Perspective-Taking in Conversation

想象一下，你坐在飞机上，在寒暄之后，你的邻座告诉你他们是飞行员。在这个时候，你可能会问这个人开飞机是什么感觉，或者他们是在哪里学会飞行的。但是，如果你的邻座告诉你她是全职妈妈，你们的对话将会进行到哪里呢？在这种情况下，你可以问她有几个孩子，他们的年龄等等。也就是说，我们可以以一种知情和有原则的方式“即时”改变谈话的进程，对其他人可能知道和相信的事情进行推断和计算。然而，我们对引导互动对话的认知机制知之甚少。迄今为止，大多数关于对话互动的文献都集中在对话伙伴知道他们共享的信息上，也称为“共同点”。但是，我们如何发现我们与对话伙伴共享的信息呢？我们怎么知道哪些信息是我们可能不会分享的？在这个项目中，研究者使用行为实验来确定人们如何形成详细的、概率的表征，即他们的对话伙伴可能知道什么，而不是他们自己知道什么。研究结果将深入了解这些表征在会话互动中是如何使用的，以及它们如何影响会话记忆。本研究探讨了人们参与日常对话的认知机制，通过探索对话伙伴在某种情况下代表自己的观点，形成对对方观点的代表，并将两者进行比较，以确定差异和相似之处。提出的12个实验测试了这种观点比较假设的预测，并将为对话伙伴如何提问和回答问题以及对话记忆如何形成提供见解。这项工作的更广泛影响包括对研究生和大批本科生的研究训练。此外，几名高中生将从头到尾参与研究过程。在当地和全国范围内招收来自代表性不足和少数群体的学生。去识别数据和代码将与研究预注册一起公开提供。研究结果将促进对心理和语言过程的理解，促进计算机对话系统的设计，并为教学研究提供见解。该奖项反映了美国国家科学基金会的法定使命，并通过使用基金会的知识价值和更广泛的影响审查标准进行评估，被认为值得支持。

Imagine taking your seat in an airplane and after exchanging pleasantries, your seatmate tells you that they are a pilot. At this point, you might ask the person something about what it's like to fly a plane or perhaps about where they learned to fly. But where would that conversation have gone if your seatmate instead told you she is a stay-at-home mom? In that case, you might ask how many children she has, their ages, etc. That is, we can change the course of a conversation “on-the-fly” in an informed and principled way, making inferences and calculations about what other people might know and believe. Yet we know little about the cognitive mechanisms that guide interactive conversation. To date, most of the literature on conversational interaction focuses on the information that the conversational partners know they share, also called the "common ground." But how do we discover what information we share with a conversational partner? And how do we figure out what information we probably don't share? In this project, the investigator uses behavioral experiments to determine how people form detailed, probabilistic representations of what their conversational partner might know, as compared with what they themselves know. Findings will provide insight into how these representations are used during conversational interactions and how they influence memory for conversation. This research probes the cognitive mechanisms that allow people to engage in everyday conversation by exploring the idea that conversational partners represent their own perspective in a situation, form a representation of their partner's perspective, and compare the two, in order to determine the differences and similarities. Twelve proposed experiments test predictions of this perspective-comparison hypothesis and will provide insights into how conversational partners ask and answer questions and how memory for conversation is shaped. Broader impacts of the work include research training for graduate students and a large cohort of undergraduates. In addition, several high school students will participate in the research process from start to finish. Students from under-represented and minoritized groups are recruited locally and nationally. De-identified data and code will be made publicly available along with study pre-registrations. Findings will advance understanding of psychological and linguistic processes, advance the design of computerized dialog systems, and offer insights into pedagogical research.This award reflects NSF's statutory mission and has been deemed worthy of support through evaluation using the Foundation's intellectual merit and broader impacts review criteria.