Search

Children’s early word learning is to a large extent driven by the prevalence of words in their language environment, with words that are spoken more often to children being learned earlier. However, children receive language from a variety of sources, including books, television, and movies meant for children, as well as speech and media that is meant for adults, but over- heard by children. Despite considerable similarity of word frequency distributions from these different input sources, there is also significant and predictable variability between them. For example, function words are far more frequent in books than in everyday speech, while early-learned nouns (e.g., ‘ball’ and ‘mommy’) are more frequent in child-directed speech than in other sources. Children receive a mixture of these different frequency distributions. The goal of this paper is to better understand the shared and unique variance in these input sources – in both English and French – and to evaluate how predictive these distributions are of children’s early word learning.

Human learning and generalization benefit from bootstrapping: we arrive at complex concepts by starting small and building upon past successes. In this paper, we examine a computational account of causal conceptual bootstrapping, and describe a novel experiment in which the sequence of training data results in a dramatic order effect: participants succeed in identifying a compound concept only after experiencing training data in a “helpful” order. Our computational model represents causal relations as reusable, modular programs, which can themselves be “chunked” and flexibly reused to tackle more complex tasks. Our specific approach is based in combinatory logic and adaptor grammars, building on previous theories that posit a “language of thought” for concept representation, but making the learning process more sensitive to a learner’s experiences than any particular choice of conceptual primitives. Crucially, we demonstrate that a caching mechanism like that used in adaptor grammars is key to explain human-like bootstrapping patterns in causal generalization.

Children do not learn language from passively analyzing correlations between language and observations, but from interaction with caregivers or peers. The non-nativist approach claims that the main driver of language learning should be to achieve communicative goals. Imitation, on the other hand, is another natural desire that many argue influences language learning. However, there are still gaps in the research on what roles communicative goals and imitating linguistic input play in language acquisition, due to the difficulty of performing comprehensive experiments with human learners. In this paper, we propose a computational framework using simulated experiments that allows us to compare the roles of the two drivers. Specifically, we simulate a two-way communication game between a speaker, corresponding to a language learner, and a listener, corresponding to a caregiver or teacher. The speaker's communicative goals are modeled as rewards for successful completion of a referential game, and imitation is performed by mimicking feedback from the listener. The listener adaptively chooses to give feedback and makes choices based on the speaker's utterances. With empirical results on naturalistic visual and language data, we find that communicative goals play an important role in driving language learning, whereas imitation accelerates the learning process. We also find that (1) models trained with communicative goals tend to use minimal vocabulary and utterances and overextend them to concepts outside the original word meanings; (2) the strategy with which the listener provides feedback also influences the learning results and speed. Code and data for replicating the experiments are available\footnote{\url{https://bit.ly/interactgym}} to spur future research on models for computational studies of language learning.

Burns & Krygier (2015) demonstrated that people could exhibit a strong bias towards the smaller first digits, in a way similar to that described by Benford’s law. This paper sought to expand the scope of this phenomenon and to test a possible explanation, the Recognition Hypothesis that a Benford bias is due to life-long environmental exposure to this statistical relationship. Participants completed three numerical tasks: A Generation Task requiring answering trivia questions; a Selection Task requiring selecting between two numerical responses; and an Estimation task requiring estimating the number of jelly beans in a jar. The results found no evidence of any first digit effect in the Recognition Task, some evidence of Benford bias in the Generation Task and strong evidence in the Estimation Task. Future research should focus on alternatives to the Recognition Hypothesis and investigate the parameters of Benford bias in generation tasks

As algorithmic tools increasingly aid experts in making consequential decisions, the need to understand the precise factors that mediate their influence has grown commensurately. In this paper, we present a crowdsourcing vignette study designed to assess the impacts of two plausible factors on AI-informed decision-making. First, we examine homophily---do people defer more to models that tend to agree with them?---by manipulating the agreement during training between participants and the algorithmic tool. Second, we considered incentives---how do people incorporate a (known) cost structure in the hybrid decision-making setting?---by varying rewards associated with true positives vs. true negatives. Surprisingly, we found limited influence of either homophily and no evidence of incentive effects, despite participants performing similarly to previous studies. Higher levels of agreement between the participant and the AI tool yielded more confident predictions, but only when outcome feedback was absent. These results highlight the complexity of characterizing human-algorithm interactions, and suggest that findings from social psychology may require re-examination when humans interact with algorithms.

A central question for language development is whether very early telegraphic speech, ungrammatical but intelligible, reflects adult-like abstract linguistic competence, or instead is the result of statistical learning without the reliance on the abstract knowledge. In this paper, we develop a simulation paradigm to evaluate the predictive fit of different accounts of three-word utterances children (<20 months) produced. The simulation subjects are a) human adults with full linguistic competence, b) a statistical model that captures local statistical regularities of language input, and c) a deep neural model that learns and processes language input with global context-aware learning mechanisms. The statistical language model predicted child three-word utterances, both grammatical and ungrammatical ones, better than the neural models and even better (but not statistically) than human adults. The findings suggest simple local statistical regularities underlying child early telegraphic speech.

In an experimental study, we investigated how social interaction dynamics affect collective creativity. Pairs of participants collaborated in a computer game, creating “beautiful and interesting” shapes by moving tiles on a large touchscreen. We identified naturally emerging interaction styles by applying k-means clustering on participants’ tile moves. The game allowed us to quantify the unfolding creative process in a well-defined search space. Pairs characterized by a single dominating member tended to visit fewer areas of the solution space, stay there longer and created on average more (but less original) shapes. In contrast, pairs that took turns with every tile move tended to explore more, stayed in each area of the solution space for less time and created fewer (but more original) shapes. While previous literature found conflicting effects of ‘creating with another’, the current paper suggests naturally emerging interaction styles as a differentiating factor underlying how collective creativity unfolds.

The paper aims to examine participants’ open-text ‘think aloud’ explanations of their reasoning while making a judgement about an ambiguous scenario. It aims to consider this data in light of frameworks such as causal modelling, intuitive theories, coherence and the story model. Consistent with these frameworks, we find that participants bring in a large amount of world knowledge to connect ambiguous evidence to unobserved, inferred variables and, via these, to the target judgement. We attempt to represent these chains of inferences using causal diagrams and find that participants interpretations of the scenario can be lumped into one of two distinct causal models, each presenting an internally coherent ‘image’ of the ambiguous scenario. Furthermore, participants’ judgement predicts which of those two models they adhere to. We discuss the limitations and merits of this methodological approach for investigating these types of frameworks.

The "myside bias'' in evaluating arguments is an empirically well-confirmed phenomenon that consists of overweighting arguments that endorse one's beliefs or attack alternative beliefs while underweighting arguments that attack one's beliefs or defend alternative beliefs. This paper makes two contributions: First, it proposes a probabilistic model that adequately captures three salient features of myside bias in argument evaluation. Second, it provides a Bayesian justification of this model, thus showing that myside bias has a rational Bayesian explanation under certain conditions.

This paper describes some epistemic cultural considerations which shape the uses of psychology. I argue the study of mind is bound by the metaphysical background of the given locale and era in which it is practiced. The epistemic setting in which psychology takes place will shape what is worth observing, how it is to be studied, how the data is to be interpreted, and the nature of the ultimate explanatory units. I argue epistemic constraints shape the praxes that arise from structural study of the mind. In order to illustrate this notion of epistemic cultural constraint, I discuss Soviet Psychology and provide a contrast between practical uses of psychoanalysis in India, Egypt, and rural Ghana. In response to these conceptual and practical epistemic limitations, psychology could adapt methods drawn from history and anthropology towards an interdisciplinary psychology.

Rules undoubtedly guide our moral lives. Simple moral rules prohibit lying, cheating, and stealing, for instance. But the moral mind is more flexible than a theory based only on rule-adherence can account for. In this paper, we look at one particular kind of flexibility: the ability to figure out when it is okay to break a moral rule. We elicit judgments of the moral acceptability of breaking a simple rule: it's wrong to cut in line. We created a video game environment in which agents attempt to gather water -- and sometimes must stand in line behind others to do so. Subjects watch clips of the game being played and make judgments of the moral acceptability of cutting in line across a wide range of spatio-temporally varied and dynamic scenarios. Our data suggests that subjects make judgments by using a generative understanding of the underlying function of the rule about waiting in line. We further show that our data cannot be accounted for by either 1) simple rule adherence or 2) utility maximization.

Geometric analogy problems remain an intriguing part of intelligence scales, which is closely correlated to many cognitive studies, such as perception, conception, memory, abstract and inductive reasoning. The problems not only target the most fundamental element --- analogy-making --- in human cognition, but also require integration of multiple components and stages: looking at the test booklet, thinking for a minute or two, and deciding the answer. Great efforts and achievements have been made to explain different individual aspects of this process. In this paper, we take a more holistic approach from the perspective of problem-solving, by modeling the entire process, from the moment the visual stimuli are received to the moment an answer is decided. Therefore, we explore how the final solution can be built upon visual inputs and necessary components that lie between the perceptual input and conceptual output. Particularly, we designed a novel similarity metric and a correspondence-finding method based on mapping and optimization. With these two basic blocks, we implemented a computational model, and report our initial results on a classical problem set.

Several popular sequence-based and pretrained language models have been found to be successful for text-driven prediction of brain activations. However, these models still lack long-term cognitive plausibility as well as insights on the underlying neural substrate mechanisms. This paper studies the influence of context representations of different language models such as sequence-based models: Long short-term memory networks (LSTMs) and ELMo, and popular pretrained Transformer language model (Longformer). In particular, we study how the internal hidden representations align with the brain activity observed via fMRI when the subjects listen to several narrative stories. We use brain imaging recordings of subjects listening to narrative stories to interpret word and sequence embeddings. We further investigate how the representations of language models layers reveal better semantic context during listening. Experiments across all language model representations provide the following cognitive insights: (i) the representations of LSTM cell states are better aligned with brain recordings than LSTM (hidden state), the cell state activity can represent more long-term information, (ii) the representations of ELMo and Longformer display a good predictive performance across brain regions for listening stimuli; (iii) Posterior Medial Cortex (PMC), Temporo-Parieto-Occipital junction (TPOJ), and Dorsal Frontal Lobe (DFL) have higher correlation versus Early Auditory (EAC) and Auditory Association Cortex (AAC).

In this paper, we have studied the impact of clickbait headlines on the distribution of visual attention on hyperlinked news articles. Visual attention is a driving factor in ad-based revenue models that support online journalism. Importantly, it is also an indicator of cognitive processes involved in reading and comprehension. We hypothesize that articles with clickbait headlines receive lesser visual attention when controlled for articles’ content. This is based on the premise that a significant proportion of clicks on clickbait headlines are driven by readers’ specific epistemic curiosity rather than knowledge acquisition. An eye-tracker setup was used to infer visual attention from the gaze-fixation analysis conducted on data from 60 participants. Our results suggest that clickbait headlines significantly reduce the visual attention on news articles. Though, article content comprehension measured by a recall test was comparable for clickbait and non-clickbait headlines. Our findings add to the discussions on the cognitive attention and the implications of using clickbait headlines for news publishers, newsreaders, and advertising agencies alike.