A comparison of statistical learning of naturalistic textures between DCNNs and the human visual hierarchy

doi:10.1007/s11431-024-2748-3

PSYCH OpenIR

	A comparison of statistical learning of naturalistic textures between DCNNs and the human visual hierarchy
	Lu, XinCheng 1; Yuan, ZiQi 3; Zhang, YiChi 3; Ai, HaiLin 1; Cheng, SiYuan 1; Ge, YiRan 1; Fang, Fang 4,5,6,7,8; Chen, NiHong 1,2,9
通讯作者	Fang, Fang(ffang@pku.edu.cn) ; Chen, NiHong(nihongch@tsinghua.edu.cn)
摘要	The visual system continuously adapts to the statistical properties of the environment. Existing evidence shows a close resemblance between deep convolutional neural networks (CNNs) and primate visual stream in neural selectivity to naturalistic textures above the primary visual processing stage. This study delves into the mechanisms of perceptual learning in CNNs, focusing on how they assimilate the high-order statistics of natural textures. Our results show that a CNN model achieves a similar performance improvement as humans, as manifested in the learning pattern across different types of high-order image statistics. While L2 was the first stage exhibiting texture selectivity, we found that stages beyond L2 were critically involved in learning. The significant contribution of L4 to learning was manifested both in the modulations of texture-selective responses and in the consequences of training with frozen connection weights. Our findings highlight learning-dependent plasticity in the mid-to-high-level areas of the visual hierarchy. This research introduces an AI-inspired approach for studying learning-induced cortical plasticity, utilizing DCNNs as an experimental framework to formulate testable predictions for empirical brain studies.
关键词	CNN perceptual learning naturalistic texture psychophysics
	2024-08-01
语种	英语
DOI	10.1007/s11431-024-2748-3
发表期刊	SCIENCE CHINA-TECHNOLOGICAL SCIENCES
ISSN	1674-7321
卷号	67 期号:8 页码:2310-2318
收录类别	SCI
资助项目	National Natural Science Foundation of China[31971031] ; National Natural Science Foundation of China[31930053] ; National Natural Science Foundation of China[32171039] ; STI2030-Major Projects[2021ZD0203600] ; STI2030-Major Projects[2022ZD0204802] ; STI2030-Major Projects[2022ZD0204804]
出版者	SCIENCE PRESS
WOS关键词	IMAGE STATISTICS ; REPRESENTATIONS
WOS研究方向	Engineering ; Materials Science
WOS类目	Engineering, Multidisciplinary ; Materials Science, Multidisciplinary
WOS记录号	WOS:001282858100010
资助机构	National Natural Science Foundation of China ; STI2030-Major Projects
引用统计
文献类型	期刊论文
条目标识符	https://ir.psych.ac.cn/handle/311026/49252
通讯作者	Fang, Fang; Chen, NiHong
作者单位	1.Tsinghua Univ, Dept Psychol & Cognit Sci, Beijing 100084, Peoples R China 2.Tsinghua Univ, McGovern Inst Brain Res, IDG, Beijing 100084, Peoples R China 3.Tsinghua Univ, Dept Comp Sci & Technol, Beijing 100084, Peoples R China 4.Peking Univ, Sch Psychol & Cognit Sci, Beijing 100871, Peoples R China 5.Peking Univ, Beijing Key Lab Behav & Mental Hlth, Beijing 100871, Peoples R China 6.Peking Univ, Key Lab Machine Percept, Minist Educ, Beijing 100871, Peoples R China 7.Peking Univ, Peking Tsinghua Ctr Life Sci, Beijing 100871, Peoples R China 8.Peking Univ, McGovern Inst Brain Res, IDG, Beijing 100871, Peoples R China 9.Chinese Acad Sci, Inst Psychol, State Key Lab Brain & Cognit Sci, Beijing 100101, Peoples R China
推荐引用方式 GB/T 7714	Lu, XinCheng,Yuan, ZiQi,Zhang, YiChi,et al. A comparison of statistical learning of naturalistic textures between DCNNs and the human visual hierarchy[J]. SCIENCE CHINA-TECHNOLOGICAL SCIENCES,2024,67(8):2310-2318.
APA	Lu, XinCheng.,Yuan, ZiQi.,Zhang, YiChi.,Ai, HaiLin.,Cheng, SiYuan.,...&Chen, NiHong.(2024).A comparison of statistical learning of naturalistic textures between DCNNs and the human visual hierarchy.SCIENCE CHINA-TECHNOLOGICAL SCIENCES,67(8),2310-2318.
MLA	Lu, XinCheng,et al."A comparison of statistical learning of naturalistic textures between DCNNs and the human visual hierarchy".SCIENCE CHINA-TECHNOLOGICAL SCIENCES 67.8(2024):2310-2318.