Search Papers | Poster Sessions | All Posters
Poster A111 in Poster Session A - Tuesday, August 6, 2024, 4:15 – 6:15 pm, Johnson Ice Rink
Novel binary categorization task to study high-level visual object classification in macaque monkeys
Han Zhang1,2, Zhihao Zheng1, Jiaqi Hu1,2, Qiao Wang1, Zixuan Li1,2, Mengya Xu1, Gouki Okazawa1,2 (); 1Institute of Neuroscience, Center for Excellence in Brain Science and Intelligence Technology, Chinese Academy of Sciences, China, 2University of Chinese Academy of Sciences, China
Humans associate visually perceived objects with abstract concepts, such as animate, mammalian, or artificial. This ability seems to depend on human language, but previous studies have reported that neural representations in higher visual areas of macaque monkeys also reflect some of these concepts. But can macaque monkeys categorize objects at this abstract level? Here, we developed a novel binary categorization task and found that monkeys quickly learned to classify images of natural objects based on abstract concepts, including animate versus inanimate, natural versus artificial, and mammalian versus non-mammalian. They generalized the learned rule to new images and made errors consistent with human classification. Since their choices could be well fit by artificial neural networks, we interpret that they could solve the tasks by extracting higher-order visual features. Our behavioral paradigm is well suited to study the capacity of macaque monkeys to visually categorize objects using various rules and stimulus sets.
Keywords: monkey behavior object recognition perceptual decision making abstract concepts