A Computational Model of Afterimage Rotation in the Peripheral Drift Illusion Based on Retinal ON/OFF Responses

Yuichiro Hayashi, Shin Ishii, Hidetoshi UrakuboPLoS 2014 · DOI

Abstract

Human observers perceive illusory rotations after the disappearance of circularly repeating patches containing dark-to-light luminance. This afterimage rotation is a very powerful phenomenon, but little is known about the mechanisms underlying it. Here, we use a computational model to show that the afterimage rotation can be explained by a combination of fast light adaptation and the physiological architecture of the early visual system, consisting of ON- and OFF-type visual pathways. In this retinal ON/OFF model, the afterimage rotation appeared as a rotation of focus lines of retinal ON/OFF responses. Focus lines rotated clockwise on a light background, but counterclockwise on a dark background. These findings were consistent with the results of psychophysical experiments, which were also performed by us. Additionally, the velocity of the afterimage rotation was comparable with that observed in our psychophysical experiments. These results suggest that the early visual system (including the retina) is responsible for the generation of the afterimage rotation, and that this illusory rotation may be systematically misinterpreted by our high-level visual system.

Introduction

The Peripheral Drift Illusion (PDI) is a famous visual phenomenon where static images—most notably the "Rotating Snakes" or the sawtooth-patterned Fraser-Wilcox stimulus—appear to move. While much research focuses on why these images move during active viewing, a particularly powerful version occurs after the image is removed: the afterimage rotation. Until now, simple adaptation models couldn't explain why the afterimage appears to rotate, nor why the direction of rotation flips depending on whether the background is light or dark. This study moves beyond purely descriptive accounts to a mechanistic computational model that aligns with human biology.

Methods: The Retinal ON/OFF Model

The researchers developed a model based on two key physiological features of the early visual system:

  1. Biphasic Temporal Impulse Response (TIR): A mathematical representation of how retinal neurons "overshoot" and "undershoot" as they adapt to changes in light.
  2. ON and OFF Pathways: The division of visual signals into two channels—one for light increments (ON) and one for light decrements (OFF).
Figure 6: Schematic of the retinal ON/OFF model showing temporal responses to light intensity changes and their projection to the high-level visual system.

The model simulates a grid of these ON and OFF units. When a stimulus disappears, the model tracks "focus lines"—spatial locations where the ON and OFF responses are balanced. These model outputs were compared to data from psychophysical experiments involving human observers who judged the direction and velocity of illusory rotation against various background luminances.

Key Insights and Results

The Crossover Effect and Background Reference

The core mechanism identified is that the time a neuron takes to recover from adaptation depends on the intensity of the preceding light. Because the Fraser-Wilcox stimulus has a luminance gradient, different parts of the retina recover at different speeds. This creates a "crossover" point between ON and OFF signals that shifts across the retina over time, which the brain interprets as motion.

The background luminance sets the "steady state" for the retina. The model demonstrated that a light background causes focus lines to rotate clockwise, while a dark background flips the recovery dynamics to counterclockwise—perfectly matching human reports. To achieve this match, the model required a logarithmic input stage, suggesting that the illusion is tied to how our eyes compress physical light intensities (Weber’s Law).

Quantitative Consistency

The model successfully reproduced the background-dependent flip in rotation direction. Human observers reached a "balanced" state (50/50 chance of seeing CW or CCW) at a relative background luminance of ~0.22 to 0.41, which the model matched once logarithmic scaling was applied. Furthermore, human observers perceived the afterimage rotating at roughly 118°/s to 148°/s. By adjusting the model's time constants to reflect fast adaptation (approx. 200ms), the model produced a rotation velocity of 125°/s, showing strong quantitative consistency with human biology.

Discussion and Limitations

These results suggest that the early visual system, including the retina, is responsible for generating the signals that lead to afterimage rotation. However, the current model struggles with sharp-edged stimuli (like the Kitaoka-Ashida stimulus) unless the image is blurred, suggesting that edge-detection or "perceptual filling-in" mechanisms are also involved. Additionally, while the model explains how the motion signal is generated, it does not explicitly model the cortical "decoder" that turns these focus lines into a conscious perception of spinning. Future work may explore the discrepancy between the model's required time constants and those found in some physiological studies, potentially due to the complexity of sustained image adaptation versus single light pulses.

Summary updated

Abstract

Human observers perceive illusory rotations after the disappearance of circularly repeating patches containing dark-to-light luminance. This afterimage rotation is a very powerful phenomenon, but little is known about the mechanisms underlying it. Here, we use a computational model to show that the afterimage rotation can be explained by a combination of fast light adaptation and the physiological architecture of the early visual system, consisting of ON- and OFF-type visual pathways. In this retinal ON/OFF model, the afterimage rotation appeared as a rotation of focus lines of retinal ON/OFF responses. Focus lines rotated clockwise on a light background, but counterclockwise on a dark background. These findings were consistent with the results of psychophysical experiments, which were also performed by us. Additionally, the velocity of the afterimage rotation was comparable with that observed in our psychophysical experiments. These results suggest that the early visual system (including the retina) is responsible for the generation of the afterimage rotation, and that this illusory rotation may be systematically misinterpreted by our high-level visual system.
BibTeX
@inproceedings{hayashi2014computational,
  author = {Yuichiro Hayashi and Shin Ishii and Hidetoshi Urakubo},
  title = {{A Computational Model of Afterimage Rotation in the Peripheral Drift Illusion Based on Retinal ON/OFF Responses}},
  booktitle = {PLoS},
  year = {2014},
  doi = {10.1371/journal.pone.0115464},
}
Figure