In recent years, a multitude of datasets of human–human conversations has been released for the main purpose of training conversational agents based on data‐hungry artificial neural networks. In this paper, we argue that datasets of this sort represent a useful and underexplored source to validate, complement, and enhance cognitive studies on human behavior and language use. We present a method that leverages the recent development of powerful computational models to obtain the fine‐grained annotation required to apply metrics and techniques from (...) Cognitive Science to large datasets. Previous work in Cognitive Science has investigated the question‐asking strategies of human participants by employing different variants of the so‐called 20‐question‐game setting and proposing several evaluation methods. In our work, we focus on GuessWhat, a task proposed within the Computer Vision and Natural Language Processing communities that is similar in structure to the 20‐question‐game setting. Crucially, the GuessWhat dataset contains tens of thousands of dialogues based on real‐world images, making it a suitable setting to investigate the question‐asking strategies of human players on a large scale and in a natural setting. Our results demonstrate the effectiveness of computational tools to automatically code how the hypothesis space changes throughout the dialogue in complex visual scenes. On the one hand, we confirm findings from previous work on smaller and more controlled settings. On the other hand, our analyses allow us to highlight the presence of “uninformative” questions (in terms of Expected Information Gain) at specific rounds of the dialogue. We hypothesize that these questions fulfill pragmatic constraints that are exploited by human players to solve visual tasks in complex scenes successfully. Our work illustrates a method that brings together efforts and findings from different disciplines to gain a better understanding of human question‐asking strategies on large‐scale datasets, while at the same time posing new questions about the development of conversational systems. (shrink)

In this paper, we investigate the informativeness of 4- to 6-year-old children’s questions using a combined qualitative and quantitative approach. Children were presented with a hierarchical version of the 20-questions game, in which they were given an array of objects that could be organized into three category levels based on shared features. We then tested whether it is possible to scaffold children’s question-asking abilities without extensive training. In particular, we supported children’s categorization performance by providing the object-related features needed to (...) ask effective constraint-seeking questions. We found that with both age and scaffolding children asked more effective questions, targeting higher category levels and therefore reaching the solution with fewer questions. We discuss the practical and theoretical implications of these results. (shrink)

One way to learn about the world is by asking questions. We investigate how younger children (7- to 8-year-olds), older children (9- to 11-year-olds), and young adults (17- to 18-year-olds) ask questions to identify the cause of an event. We find a developmental shift in children’s reliance on hypothesis-scanning questions (which test hypotheses directly) versus constraint-seeking questions (which reduce the space of hypotheses), but also that all age groups ask more constraint-seeking questions when hypothesis-scanning questions are least likely to pay (...) off: When the solution is one among equally likely alternatives (Study 1) or when the problem is difficult (Studies 1 and 2). These findings are the first to demonstrate that even young children dynamically adapt their strategies for inquiry to increase the efficiency of information search. (shrink)

Children ask many questions, but do not always receive answers to the questions they ask. We were interested in whether the act of generating questions, in the absence of an answer, is related to children’s later thinking. Two experiments examined whether children retain the questions they ask in working memory, and whether the type of questions asked relate to their categorization. Four to ten-year-old children were shown 12 novel objects, asked three questions about each, and did not receive answers to (...) their questions. Children recalled their questions in the first experiment and categorized variants of the novel objects in the second experiment. We found that children have robust working memory for their questions, indicating that these questions may relate to their subsequent thinking. Additionally, children generalize category boundaries more narrowly or broadly depending on the type of question they ask, indicating that children’s questions may reflect an underlying bias in how they think about the world. These findings suggest that future research should examine questions in the absence of answers to understand how inquiry affects children’s cognitive development. (shrink)