The role of statistical summary representations in visual learning

直觉告诉我们，大脑应该用代表单个物品的方式来代表一组物品。与这种直觉相反，研究人员发现，信息集是根据它们的统计特性来表示的，比如它们的平均大小或颜色的可变性。在过去的15年中，研究人员已经确立了统计摘要表征（SSRs）在感知和注意中的重要作用。这个研究项目的目的是确定ssr在视觉学习中是否也有作用，如果有的话，了解促进这种学习的条件。为了解决这些问题，我将开展旨在探索ssr在三个日常任务中对学习的贡献的实验：分类、搜索和学习识别面孔。这些发现有可能推进这些过程的模型，这些模型目前还没有纳入SSRs的作用。同样重要的是，它们有可能为医疗决策、人工导航系统和面部识别软件的进步提供信息。

Intuition would suggest that the brain should represent sets of items in the same way it represents individual items. Contrary to this intuition, researchers have discovered that sets of information are represented according to their statistical properties, such as their average size or variability in colour. Over the past 15 years, researchers have established important roles for such statistical summary representations (SSRs) in perception and attention. The purpose of this program of research is to determine whether there might also be a role for SSRs in visual learning, and, if so, to understand the conditions that facilitate such learning. To address these questions, I will carry out experiments designed to explore the contribution of SSRs to learning in three everyday tasks: categorization, search, and learning to recognize faces. The findings have the potential to advance models of these processes, which currently do not incorporate a role for SSRs. Of equal importance, they have the potential to inform advances in medical decision making, artificial navigation systems, and face recognition software.