Eye, Brain, and Vision
Home
Book
Illusions
Biography
Publications
David Hubel's
Home
Book
Illusions
Biography
Publications
HMS
 
     
                                                                                    
   
 

inhibitory input from another set, as in the diagrams on the two preceding pages, or by convergent input from many end-stopped simple cells.
The optimal stimulus for an end-stopped cell is a line that extends for a certain distance and no further. For a cell that responds to edges and is end stopped at one end only, a corner is ideal; for a cell that responds to slits or black bars and is stopped at both ends, the optimum stimulus is a short white or black line or a line that curves so that it is appropriate in the activating region and inappropriate—different by 20 to 30 degrees or more—in the flanking regions, as shown in the diagram of the curved contour on this page.
We can thus view end-stopped cells as sensitive to corners, to curvature, or to sudden breaks in lines.



                                     THE IMPLICATIONS                               OF SINGLE-CELL PHYSIOLOGY                                          FOR PERCEPTION
The fact that a cell in the brain responds to visual stimuli does not guarantee that it plays a direct part in perception. For example, many structures in the brainstem that are primarily visual have to do only with eye movements, pupillary constriction, or focusing by means of the lens. We can nevertheless be reasonably sure that the cells I described in this chapter have a lot to do with perception. As I mentioned at the outset, destroying any small piece of our striate cortex produces blindness in some small part of our visual world, and damaging the striate cortex has the same result in the monkey. In the cat things are not so simple: a cat with its striate cortex removed can see, though less well. Other parts of the brain, such as the superior colliculus, may play a relatively more important part in a cat's perception than they do in the primate's. Lower vertebrates, such as frogs and turtles, have nothing quite like our cortex, yet no one would contend that they are blind.
We can now say with some confidence what any one of these cortical cells is likely to be doing in response to a natural scene. The majority of cortical cells respond badly to diffuse light but well to appropriately oriented lines. Thus for the kidney shape shown in the illustration on the next page, such a cell will fire if and only if its receptive field is cut in the right orientation by the borders.
Cells whose receptive fields are inside the borders will be unaffected; they will continue to fire at their spontaneous rate, oblivious to the presence or absence of the form.
   This is the case for orientation-specific cells in general. But to evoke a response from a simple cell, a contour must do more than be oriented to match the optimum orientation of the cell; it must also fall in the simple cell's receptive field, almost exactly on a border between excitation and inhibition, because the excitatory part must be illuminated without encroachment on the inhibitory part. If we move the contour even slightly, without rotating it, it will no longer stimulate the cell; it will now activate an entirely new population of simple cells. For complex cells, conditions are much less stringent because whatever population of cells is activated by a stimulus at one instant will remain unchanged if the form is moved a small distance in any direction without rotation. To cause a marked change in the population of activated complex cells, a movement has to be large enough for the border to pass entirely out of the receptive fields of some complex cells and into the fields of others. Thus compared to the population of simple cells, the population of activated complex cells, as a whole, will not greatly change in response to small translational movements of an object.
Finally, for end-stopped cells, we similarly find an increased freedom in the exact placement of the stimulus, yet the population activated by any form will be far more select. For end-stopped cells, the contour's orientation must fit the cell's optimum orientation within the activating region but must differ enough just beyond the activating region so as not to annul the excitation. In short, the contour must be curved just enough to fit the cell's requirements, or it must terminate abruptly, as shown in the diagram of the curve (page 21).
One result of these exacting requirements is to increase efficiency, in that an object in the visual field stimulates only a tiny fraction of the cells on whose receptive fields it falls. The increasing cell specialization underlying this efficiency is likely to continue as we go further and deeper into the central nervous system, beyond the striate cortex. Rods and cones are influenced by light as such. Ganglion cells, geniculate cells, and center-surround cortical cells compare a region with its surrounds and are therefore likely to be influenced by any contours that cut their receptive fields but will not be influenced by overall changes in light intensity. Orientation-specific cells care not only about the presence of a contour but also about its orientation and even its rate of change of orientation—its curvature. When such cells are complex, they are also sensitive to movement. We can see from the discussion in the last section that movement sensitivity can have two interpretations: it could help draw attention to moving objects, or it could work in conjunction with microsaccades to keep complex cells firing in response to stationary objects.
I suspect light-dark contours are the most important component of our perception, but they are surely not the only component. The coloring of objects certainly helps in defining their contours, although our recent work tends to emphasize the limitations of color in defining forms. The shading of objects, consisting of gradual light-dark transitions, as well as their textures, can give important clues concerning shape and depth. Although the cells we have been discussing could conceivably contribute to the perception of shading and texture, we would certainly not expect them to respond to either quality with enthusiasm. How our brain handles textures is still not clear. One guess is that complex cells do mediate shades and textures without the help of any other specialized sets of cells. Such stimuli may not activate many cells very efficiently, but the spatial extension that is an essential attribute of shading or texture may make many cells respond, all in a moderate or weak way. Perhaps lukewarm responses from many cells are enough to transmit the information to higher levels.
Many people, including myself, still have trouble accepting the idea that the interior of a form (such as the kidney bean on the facing page) does not itself excite cells in our brain—that our awareness of the interior as black or white (or colored, as we will see in Chapter 8) depends only on cells sensitive to the borders. The intellectual argument is that the perception of an evenly lit interior depends on the activation of cells having fields at the borders and on the absence of activation of cells whose fields are within the borders, since such activation would indicate that the interior is not evenly lit. So our perception of the interior as black, white, gray, or green has nothing to do with cells whose fields are in the interior—hard as that may be to swallow. But if an engineer were designing a machine to encode such a form, I think this is exactly what he would do. What happens at the borders is the only information you need to know: the interior is boring. Who could imagine that the brain would not evolve in such a way as to handle the information with the least number of cells?
After hearing about simple and complex cells, people often complain that the analysis of every tiny part of our visual field—for all possible orientations and for dark lines, light lines, and edges—must surely require an astronomic number of cells. The answer is yes, certainly. But that fits perfectly, because an astronomic number of cells is just what the cortex has. Today we can say what the cells in this part of the brain are doing, at least in response to many simple, everyday visual stimuli. I suspect that no two striate cortical cells do exactly the same thing, because whenever a microelectrode tip succeeds in recording from two cells at a time, the two show slight differences—in exact receptive field position, directional selectivity, strength of response, or some other attribute. In short, there seems to be little redundancy in this part of the brain.
How sure can we be that these cells are not wired up to respond to some other stimulus besides straight line segments? It is not as though we and others have not tried many other stimuli, including faces, Cosmopolitan covers, and waving our hands. Experience shows that we would be foolish to think that we had exhausted the list of possibilities. In the early 1960s, just when we felt

   
 
 
Contents
Next Page

 
 
 
 
 


For an end-stopped cell such as the one shown on the previous page, a curved border should be an effective stimulus.
How arc cells in our brain likely to respond to some everyday stimulus, such as this kidney-shaped uniform blob? In the visual cortex, only a select set of cells will show any interest.
Next Page
Contents