in their predictive capabilities, the implicit notion of "human" that they rely upon often glosses over individual-level differences in beliefs, attitudes, and associations, as well as group-level cultural constructs. In this paper, we model shared representations of food healthiness by aligning learned word representations with the consensus among a group of respondents. To do so, we extend Cultural Consensus Theory to include latent constructs structured as fine-tuned word representations. We then apply the model to a dataset of people's judgments of food healthiness. We show that our method creates a robust mapping between learned word representations and culturally constructed representations that guide consumer behavior.

When faced with a polar question, speakers often provide overinformative answers going beyond a simple “yes” or “no”. But what principles guide the selection of additional information? In this paper, we provide experimental evidence from two studies suggesting that overinformativeness in human answering is driven by considerations of relevance to the questioner’s goals which they flexibly adjust given the functional context in which the question is uttered. We take these human results as a strong benchmark for investigating question-answering performance in state-of-the-art neural language models, conducting an extensive evaluation on items from human experiments. We find that most models fail to adjust their answering behavior in a human-like way and tend to include irrelevant information. We show that GPT-3 is highly sensitive to the form of the prompt and only achieves human-like answer patterns when guided by an example and cognitively-motivated explanation.

A long tradition of studies in psycholinguistics has examined the formation and generalization of ad hoc conventions in reference games, showing how newly acquired conventions for a given target transfer to new referential contexts. However, another axis of generalization remains understudied: how do conventions formed for one target transfer to completely distinct targets, when specific lexical choices are unlikely to repeat? This paper presents two dyadic studies (N=240) that address this axis of generalization, focusing on the role of nameability --- the a priori likelihood that two individuals will share the same label. We leverage the recently-released KiloGram dataset, a collection of abstract tangram images that is orders of magnitude larger than previously available, exhibiting high diversity of properties like nameability. Our first study asks how nameability shapes convention formation, while the second asks how new conventions generalize to entirely new targets of reference. Our results raise new questions about how ad hoc conventions extend beyond target-specific re-use of specific lexical choices.

Analogical reasoning is central to thought and learning, with a key focus on similarities where the same relations hold across different sets of elements. The current research investigated analogical reasoning in first-order and second-order relations, represented in word-like symbols following syntactic rules. Since most previous studies in this field have focused on visuospatial and semantic analogy, the primary goal of this paper was to specify the neural substrates for analogical reasoning based on syntactic rules. We found the activation locates in both anterior and posterior MFG, and these two regions correspond to parts of rlPFC and dlPFC respectively. The results reveal that although analogical reasoning with syntactic rules recruits similar brain regions as those in visuospatial and semantic analogy, their activation patterns are not the same. The role of rlPFC, the core area for analogical reasoning, needs to be reconsidered, and the complexity of syntactic-related activation is worth more attention.

Generic statements ('Tigers have stripes') are pervasive and early-emerging modes of generalization with a distinctive linguistic profile. Previous experimental work found that generics display a unique asymmetry between their acceptance conditions and the implications that are typically drawn from them. This paper presents evidence against the hypothesis that only generics display an asymmetry. Correcting for limitations of previous designs, we found a generalized asymmetry effect across generics, various kinds of explicitly quantified statements ('most', 'some', 'typically', 'usually'), and variations in types of predicated properties (striking vs. neutral). We discuss implications of these results for our understanding of the source of asymmetry effects and whether and in which ways these effects might introduce biased beliefs into social networks.

Probabilistic models of language understanding are valuable tools for investigating human language use. However, they need to be hand-designed for a particular domain. In contrast, large language models (LLMs) are trained on text that spans a wide array of domains, but they lack the structure and interpretability of probabilistic models. In this paper, we use chain-of-thought prompts to introduce structures from probabilistic models into LLMs. We explore this approach in the case of metaphor understanding. Our chain-of-thought prompts lead language models to infer latent variables and reason about their relationships in order to choose appropriate paraphrases for metaphors. The latent variables and relationships chosen are informed by theories of metaphor understanding from cognitive psychology. We apply these prompts to the two largest versions of GPT-3 and show that they can improve performance in a paraphrase selection task.

Both network (graph) and vector space are two representational forms commonly used in modeling knowledge structures and processes, but their relationship has not been extensively explored. In this paper, we conceptualize and formally show that these two types of models can be related in terms of both structure and process. In particular, the traditional and `higher order' cosine similarity in the vector space is mathematically equivalent to intersecting activation spread on a network after traversing direct and indirect paths. Inspired by this equivalence, we transfer graphical techniques to vector space modeling and demonstrate that the `higher-order' information embedded in the vector space can be used to create more powerful representations and accelerate learning in neural networks. Our result may have profound implications for both cognitive representational theory and machine learning practice.

In recent years, a number of methodological approaches has been developed that allow for identifying patterns in first language acquisition. In most cases, such methods have been used to analyze monolingual acquisition. In a global perspective, however, monolingual acquisition is the exception, and multilingualism is the norm. This raises the question to what extent such pattern-detection methods can be used for multilingual data as well, especially in the case of code-mixing, i.e. the use of more than one language in a single utterance. In this paper, we explore the potential of a data-driven approach to child language data proposed by McCauley & Christiansen (2019a), viz. chunk-based learning, to account for multilingual acquisition data. In particular, we compare the patterns found in the children’s code-mixed utterances with those in their monolingual utterances to test the hypothesis that both types of utterances are ultimately constructed from the same “building blocks”.

The present paper reports an experiment (N = 348) with a two-year-delayed (M = 695 days) follow-up that tests an approach to raising willingness to take climate actions. Here we focus on the longitudinal results. Our experimental materials were designed to harness the power of two cognitive constraints — coherence and causal invariance, which map onto two narrative proclivities that anthropologists have identified as universal — to promote climate action across the political spectrum. Towards that goal, the essential role of these constraints in belief formation predicts that climate-change information would be more persuasive when it is embedded in a personal climate-action narrative, the evocation of which can benefit from exposure to parsimonious scientific explanations of indisputable everyday observations, juxtaposed with reasoners’ own, typically less coherent explanations, occurring in a context that engages their moral stance. Our brief one-time intervention, conducted in U.S. states with the highest level of climate skepticism, showed that across the political spectrum, our materials raised willingness to take climate actions in the immediate assessment. It also raised the likelihood of reports two years later of having taken or would have taken those actions had the opportunity existed, suggesting long-lasting effects. Our approach adopts the framework that conceptions of reality are representations, and adaptive solutions in that infinite space of representations require cognitive constraints to narrow the search.