|
Chapter 9
Acting Mind
Now we shift our emphasis from perception to
action, to consider the nature of the body movements that make up our behaviors.
This chapter starts by noting further how our minds, body actions, and environment
are linked in a seamless whole and then considers how mind and consciousness
might be defined from the point of view of movement. We next review some aspects
of our experience of movement of which we are usually unaware, using several
simple exercises to sense the hierarchical and habitual nature of the neuronal
control of our motion, exercises that can also be used to refine and change
that control. A further section takes a brief look at the neuronal pathways
underlying action, and a final section considers how movement might be regulated
by a Darwin Machine mechanism.
Action---The Interface of Mind and Environment
As we go about our daily lives, a significant
fraction of our experience consists of preoccupation with the thoughts flowing
through our heads, the chatter of our internal discourse. Occasionally, this
noise may diminish, as though the brain is taking a brief time-out from constantly
talking to itself. At such times, we can quietly notice the rest of our body
and its experience as we move through our activities. Do you ever feel as though
an "it" rather than an "I" is doing the moving? And do
you ever wonder if this is what it would be like to be an animal, not self-conscious,
moving through its terrain? In this state of mind, you probably notice many
small signals from your body, and also from the environment, that you usually
ignore. When you pass either an attractive or a menacing person, subtle somatic
episodes or emotions associated with approach or avoidance appear and then
disappear. If the menacing person suddenly says, "Got any spare change?" you
note a more obvious tensing of the whole body and an increase in heartrate,
as though it is preparing for flight. These multiple body flickerings are
present at all times, but normally you tune them out.
Such experiences reflect the processes going
on as movement of your body leads to construction of a representation of the
external world from the perturbations caused as that world touches it and showers
it with odors, sounds, and light. It is easy to lose sight of this embodiment
of our brain-mind because we spend most of the hours of our waking consciousness
living in our heads---thinking about, reading about, and talking about ideas.
Many of us usually tune out awareness of what is going on in the rest of our
bodies as we move about and as our feelings change. The brain, however, does
not tune out. Outside the narrow spotlight of our awareness, a massive and
constant stream of information flows back and forth between brain and body.
The global state of parasympathetic/sympathetic activation in the peripheral
autonomic nervous system that innervates our viscera is being monitored, and
the outcome affects the emotional balance between approach and avoidance in
every passing moment.
***************
DESIGN NOTE: IMPORTANT POINT
It is our whole organism---body and brain---that
interacts with our environment. Mind is about the rest of the moving body,
about what to do with it, about where to move next.
***************
The image conjured up by science fiction writers
and philosophers of a "brain in a vat," isolated but otherwise identical
to the normal brain, is unrealistic in more than just the obvious ways. The
vat makers would have to reproduce the subtle and myriad sensory inputs to
this brain, its responding actions on the body, and the brain's subsequent
sensing of the effects of those actions---a formidable proposition! It is unlikely
that there are autonomous symbolic brain structures that do not require, or
are not regulated by, the exchange of information with the rest of the body.
The body is an essential component of the content of our mind.
Acting on and perceiving the environment require
constant modifications of body musculature. Goal-directed movements are constantly
checked and refined as sensory receptors monitor their progress. The body is
constantly changing itself in accordance with the visual world that is present
and the actions being contemplated in it, part of a massive background calculation
that yields each moment's answer to the question "Do I like or dislike,
approach or avoid this situation?" Perceiving is thoroughly about acting,
and acting is completely about perceiving, a rich relationship that can be
obscured by the arbitrary separation of the two in most accounts and textbooks. The
self we all experience as we move through a day requires a continually reconstructed
or updated image of our perceptions, feelings, and actions. What we think is
going on "out there" is not just the photograph-like image we are
most aware of. It is also a subtle symphony of brain representations of our
viscera, our moving muscles, and the biochemical regulators of our hypothalamus.
Our minds are ensembles of physiological operations
linking the muscular, endocrine, immune, and nervous systems (central, peripheral,
voluntary, and autonomic). This ensemble that has evolved to guide action specific
to our species (over geological time) and our culture (over the lifetime of
the individual). Our mind-brain must be defined in a way that considers the
whole organism, its movement, and its feelings. It
must also be described in terms of ongoing reciprocal interactions with the
environment. Our actions change our environment, and those environmental changes
in turn affect our actions in a continuous loop. This is the heart of cognition,
and the explicit data storage and logical manipulation that are the focus of
much cognitive neuroscience play the role of supporting this reciprocal interaction.
Mind is much more than a filing cabinet or a logic machine. And
we need to keep all of this in mind as we try to achieve an understanding of
action or movement sequences, whether spontaneous or learned.
A Movement-Oriented View of Mind and Consciousness.
Starting with the first twitches of a developing
embryo, the constant currency of animal life is action and movement. The innate
movements of the embryo are the earliest building blocks of motor behavior.
Those movements begin before the development of the nervous system that will
ultimately control them, and they help shape the formation of that system.
It is through testing the appropriateness of action that motor and sensory
pathways develop and are coordinated. Most textbooks offer a description of
the sort shown in part (a) of Figure 9-1 to represent the basic relationship
between acting and sensing. A linear path from sensing to acting is portrayed,
and the effect of action on subsequent perception, shown by the dashed line,
is not emphasized.
Figure 9-1
(a) This conventional block diagram of body
function emphasizes a one-way sequence from perception through analysis to
action. (b) A more appropriate diagram shows a circular and continuous loop:
Acting on the environment perturbs it, and perceiving the effect of that perturbation
informs further action.
We could also present the sequence, in a subtly
different but perhaps more useful way, as a circle or cycle wherein any point
could be the start. For example, let's turn it around, as in part (b) of Figure
9-1, to make movement the primary step. Perception of the resulting perturbation
of the environment then informs subsequent movement. In this view, we might
consider consciousness a mechanism by which an organism initiates movement
to acquire information about its present and past environment. The
perceived unity of our consciousness may have something to do with the fact
that we are constantly generating coordinated movement patterns in the face
of diverse sensory stimuli. Thus, in seeking the physical location of the neural
correlates of consciousness, we might do well to look closely at the higher
motor levels, located in frontal cortex and cingulate cortex.
Thinking can be viewed as the activity of deciding "what
movements I make next." The most complex brains are found in animals that
make the most complex motions. The fact that motor activity is commonly assumed
to be a lower cortical function than those subserving "pure" thought
reflects the tradition, found in many cultures, of separating mind and body
and assuming that the latter is somehow lesser. But how can we say that the
performance of an acrobat is less "intellectual" than that of a mathematician?
Earlier in this century, investigators seriously discussed---but later dismissed---a
motor theory of thought, which suggested that conscious thought correlated
with brain activity underlying actual or imaged movement. Now
modern brain imaging has revealed that conscious planning and motor imagery
engage the same brain areas, a result that makes such ideas more plausible
(see below).
***************
DESIGN NOTE: IMPORTANT POINT
It is possible to view mental activity as a
means to the end of executing actions, rather than depicting motor activity
as something subsidiary, designed to satisfy the demands of the higher centers.
***************
Kinesthetic Intelligence
An evolving kinesthetic intelligence may have
led to our narrative self consciousness. The neurophysiologist William Calvin
suggests that evolution of our motor skills, as in making and using tools and
spears, established sequence-handling machinery in the premotor cortex that
became Broca's area, which is crucial for generating the sequences of syllables
used in talking to oneself and others. Speaking
is itself an intricate physical feat. From the point of view of the premotor
cortex, articulating long strings of syllables might be a lot like climbing
a rock face or pitching a baseball---or throwing a spear. Our fundamental kinesthetic
experiences of up-down, right-left, inside-outside might provide a physical
basis for the development of symbols and metaphors used in language (see Chapter
11 for more on this topic).
Consider the possibility that many of your conscious
thoughts are associated with subtle acts of moving and sensing your musculature.
Some experienced meditators report this to be the case. (Can you really think
well without tensing your neck? Can you do math problems lying down?) We all
know that feedback from muscle tension and limb position can affect our conscious
feelings. Forming our facial muscles into the shape of a smile can actually
lighten our mood, and making a frown can do the opposite. Calvin mentions the
account of Oliver Sacks, who reported having his shoulder muscles electrically
stimulated to generate the gesture of a shrug---whereupon he felt as though
he wanted to shrug his shoulders to express nonchalance, as in "So?"
Action Repertoires
The consequence of a long developmental history
that teaches effective patterns of acting and sensing is the construction of
an elaborate hierarchy of neural agents that filter and process both output
and input. Our brains are an archive of images or representations of both perceiving
and acting, accumulated during our developmental history. The
activity of the brain reflects expectations set up by years of experience.
We have a large array of motor repertoires---movement patterns built up from
simpler subunits that we automatically execute, altering them through feedback
if an intended action was not accomplished. Luria,
the famous Russian psychologist, has used the term kinetic melodies to describe
such movement patterns. Many of us have such automatic routines for dressing,
unlocking a door, or moving through a familiar house that we can be completely
unaware of carrying out the actual movement sequences while we daydream about
something else.
Parallel Actions
With training, we can split ourselves into parallel
steams of activity, becoming more muted versions of the split-brain patients
mentioned in Chapter 7. One example of parallel streaming is the "automatic
writing" that became a fad early in the 20th century. Gertrude Stein,
working in William James's laboratory, trained herself to write down dictated
words while reading something else. She recalled the reading but recalled nothing
of the words she had written down by dictation. A similar kind of parallel
processing happens in kinesic communication, or body language, which we discussed
briefly in Chapter 5. We can be aware of, and consciously respond to, the verbal
content of a discussion with someone, while at the same time unconsciously
reacting to signals sent by that person's posture and tone of voice, and all
the time we are making equally unconscious movements of our own. During acting
and sensing, we are conscious of only a minute fraction of the total information
flow. If this were not the case, we would be overwhelmed---like someone who
tried to watch everything on 50 different cable TV channels at once or to read
every message going over the Internet.
Movement Exercises That Reveal Underlying Mechanisms
You can become partly aware of some of the complex
layers of activity underlying an action---some aspects of the kinetic melodies
that underlie our daily acts. This requires that you be willing to slow down
enough to engage the subtle sensing requested in the exercises that follow.
You can experience what we see reflected by modern imaging and electrical recording
experiments that monitor the brain's activity during the planning and execution
of discrete actions. The first exercise deals with the planning that goes on
before movement actually is executed. Both imagining and initiating a movement
happen over a time scale of about a third of a second.
********
DESIGN NOTE: SELF-EXPERIMENT
This exercise works best if someone is giving
you these instructions, but you also can do it alone. Sit quietly with eyes
your closed, letting both arms rest on the arms of your chair or on a desk.
Try to relax your arms completely. Now imagine that someone is going to ask
you to lift your right arm gently. Think about lifting your right arm, but
don't actually do it. Then stop, and completely relax your right arm. Do you
feel the "letting go" of the muscles you had already tensed slightly,
the whole configuration you had prepared for lifting? It is really all there,
planned before the gross movement occurs. Now repeat the experiment and actually
go ahead and lift your arm. What do you feel?
*********
MRI imaging shows that when subjects are asked
to think about moving a hand, the premotor cortex and other areas adjacent
to the motor cortex become most active; this indicates that there is a distinction
between parts of the brain that prepare for movements and parts that carry
them out. Just as imagining an object engages the same areas of visual cortex
that actually viewing the object activates, so motor images engage regions
that are active during actual movement. Recent
work has shown that motor imagery and motor execution involve activation of
very similar cerebral structures at all stages of motor control, except that
the final motor output is not expressed during motor imagery.
A second exercise deals with noting the various
components of a movement pattern in more detail. Many of our habitual movement
patterns contain extraneous contractions, and you can learn about this in an
interesting way. You can sometimes sense more about what goes into a movement
pattern by imagining that pattern than by actually carrying it out.
*******
DESIGN NOTE: SELF-EXPERIMENT
Repeat the previous arm-lifting exercise, but
don't do the actual movements. Close your eyes and imagine yourself lifting
your arm from the surface it is resting on. Gently note and feel all the muscles
getting ready to lift the arm. Now note your neck muscles. Do you really need
to tense them to lift your arm? No. The muscles that contract to lift your
arms are down in your chest and back; the neck muscles are not needed. Imagine
relaxing your neck while you lift your arm. Now, slowly, actually do the movement.
***********
The subtle sensing that you are using in these
exercises is at the core of some training techniques used by musicians, dancers,
and athletes to refine their movement---techniques that tell us much about
the relationship between moving and sensing. They
also enable us to detect chronic inappropriate movement patterns and retrain
ourselves to move more efficiently. A further trick is to go beyond imagining
a movement and to perform it very slowly and very gently (so that feedback
can register to shape, in turn, the central program that is reading out). This
works because our sensory receptors are most sensitive to stimuli of low to
medium intensity. If a motion is made in a gross or more strenuous manner,
the receptors don't report what is happening so effectively. In slow, gentle
movements, one can sense and play with new configurations, instead of locking
into a habitual pattern. If a task can be done by a group of, say, four muscles,
but our habit is to let one muscle strongly predominate over the others, the
brain listens to feedback from this one and ignores the rest. If appropriate
use and interaction of these muscles are to be sensed, then small movements
must be employed so that one senses tonus and stretch from all more or less
equally. You can feel this by trying the following experiment.
*********
DESIGN NOTE: SELF-EXPERIMENT
Press your hand down hard on the arm of your
chair, and note how much you can feel. Now release and shake out your arm and
hand. Now press in the same way, but begin very gently and slowly---and see
how much more you sense.
**********
These exercises---imagining a movement or carrying
it out very gently and slowly---can be a basis for retraining muscle movement
patterns, for laying down new procedural memories and their corresponding long-term
changes in the wiring of our brains. This is relevant to more people than just
dancers and athletes. As we age and begin to have more difficulty jumping over
fences and getting into the back seats of compact cars, it is in part because
we waste so much energy and effort. We usually learn movement patterns when
we are younger and more energetic, and we refine them only until they are "good
enough." Only later do we discover that they really don't work very well.
Retraining to refine our inefficient movement patterns can enable us to enjoy
our customary activities for a longer time.
Neuronal Pathways Underlying Action
How are appropriate action and sensing linked?--by
specialized committees of neurons on many levels of organization, from the
very simple to the very complex. Committees of such agents in the spinal cord
can direct walking in the absence of the brain, but the higher centers are
required for purposeful action. These committees, which often have overlapping
memberships, can organize themselves if they are given some feedback about
how well they are doing. Acting and sensing are linked by many feedback loops.
Motor nerves descending from brain to spinal cord send branches to ascending
sensory pathways, and these branches serve to adjust sensory bias or communicate
an expected sensory input from the about-to-be-ordered movement so that it
can be compared to what actually happens. This is the system that permits motor
learning by trial and error. Think
about learning how to write when you were very young. Can you remember how
hard it was to get your hand to make fine motor movements---how your letters
were very large and crude at first? Years of variation (trying different muscle
movements) and then selection (noting which movements resulted in the best
and smallest letters), went into establishing the stereotyped movements you
now use when you write a letter by hand.
To exert motor control, our brains approximately
reverse the transformations that we discussed in Chapter 8. Instead of building
up an image of what is outside, we "realize," or carry out, the image
of a desired action on the outside world. This
process has to be instructed by the "what" and "where" systems
of the temporal and parietal lobes mentioned in Chapter 8. These
systems define the world in which the movement is occurring. World-centered
and person-centered coordinates of movement must be established on the basis
of maps of the environment in the parietal cortex and the hippocampus, which
are themselves built from sensory input. These coordinates are used by the
supplementary motor areas and the premotor cortex as they prepare instructions
for the primary motor cortex and other movement centers in the brain. It is
important to distinguish movement kinematics, the sequence of positions that
a limb is expected to occupy at different times, from movement dynamics, the
actual control of the movement plan as it is carried out under differing environmental
conditions.
***************
DESIGN NOTE: IMPORTANT POINT
Whereas perception builds up an internal representation
of the external world, action starts with an internal image of the desired
outcome of a movement and then proceeds to execute that movement.
***************
It is presently thought that the supplementary
motor area contains an action programming subsystem that sends coordinates
of each component movement, in the proper order, to an instruction generation
subsystem associated with the premotor cortex, which then instructs a movement
execution subsystem of the primary motor cortex. The
location of these cortical areas is shown in Figure 9-2. The complexity of
these processes is revealed by the variety of apraxias (breakdowns in skilled
movements) that can be caused by brain lesions. Patients with ideomotor apraxia
are able to perform a sequence of movements, but they have difficulty with
the individual components. Individuals with ideational apraxia can perform
an activity such as lighting a candle if the lights go out, but they are unable
to follow an instruction to do so. Disorders sometimes focus on specific activities,
as in dressing apraxia and oral apraxia (inability to make skilled gestures
with facial muscles.) The agraphias, a series of specific disruptions in the
ability to write, can occur in the presence or absence of the ability to read.
Some of these disorders reflect problems in producing the movements per se,
and others are caused by problems in language processing.
Figure 9-2
Cortical areas involved in the planning and
execution of movement. Imaging experiments reveal that very simple habitual
movements activate mainly the primary motor cortex. Activity in the supplementary
motor cortex increases during more complex movements, whether they are actually
carried out or are just mentally rehearsed without execution.
A massive cable of axons carries instructions
from motor cortex to motor neurons in the spinal cord that direct fine movements
of our limbs, hands, and fingers. This direct pathway from cortex to the spinal
cord appears first in mammals, and increasingly detailed motor and sensory
maps of the body appear in the somatosensory and motor cortexes of higher mammals
and primates. The human cortex differs strikingly from that of the chimpanzee
in that much larger sensory and motor areas are devoted to the face and hands.
Dominance of our motor activity by one cortex, usually the left, is much more
striking than in monkeys and apes. We also go to greater lengths in employing
one arm and hand, usually the left, to support and provide context for fine
movements, while the other hand, usually the right, carries them out.
The cortical system for voluntary movements
is layered on top of two other levels of movement control: descending systems
of the brainstem and the spinal cord. Medial (toward the center) brainstem
pathways are important in combining somatosensory, visual, and vestibular (balance)
information to control posture and the axial core muscles that must position
themselves to support movement of the limbs. Lateral (toward the edges) brainstem
pathways control distal muscles of the limbs and are important for goal-directed
action. Thus it is the medial pathways that are most involved when your posture
is automatically adjusting itself against gravity as you walk or turn, and
the lateral pathways come into play when you reach out and pluck fruit off
a tree. The spinal cord, the lowest level of motor control, contains local
circuits that carry out rhythmic motor patterns, such as walking or running,
and also reflexive withdrawals from aversive stimuli. These local spinal circuits
are modulated by the higher centers.
The three-level hierarchy of cortex, brainstem,
and spinal cord is regulated by two further subcortical systems that report
back to the cortex via the thalamus (see Figure 9-3). These are the basal ganglia
and the cerebellum. The function of these complex structures is not well understood.
The basal ganglia are a complex set of nuclei that receive inputs from all
cortical areas. Until recently it was thought that these nuclei sent information
back mainly to the motor planning areas of the frontal cortex, but recent work
has shown they make specific closed loops with virtually all cortical areas.
Electrical recording and brain imaging studies suggest that basal ganglia structures
are involved in establishing the context for motor behaviors and also some
purely cognitive operations. Activity is observed that corresponds to motivations,
expectations, or rewards. Diseases of the basal ganglia cause motor abnormalities
such as involuntary movements and loss of spontaneous movements. One of the
nuclei, the substantia nigra (Latin for "black stuff") is dark because
its neurons contain a derivative of the neurotransmitter dopamine. It is the
death of these dopaminergic neurons that causes Parkinson's disease. Abnormality
in the dopamine system of another center, the caudate nucleus, appears to be
associated with the faulty neurological braking that causes the chronic tic
disorder Tourette's syndrome.
***************
DESIGN NOTE: IMPORTANT POINT
Although the function of the basal ganglia is
not well understood, several disorders result from abnormalities of these areas.
Parkinson's disease results from the loss of dopaminergic neurons in the substantia
nigra, and Tourette's syndrome can be traced to abnormalities in the caudate
nucleus.
***************
The cerebellum receives sensory input that enables
it to compare actual performance with intended movements and to compensate
when things do not go according to plan. Lesions here disrupt balance and coordination.
A recent interpretation is that the cerebellum is not activated by the control
of movement per se but rather serves as an error-checking device during motor,
perceptual, and cognitive performances, processing sensory data related to
those activities. There is
strong evidence that the cerebellum is involved in motor learning, because
such learning can be blocked by lesions in one of its three main subdivisions. Genetic
manipulation of mice to remove a particular form of the postsynaptic receptor
acted on by glutamic acid can make them deficient in one type of cerebellum-dependent
motor learning. The cerebellum
receives input from the periphery as well as from the sensory and motor cortexes
and sends output mainly to the motor regions of the brainstem and cerebral
cortex. The cerebellum's input and output happen at a fixed rate, 10 times
a second, so we actually move in a somewhat jerky way at about this frequency. There
is accumulating evidence that the cerebellum does much more than just effect
motor coordination and control. Different regions of it become active during
different cognitive tasks (matching verbs with nouns, for example).
Figure 9-3
Modulation of the motor control hierarchy of
cortex, brainstem and spinal cord by subcortical systems. Most cortical regions
send information to both the basal ganglia and the cerebellum. After being
processed by the lower structures, this information is fed back via the thalamus
to those same cortical regions. It may also be sent to other targets. The basal
ganglia seem to deal mainly with the motivation and context for action, whereas
the cerebellum monitors the status of action. Recent experiments implicate
both the basal ganglia and the cerebellum in higher cognitive functions such
as language.
If this is beginning to sound complicated, it
is. We know much less about motor systems than about the visual system (this
is why textbooks typically devote twice as much space to sensation as to motion).
The descriptions of motor pathways that we have, of the last ten or so layers
of synapses before a muscle contraction, are not as detailed (or as simple)
as the description we can give of the first ten or so synapses of sensory input
in the visual system. Recently, however, progress has been made in understanding
the coordination of movement sequences at the cortical level. Electrical
recordings from single neurons in the primary motor cortex of monkeys show
that firing can correlate with several different aspects of a movement: its
direction, its force, or the visual or other stimuli that are used as the instruction
to move. In a monkey trained
to move its arm in the direction of a visual stimulus such as a point of light
or a spiral, the firing of cells that calculate the movement can be detected
just before the actual motion begins. Although
large areas of motor cortex are organized into areas of spatial representation
(legs, arms, torso, head, and so on), finer-scale maps seem to represent combinations
of muscles used in coordinated actions (for example, the muscles of the torso,
shoulder, upper arm, and lower arm that must be activated in a precise and
smooth sequence for the animal to reach out and grasp an object).
Recordings from the ventral premotor cortex
of the monkey have demonstrated the existence of neurons that fire both when
the monkey grasps or manipulates an object and when it observes the experimenter
making a similar action. These neurons, which are sometimes called mirror neurons,
appear to be part of a system that matches observed events to similar internally
generated actions, and thus they form a link between observer and actor. Imaging
studies on humans show corresponding activation of premotor cortical areas
both when an action is executed and when another person is observed carrying
out the same action. These
mimetic phenomena may represent an observation-execution matching system, an
information exchange mechanism that underlies the more complicated mimetic
and linguistic human intelligences that we discussed in Chapter 5.
Movement Generation by Darwin Machines
How might the hierarchy of subroutines that
go into making our movement be regulated and refined, especially those supporting
movements that are very rapid? Reaching, throwing, and kicking are ballistic
movements carried out so fast that there isn't time for feedback from sensory
receptors. The minimum time for a nerve signal to make the round trip from
our arm to our brain and back to our arm is about 110 milliseconds. A dart
throw is over in about 120 milliseconds, so there is no time for us to receive
and process feedback on how things are going. The instructions for the sequential
muscle movements must be stored and then "read out" in some way.
You got a sense of this "get set" storage in the first self-experiment
in this chapter.
The neurophysiologist William Calvin describes
a movement sequence as analogous to a row of railroad cars in a train, where
each car represents a component of the movement. Each component could consist,
for example, of a particular set of nerve cell firings that cause movement
of a subset of the muscles engaged in the total movement sequence. How is the
best movement sequence (arrangement of railroad cars) chosen? Imagine
that many trains are lined up in parallel, each with a slightly different sequence
of cars (movement instructions), as in a railroad yard that contains many parallel
train tracks. (Figure 9-4 illustrates the oversimplified case of three tracks.)
All of these tracks converge on a track that exits from the train yard (the
movement sequence actually performed). Now, let each train be tested for how
effectively its sequence of cars works. Let those trains that worked best duplicate
themselves, displacing some trains that were less effective. Repeat the cycle
until identical trains are running on every track. Finally, put all these tracks
to work together to provide refined control for a series of motions. (Another
analogy might be that of a number of people starting to sing randomly but eventually
coming together into a chorus.) We are essentially describing yet another version
of the Darwin Machine.
***************
DESIGN NOTE: IMPORTANT POINT
Movement may be controlled by the application
to neural networks of a version of the same abstract Darwin Machine that is
used on the longer time scales of species evolution and organismal development.
In all of these contexts, variation and selection act together as though to
attain a goal.
**************
Figure 9-4
A Darwin Machine model for generating and refining
a movement sequence. Let each symbol represent one stage in an overall movement
sequence. (a) These three sequences of symbols represent parallel but slightly
different routes to generating the overall movement. Each sequence is executed
and "graded" in terms of its effectiveness in the current environment
and by comparison with memories of similar past sequences. This grade is indicated
by the number at the left of each sequence. (b) A sequence that is ranked as
less effective can be replaced by a duplicate of a more successful sequence
(arrow). Sometimes a slight alteration that occurs during this duplication
might produce an even more effective sequence. (c) Repetition of this process
many times finally generates a unified "chorus" of parallel and similar
sequences that, when executed together, can control a complicated movement
with great precision. Of course, the brain would use a great many more sequences
than the three illustrated here.
Why do we suggest that there are many trains
carrying out the same sequence of instructions? Because for any one train by
itself, there is a problem. The range within which a nerve cell may fire after
it is triggered is from 1 to 10 milliseconds. Calvin points out that this is
too imprecise even for such an activity as throwing a rock at a stationary
prey 4 meters away. There is an 11-millisecond launch window in which a rock
can be thrown to hit a rabbit at 4 meters. A rock released too early will overshoot;
a rock released too late will land in front of the rabbit. The launch window
shrinks to 1.4 milliseconds if the target is 8 meters away. How do we guarantee
a launch within a 1-millisecond interval using nerve cells that have a 1- to
10-millisecond "jitter"? Part of the answer may be that trains with
identical sequences of cars can emerge simultaneously from the marshaling yard
and have their firing times averaged. When the firings of many motor neurons
are averaged, the motor effect is much more reliable than that elicited by
the firing of any of the individual neurons. (Think of an engineer who wants
to measure the time at which midnight occurs and has a 100 different clocks,
each of which is off by a few seconds. Noting when half of the clocks have
struck midnight gives a much more accurate estimate than listening to just
one.)
Calvin further proposes that a cerebral Darwin
Machine might underlie our ability to carry a small-scale model of external
reality in our heads, all the while continuing to test it against various alternatives,
continuously revising our assessment of which is best, predicting future situations,
and so on. His suggestion is that a random sequencer arranges templates into
parallel strings, each of which is "graded" by memories of how similar
strings performed in the past and weighted for emphasis by the present environment.
Dozens or hundreds of such sequences might be tried out simultaneously, with
the most appropriate new generation shaping up in milliseconds, thus initiating
insightful action without overt trial and error. It is interesting that the
shaping-up or selective mechanisms appear to be damaged in some patients with
frontal lobe lesions. These patients can spin scenarios of action but cannot
easily choose among them.
Ideas that flow through our heads could be selected
as candidates for storage in long-term memory by similar mechanisms. Most of
the candidate sequences would be held in our unconscious short-term memory
and thrown away after use. Transfer to long-term memory would be necessary
to establish the ability to recall such sequences. Only those that survived
a thorough reality test would be retained. The sequences---at least those involved
in the short-term generation of language---might contain about seven chunks
of information each. (We can hold about seven digits, such as a phone number,
in short-term memory long enough to repeat them or dial a phone.) This could
be why, when deciding what to say next, we plan ahead by no more than about
half a dozen chunks, words, or concepts. We usually don't know how our sentences
are going to end when we start them.
Summary
Trying to understand action is a more daunting
task---and probably a more complex one---than we face with perceptions, the
subject of Chapter 8. Our self conscious cognitions rest on a vast background
of moving muscles and viscera that are continually monitored as the organism
moves about in its physical and social environment. The core event of our daily
activities is a continuous feedback cycle in which we initiate movements of
these muscles and viscera to acquire information about, and change, our internal
and external environments, and perception of these changes instructs further
action. Our habitual movements are regulated by a vast repertoire of automatic
movement routines learned during development. These are orchestrated by a complex
array of cortical mechanisms that proceed from an image of an intended action
to its execution. Electrical recordings of brain cell activities and imaging
studies reveal some of building blocks of this movement control---a collaborating
hierarchy that extends from cortex to subcortical, brainstem, and spinal cord
mechanisms. Specific brain lesions can impair different aspects of the sequences
involved in planning, instruction generation, and execution of movement. One
model suggests that the generation and refinement of movement utilize a Darwin
Machine mechanism in which more successful sets of instructions are duplicated
at the expense of less useful ones.
The reasons that an array of nerve cells might
have for firing in the supplementary motor or premotor cortex are much more
numerous than those that affect an array of cells in the visual cortex. The
visual cells, even though they are subject to feedback from higher centers,
have the fairly definite job of analyzing information about visual sensory
stimuli, whereas the motor cells may participate in the resolution of many
decisions about why and how a reaction to a particular visual stimulus is taking
place. Goals, memories, and intentions have more influence on actions than
on perceptions. Further, as we consider the constant feedback between action
and perception that links our brain, body, and world together, yet another
set of actors must be integrated into our description: the array of emotions
that provide rapid response systems and that also bias and pattern many of
our behaviors. These emotions are the subject of the next chapter.
Questions for Thought
1. The first section of this chapter reexamines
the idea of an extended definition of mind, first mentioned in Chapter 1, a
definition encompassing body and environment. This can be contrasted with a
definition that restricts the idea of mind to the activities of the brain.
What are the merits of these two approaches? Which do you think is more useful?
2. This chapter gives several examples of our
ability to carry out several sensory-motor sequences, or kinetic melodies,
at the same time. Some we are conscious of, some not. Please continue the thinking
about selves that we started in Chapter 7 and suggest how you might incorporate
such information into your idea of what a "self" is.
3. Electrophysiological experiments have shown
that nerve cell firing in preparation for a movement can begin slightly before
our conscious awareness of intending to make that movement. How can you reconcile
such information with the assumption that consciousness plays a role in controlling
our actions?
4. Can you think of experiments that might test
the Darwin Machine model that Calvin has suggested underlies complex action
sequences (including sequences of thought)? For example, could the complex
sequence of speaking language provide a context for testing this idea?
Suggestions for Further General Reading
Clark, A. 1997. Being There. Cambridge, MA:
M.I.T. Press. An extended discussion of how mind must be defined in terms of
an acting body and its interactions with the external environment.
Damasio, A.R. 1994. Descartes' Error---Emotion,
Reason, and the Human Brain. New York: Putnam's. The first section of this
chapter draws on Damasio's Chapter 10, "The Body-Minded Brain."
Minsky, M. 1986. The Society of Mind. New York:
Simon & Schuster. Minsky's book has interesting descriptions of the sorts
of agents that are necessary to underlie and support action: hierarchies of
stable subassemblies.
Calvin, W.H. 1990. The Cerebral Symphony. New
York: Bantam. The material on Darwin Machine models of movement is drawn from
Chapters 10 and 11 of Calvin's book.
\eoc\Reading on More Advanced or Specialized
Topics
Crammond, D. 1997. Motor imagery: Never in your
wildest dream. Trends in Neuroscience 20:54--57. This article describes experiments
that monitor brain activity during the imagining of movement as well during
its actual execution.
Gazzaniga, M.S., ed. 1995. The Cognitive Neurosciences.
Cambridge, MA: M.I.T. Press. Part IV of this volume is devoted to motor control.
Bear, M., Connors, B., & Paradiso, M. 1996.
Neuroscience: Exploring the Brain. Baltimore, MD: Williams & Wilkins. Chapters
13 and 14 of this book review spinal cord and brain control of movement.
Kosslyn, S.M., & Koenig, O. 1992. Wet Mind---The
New Cognitive Neuroscience. New York: Free Press. Chapter 7 discusses brain
control of movement.
|
