Predictive coding feedback results in perceived illusory contours in a recurrent neural network. (December 2021)

• Record Type: Journal Article
• Authors: Pang, Zhaoyang; O'May, Callum Biggs; Choksi, Bhavin; VanRullen, Rufin
• Is Part Of: Neural networks. Volume 144(2021)
• Page Start: 164
• Page End: 175
• Publication Date: 2021-12
• Subjects: Illusory contours -- Predictive coding -- Deep learning -- Kanizsa squares -- Feedback -- Generative models; Neural computers -- Periodicals; Neural networks (Computer science) -- Periodicals; Neural networks (Neurobiology) -- Periodicals; Nervous System -- Periodicals; Ordinateurs neuronaux -- Périodiques; Réseaux neuronaux (Informatique) -- Périodiques; Réseaux neuronaux (Neurobiologie) -- Périodiques; Neural computers; Neural networks (Computer science); Neural networks (Neurobiology); Periodicals; 006.32
• DOI: 10.1016/j.neunet.2021.08.024
• Languages: English
• ISSNs: 0893-6080

Abstract

Abstract: Modern feedforward convolutional neural networks (CNNs) can now solve some computer vision tasks at super-human levels. However, these networks only roughly mimic human visual perception. One difference from human vision is that they do not appear to perceive illusory contours (e.g. Kanizsa squares) in the same way humans do. Physiological evidence from visual cortex suggests that the perception of illusory contours could involve feedback connections. Would recurrent feedback neural networks perceive illusory contours like humans? In this work we equip a deep feedforward convolutional network with brain-inspired recurrent dynamics. The network was first pretrained with an unsupervised reconstruction objective on a natural image dataset, to expose it to natural object contour statistics. Then, a classification decision head was added and the model was finetuned on a form discrimination task: squares vs. randomly oriented inducer shapes (no illusory contour). Finally, the model was tested with the unfamiliar "illusory contour" configuration: inducer shapes oriented to form an illusory square. Compared with feedforward baselines, the iterative "predictive coding" feedback resulted in more illusory contours being classified as physical squares. The perception of the illusory contour was measurable in the luminance profile of the image reconstructions produced by the model, demonstrating that the model really "sees" the illusion. Ablation studies revealed that natural image pretraining and feedback error correction are both critical to the perception of the illusion. Finally we validated our conclusions in a deeper network (VGG): adding the same predictive coding feedback dynamics again leads to the perception of illusory contours. Highlights: Recurrent networks with predictive coding dynamics perceive illusory shapes as humans. Equivalent feedforward convolutional networks do not perceive illusory shapes. The recurrent networks classify illusory shapes as real ones. The networks' generative pathway reconstructs non-existent illusory edges. Information from feedback connections is necessary for illusory perception.