Most people assume that what you see is pretty much what your eye sees and reports to your brain. In fact, your brain adds very substantially to the report it gets from your eye, so that a lot of what you see is actually “made up” by the brain (see Seeing more than your eye does). Perhaps even more interestingly, the eye actually throws away much of the information it gets, leaving it to the rest of the brain to fill in additional information in its own ways. A characteristic pattern of connections among neurons (nerve cells) in the eyes of most animals (including humans), termed a “lateral inhibition network”, is a significant way information is thrown away. Lateral inhibition helps to explain a number of “optical illusions” and, more importantly, provides an excellent example of how the brain is organized to actively “make sense” of the information it gets, rather than to simply absorb and respond to it. In so doing, it provides some valuable insights into the sources of our sense of “reality”.


To understand lateral inhibition networks and their function, one needs to know a bit about the gross anatomy, optics, and initial neural signals of the eye. As shown in the figure, a simple (but adequate) way to think of the eye is that it is a hollow sphere with a hole in the front (which corresponds to the black spot, or “pupil” of a real eye, located within the colored “iris”). The pupil is the opening through which light enters the eye. The pupil, together with two curved transparent structures (an overlying “cornea” and a “lens” just inside the pupil, neither of which are shown in the figure), also controls the path taken by light, much as does the aperature and lens of a camera, creating on the back of the eye an inverted image of whatever is being observed. The image falls on the “retina” (the grey arc in the figure), a sheet of neurons which includes a layer of photoreceptors, neurons specialized to measure light intensity and translate it into signals which the rest of the nervous system can understand. Hence, at this first processing step, each photoreceptor generates a signal related to the intensity of light coming from a corresponding point of the observed object. Photoreceptors corresponding to brighter arrays of the object (yellow) receive more light and generate larger signals than those corresponding to darker areas (black).


The signals resulting from light falling on the photoreceptors are not sent directly to the brain in the optic nerve (grey line leaving the back of the ye in the figure above) but are instead first processed in a number of ways by a variety of interactions among neurons within the retina, of which the lateral inhibition network is an instance. Such a network is shown schematically below (the anatomical and physiological details of retinal organization are more complicated than shown in the schematic and vary substantially from organism to organism; humans, and other vertebrates, have three layers of neurons in the retina, rather than the two shown in the figure, but the functional outcome is the same).

The schematic illustrates a small portion of the retina, one of those at which the overlying light pattern changes from darker to lighter, as shown at the top of the figure. The green rectangles represent photoreceptors, each generating a signal appropriate for the amount of light falling on it. The red circles represent output neurons of the retina, whose signals will go to the brain through the optic nerve. Each output neuron is shown as receiving excitatory input from an overlying photoreceptor (vertical black lines) as well as inhibitory input from adjacent photoreceptors (angled blue lines). It is this laterally spread inhibition that gives “lateral inhibition” networks their name. At the bottom of the schematic is a representation of the signals in the output neurons (purple lines). Output neurons well to the right of the dark/light border are excited by an overlying photoreceptor but also inhibited by adjacent, similarly illuminated photoreceptors. The same is true far to the left of the dark/light border. Hence, assuming that the network is organized so that equal illumination of exciting and inhibiting photoreceptors balances out, output neurons far from the edge in either direction will have the same output signals. Only output neurons near the dark/light border will have different output signals. As one approaches the dark/light border from the left, the signals will decrease, because inhibition from more brightly lit photoreceptors to the right will outweigh the excitation from the overlying dimly lit photoreceptors. As one approaches the dark/light border from the right, the signals will increase because excitation from brightly lit photoreceptors is not completely offset by inhibition from the dimly lit photoreceptors to the left.


Clearly, output neurons are not telling the brain exactly what the light intensity is at each point on the retina. They are instead telling the brain which regions of the retina have “edges” (areas where light intensity changes quickly). The output neurons also provide information about whether the quick change in intensity is increasing or decreasing (see figure above) and by how much (not shown above, but you can check this out with our lateral inhibition simulator). Why then does one see solid squares of black and white in the checkerboard above? Why does one not instead see the same brightness at the center of both the black and the white squares, with only the borders between them looking different? Part of the answer is a “filling-in” quite analagous to that which occurs for the blindspot. Knowing that there is a particular change of light intensity at a particular place and no change in light intensity until another particular place, the brain can “fill-in” the intensity between the two. If you look just right at the checkerboard, however, you can see clear signs of the operation of the lateral inhibition network. The centers of the white squares look a little darker than the edges, and the centers of the dark squares a little lighter than the edges, just as you would expect given a lateral inhibition network.

Whether you can or can’t see the smudged whites and cloudy blacks in the checkerboard, you can’t miss the evidence for a lateral inhibition network in the figure below. The small grey squares are exactly the same intensity, but the one on the left looks darker and the one on the right looks lighter. The reason for this is that ouput cells near the edge of the left small grey square are signalling decreasing light intensity as one moves into the square, while those near the edge of the right small grey square are signalling increasing light intensity as one moves into the square. Clearly, the surrounding intensities dramatically affect what one perceives, just as one would expect with a lateral inhibition network (this too you can demonstrate for yourself if you have our lateral inhibition simulator open or you open it).


An obvious question is why is the nervous system organized such that it can be so easily fooled by the grey squares on light and dark backgrounds? In fact, its not in general so easy to “fool” the nervous system. That’s one reason why the few cases where we can show that it happens, optical illusions, are so much fun. But optical illusions are even more interesting for the fundamental insights they offer into how the nervous system works, something you’re less likely to ask about unless you’re fooled. And so the question remains: why do we (and other animals) have lateral inhibition networks in our visual systems? Why throw away information about brightness and then rebuild it inside the brain? The figure below will help us work towards an answer to that question.

Though we don’t usually notice, we live in a world of constantly varying light intensity. And though we don’t usually think about it, what this means is that the light reaching our eyes from any given object in the world is changing all of the time. Under bright sun, as shown to the left in the figure, the checkerboard (upper light/dark line) casts a certain amount of light on our photoreceptors (lower light/dark line, notice that it is of course left/right inverted by the pupil/cornea/lens). On a cloudy day, as shown to the right, the checkerboard is the same checkerboard but the amount of light reflected to our eye from both the light and the dark squares is reduced. A very interesting and important property of a lateral inhibition network (compare the output plots at the bottom of both figures) is that their outputs will be pretty much the same in the two cases. If you have the lateral inhibition simulator open, you can verify this for yourself (or you can open it). Since the network detects and reports only locations of large changes in brightness, the brain gets more or less the same signals in both cases, and one sees more or less the same checkerboard.

What changes without our noticing it is not only overall intensity but also intensity variations. If a light is located to one side of a checkerboard, as in the figure above, more light will be reflected to the eye from the side closer to the light and less from points further from the light. So what will appear on the photoreceptors is relatively bright at the left, steadily darker moving toward the right, then brighter, and then steadily darker again. Because of the lateral inhibition network, he relatively slow changes in light intensity due to the direction of the light source won’t much affect the output neurons, and once again the brain gets more or less the same signals and you see more or less the same checkerboard. This too you can test if you have the lateral inhibition simulator open (or you can open it).


Clearly, what we see is not “what is out there”, but rather something which is significantly affected by the structures of our brains. That’s an important and quite general (lateral inhibition networks are only one of myriad examples) lesson to keep in mind, both for oneself in trying to make sense of the world and when arguing with other people (whose brain structures may differ) about what it is REALLY like out there. Its a particularly important lesson to keep in mind since the lateral inhibition network (and similar brain structures) are having their effects in ways that we are normally totally unaware of, until they are called to our attention, and even then are more than likely to stop being aware of when we stop thinking about them. Lateral inhibition networks are operating as part of the “unconscious” brain, and largely without providing any information to the “conscious” part of the brain about what they are doing.

The organization of the brain is such as to create “abstractions”, rather than to simply take input at face value. A checkerboard is a checkerboard is a checkerboard not because the input reaching our eyes is the same at all times but rather because the nervous system is organized to reject some information and replace it with other information. The “booming, buzzing confusion” of the external world is rendered stable and comprehensible by the organization of the nervous system. That organization represents information added by the nervous system to the information is receives, and constitutes a presumption that there exist stable, constant external forms with well-defined boundaries. The presumption is strong enough so the nervous system actually creates boundaries where none in fact exist, as shown by the figure above. Such a presumption probably reflects genetic information (the experience of innumerable generations of ancestral organisms) about the nature of “reality”, and might be absent in organisms evolving under other circumstances where they do not interact with stable, constant, spatially bounded external forms. Such presumptions also probably provide a basis for the human experience of “ideal forms”. Such “ideal forms” are not, as Plato might have imagined, properties of the real world dimly glimpsed by imperfect humans, but rather abstractions created by the brain. The misunderstanding may have arisen because the abstractions are properties of the “unconscious” rather than the “conscious” parts of the brain.

“Contour and Contrast” by Floyd Ratliff (Scientific American, June, 1972) describes some of the early neurophysiological work on the horseshoe crab which led to modern understanding of lateral inhibition networks. That article, and Floyd Raliff’s Mach Bands (Holden-Day, San Francisco, 1965) nicely connects the neurobiology to earlier intuitive understandings of visual processing by artists as well as to earlier psychophysical explorations of the relationship between visual perception and external reality.

For more complete descriptions of retinal structure and function, and the structure and function of the visual system generally, see C.R. Michael, “Retinal processing of visual images”, Scientific American, May, 1969, and D.H. Hubel and T.N. Wiesel, “Brain mechanisms of vision”, Scientific American, September, 1979.

For more on the eye, including the complexity of the real retina:

Some additional interesting links on visual illusions:

Created by Paul Grobstein. Applet by Bogdan Butoi


Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s