Similarly, when you see a person’s two arms forming a cross, there are two possible interpretations. A malicious surgeon may have amputated an arm and pasted the two halves on either side of the intact arm—or one arm may simply be placed perpendicularly in front of the other. Your visual system instantly sees that the latter is true; you do not even consider the former interpretation. Again, this is not because of high-level knowledge about the improbability of amputated arms—note that the brain has the same instantaneous reaction when the cross is made of wood, which could quite easily and bloodlessly have been sawed into pieces.
Borderline cases exist, however, such as the bear you “hallucinate” behind a tree. This drawing seems to show only circles bisected by lines, until the addition of what appear to be claws makes the dot at the top right morph into a nose and the circles into paws. Such examples blur the distinction between seeing and knowing. For instance, if you observe a fast-moving toy train go into a short tunnel and emerge on the other side within a third of a second, you will actually “see” the motion of the train, as if the tunnel were transparent. You have modally completed the motion across the tunnel—a phenomenon first pointed out by Gestalt psychologist Albert Michotte (1881–1965).
If the train is slow, on the other hand, taking a minute or more to traverse the tunnel, you still know that a single train entered and then emerged on the other side, but this time it is a logical inference rather than a visual perception. At speeds of about a second, however, you are in a borderline state between perception and logic, and the question of whether you actually “see” the train’s movement comes perilously close to being a philosophical one.
Elongated Cats
The tendency to anticipate contours is so strong it overrides our knowledge of how the world actually works—as demonstrated, for example, when a cat seems unrealistically stretched around a tree: the brain is responding to continuity, independent of whether it makes sense or not.
Such visual anomalies occur because these rules are evolutionarily ancient and were not designed to handle improbable juxtapositions created by scientists. Programming sophisticated object knowledge into the system would have been too demanding—and unnecessary. Only in myth and fantasy do animals abruptly morph into unaccustomed shapes.
According to hierarchical views of visual processing, the detection of edges in a two-dimensional drawing is a relatively simple process that necessarily precedes the act of constructing high-level 3-D representations. Other figures designed by Tse challenge this conclusion.
The simplest is his lab’s logo. It can be seen either as two flat bird heads (one of them upside down) or as a 3-D black worm wrapped around a white cylinder (the worm is amodally completed by the presence of the cylinder). Unlike the Kanizsa triangle, in which the three disk regions align, implying the existence of edges, in this Tse figure there is no direct continuity of luminous edges or physical contours. And yet the brain perceives the 3-D worm. These illusions suggest that amodal completion is not only a matter of filling in continuous contours. The visual system is cleverer than that. In fact, in another Tse creation, objects complete amodally behind contours without their exact shape even being specified.
Object Lessons
In their pioneering work in the 1960s, neurobiologists David H. Hubel and Torsten N. Wiesel of Harvard University showed that brain cells in the primary visual cortex respond principally to the dark/light edges that convey the contours of an object or creature. Rudiger von der Heydt of Johns Hopkins University has subsequently shown that cells in the secondary visual cortex respond to illusory contours such as those of the Kanizsa triangle.
Rights & Permissions
See what we're tweeting about
3 Comments
Add CommentAs I understand, the brain includes a specialized pattern recognition system that can recognize faces, familiar shapes, etc., much faster than we can process language, for example. It seem quite feasible that this 'dedicated hardware' is simply invoked many times in a brute force attempt to identify a discrete number of critical but incomplete shapes within visual data.
Reply | Report Abuse | Link to thisGeneral purpose computers serially process data, whereas parallel architectures can (once required data is in place) can perform a given function on an array of like data (such as an image) in the time required to process a single element (pixel). A specialized parallel image comparator could very quickly provide assessments of partial images, whereas a similar process implemented using serial functions would require an enormous amount of processing logic, move a lot of data and take a very long time to complete.
This image recognition function is very important and useful to living organisms...
I think it's likely we'll see the same generalized amodal completion algorithm at work in all forms of sensory processing and cognition: language, proprioception, etc.
Reply | Report Abuse | Link to thisi am looking for special edition 169 best illusions i have called r customer service number it is disconnected please tell me where i can buy this issue or can i order it? dnance56@att.net
Reply | Report Abuse | Link to this