Eye, Brain, and Vision

To speak, as I do here, of separate stages immediately raises our problem
with genealogy. In the retina, as we will see in Chapter 3, the minimum
number of stages between receptors and the output is certainly three, but
because of two other kinds of cells, some information takes a more diverted
course, with four or five stages from input to output. For the sake of conven-
ience, the diagram ignores these detours despite their importance, and makes
the wiring look simpler than it really is. When I speak of the retinal ganglion
cells as "stage 3 or 4", it's not that I have forgotten how many there are.
To appreciate the kind of transfer of information that takes place in a net-
work of this kind, we may begin by considering the behavior of a single retinal
ganglion cell. We know from its anatomy that such a cell gets input from
many bipolar cells—perhaps 12,100, or 1000—and that each of these cells is in
turn fed by a similar number of receptors. As a general rule, all the cells
feeding into a single cell at a given stage, such as the bipolar cells that feed into
a single retinal ganglion cell, are grouped closely together. In the case of the
retina, the cells connected to any one cell at the next stage occupy an area i to
2 millimeters in diameter; they are certainly not peppered all over the retina.
Another way of putting this is that none of the connections within the retina
are longer than about i to 2 millimeters.
If we had a detailed description of all the connections in such a structure and
knew enough about the cellular physiology—for example, which connections
were excitatory and which inhibitory—we should in principle be able to de-
duce the nature of the operation on the information. In the case of the retina
and the cortex, the knowledge available is nowhere near what we require. So
far, the most efficient way to tackle the problem has been to record from the
cells with microelectrodes and compare their inputs and outputs. In the visual
system, this amounts to asking what happens in a cell such as a retinal ganglion
cell or a cortical cell when the eye is exposed to a visual image.
In attempting to activate a stage-3 (retinal ganglion) cell by light, our first
instinct probably would be to illuminate all the rods and cones feeding in, by
shining a bright light into the eye. This is certainly what most people would
have guessed in the late 1940s, when physiologists were just beginning to be
aware of synaptic inhibition, and no one realized that inhibitory synapses are
about as plentiful as excitatory ones. Because of inhibition, the outcome of any
stimulation depends critically on exactly where the light falls and on which
connections are inhibitory and which excitatory. If we want to activate the
ganglion cell powerfully, stimulating all the rods and cones that are connected
to it is just about the worst thing we can do. The usual consequence of stimu-
lating with a large spot of light or, in the extreme, of bathing the retina with
diffuse light, is that the cell's firing is neither speeded up nor slowed down—in
short, nothing results: the cell just keeps firing at its own resting rate of about
five to ten impulses per second. To increase the firing rate, we have to illumi-
nate some particular subpopulation of the receptors, namely the ones con-
nected to the cell (through bipolar cells) in such a way that their effects are
excitatory. Illuminating only one such receptor may have hardly any detecta-
ble effect, but if we could illuminate all the receptors with excitatory effects,
we could reasonably expect their summated influences to activate the cell—
and in fact they do. As we will see, for most retinal ganglion cells the best
stimulus turns out to be a small spot of light of just the right size, shining in
just the right place. Among other things, this tells you how important a role
inhibition plays in retinal function.

VOLUNTARY MOVEMENT
Although this book will concentrate on the initial, sensory stages in
the nervous system, I want to mention two examples of movement, just to
convey an idea of what the final stages in the diagram on page 24 may be
doing.
Consider first how our eyes move. Each eye is roughly a sphere, free to
move like a ball in a socket. (If the eye did not have to move it might well have
evolved as a box, like an old-fashioned box camera.) Each eye has six extraocu-
lar muscles attached to it and moves because the appropriate ones shorten.
How these muscles all attach to the eye is not important to us here, but we can
easily see from the illustration that for one eye, say the right, to turn inward
toward the nose, a person must relax the external rectus and contract the
internal rectus muscles. If each muscle did not have some steady pull, or tone,
the eye would be loose in its socket; consequently any eye movement is made
by contracting one muscle and relaxing its opponent by just the same amount.
The same is true for almost all the body's muscle movements. Furthermore,
any movement of one eye is almost always part of a bigger complex of move-
ments. If we look at an object a short distance away, the two eyes turn in; if
we look to the left, the right eye turns in and the left eye turns out; if we look
up or down, both eyes turn up or down together.
All this movement is directed by the brain. Each eye muscle is made to
contract by the firing of motor neurons in a part of the brain called the brain-
stem. To each of the twelve muscles there corresponds a small cluster of a few
hundred motor neurons in the brainstem. These clusters are called oculomotor
nuclei. Each motor neuron in an oculomotor nucleus supplies a few muscle
fibers in an eye muscle. These motor neurons in turn receive inputs from other
excitatory fibers. To obtain a movement such as convergence of the eyes, we
would like to have these antecedent nerves send their axon branches to the
appropriate motor neurons, those supplying the two internal recti. A single
such antecedent cell could have its axon split, with one branch going to one
oculomotor nucleus and the other to its counterpart on the other side. At the
same time we need to have another antecedent nerve cell or cells, whose axons
have inhibitory endings, supply the motor neurons to the external recti to
produce just the right amount of relaxation. We would like both antecedent
sets of cells to fire together, to produce the contraction and relaxation simulta-
neously, and for that we could have one master cell or group of cells, at still
another stage back in the nervous system, excite both groups. This is one way
in which we can get coordinated movements involving many muscles
Practically every movement we make is the result of many muscles con-
tracting together and many others relaxing. If you make a fist, the muscles in
the front of your forearm (on the palm side of the hand) contract, as you can
feel if you put your other hand on your forearm. (Most people probably think
that the muscles that flex the fingers are in the hand. The hand does contain
some muscles, but they happen not to be finger flexors.) As the diagram on
this page shows, the forearm muscles that flex the fingers attach to the three
bones of each finger by long tendons that can be seen threading their way
along the front of the wrist. What may come as a surprise is that in making a
fist, you also contract muscles on the back of your forearm. That might seem
quite unnecessary until you realize that in making a fist you want to keep your
wrist stiff and in midposition: if you flexed only the finger flexor muscles,
their tendons, passing in front of the wrist, would flex it too. You have to
offset this tendency to unwanted wrist flexion by contracting the muscles that
cock back the wrist, and these are in the back of the forearm. The point is that
you do it but are unaware of it. Moreover, you don't learn to do it by attend-
ing 9 A.M. lectures or paying a coach. A newborn baby will grasp your finger
and hold on tight, making a perfect fist, with no coaching or lecturing. We
presumably have some executive-type cells in our spinal cords that send excit-
atory branches both to finger flexors and to wrist extensors and whose func-
tion is to subserve fist making. Presumably these cells are wired up completely
before birth, as are the cells that allow us to turn our eyes in to look at close
objects, without thinking about it or having to learn.