Title:

OS23-10 Influence of CNN Layer Depth on Spiral Visual Illusions

Publication: ICAROB2025
Volume: 30
Pages: 649-652
ISSN: 2188-7829
DOI: 10.5954/ICAROB.2025.OS23-10
Author(s): Kenji Aoki, Makoto Sakamoto
Publication Date: February 13, 2025
Keywords: Convolutional Neural Network, CNN, Visual Illusion, Spiral Illusion
Abstract: Understanding the mechanism of visual illusion generation through Convolutional Neural Networks (CNNs) that mimic the receptive fields of the visual cortex can contribute to elucidating the mechanisms of visual information processing in the brain. In our previous research, we demonstrated the potential for Fraser's spiral illusion to manifest in CNNs. In this study, we focused on the depth of the CNN layer structure and examined the impact of the number of layers on the manifestation of the visual illusion. We provided 14 types of spiral illusion images to three different CNN patterns with varying layer structures and tasked them with distinguishing between concentric circles and spirals. The results indicated that CNNs with fewer layers were more prone to the illusion, whereas CNNs with more layers were less likely to exhibit the illusion. These results suggest that the number of layers in a CNN influences the manifestation of visual illusions.
PDF File: https://alife-robotics.co.jp/members2025/icarob/data/html/data/OS/OS23/OS23-10.pdf
Copyright: © The authors.
This article is distributed under the terms of the Creative Commons Attribution License 4.0, which permits non-commercial use, distribution and reproduction in any medium, provided the original work is properly cited.
See for details: https://creativecommons.org/licenses/by-nc/4.0/

ALife Robotics Corporation Ltd.

HOME

 

 

(c)2008 Copyright The Regents of ALife Robotics Corporation Ltd. All Rights Reserved.