Brooks Daniel I, Ng Ka H, Buss Eric W, Marshall Andrew T, Freeman John H, Wasserman Edward A
Department of Neuroscience, Brown University, Providence, RI 02912, USA.
J Exp Psychol Anim Behav Process. 2013 Jan;39(1):85-92. doi: 10.1037/a0030404.
Strong interest exists in developing a rodent model of visual cognition to conduct research into the neural mechanisms of visual categorization. Yet, doubt remains as to whether rats perform visual categorization tasks as do humans and nonhuman primates. Here, we trained eight rats on two visual categorization tasks using photographs of eight objects from each of four basic-level categories: chairs, flowers, cars, and humans. In Experiment 1, rats learned to categorize chairs versus flowers; in Experiment 2, rats learned to categorize cars versus humans. After rats learned each discrimination, we tested them with eight novel pictures from each of the categories. The rats performed at reliably above-chance levels during these generalization tests. To determine which dimension(s) of the stimuli controlled the rats' behavior, we conducted regression analyses using several image dimensions. The chair versus flower discrimination was mainly controlled by the convexity of the stimuli, whereas the car versus human discrimination was mainly controlled by the aspect ratio of the stimuli. These results demonstrate that rats can categorize complex visual objects using shape-based properties of photographs.
人们对开发一种视觉认知的啮齿动物模型以研究视觉分类的神经机制有着浓厚的兴趣。然而,对于大鼠是否能像人类和非人类灵长类动物那样执行视觉分类任务,仍存在疑问。在此,我们使用来自四个基本类别(椅子、花朵、汽车和人类)中每个类别的八个物体的照片,对八只大鼠进行了两项视觉分类任务的训练。在实验1中,大鼠学习对椅子和花朵进行分类;在实验2中,大鼠学习对汽车和人类进行分类。在大鼠学会每种辨别后,我们用来自每个类别的八张新图片对它们进行测试。在这些泛化测试中,大鼠的表现可靠地高于随机水平。为了确定刺激的哪些维度控制了大鼠的行为,我们使用几个图像维度进行了回归分析。椅子与花朵的辨别主要由刺激的凸度控制,而汽车与人类的辨别主要由刺激的长宽比控制。这些结果表明,大鼠可以利用照片基于形状的属性对复杂视觉对象进行分类。