Look at the image above. Looks like you’re falling into it, doesn’t it? Like the darkness is approaching, threatening to overtake you if you stare too long. Why is that?
A new study has an explanation – and it all comes down to how the individual cells in our eyes interact with each other.
The expanding hole illusion
It’s a classic of the optical illusion genre: “The Expanding Hole Illusion challenges traditional views of motion perception by demonstrating how static images can evoke strong sensations of movement,” explains the paper, currently available as a preprint and thus not yet peer-reviewed.
It’s a trick powerful enough to even elicit physical responses: stare at this fake-expanding hole, and your pupils will actually dilate as if to cope with the increase in darkness. But what’s going on inside our eyes and brain to cause all this?
The key appears to be in how our eyes work at a very basic level, the authors report. “Through a combination of previously reported psychophysical experiments and our work in bioderived modelling, we have shown that this illusion likely arises from contrast-dependent lateral interactions in early visual areas,” they write.
To understand what that means, we need to explain how our eyes figure out what we’re seeing. Specifically, we’re talking about the retinal ganglion cells, or RGCs – neurons which sit near the inner surface of the retina and are responsible for sending visual stimuli to the brain.
Now, these cells aren’t just all firing off equally: “retinal ganglion cells have receptive fields that have a very basic organization, which resembles two concentric circles,” explains the University of Minnesota’s Introduction to Sensation and Perception. “This concentric receptive field structure is usually known as center-surround organization.”
So, based on where a visual stimulus is in front of your eyes – where in these two concentric circles it falls – different RGCs are going to respond. And they do so in specific ways: “On-center retinal ganglion cells respond to light spots surrounded by dark backgrounds like a star in a dark sky,” the book explains, while “off-center retinal ganglion cells respond to dark spots surrounded by light backgrounds like a fly in a bright sky.”
However, “if light falls on the entire receptive field, and not just the center, the cell will not increase its firing rate above baseline,” it adds.
What this adds up to is that our RGCs are constantly on the lookout – no pun intended – for the edges of things. In fact, they’re so intent on finding where visual stimuli start and finish in the world that they also do something called “lateral inhibition” – a process wherein once one cell notices there’s something to fire off a signal for, it will also send a message to the cells surrounding it, telling them to quieten down for a bit.
The edge of darkness
This is where the expanding hole illusion comes in. Look at the image (if you can stand it) and you’ll immediately notice that, well, there really aren’t many hard edges at all, are there? So how are our RGCs meant to understand it?
Well, as it turns out, there’s a pretty ingenious way to figure that out – and it’s surprisingly similar to the way you made your middle-school art projects pop.
“Applying a Gaussian filter to an image creates a smoothed or blurred version of it,” notes the paper.
“The Difference of Gaussians (DoG) filter is a widely used computational approach to mimic the receptive field properties of retinal ganglion cells,” it explains. “The Difference of Gaussians (DoG) output is obtained by subtracting two differently blurred versions of the same image, effectively functioning as a band-pass filter.”
In other words… well, you know those images that have a filter applied to show “how your dog sees the world” or “how colorblind people see things”? This effect is kind of like that, except the result is “how some of your RGCs see the world”. And depending on which particular DoG filter you use, you can change which type of RGC you’re viewing things as – and also, therefore, which types are being inhibited.
Over five different DoG filters, the effect of increasing the radius of firing RGCs had a very clear effect: “The scale of the centre Gaussian […] increases from 4 to 20 in increments of 4, intensifying the DoG filter’s sensitivity to contrasts in the central area and simulating the dynamic effect of forward motion,” the paper notes. “This emphasised response visually reinforces the perceived expansion in the original pattern.”
What this means, basically, is that different RGCs are interpreting the image differently – some are seeing quite a small black area; some are seeing a larger one. But they’re all reporting these images to the brain at once, resulting in a confused sort of “well, maybe it’s a moving object?” conclusion.
“The findings indicate that the expanding hole illusion is a result of complex neural interactions, specifically involving contrast-based lateral inhibition within the visual cortex,” the researchers write. “The alignment of perceptual expansion strength with physiological pupil responses suggests that the illusion is not purely perceptual but involves coordinated processing across different neural circuits.”
“Our model aligns with previous studies on visual motion illusions, extending these insights to explain the Expanding Hole Illusion,” they add.
Why is this important?
So, here’s a question: why should we care about this? Optical illusions are cool, but they’re hardly rocket surgery – why are serious researchers wasting their time explaining why we sometimes, given the right conditions, see a static image as “moving”?
Well in fact, there are a few reasons we should take note of this research – and, as is so often the case, the actual conclusion is only one of them. Even the other illusions it illuminates aren’t the most important takeaway – although the paper makes note of quite a few famous examples that their results can help to explain, and other experts already have ideas on how it might apply to their own areas of research.
For example, “I might be able to use it to understand the patterns that we see in nature, which makes me very excited,” Jolyon Troscianko, a visual ecologist at the University of Exeter who was not involved in the new paper, told New Scientist. “Zebra stripes and butterfly wing patterns, and all of these kinds of things that are often very poorly understood.”
But the study’s most widely applicable benefit? It’s conceptual. So far, explanations for how this kind of illusion works have been “very high level and arm-wavey,” Troscianko said, involving ideas about our brains getting confused over trying to process a 2D image as a 3D “hole”.
This new theory, though, is simpler – both as an explanation, and as a testable hypothesis. Moreover, it presents a new way of thinking about some optical illusions: they can be just a physical result of basic neuronal reactions, the paper suggests, rather than always the result of higher-order cognitive processes.
Of course, those more complex explanations “might be true – you can’t disprove them,” Troscianko told New Scientist. “But if you come up with an explanation that relies on fairly early visual processing stuff in your brain, then that’s – for me – more useful.”
The study, which has been posted as a preprint and has not yet been peer-reviewed, is available via arXiv.
