Course Home
Lecture Notes

Comments? Idea?

On Symmetry, Illusory Contours and Visual Perception

Final project by

Liana Diesendruck

nigri at bgu dot ac dot il


Shape completion and illusory contours

Shape completion, i.e. deciding which contours should “go together” in order to form a figure, is still an open question and its process isn’t completely understood yet.

In its natural habitat, an animal's perception of an object has to overcome some ratter degraded visual conditions, such as bad illumination and occlusion of an object by other. Adding to those bad conditions, the retinal projection of the animal's visual field is confined into a two-dimensional space, while the real world is (so much!) three-dimensional.

Therefore, image projections in which the boundaries are not continuous or/and are not notably separable by luminance difference may be naturally generated; the completion of this “gaps” would be fundamental to a “real” perception of the world. Boundary completion may be a mechanism of assuring the animal awareness of the real condition/situation in front of its eyes, preventing importune surprises (such as predators) and facilitating the navigation through its environment.

Illusory contours(illusory shape completion) occurs when borders are perceived, despite of being inexistent in the visual stimulus, by a brightness enhancement of the completed shape. See bellow some examples of visual illusions that are based on illusory contours. Notice that the first one is the famous Kanisza’s triangle.

Unilateral Visual Neglect

“Neglect is a neurological condition under which patients with unilateral (often right) parietal lesion fail to consciously perceive stimulus shown in the contra-lateral (left) visual field.” - Driver & Mattingley, 1998.

Hemineglect or unilateral neglect is the most common attention impairment disorder and has many neurological symptoms. It is associated with injuries in the right parietal lobe, many times following a stroke. The condition doesn’t involve damage to early visual areas and also affects the capacity of describing from memory. Thus it is assumed to be a imagery, perception and attention deficit.

People suffering from unilateral neglect may fail to detect objects on the opposite side to their injury. A patient suffering from this condition, following an injury in the right hemisphere, may fail to notice or hear people standing in their left side, eat just what is in the right side of his plate, shave just the right side of his face, hold his head slightly inclined to the right, etc.

Neglect is diagnosed by a series of simple clinical tests; bellow we can see some replications of the result of these tests in a person diagnosed as suffering from this condition.


Symmetry plays an important role in nature from the evolutionary point of view; particularly, in the prey-predator relationship, we may see good use of symmetry detection in order to “crack” camouflage as most animals have bilateral symmetry.

Accordingly, there are brain mechanisms that respond stronger to symmetrical objects/shapes, seeming to be an evolution-preserved mechanism in most animals.

Symmetry has been proved to be detected by the human brain even before we consciously pay attention to the scene.

Approach and Method

Visual neglect patients present us with a great opportunity to understand high-level sensory-perceptual mechanisms as they have no impairment in the early visual processes, and nevertheless they fail to “see” part of a presented scene.

Contour completion use not well-understood brain mechanisms and people with visual neglect may help us enlighten this phenomenon. The use of visual illusions (particularly referring to illusory contours) as cues in psychophysical tasks to visual neglect patients may lead to the understanding of these mechanisms. 

It is not clear, theoretically, if visual neglect patients should be able to percept visual illusions as healthy individual do; moreover, even if they fail to report the visual illusion, it doesn’t imply that they didn’t “compute” it through the perceptual mechanism.

First experiment

Objective: In this experiment we want to verify if and when patients with visual neglect may be aware of illusory contours.

Method: The subjects will be randomly presented with pictures of visual illusions based on illusory contours and their derivations that don’t form illusory contours; and will have to describe what they see in the picture.

Examples of pictures used:

Unilateral neglect patients have difficulty to move attention from the ipsilateral (same side of the injury) to the contralateral (opposite to the injury) side of a picture. There are findings, though, that if the cue in the contralateral side is big enough, an attention shift is observed.

Although, the patient should be unable to report the illusion, as he doesn’t perceive the inducer, the boundary completion may have been successfully completed; so, the outcome of the experiment isn’t as obvious as it seems to be.

We expect to receive different descriptions for pictures with big ipsilateral inducers and with big contralateral. Although, it is not clear whether the visual illusion will be reported, we expect to receive always a description (big/small) of the ipsilateral inducer; the contralateral inducer may be described or not (probably) depending on its size.

Results: Please, come back later.

Second experiment

Objective: The objective here is use the paradigm of Vuilleumier and Landis[8] in the paper “Illusory contours and spatial neglect”, adding totally non-symmetrical figures. We’ll try to reproduce their results.

Method: The subjects will be randomly presented with pictures of regular figures, visual illusions based on illusory contours and their derivations that don’t form illusory contours; the patients will have to perform an adapted line bisection task; that is, the subjects should indicate where the middle point of the figures is.

In addition to the pictures used in the previous experiment will be used: a picture of regular line; new totally asymmetric pictures; and the same pictures with real boundaries instead of illusory ones.

Examples of new pictures used:

Vuilleumier and Landis showed that the neglect line-bisection effect is received in illusory lines in conform to the “real” line-bisection effect in neglect patients. But they used symmetrical geometric shapes in their experiments and it may have influenced their result: Driver[9] proved that the human brain process preattentively symmetrical shapes; that said, it is possible that Vuilleumier’s results were influenced by the symmetrical shape of the illusion inducers.

Results: Please, come back later.

Third experiment

Objective: To assert whether or not there is some type of symmetrical shape completion done by the brain in order to compensate for the (poorly) perceived image so that it fits more properly the visual scene.

Method: The subjects will be randomly presented, for a small amount of time, with pictures that if they were mirrored, they would infer a visual illusion based on illusory contours, their respective illusion, and figures with geometrical shapes in the ipsilateral side; they will be asked to repeat the last first two experiments with the new figures.

The following picture will illustrate the idea of the mirroring:

In (a) we see the original inducer, whereas in (b) we see the complete illusion received by mirroring the inducer.

Usually, when a neglect patient is asked to draw from memory/copy symmetrical figures, they tend to ignore the contralateral side of the object. When they are asked to do the same with non-symmetrical objects, two possible behaviors are seen: they ignore the contralateral side of the object, or they “squeeze” all its elements into the ipsilateral side of the drawing.

If there is some kind of symmetry-completion, we will observe similar results for the “half illusion” and “complete illusion” pictures; the geometrical shapes should give “normal” results.

Results: Please, come back later.


The experiment wasn’t run yet. Give us some time and check here again for results.


As we still don’t have the results, for now we shall enjoy some rhetorical thinking here.

Suppose that the predicted behavior was observed in this battery of experiments. What does it mean?

First of all, that boundary completion really is in a preattentive stage of the “perception path”; maybe even in a parallel stage to symmetry perception (just a speculation, more research would be required to confirm/deny this hypothesis).

In the computer vision field, this fact may have great influence in the designing of novel techniques of image segmentation, highlighting the importance of top-down mechanisms.

In the cognitive science/neuroscience areas, the results should contribute to a better understanding of how the brain organizes the different stages of information process in order to perceive the world.

Second, it shows that the brain compensates for the hemineglect condition by using symmetry perception mechanisms. This finding would help the development of new rehabilitee tools for the condition.

Additional Information


[1] Visual perception: physiology, psychology and ecology / Vicki Bruce, Patrick R. Green, Mark A. Geor

[2] High-level vision: object recognition and visual cognition / Shimon Ullman

[3] Active Perception / Yiannis Aloiminis

[4] Vision science: photons to phenomenology / Stephen E. Palmer

[5] Fearful symmetry: is God a geometer? / Ian Stewart and Martin Golubitsky

[6] Cognitive Neuroscience: The Biology of the Mind / M.S.Gazzaniga,R.Ivry,G.R.Mangun

[7] Seeing is deceiving: the psychology of visual illusions / Stanley Coren, Joan Stern Girgus

[8] Vuilleumier, Landis. Illusory contours and spatial neglect. Cognitive Neuroscience 1997.

[9] Driver J, Baylis GC, Rafal RD. Preserved figure-ground segregation and symmetry perception in visual neglect. Nature 1992; 360: 73–75.