This paper reflects the research and thoughts of a student at the time the paper was written for a course at Bryn Mawr College. Like other materials on Serendip, it is not intended to be "authoritative" but rather to help others further develop their own explorations. Web links were active as of the time the paper was posted but are not updated.
2001 Second Web Report
The family and medical staff who attended Vincent, blinded since childhood by thick cataracts, had high hopes that, for the first time in nearly 45 years, he would be able to see following a surgery to remove the cataracts. When the bandages came off, Vincent saw colors, movement and shape. He even saw details and isolated features of objects. What he could not do, to their dismay, was to make sense of what he saw: he could not form coherent perceptions of objects in his world from the parts and features, and he had no sense of space, depth, or distance. (1)
That Vincent could perceive color, motion and shape directly following surgery is not surprising, given that the eye comes equipped to transduce signals relating to color, motion and shape. (2)These signals leave the eye, converging and diverging along neural pathways to other visual processing centers in the brain. (3) Eventually, many end up at the occipto-temporal region, which is specialized for object recognition, and the posterior parietal area, which is specialized for spatial perception. Interestingly, Vincent's pattern of impairment seems to roughly correspond to these "what" and "where" processes in the brain. (8) What interests me mainly is the failure of Vincent's "what" process, his inability to organize the various percepts into meaningful, whole representations. (4)
Early in the 20th century, Gestalt psychologists explored how organization governs perception by grouping parts into coherent wholes. They discovered laws of grouping, including similarity, proximity, good continuation, and closure. (5) From his pattern of visual perceptive deficits, it appears that Vincent's brain does not group visual input into wholes according to the Gestalt principles. The visual stimulus remains fragmented, disorganized. If there is a center in the brain for perceiving objects (like the "what center"), is it somehow related to the Gestalt-like integration of objects into wholes? And if so, how does that mechanism work, and why does it not seem to work for Vincent?
Unfortunately, although the Gestalt principles did predict with great accuracy how objects tend to be perceptually grouped as wholes, it was mainly a descriptive theory. It didn't try to link the observed regularities with biological substrates or explain how the phenomena arose. It could be that the brain comes equipped with Gestalt-like mechanisms (such as a bias to group pieces into a whole based on how much relative space is between the pieces - the principle of "proximity"). Or it could also be that the brain is equipped with neutral circuitry that is shaped by experience. Such a theory might explain the ubiquity of Gestalt-like perception this way: because the world is full of objects, people generally have very similar visual learning histories, and so we all end up with a similar set of perceptual tools (like grouping by similarity, proximity, etc) to help us make sense of the signals we get from our eye.
In other words, the competing hypotheses boil down to the old nature nurture debate: does the brain come equipped to perceive objects as wholes, or does it learn the grouping laws through experience? Oliver Sachs, the recorder of Vincent story, would probably argue that Vincent's pattern of deficits clearly implies grouping principles are a result of experience. (1) The argument goes: Vincent has had no visual experience, Vincent does not segment his visual field into groups/wholes, therefore experience must be what causes us to see the world in that way. But is this necessarily so?
Perhaps Vincent's pattern of deficits can be explained in other ways. Another hypothesis: Vincent's brain (like everyone's) has built in mechanisms, or biases, to perceive objects as coherent wholes. Most people's sensory input is largely visual, and so visual signals are most often routed through brain's object constructing biases. Because the habitual signal through the circuitry is visual, it's the visual signal that becomes expert at making its way through. However, Vincent's sensory input was mainly tactile and auditory. In turn, his object recognition tool bag became sensitive to tactile input. Naturally, when his eyes were fixed, the system was no longer sensitive to visual input anymore, it was expert at negotiating tactile and auditory input instead.
In other words, experience does have something to do with how we perceive the world, but it does not necessarily imply cause. The main problem with Sach's argument is that it's not that Vincent has had NO experience, it's that he's had a DIFFERENT set of experiences than most of us. We cannot know what's in the system or not prior to experience, through Vincent's story, and Sachs makes a leap in assuming we can. To make Sach's argument credible, we would have to look at the brain prior to ANY experience and show that without it, there is an inability to perceive the world as integrated wholes.
But babies, who have very little experience in perceiving the world, do seem perceive it in terms of objects. Using a habituation procedure, infants were shown a rod moving back and forth behind a block, partially hiding the rod. When the block was removed, the rod was actually two disjointed pieces, and the babies dishabituated. In other words, they expected to see "an object" not two. They were exhibiting the Gestalt principle of "common fate," that is, grouping by common movement. (4) We can infer that they have a sense of "object-ness" that was violated by the disjointed rods. This is evidence against Sach's experience argument.
And yet, a stronger argument for the biological bases of object perception should really include direct neurological evidence for Gestalt perceptual grouping. Winfried Fellenz has been looking for such evidence.
Fellenz says that synchronization of cortical neurons can account for many of the Gestalt grouping principles. (6) Specifically, she argues that there are 4 stages in object recognition. The first two stages are entirely data driven, pre-attentive, and have to do with opposing operations: stage one segments-- it defines edges and endpoints, and stage two groups-- it pulls like stuff together. The third and fourth stages are concept driven, and have to do with highlighting relevant features in the service of object recognition.
In the first stage, lateral inhibition essentially throws away a lot of information about light on the retina, reducing it to information about edges and endpoints of incoming stimuli. In the second stage, that information is extracted into perceptual groups by mechanisms that Gestalt has labeled as grouping by "proximity" and "good continuation". Basically, as lateral and vertical information is shared by nearby neurons, they synchronize and de-synchronize phase information. In doing so, bad responses from the earlier "edge information" stage are eliminated, and the input is grouped-smoothing together that which is similar and breaking apart that which is not. Fellenz argues that this process can account for Gestalt phenomenon, such as gap closing and completing contours ("closure" and "good continuation" respectively).
The third and fourth stages are cognitively, or top-down driven. They incorporate an attentional filter that highlights some perceptual groups over others in the service of recognizing objects. It seems as though Fellenz' model makes a distinction between object recognition, which is about assigning meaning to a percept, and object perception, which has more to do with the sensory grouping into object-ness, apart from any meaning attached to it.
Assuming that Vincent has all of the same neural equipment we do, can this model account for his inability to see objects as more than a disorganized muddle of color, shape and movement? Earlier I hypothesized that perhaps his normal pathways were somehow rewired, but I did not propose what mechanism or path was rewired, and it seems to me that the attentional mechanism might be a good fit. The purpose of attention is to extract relevant features of the preattended input. Perhaps this process, by definition, requires experience. How would the attentional system "know" what is a "relevant" feature and what is not, without prior experience?
But Fellenz' model argues that the Gestalt phenomena occur before the attention stage. If I were Fellenz, I would be interested in seeing what Vincent's visual system would do with simple visual information, such as the figures presented in her experiments. (6) Regardless of his inability to make sense of complex scenes and extract relevant information from them, would his perceptual system group dots in the same way that our brains do? Would it close small gaps in figures like ours would? These processes seem more basic than what happens when putting together complex visual scenes, which seemed to be particularly overwhelming to him.
I want to make a diversion here, and point out that before his operation, Vincent DID have ways of recognizing objects as objects. For example, he first recognized a face when it spoke, even though the "face" was to him a jumble of colors and movements that had no "objectness" to them. He had a "concept" of face as a bounded entity, though we do not know whether he had a mental image of a face as we do. And when he saw black and white spots moving about near him, he recognized them as his cat and dog when they barked, meowed, or he touched them. Again, he had a "concept" of cat and dog as bounded entities. Even if you put the two of them (both were black and white), right next to each other, if Vincent could touch or hear them, he would not assume that they were one continuous object. He would perceive them as separate objects no matter what his new disorganized visual input told him.
The point here is that, for his 45 years of blindness, Vincent had been bottom up processing sensory input, (like stages one and two, though maybe in different centers than the visual centers implicated in Fellenz' model). It likely grouped with similar gestalt principles. His top down system learned to attend to the relevant cues in the input, and in combination, these data driven and conceptually driven systems did just what our visual system does: recognize objects as objects. He formed coherent representations of his world, though not visually.
Where I'm going with this, is simply that perhaps the gestalt principles and their neural substrates are described by Fellenz and others solely (or mainly) using visual information, but that does not mean that the gestalt principles apply only to vision, or that they can be found only in visual substrates. It makes sense that since most of us use vision as our primary source of information about the world (and a lot of cortex is devoted to processing it), we associate object recognition, and the gestalt principles primarily with vision, and conceive of other systems as separate and possibly even unrelated. But the gestalt psychologists were not nearly so specific or restrictive in their hypotheses - they set out to explain a general principle of mental organization for all kinds of perception.
Others have demonstrated gestalt principles in other modalities. For example, in the auditory system, the brain completes words, and sounds with gaps, in the same way that we complete visual gaps. When a sentence is heard in an attended ear and then switched to the unattended ear, we unconsciously splice the fragments together as in "good continuation." We segment words and sentences from each other unconsciously. These data suggest that the brain has a flexible "entity recognition" system, able to utilize input from all of sensory modalities.
Perhaps the modalities even cooperate and increase efficiency or accuracy of object recognition. Beatrice de Gelder has provided evidence of cross-modality cooperation. She has shown that when a touch is applied to the body on the same side as a visual cue, vision is enhanced. (7)If tactile sensation helps a subject to recognize an object in space, what might that say about tactile sensation and its link to the biological substrates of object recognition?
What is most unfortunate for Vincent was that those who loved him most, and who competently cared for him in the hospital, missed the point: Vincent didn't need to be "fixed." His blindness kept him from seeing visual input, but his brain, an incredible organ, kept doing what brains do best: integrating the input it got, making representations of the world, so that he could act on it. The moment they gave him sight, he became impaired: his brain had become expert at processing some kinds of sensory input, but it was a novice with vision, and did not know how to handle the information. Not unlike a left handed person trying to write with her right hand after 45 years of left-hand dominance. He eventually decompensated and returned, to his loved ones' disappointment, to the life of a blind man.
What does Vincent's story and Fellenz' research tell us about gestalt and the nature/nurture debate? I think, taken together, it suggests that the brain is built to gather information and synthesize it, in order to make sense of the world, and that experience makes the system expert in some modalities, and in doing so necessarily leaves it relatively inexperienced, even handicapped in others.
2) Conversations with Neil's Brain, by William Calvin
3) More than the Sum of its Parts
4. Psychology, 4th ed. Henry Gleitman. New York: W.W. Norton, 1995.
5)6) Neural Dynamics for Preattentive Perceptual Grouping , by Winfried Fellenz
7) More to Seeing than Meets the Eye , More to Seeing than Meets the Eye
8. Cognitive Neuroscience. Michael Gazzaniga. New York: W.W. Norton, 1998.
| Course Home Page
| Forum | Brain and
Behavior | Serendip Home |