How Does Our Perception Remain Stable Despite Our Constantly Moving Eyes?

Introduction

The average person rapidly moves their eyes about three times every second. These movements, called 'saccades' can be small, or very large, and can occur at speeds that cover up to 475-525° per second, which is incredibly quick if you consider the relative distance a nearby object must 'appear' to travel across your visual field in such a short time.

So then, with all these rapid eye movements constantly occurring, you've got to wonder; how come we don't really notice them? To us, the world seems pretty stable - it doesn't feel like we're constantly getting thrown around like on a rickety rollercoaster... While you're reading, the book or screen doesn't appear to rocket to the left or right - the words on the page don't appear to be moving (depending on how many glasses of wine you've indulged in). How come?

In the following, I'm going to explore two broad categories of theories that seek to explain this phenomenon.

The World-Centred Map Hypothesis

Overview

One of the leading theories in understanding visual stability revolves around the concept of an allocentric, or world-centred, map in the brain. This theory posits that our brain constructs and updates a map of our surroundings independent of our eye movements.

How It Works

Imagine looking around a room. As your eyes move from object to object, this theory suggests that your brain updates a mental map of the room's layout. Each saccade doesn't disrupt this map; instead, the new visual information is integrated, helping maintain a continuous and stable perception of your environment. In essence, it's the map and your relationship to it that forms your perception, and not the direct input from your eyes. When we look at an object, the map is referenced, the relevant features that are new or have changed get updated. That information is then fed back to us to make sense of what we are seeing.

Thus, our perception of the environment is largely based on this internal map and our relationship to it. This map is continuously updated with new visual information as we move our eyes and body, helping us orient ourselves in space and maintain a consistent understanding of our surroundings. We're not subject to the rapid movements of our eyes, for the map itself is stable, and it's that stable map that makes it through to our conscious view of the world.

Criticisms

One downside to theories based on this allocentric map idea is that it fails to account for errors in our visual processing. For instance, we are not always aware of the changes made to our environment, as you can see from any number of illusion videos or selective attentional tasks online. For instance, why would counting ball bounces impact your awareness of 'surprising environmental additions' if the map was updated via all incoming visual information - it should make it to our awareness regardless, right? Perhaps most concerning for the advocates of this theory however is that no evidence of this allocentric map or the brain signals carrying eye position information necessary for updating such a map has been found.

The Shifting Receptive Fields Theory

Overview

Contrasting the world-centred approach, the second theory suggests that our perception of space remains tied to retinotopic, or eye-centred, coordinates. In this case, there is no map with which to interpret the incoming data, and so we need some other mechanism by which to smooth out the rapid eye movements.

There are numerous theories that fall under this broad category of saccade-adjustment explanations. Nevertheless, the vast majority of these theories rely upon the concept of anticipation, or prediction of incoming data (predictive coding/predictive processing for those of you who have read my other posts or are familiar with the material). One of the most famous of these theories (shifting receptive fields) comes from an observation of how neurons in the parietal cortex behave; how they adapt their receptive fields in anticipation of eye movements.

A Practical Example

Let's use a simple example to illustrate this concept. For the sake of argument, imaging your retina (the layer of photoreceptors cells at the back of your eye upon which light is focused) is a large canvass with a cartesian grid reference system on it. On the eye, there is an area called the macula, which is responsible for our central vision. At the centre of that is the fovea, where the highest visual acuity is found due to its higher density of cones. Because visual acuity is at its highest here and drops off the further out from the fovea you get, that means that the object of your focus usually matches up with this area. As such, we'll call it the centre point on our gridded canvass at (5,5).

Now, picture yourself at a park, observing a treetop in the background. As you plan to shift your gaze to squirrel beneath the tree (let's say at grid reference (4,1)), the neurons in the parietal cortex that were originally selective of the features found at (5,5) begin to adapt even before your eyes move. You're still looking at the treetop, but the neurons originally processing this region have began preparations to process the area around the squirrel. This shift in neuronal focus prepares your visual system to process information from the new focal point seamlessly. Once you do move your eyes, the area around the squirrel is now seated at the centre of the canvass at (5,5), and the neuronal group responsible for this area continues to process it as it normally would, its receptive field remaining unchanged - but that anticipatory preparation helped to smooth out the transition so as to avoid any perceived jerks of motion caused by the saccade.

Criticisms

A common criticism of predictive theories such as these is that they are difficult to prove. We can identify some change in neuronal activity that precedes a behavioural change such as an eye saccade, but it remains difficult to specify exactly what this change entails. For instance, are the neurons preparing to encode information from the intended destination of the saccade, or are they simply preparing for some change - any change - regardless of intended destination?

Gradually, we're beginning to see evidence from fMRI and other neuroimaging studies that seem to suggest there is some pre-processing indicative of a predictive response, such as occlusion studies making use of multivariate (or multivoxel) pattern analysis (MVPA) to determine what the information the pattern of neural activation most likely represents, though as far as I am aware, this is mostly limited to the early visual system. However, the question remains whether these predictive mechanisms are present throughout the brain, or are confined only to the areas we've currently observed them in.

Thus, the common criticism of predictive theories such as these is that they tend to beg the question, assuming predictive processing is already established and proven, and then extrapolate out to explain other aspects of cognition before we've established the fundamentals first.

Conclusion: A World of Perception

These theories, while distinct, share a common goal: to explain how we experience a stable and coherent visual world. Whether through a world-centred map or shifting neuronal patterns, the human brain's ability to adapt and process visual information remains an area of awe and wonder. It's my hope that we will one day be able to accurately examine and understand exactly how the brain pulls off these amazing feats of processing, and perhaps take advantage of our understanding to create artificial or complementary systems to improve our own lives, whether that be by correcting for dysfunctional maladies or by improving our lives through advanced technologies that leverage such understanding.

A Note on my Sources:

I've been pretty poor for keeping track of my posts' references thus far, as these posts are intended to be more of a creative outlet for myself with the side-benefit of perhaps helping others understand these complex topics, rather than being pieces of academic writing.

Nevertheless, recognising the work of others and signposting where to learn more about a topic is, and will always remain, important. As such, I shall put more of an effort into keeping track of where my information comes from and providing the necessary links where appropriate going forward.

For this post, the bulk of the structure comes from a doctoral thesis by Tao He entitled "Perceiving the future: Predictive processing from the eyes to the brain". You'll notice the examples given are similar to those presented on pages 14-16. While similar in content, I hope that I have sufficiently expanded on the content and explained it in a manner that makes it more accessible to the everyday reader.

Information about the eye and about the criticisms of predictive theories come from my own understanding formed over the last two years. Much of it can be found in my MSc dissertation entitled "Sharpening or Dampening: A Pilot fMRI Paradigm Investigating Predictive Feedback". I am unsure whether I have the rights to post this online as it may be the property of the university, and so I shall endeavour to break down the relevant information found within into singular blog posts in the future with adequate referencing, so that's something to keep an eye out for going forward. If you're keen to read the dissertation however, get in touch!

Finally, all images on this site are either generated using generative AIs which confer the right for use in this medium, or are sourced from royalty-free sources unless otherwise specified.