Collaborative Research: Knowledge transmission through iterated learning

人类的许多知识不是从我们周围的物质世界学到的，而是从其他人那里学到的。例如，每一代新物理学家都不必重新发现牛顿力学定律；这些定律从一个物理学家传到另一个物理学家。同样的“迭代学习”也发生在语言、音乐和其他不容易从头开始学习的复杂知识上。尽管迭代学习在人类文化和社会中无处不在，但人们对知识在人与人之间传播时发生变化的认知过程知之甚少。在美国国家科学基金会（National Science Foundation）的支持下，卡利什博士和格里菲斯博士将通过测试和扩展将文化进化的数学与人类认知过程统一起来的理论，来研究这个问题。他们将进行一系列实验，旨在在实验室环境中重现迭代学习的过程，并通过开发计算模型来补充这些实验，这些模型旨在根据观察到的学习数据来测试理论预测。主要问题是，在传播过程中可能发生的相对较小的知识扰动是否会在许多迭代过程中累积起来产生大的影响。这些实验的结果有望为人类认知理论以及知识的文化传播理论提供一些重要而普遍的约束。

Much of human knowledge is not learned from the physical world around us, but from other people. For example, each new generation of physicists does not have to rediscover the laws of Newtonian mechanics; these laws are passed on from one physicist to the next. This same kind of "iterated learning" occurs for language, music, and other kinds of complex knowledge that are not easily relearned from scratch, so to speak. Despite the ubiquity of iterated learning in human culture and society, very little is known about the cognitive processes by which knowledge changes when it is transmitted from one person to the next. With support from the National Science Foundation, Dr. Kalish and Dr. Griffiths will investigate this issue by testing and extending theories that unify the mathematics of cultural evolution with the processes of human cognition. They will conduct a series of experiments that are designed to reproduce the processes of iterated learning in a laboratory environment, and the experiments will be complemented by the development of computational models designed to test the theoretical predictions against the observed learning data. The main issue is whether relatively small perturbations in knowledge that may occur in the act of transmission will or will not add up to have large effects over the course of many iterations. The results of these experiments promise to provide us with some important and general constraints on theories of human cognition, as well as theories of cultural transmission of knowledge.