Selasa, 29 Maret 2016

motor programing brief action

Chapter Outline
Motor Program Theory
Evidence for Motor Programs
Motor Programs and the Conceptual Model
Problems in Motor Program Theory: The Novelty and Storage
Problems
Generalized Motor Program Theory
Summary
160


Chapter Objectives
Chapter 5 describes how motor programs are used in the control of
movement. This chapter will help you to understand
motor control as an open-loop system and the role of motor
programs,
experimental evidence for motor programs,
limitations and problems in the simple motor program concept, and
generalized motor programs and evidence for this expanded
concept.
Key Terms
central pattern generator (CPG)
deafferentation
generalized motor program (GMP)
invariant features
motor program
novelty problem
open-loop control
parameterized
parameters
reflex-reversal phenomenon
relative timing
sensory neuropathy
startle RT
storage problem
surface feature
161


Watching a guitarist pluck a series of notes with blazing speed, or a pianist run
trills up and down the keyboard, reminds us that in many skills, a number of
separate actions can appear in very quick sequence. Yet these separate actions
are produced while maintaining a specific rhythm to the sequence that leaves the
impression of a single, fluid, coordinated motion. How does the skilled musician
produce so many movements so quickly? What controls them, and how are they
combined to form a whole? The skilled musician gives us the impression that
these quick movements might be organized in advance and run off without much
feedback control.
This chapter investigates the idea of open-loop control, introducing the concept
of the motor program as responsible for this kind of movement control. Then the
various feedback pathways discussed in the previous chapter are examined as to
their interaction with motor programs, giving a more complete picture of the
interplay of central and peripheral contributions to movements. The chapter also
focuses on the concept of a generalized motor program (GMP), a theory that
can account for the common observation that movements can be varied along
certain dimensions—for example, playing the guitar or piano sequence slower or
faster (or louder or softer) without sacrificing their underlying structure (i.e., the
rhythm).
In many actions, particularly quick ones produced in stable and predictable
environments (e.g., springboard diving, hammering), most people would assume
that a performer somehow plans the movement in advance and then triggers it,
allowing the action to run its course without much modification or awareness of
the individual elements. Also, the performer does not seem to have much
conscious control over the movement once it’s triggered into action; the
movement just seems to “take care of itself.” Perhaps this is obvious. Certainly
you cannot have direct, conscious control of the thousands of individual muscle
contractions and joint movements—all the degrees of freedom that must be
coordinated as the skilled action is unfolding. There is simply too much going on
for the limited-capacity attentional mechanisms (which we have discussed in
chapters 3 and 4) to control any one of them individually.
If these individual contractions are not controlled directly by processes of which
you are aware, how then are they controlled and regulated? In many ways, this
question is one of the most fundamental to the field of motor behavior because it
goes to the heart of how biological systems of all kinds control their actions. This
162


chapter focuses on the ways the central nervous system is organized functionally
before and during an action and how this organization contributes to the control
of the unfolding movement. As such, this chapter is a close companion to chapter
4, which considered the ways sensory information contributes to movement
production. This chapter adds the idea of centrally organized commands that
sensory information may modify somewhat. First, though, comes the important
concept of a motor program, which is the prestructured set of movement
commands that defines and shapes the movement.
Motor Program Theory
The concept of the motor program, which is central to this entire chapter, is based
on a kind of control mechanism that is in some ways the opposite of the closed-
loop system discussed throughout chapter 4. This type of functional organization
is called open-loop control.
Open-Loop Control
The basic open-loop system is illustrated in figure 5.1, and consists essentially of
two parts: an executive and an effector. This open-loop structure has two of the
main features used in closed-loop control (figure 4.1), but missing are feedback
and comparator mechanisms for determining system errors. Open-loop control
begins with input about the desired state being given to the executive (or
decision-making) level, whose task it is to define what action needs to be taken.
The executive then passes instructions to the effector level, which is responsible
for carrying out these instructions. Once the actions are completed, the system's
job is over until the executive is activated again. Of course, without feedback, the
open-loop system is not sensitive to whether or not the actions generated in the
environment were effective in meeting the goal; and since feedback is not
present, modifications to the action cannot be made while the action is in
progress.
163


Figure 5.1 A basic open-loop system.
This kind of control system can be observed in many different real-world
mechanisms. For example, an open-loop system is used in most traffic signals,
where it sequences the timing of the red, yellow, and green lights that control the
traffic flow. If an accident should happen at that intersection, the open-loop
system continues to sequence the lights as if nothing were wrong, even though the
standard pattern would be ineffective in handling this new, unexpected traffic
flow problem. Thus, the open-loop system is effective as long as things go as
expected, but it is inflexible in the face of unpredicted changes.
A microwave oven is another example of an open-loop system. The user places a
frozen entree in the oven and programs it to defrost for 5 min, and then cooks on
high power for another 2 min. Here, the program tells the machine what
operations to do at each step and specifies the timing of each operation. Although
some microwave ovens are sensitive to the temperature of the item being cooked,
many are not, and these latter machines follow the instructions without any regard
for whether they will result in the desired state (an entree that is ready to eat).
Generally, the characteristics of a purely open-loop control system can be
summarized as follows:
Advance instructions specify the operations to be done, their sequencing,
and their timing.
Once the program has been initiated, the system executes the instructions,
essentially without modification.
There is no capability to detect or to correct errors because feedback is not
involved.
164


Open-loop systems are most effective in stable, predictable environments in
which the need for modification of commands is low.
Motor Programs as Open-Loop Systems
Many movements—especially ones that are rapid, brief, and forceful, such as
kicking and key pressing—seem to be controlled in an open-loop fashion, without
much conscious control once the movement is under way. The performer in these
tasks does not have time to process information about movement errors and must
plan the movement in its entirety before movement initiation. This is quite
different from the style of control discussed in the previous chapter, where the
movements were slower (or longer in time) and were largely based on feedback
processes of various kinds.
Open-loop control seems especially important when the environmental situation
is predictable and stable. Under these circumstances, human movements appear
to be carried out without much possibility of, or need for, modification. This
general idea was popularized more than a century ago by the psychologist
William James (1891) and has remained as one of the most important ways to
understand movement control.
Consider a goal such as hitting a pitched baseball. The executive level, which
consists of the decision-making stages of the system defined in chapter 2,
evaluates the environment in the stimulus identification stage, processing such
information as the speed and direction of the ball. The decision about whether or
not to swing is made in the response selection stage. Then, the movement is
programmed and initiated in the movement programming stage, in which details
about the swing's speed, trajectory, and timing are determined.
Control is then passed to the effector level for movement execution. The selected
motor program now carries out the swing by delivering commands to the spinal
cord, which eventually directs the operations of the skeletal system involved in
the swing. This movement then influences the outcome—resulting in the desired
movement (hitting the ball squarely) or not (e.g., missing the ball, popping the
ball up).
Although the decision-making stages determine what program to initiate and have
some role in the eventual form of the movement (e.g., its speed and trajectory),
165


movement execution is not actually controlled by the conscious decision-making
stages. Therefore, the movement is carried out by a system that is not under direct
conscious control. On this view, the motor program is the agent determining
which muscles are to contract, in what order, when, and for how long (timing).
Practice, which leads to learning skilled actions, is thought of as “building” new,
more stable, more precise, or longer-operating motor programs (or some
combination of these). Initially a program might be capable only of controlling a
short string of actions. With practice, however, the program becomes more
elaborate, capable of controlling longer and longer strings of behavior, perhaps
even modulating various reflexive activities that support the overall movement
goal. These programs are then stored in long-term memory and must be retrieved
and prepared for initiation during the response programming stage.
Open-Loop Control in the Conceptual Model
How does this concept of open-loop control and the motor program fit with the
conceptual model of human performance? Figure 5.2 shows the conceptual model
developed in chapter 4 (figure 4.10), now with the portions highlighted (light-
green shading) that comprise the open-loop components. The conceptual model
can here be thought of as an open-loop control system with feedback added (the
parts not shaded) to produce corrections through the other loops discussed
previously. This more complete conceptual model has two basic ways of
operating, depending on the task. If the movement is very slow or of long duration
(e.g., threading a needle), the control is dominated by the feedback processes. If
the movement is very fast or brief (e.g., a punch or kick), then the open-loop
portions tend to dominate. In most tasks, motor behavior is not either open or
closed loop alone but a complex blend of the two.
166


Figure 5.2 Conceptual model with the open-loop processes highlighted in light
                                   green.
For very fast and or brief actions, the theory of motor programs is useful because
it gives a set of ideas and a vocabulary to talk about a functional organization of
the motor system. If a given movement is said to be “a programmed action,” it
appears to be organized in advance, triggered more or less as a whole, and
carried out without much modification from sensory feedback. This language
describes a style of motor control with central movement organization, where
movement details are determined by the central nervous system and are then sent
to the muscles, rather than controlled by peripheral processes involving
feedback. Of course, both styles of control are possible, depending on the nature
of the task, the time involved, and other factors.
167


Evidence for Motor Programs
A number of separate lines of evidence converge to support the existence of
motor program control. This evidence comes from some rather diverse areas of
research: (a) studies of reaction time in humans; (b) experiments on animals and
case studies involving animals and humans in which feedback has been removed;
(c) the impact on performance when movement is unexpectedly blocked; (d) the
analysis of behaviors when humans attempt to stop or change an action; and (e)
studies of movements initiated by startling stimuli.
Reaction-Time Evidence
In RT experiments, the duration of the RT delay was slowed when more
information needed to be processed (e.g., Hick’s Law), when processing was not
“natural” (e.g., in S-R incompatible situations), and so on. Generally, RT was
determined mainly by the slowness of the stimulus identification and response
selection stages. In this section we review evidence that RT is also influenced by
the factors affecting the movement programming stage.
Response Complexity Effects
Subjects in RT experiments are typically asked to respond to a stimulus by
initiating and carrying out a predetermined movement as quickly as possible (as
discussed in chapter 2). Duration of the RT delay is measured as the interval
from the presentation of the stimulus until the movement begins, so any added
time for the movement itself does not contribute directly to RT. However,
beginning with the work of Henry and Rogers (1960; see Focus on Research 5.1),
many experimenters have shown that RT is affected by several features of the
movement to be performed, presumably by influencing the complexity (and
duration) of the movement programming stage.
Henry and Rogers (1960) found that simple RT was elevated with increases in
the complexity of the movement to be performed after the response was initiated.
This, plus much more research on this important finding since the publication of
Henry and Rogers’ work, has produced the following set of findings (Klapp,
1996):
168


RT increases when additional elements in a series are added to the action
(e.g., a unidirectional forward stroke in table tennis would likely be
initiated with a shorter RT than a backswing plus a forward stroke).
RT increases when more limbs must be coordinated (e.g., a one-handed
piano chord would be initiated with a shorter RT than a two-handed chord).
RT increases when the duration of the movement becomes longer (e.g., a
100 ms bat swing would be initiated with a shorter RT than a 300 ms bat
swing).
The interpretation is that when the to-be-produced movement is more “complex”
in any of these ways (number of elements, number of limbs involved, the overall
duration of the action), RT is longer because more time is required to organize
the motor system before the initiation of the action. This prior organization
occurs, as discussed in chapter 2, in the movement programming stage. The effect
on RT of the nature of the to-be-performed movement provides evidence that at
least some of the action is organized in advance, just as a motor program theory
would expect.
Focus on Research 5.1
The Henry–Rogers Experiment
One of Franklin Henry's many important contributions was a paper that he
and Donald Rogers published in 1960. The experiment was simple, as
many important experiments are. Subjects responded as quickly as
possible to a stimulus by making one of three movements that were
prepared in advance. Only one of these movements would be required for
a long string of trials, so this was essentially a simple-RT paradigm (see
chapter 2). The movements, designed to be different in complexity, were
(a) a simple finger lift, (b) a simple finger lift plus a reach to slap a
suspended ball, and (c) a movement requiring a finger lift followed by
slapping the most-distant ball with the back of the hand, then moving to
the push button, and then grasping the near ball (see Fischman, Christina,
& Anson, 2008, for details; see also figure 5.3).
169


The seated subject would begin with his finger on the release key
(labeled as “D” in Figure 5.3). For the most complex action, the subject
would respond to the stimulus lights by lifting his finger from the release
key, reaching forward to slap the far tennis ball, moving down to push the
button on the base (E), then, finally, reaching forward and upwards to
slap the second tennis ball; for the movement of intermediate complexity,
the subject responded to the stimulus lights by lifting his finger off the
release key, and then reaching to slap the far tennis ball; for the simplest
movement, the subject only had to lift his finger off the release key. Each
of these actions was to be done as quickly as possible.
Figure 5.3 Apparatus used by subjects in the Henry and Rogers (1960)
experiment. Relevant parts are: A = tennis balls; D = release key; E =
                  push button; H = stimulus lights.
Reprinted from Howell 1953.
Henry and Rogers measured the RT to initiate each of these actions—the
interval from the presentation of the stimulus until the beginning of the
required movement. (Remember that the RT does not include the time to
complete the movement itself.) They found that the time to initiate the
movement increased with added movement complexity. The finger-lift
movement (a) had an RT of 150 ms; the intermediate-complexity
movement (b) had an RT of 195 ms; and the movement with two reversals
in direction (c) had an RT of 208 ms.
Notice that in each case, the stimulus to signal the movement (processed
during the stimulus identification stage) and the number of movement
choices (processed during the response selection stage) remained
constant across the different conditions. Thus, because the only factor that
varied was the complexity of the movement, the interpretation was that
the elevated RTs were somehow caused by increased time for movement
170


programming to occur before the action. This notion has had profound
effects on the understanding of movement organization processes and has
led to many further research efforts to study these processes more
systematically (reviewed in Christina, 1992). Most importantly, these
data support the idea that rapid movement is organized in advance, which
is consistent with the motor program concept.
Exploring Further
1. Analyze the differences in actions required for the three movements
   in the Henry and Rogers study. Describe at least three differences in
   the movements’ requirements that might have led to increases in the
   complexity of the motor program.
2. What additional changes could be made to the action requirements
   of the most complex movement (C) that might be expected to
   increase movement programming time?
Startled Reactions
In the previous section we discussed the idea that RT becomes longer with
increases in the “complexity” of the to-be-performed movement. Here, we focus
on research showing that RT can be dramatically shortened under certain
conditions.
We have all been in situations in which a completely unexpected event, such as a
very loud noise or very bright light, caused a severe reaction—we were startled.
The response is often accompanied by contractions in the muscles of the face and
neck and protective movements of the upper limbs. A very interesting property of
the startle response (RT) is that these movements are initiated much faster than
can be accounted for by voluntary responses to a stimulus.
An innovative series of studies involved the startle RT as a paradigm to reveal
insights about movement programming. In these studies (reviewed in Carlsen et
al., 2011; Valls-Solé, Kumru, & Kofler, 2008), the subject is typically asked to
prepare to make a rapid, forceful, sometimes complex response to a moderately
intense stimulus (auditory or visual). Occasionally, the stimulus is accompanied
by an extremely loud acoustic signal (e.g., 130 decibels [dB]; by comparison, the
171


sound of a chainsaw is about 110 dB).The loud acoustic signal usually produces
the typical startle indicators (clenched neck and jaw muscles, among other
reactions). However, what also happens is that the prepared movement is
produced normally, but with an RT that may be up to 100 ms shorter than on the
control trials without the loud stimulus. The pattern of the actions remained
unchanged.
These findings fit quite well into the motor program concept. The idea here is that
the executive has prepared a motor program in advance of the stimulus to
respond, which is normally released by a voluntary, internal “go” signal from the
executive to the effectors. The startle RT has the effect of hastening the release of
this signal, by means of either speeding up the executive’s processing time or
perhaps even bypassing the executive altogether. The research is unclear at this
point as to exactly why the same movement is initiated much faster on startled
trials than on normal, unstartled trials, but the role of the motor program in
carrying out the response is clearly implicated.
Deafferentation Experiments
In chapter 4, we mentioned that information from the muscles, the joints, and the
skin are collected together in sensory nerves, which enter the spinal cord at
various levels. A surgical technique termed deafferentation involves cutting
(via surgery one or more of) an animal’s afferent nerve bundles where they enter
the cord, so the central nervous system no longer can receive information from
some portion of the periphery. The motor pathways are not affected by this
procedure as information about motor activity passes through the (uncut) ventral
(front) side of the cord. Sensory information from an entire limb, or even from
several limbs, can be eliminated by this procedure.
What are experimental animals capable of doing when deprived of feedback from
the limbs? Films of monkeys with deafferented upper limbs reveal that they are
still able to climb around, playfully chase each other, groom, and feed themselves
essentially normally. It is indeed difficult to recognize that these animals have a
total loss of sensory information from the upper limbs (Taub, 1976; Taub &
Berman, 1968). The monkeys are impaired in some ways, however; they have
difficulty in fine finger control, as in picking up a pea or manipulating small
objects. On balance, though, it is remarkable how little impaired these animals
are in most activities.
172


If the movement is quick enough, the motor program controls the entire action; the
movement is carried out as though the performer were deprived of feedback. The
capability to move quickly thus gives additional support to the idea that some
central program handles the movement control, at least until feedback from the
movement can begin to have an effect.
Case studies of humans also support this general conclusion. Lashley (1917)
found that a patient with a gunshot wound to the back, who was without sensory
feedback information from the legs, could still position his knee at a specified
angle without feedback. And individuals who have lost much of their sensory
feedback (so-called sensory neuropathy patients) are able to perform quite well
in their environments as long as visual information is available (Blouin et al.,
1996).
These studies show that sensory information from the moving limb is certainly
not absolutely critical for movement production, and it is clear that many
movements can occur nearly normally without it. This evidence suggests that
theories of movement control must be generally incorrect if they require sensory
information from the responding limb. Because feedback-based theories cannot
account for these actions, many theorists have argued that the movements must be
organized centrally via motor programs and carried out in an open-loop way, not
critically dependent on feedback (e.g., Keele, 1968). In this sense, the
deafferentation evidence supports the idea that movements can be organized
centrally in motor programs.
Central Pattern Generator
The idea of motor programs is similar to that of the central pattern generator
(CPG), which was developed to explain certain features of locomotion in
animals, such as swimming in fish, chewing in hamsters, and slithering in snakes
(Grillner, 1975). A genetically defined central organization is established in the
brainstem or the spinal cord. When this organization is initiated by a brief
triggering stimulus from the brain, sometimes called a command neuron, it
produces rhythmic, oscillating commands to the musculature as if it were defining
a sequence of right–left–right activities, such as might serve as the basis of
locomotion. These commands occur even if the sensory nerves are cut
(deafferented), suggesting that the organization is truly central in origin.
173


An example of a simple network that could account for the alternating flexor–
extensor patterns in locomotion is shown in figure 5.4. Here, the input trigger
activates neuron 1, which activates the flexors as well as neuron 2. Then neuron
3 is activated, which activates the extensors. Neuron 4 is then activated, which
activates neuron 1 again, and the process continues. This is, of course, far too
simple to account for all of the events in locomotion, but it shows how a
collection of single neurons could be connected to each other in the spinal cord to
produce an alternating pattern.
Figure 5.4 A simplified illustration of a central pattern generator.
Reprinted by permission from Schmidt and Lee 2011.
The notion of the CPG is almost identical to that of the motor program. The main
difference is that the motor program involves learned activities that are centrally
controlled (such as kicking and throwing), whereas the CPG involves more
genetically-defined activities, such as locomotion, chewing, and breathing. In any
case, there is good evidence that many genetically defined activities are
controlled by CPGs (Zehr, 2005).
The concept of a central pattern generator is used to describe simple, genetically-
  defined activities such as walking, whereas motor program theory applies to
                      learned skills such as riding a bicycle.
174


Inhibiting Actions
Another line of evidence to support motor program control can be found in
experiments in which subjects are required to inhibit or stop a movement after
having initiated the process of making the action. This is the kind of activity that
one sees quite frequently in baseball batting (see Focus on Application 5.1 on
checked swings). The question asked by researchers concerns the “point of no
return”—at what point after starting the processing stages that lead to a movement
is one committed to making, or at least starting, the action? In other words, at
what point is the signal released to send the motor program to the muscles?
The “stop-signal” paradigm is the method most frequently used to study action
inhibition, and an early contribution to this research was provided by Slater-
Hammel (1960), described in detail in Focus on Research 5.2. The findings of
this study, which involved a very simple finger lift off a key (presumably with
175


little biomechanical delay), suggested that the point of no return occurred about
150 to 170 ms before the time when the movement was initiated. An action such
as a baseball swing has a much longer movement completion time than the finger
lift used by Slater-Hammel. Nevertheless, considerable evidence suggests that a
motor program is released that is responsible for initiating the action in tasks like
this and that serves to carry out the entire action unless a second stop-signal
program is initiated in time to arrest its completion (see Focus on Research 5.2;
also Verbruggen & Logan, 2008).
Focus on Application 5.1
Checked Swings in Baseball
The bat swing in baseball is a good example of a motor program in
action. The typical swing consists of a coordinated action involving a
step with the lead foot toward the oncoming ball, followed by a rapid
rotation of the trunk and shoulders, propelling the bat with a large angular
velocity and minimum overall movement time. There is good reason to
believe that the step and swing are part of a single motor program,
initiated by good batters on almost every pitch; but parts of this swing can
be inhibited before a full execution on many of those pitches. How do
batters do this, and how successful are they at doing it?
The physics of baseball tell us that there is very little time available for a
major league batter to hit a baseball. For pitches in the range of 85 to 95
mph (137 to 153 kph), the ball takes less than a half second (500 ms) to
reach the hitting zone after being released from the pitcher’s hand. The
batter typically prepares for the pitch and may initiate the step before the
pitcher has actually released the ball. And at some point along the way,
usually before the ball reaches the midpoint in its flight toward the plate,
the batter must decide whether to proceed with the swing (including
where to aim the bat for its intended collision with the ball) or inhibit its
execution. The result is four different types of batter responses (see Gray,
2009, for much more on these ideas):
176


1. The batter successfully inhibits the motor program, and the swing is
   never initiated.
2. The batter starts the swing but inhibits the completion of the motor
   program, resulting in the bat stopping before it crosses the plate
   (which defines it as a “nonswing”).
3. The batter starts the swing but fails to inhibit the motor program in
   time, resulting in a slowed velocity of the bat as it crosses the plate
   (resulting in a “completed swing,” according to the rules of
   baseball).
4. The batter starts and completes the motor program without
   attempting to inhibit the swing—a classic example of a completed
   swing.
Focus on Research 5.2
Initiating a Motor Program
Not so long ago, races like the 100 m sprint were timed by hand, with a
stopwatch. The timing judge started her stopwatch when she saw the
smoke of the starter’s pistol and stopped it when the runners crossed the
finish line. But, let’s consider how she stopped her stopwatch. If she
stopped it when she saw the runner cross the line, then the clock would
actually be stopped a short time later, because completing her action
would be delayed by two factors: (1) the amount of time required to send
the motor instructions to the hand holding the watch and (2) the
biomechanical delays in pushing the button.
Our interest, of course, is the first concern—how long does it take to send
the motor instructions? To answer this question, Arthur Slater-Hammel
(1960) conducted an experiment that was similar to the example of the
timing judge just presented. Subjects held a finger on a key while
watching an analog timer moving at one revolution per second; lifting off
the key brought the sweep hand to an instantaneous stop. Subjects were
instructed to lift their finger from the key such that the clock hand would
177


stop at exactly the point marked 800, or roughly at the 10 o-clock position
on the clock (800 refers to a lapse of 800 ms after the start from the 12:00
position; see figure 5.5a). Note that in order to do this task accurately,
they would need to initiate the action at some point before the clock hand
actually reached the 800 position (just as the timing judge would need to
initiate the action of stopping her timer before the sprinter crossed the
finish line).
Figure 5.5 Slater-Hammel’s (1960) task (a) and results (b) .
Reprinted by permission from Schmidt and Lee 2011. Part b data from Slater-Hammel 1960.
An important aspect of the Slater-Hammel study was the insertion of
special (probe) trials that occurred rarely and unpredictably. On these
probe trials the experimenter would stop the clock hand at various
locations before it reached 800. If this happened, the subjects’ job was
simply to keep their finger on the key; thus, the probe trials required an
inhibition of the normal task of lifting the finger to stop the sweep hand.
The rationale was simple—if the motor program had not yet been sent to
the muscles when the sweep hand stopped, then the subject should be
able to inhibit the finger lift successfully. But there would be little chance
of changing such a short, ballistic action if the sweep hand stopped after
178


the motor program to lift the finger that had already been sent.
Slater-Hammel plotted the probability of inhibiting the action
successfully as a function of the time interval between 800 and where the
clock hand had stopped. The data are shown in figure 5.5b. When the
interval before the intended finger lift was relatively large (greater than
210 ms), stopping the clock hand resulted in inhibiting the movement
successfully almost all the time. However, as this interval decreased, the
subjects would lift the finger more and more often, to the point that when
the clock hand stopped at −700 (100 ms before the 800 position), the
subject could almost never inhibit the movement. Generally, when the
clock hand was stopped about 150 to 170 ms before the intended finger
lift, the subject could inhibit the movement successfully about half the
time. This finding can be interpreted to mean that the internal “go” signal
is issued about 150 to 170 ms before the intended action. This “go”
signal is a trigger for action, after which the movement occurs even
though new information indicates that the movement should be inhibited.
Exploring Further
1. Slater-Hammel’s estimate of the time required to anticipate the
   sweep hand’s arrival at the 800 position is complicated by the fact
   that subjects had a +26 ms constant error (CE) on the normal trials.
   What are the implications of this positively biased CE?
2. How could this stop-signal paradigm be adapted to examine the time
   required to make anticipatory actions in sport tasks such as batting a
   baseball?
Muscle Response Patterns
The final line of evidence supporting motor program control comes from
experiments in which patterns of muscle activity are examined when a performer
is instructed to make a brief limb action (moving a lever in the extension
direction from one position to another). Figure 5.6 shows integrated
electromyogram (EMG) tracings from a quick elbow extension movement
(Wadman et al., 1979). In the normal movement (red traces), first there is a burst
of the agonist (here, the triceps) muscle; then the triceps turns off and the
179


antagonist muscle (the biceps) is activated to decelerate the limb; and finally the
agonist comes on again near the end to stabilize the limb at the target area. This
triple-burst (agonist–antagonist–agonist) pattern is typical of quick movements of
this kind.
 Figure 5.6 Electromyographic results from agonist (upper traces) and antagonist
(lower traces) muscles when the subject actually produced the movement (normal
    trials— red lines) and when the movement was blocked by a mechanical
                     perturbation (blocked trials—blue lines).
Reprinted by permission from Wadman et al. 1979.
Occasionally, and quite unexpectedly, on some trials the lever was blocked
mechanically by the experimenter so that no movement was possible. Figure 5.6
also shows what happens to the EMG patterning on these blocked trials (blue
lines). Even though the limb does not move at all, there is a similar pattern of
muscular organization, with the onset of the agonist and the antagonist occurring
at about the same times as when the movement was not blocked. Later, after about
120 ms or so, there is a slight modification of the patterning, probably caused by
the reflex activities (e.g., stretch reflexes) discussed in chapter 4. But the most
important findings are that the antagonist (biceps) muscle even contracted at all
when the movement was blocked and that it contracted at the same time as in the
normal movements.
180


The feedback from the blocked limb must have been massively disrupted, yet the
EMG patterning was essentially normal for 100 ms or so. Therefore, these data
contradict theories arguing that feedback from the moving limb (during the action)
acts as a signal (a trigger) to activate the antagonist muscle contraction at the
proper time. Rather, these findings support the motor program idea that the
movement activities are organized in advance and run off unmodified sensory
information for 100 to 120 ms, at least until the first reflexive activities can
become involved.
Motor Programs and the Conceptual Model
Motor programs are a critical part of the conceptual model seen in figure 5.2,
operating within the system, sometimes in conjunction with feedback, to produce
flexible skilled actions. The open-loop part of these actions provides the
organization, or pattern, that the feedback processes can later modify if
necessary. The following are some of the major roles of these open-loop
organizations:
To define and issue the commands to musculature that determine when, how
forcefully, and for how long muscles are to contract and which ones are to
contract
To organize the many degrees of freedom of the muscles and joints into a
single unit
To specify and initiate preliminary postural adjustments necessary to
support the upcoming action
To modulate the many reflex pathways to ensure that the movement goal is
achieved
In the following sections we see research examples of how motor programs use
anticipatory and feedback information to regulate movement control.
Anticipatory Adjustments
Imagine that you are standing with your arms at your sides and an experimenter
gives you a command to raise an arm quickly to point straight ahead. What will
be the first detectable EMG (muscular) activity associated with this movement?
Most people would guess that the first contraction would be in the shoulder
181


muscles that raise the arm. But, in fact, the EMG activity in these muscles occurs
relatively late in the sequence. Rather, the first muscles to contract are in the
lower back and legs, some 80 ms before the first muscle in that shoulder
(Belen'kii, Gurfinkel, & Pal'tsev, 1967).
This order may sound strange, but it is really quite a “smart” way for the motor
system to operate. Because the shoulder muscles are mechanically linked to the
rest of the body, their contractions influence the positions of the segments
connected to the arm—the shoulder and the back. That is, the movement of the
arm affects posture. If no compensations in posture were first made, raising the
arm would cause the trunk to flex, as well as to shift the center of gravity
forward, causing a slight loss of balance. Therefore, rather than adjust for these
effects after the arm movement, the motor system compensates before the
movement through “knowing” what postural modifications will soon be needed.
There is good evidence that these preparatory postural adjustments are really just
a part of the movement program for making the arm movement (W.A. Lee, 1980).
When the arm movement is organized, the motor program contains instructions to
adjust the posture in advance as well as the instructions to move the arm, so that
the action is a coordinated whole. Thus, we should not think of the arm movement
and the posture control as separate events; rather, these are simply different parts
of an integrated action of raising the arm and maintaining posture and balance.
Interestingly, these preparatory adjustments vanish when the performer leans
against a support, because postural adjustments are not needed here.
Integration of Central and Feedback Control
Although it is clear that central organization of movements is a major source of
motor control, it is also very clear (see chapter 4) that sensory information can
modify these commands in several important ways, as seen in the conceptual
model in figure 5.2. Thus, the question becomes how and under what conditions
these commands from motor programs interact with sensory information to define
the overall movement pattern. This is one of the most important research issues
for understanding motor control.
182


Reflex-Reversal Phenomenon
In addition to the various classes of reflex mechanisms discussed in chapter 4
that can modify the originally programmed output (figure 4.10), another class of
reflexive modulations exists that has very different effects on the movement
behavior. Several experiments show how reflex responses are integrated with
open-loop programmed control.
In one study, for example, the experimenter applies a light tactile stimulus to the
top of a cat’s foot while it is walking on a treadmill. If this stimulus is applied as
the cat is just placing its foot on the surface of the treadmill (in preparation for
load bearing), the response is to extend the leg slightly, as if to carry more load
on that foot. This response has a latency of about 30 to 50 ms and is clearly
nonconscious and automatic. If exactly the same stimulus is applied when the cat
is just lifting the foot from the surface (in preparation for the swing phase), the
response is very different. The leg flexes upward at the hip and the knee so the
foot travels above the usual trajectory in the swing phase. Thus, the same
stimulus has different (reversed) effects when it is presented at different
locations in the step cycle.
These alterations in the reflex—reversing its effect from extension to flexion (or
vice versa) depending on where in the step cycle the stimulus is applied—has
been called the reflex-reversal phenomenon (Forssberg, Grillner, & Rossignol,
1975). It challenges our usual conceptualizations of a reflex, which is typically
defined as an automatic, stereotyped, unavoidable response to a given stimulus:
Here the same stimulus has generated two different responses.
These variations in response must occur through interactions of sensory pathways
and the ongoing movement program for locomotion (the CPG, discussed earlier).
The CPG is responsible for many of the major events, such as muscle
contractions and their timing, that occur in locomotion and other rhythmical
activities. In addition, the CPGs are thought to be involved in the modulation of
reflexes, enabling responses such as the reflex-reversal phenomenon. The logic is
that the CPG determines whether and when certain reflex pathways can be
activated in the action, as illustrated in figure 5.7a and b. During the part of the
action when the cat's foot is being lifted from the ground (swing phase), the CPG
inhibits the extension reflex and enables the flexion reflex (i.e., allows it to be
183


activated, Figure 5.7a). If the stimulus occurs, it is routed to the flexion
musculature, not to the extension musculature. When the foot is being placed on
the walking surface, the CPG inhibits the flexion reflex and enables the extension
reflex (Figure 5.7b). It does this all over again on the next step cycle. Finally,
notice that if no stimulus occurs at all, there is no reflex activity, and the CPG
carries out the action “normally” without the contribution of either reflex.
 Figure 5.7 Role of CPGs in reflex reversals. In (a) , the application of a tactile
 stimulus at the start of the swing phase of a CPG results in movement flexion; in
(b) , the application of the same tactile stimulus at the start of the stance phase of
     a CPG results in movement extension. The effect of the stimulus has been
                                      reversed.
Movement Flexibility
There is much more to be learned about these complex reflex responses, but they
undoubtedly play an important role in the flexibility and control of skills. The
cat's reflexes are probably organized to have an important survival role.
Receiving a tactile stimulus on the top of the foot while it is swinging forward
probably means that the foot has struck some object and that the cat will trip if the
foot is not lifted quickly over the object. However, if the stimulus is received
during the beginning of stance, flexing the leg would cause the animal to fall
because it is swinging the opposite leg at this time. These can be thought of as
temporary reflexes in that they exist only in the context of performing a particular
part of a particular action, ensuring that the goal is achieved even if a disturbance
is encountered. Analogous findings have been produced in speech, where slight,
unexpected “tugs” on the lower lip during the production of a sound cause rapid,
reflexive modulation, with the actual responses critically dependent on the
184


particular sound being attempted (Abbs, Gracco, & Cole, 1984; Kelso et al.,
1984). The critical goal for the motor system in such situations seems to be to
ensure that the intended action is generated and that the environmental goal (in
this case, the desired speech sound) is achieved.
This adaptable feature of a movement program provides considerable flexibility
in its operation. The movement can be carried out as programmed if nothing goes
wrong. If something does go wrong, then appropriate reflexes are allowed to
participate in the movement to ensure that the goal is met.
Problems in Motor Program Theory: The
     Novelty and Storage Problems
Open-loop control occurs primarily to allow the motor system to organize an
entire, usually rapid, action without having to rely on the relatively slow
information processing involved in a closed-loop control mode. Several
processes must be handled by this prior organization. At a minimum, the
following must be specified in the programming process in order to generate
skilled movements:
The particular muscles that are to participate in the action
The order in which these muscles are to be involved
The forces of the muscle contractions
The relative timing and sequencing among these contractions
The duration of each contraction
Most theories of motor programs assume that a movement is organized in
advance by the establishment of a neural mechanism, or network, that contains
time and event information. A kind of movement “script” specifies certain
essential details of the action as it runs off in time. Therefore, scientists speak of
“running” a motor program, which is clearly analogous to the processes involved
in running computer programs.
However, motor program theory, at least as developed so far in this chapter, does
not account for several important aspects of movement behavior. Perhaps the
most severe limitations of motor program theory are (1) the failure to account for
how novel movements are produced in the first place, and (2) lack of the
185


efficiency that would be required to store the massive number of motor programs
that would be required in order to move.
This capability for producing novel actions raises problems for the simple motor
program theory as we have developed it to this point in the chapter. On this view,
a given movement is represented by a program stored in long-term memory.
Therefore, each variation in a tennis stroke, for example, associated with
variations in the height and speed of the ball, the position of the opponent, the
distance to the net, and so on, would need a unique and separate program stored
in memory because the instructions for the musculature would be different for
each variation. Extending this view further suggests that we would need literally
a countless number of motor programs stored in memory just in order to play
tennis. Adding to this the number of movements possible in all other activities,
the result would be an absurdly large number of programs stored in long-term
memory. This leads to what has been called the storage problem (Schmidt,
1975), which concerns how all of these separate programs could be stored in
memory.
There is also the novelty problem. For example, when I am playing tennis, no
two strokes, strictly speaking, are the same. That is, every stroke requires a very
slight difference in the amount of contraction of the participating muscles. In this
sense, then, every stroke I hit is ‘novel’, implying that the system would need a
separate program for every shot.” If motor programs that are stored in memory
are responsible for all such rapid movements, then how could something
essentially novel be performed with elegance and skill without violating the
storage problem mentioned earlier? The simple motor program theory, as
presented here to this point, is at a loss to explain the performance of such novel
actions.
To summarize, these observations raise two problems for understanding everyday
movement behavior:
1. How (or where) do humans store the nearly countless number of motor
   programs needed for future use the storage problem?
2. How do performers produce truly novel behavior such as performing a
   variant of a tennis swing that you have never performed previously? The
   program for such an action cannot be represented in an already stored motor
   program: the novelty problem.
186


Many years ago the British psychologist Sir Fredrick Bartlett (1932), in writing
about tennis, said this: “When I make the stroke, I do not . . . produce something
absolutely new, and I never repeat something old” (p. 202). What did he mean?
The first part of his statement means that, even though a movement is in some
sense novel, it is never totally brand-new. Each of his ground strokes resembles
his other ground strokes, possessing his own style of hitting a tennis ball. The
second part of Bartlett's statement conveys the idea that every movement is novel
in that it has never been performed exactly that way before.
The novelty and storage problems for motor program theory discussed in the
previous section, and indeed, in explaining Bartlett’s keen insight regarding the
tennis stroke, motivated a search for alternative ways to understand motor
control. There was a desire to keep the appealing parts of motor program theory,
but to modify them to solve the storage and novelty problems. The idea that
emerged was that movement programs can be generalized (Schmidt, 1975). This
generalized motor program (GMP) consists of a stored pattern, as before. The
generalized program stored in memory is thought to be adjusted at the time of
movement execution, allowing the action to be changed slightly to meet the
current environmental demands.
Generalized Motor Program Theory
The quote from Bartlett captures the essence of GMP theory: Some features of the
tennis stroke remain the same from shot to shot, and some features of the stroke
are changed each time. According to GMP theory, what remains the same reflects
the invariant features of a motor program—those features that make the pattern
appear the same, time after time. Invariant features are the reason our unique
writing style appears the same regardless of whether we are using a pen to write
in a notebook, a marker to write large enough on a whiteboard for everyone in a
large class to read, or our toe to write something in the beach sand at Marina del
Rey.
The aspect that allows changes from stroke to stroke (in Bartlett’s quote) is
represented in GMP theory as the relatively superficial, or surface features of
the movement. If the pattern represents the invariant features of your writing style,
then modifying what are called parameters determines how it is executed,
representing its surface features. Writing something slow or fast, large or small,
187


on paper or in the sand, and with a pen or a toe, represents how the GMP is
executed at any one time. The word parameter comes from mathematics, and
represents numerical values in an equation which do not change the form of the
equation. For example, in a linear equation, whose general form is Y = a + bX,
the values a and b are parameters—Y and X are related to each other in the same
way for any values of a and b. The unique performance that occurs when certain
parameters are changed does not alter the invariant characteristics of the GMP—
the parameters change only how the GMP is expressed at any given time.
In GMP theory, movements are thought to be produced as follows. As determined
via sensory information processed in the stimulus identification stage, a GMP for,
say, throwing (as opposed to kicking) is chosen during the response selection
stage. This GMP is then retrieved from long-term memory, much the same way
that you retrieve your friend's telephone number from memory. During the
movement programming stage, the motor program is prepared for initiation, or
parameterized.
One of the necessary processes here is to define how to execute this program.
Which limb to use, how fast to throw, which direction to throw, and how far to
throw must be decided based on the environmental information available just
before action. These decisions result in the assignment (probably in the response
selection stage) of movement parameters—characteristics that define the nature
of the program's execution without influencing the invariant characteristics (that
determine its form) of the GMP. Parameters include the speed of movement, its
amplitude (overall size), and the limb used. Once the parameters have been
selected and assigned to the program, the movement can be initiated and carried
out with this particular set of surface features.
According to GMP theory, the key variables to consider are what constitute the
invariant features of the GMP and what the parameters, or surface features, are.
These important issues are discussed in the next sections.
Generalized motor program theory suggests that the motor program for signing
 your name retains its invariant features, no matter what you are writing on.
188


Invariant Features of a GMP
To begin to discover the nature of GMP representations, we need to know what
features of the flexible movement patterns remain invariant, or constant, as the
more surface features (such as movement speed, movement amplitude, and
forces) are altered. When movement time is altered, for example, almost every
other aspect of the movement changes too: The forces and durations of
contractions, the speed of the limbs, and the distances the limbs travel all can
change markedly as the movement speeds up.
However, what if some aspects of these movements could be shown to remain
constant even though just about everything else was changing? If such a value
could be found, scientists argued, it might indicate something fundamental about
the structure of the GMP that serves as a basis for all of these movements, thus
providing evidence for how motor programs are organized or represented in
long-term memory. Such a constant value is termed an invariance, and the most
important invariance concerns the temporal structuring of the pattern (or the
pattern’s “rhythm”).
Relative Timing
Rhythm, or relative timing, is a fundamental feature of many of our daily
189


activities. Of course, rhythm is critically important in such activities as music and
dance. But timing is also a key feature of many sporting activities (such as the
golf swing) and work activities (e.g., typing, hammering). There is strong
evidence to suggest that relative time is an invariant feature of the GMP. An
example is the evidence provided in the Armstrong (1970) study, discussed in
Focus on Research 5.3.
The graph in figure 5.8 shows a sample trial in which one of Armstrong’s (1970)
subjects produced a pattern from memory that was made too quickly (red trace,
overall time about 3.2 s rather than the goal movement [blue trace], to be done in
about 4.0 s). But compare the red movement pattern with the blue goal pattern in
this figure—you will notice that the whole movement appeared to speed up as a
unit. That is, each of the peaks (movement reversals) occurred sooner and sooner
in real time, but occurred at about the same time relative to the overall time of the
pattern; hence the term relative time is used to refer to the constant occurrence of
these peaks (see Gentner, 1987, or Schmidt & Lee, 2011, for more on these
issues).
Figure 5.8 Subjects learned to make a timed movement of a lever with their right
  arm. The goal movement pattern is depicted by the blue line. The trace in red
  represents a trial in which the movement is made too rapidly; the error in the
timing of the reversals increases as the movement unfolds, which is what would
   happen if the red trace were simply a speeded-up version of the blue trace.
Adapted by permission from Armstrong 1970.
Relative timing is the fundamental temporal structure of a movement pattern that
is independent of its overall speed or amplitude. Relative timing represents the
movement's fundamental “deep structure,” as opposed to the “surface” features
190


seen in the easily modified alterations in movement time. This deep temporal
organization in movements seems to be invariant, even when the actions are
produced at different speeds or amplitudes.
More specifically, relative timing refers to the set of ratios of the durations of
several intervals within the movement, as illustrated in figure 5.9. Consider two
hypothetical throwing movements, with movement 1 being performed with a
shorter movement time than movement 2. Imagine that you measure and record the
EMGs from three of the important muscles involved in each action (in principle,
nearly any feature of the movement could be measured). If you measure several of
these contraction durations, you can define relative timing by a set of ratios, each
of which is the duration of a part of the action divided by the total duration. For
example, in movement 1, the ratios b/a = .40, c/a = .30, and d/a = .60 can be
calculated from the figure. This pattern of ratios is characteristic of this throwing
movement, describing its temporal structure relatively accurately.
 Figure 5.9 Hypothetical relative timing of EMG traces from three muscles for
two hypothetical throwing movements. The relative-time ratios are computed by
dividing the muscle EMG durations (i.e., b, c, and d) for each muscle (i.e., 1, 2,
  and 3) by the overall movement time (i.e., a). Notice that these ratios remain
 roughly constant when the movement time of the action changes (upper versus
                                 lower panel).
Reprinted by permission from Schmidt and Lee 2011.
191


This set of ratios (the relative timing) stays the same for movement 2 (even
though the duration of movement 2 is longer), because the values of b/a, c/a, and
d/a are the same as in movement 1. When this set of ratios is constant in two
different movements, we say that the relative timing was invariant. Notice that
movement 2 seems to be simply an elongated (horizontally “stretched”) version
of movement 1, with all of the temporal events occurring systematically more
slowly. This will always be found when relative timing is invariant. According
to the GMP theory, movement 2 was produced with a slower timing parameter
than movement 1, so the whole movement was slowed down as a unit but its
relative timing was preserved.
One of the important principles of movement control is that, when a brief, rapid
movement is changed in terms of the speed of the action (a fast vs. a slow throw),
the size of the action (making your signature large or small), or the trajectory of
the action (throwing overarm vs. sidearm), these alterations seem to be made
with an invariant relative timing. Relative timing is invariant across several
different kinds of “surface” modifications, so the form of the movement is
preserved even though the superficial features of it may change. There is some
controversy about whether relative timing is perfectly invariant (Gentner, 1987;
Heuer, 1988), but there can be no doubt that relative timing is at least
approximately invariant.
Focus on Research 5.3
Invariances and Parameters
An important contribution to the development of the GMP theory was
made by Armstrong (1970) in analyzing the patterns of movements that
subjects made in one of his experiments. In Armstrong's experiment,
learners attempted to move a lever (figure 5.8) from side to side in such a
way that a pattern of movement at the elbow joint (defined in space and
time) occurred, as depicted by the blue line in figure 5.8. This goal
movement (blue line) had four major reversals in direction, each of
which was to be produced at a particular time in the action, with the total
192


movement occupying about 4 s.
Armstrong noticed that when the learner made the first reversal
movement too quickly (red trace), the whole movement was also done too
quickly. Notice that the red line’s first peak (at reversal) was just a little
early (at .66 rather than at .75). The discrepancy between the actual and
goal reversal times increased roughly proportionally as the movement
progressed (1.72 vs. 1.95; 2.28 vs. 2.90; and 2.94 vs. 3.59). This gives
the impression that every aspect of the movement pattern was produced
essentially correctly but that the entire pattern was simply run off too
quickly.
Armstrong’s findings provided an early impetus to the development of the
idea that the motor program can be generalized (Schmidt, 1975). Here,
the program controlled the relative timing of the movement reversals.
When an early reversal appeared sooner or later than the goal time, then
all of the subsequent reversals sped up or slowed proportionally.
Exploring Further
1. In Armstrong’s figure (figure 5.8), sketch a graph of how you predict
   an action with a 4.5 s overall movement time would look if the
   subject had preserved the same relative-timing structure.
2. Suppose Armstrong’s subjects had performed the pattern again, one
   month after the original sessions of practice. Which do you think
   would be remembered better, the overall timing or the relative-time
   structure of the pattern? Give reasons for your answer.
Classes of Movements
You can think of an activity like throwing as a class consisting of a nearly infinite
number of particular movements (e.g., throwing a lighter object overarm,
throwing more rapidly). The theory holds that the entire class is represented by a
single GMP, with a specific, rigidly defined relative-timing structure. This
program can have parameters in several dimensions (e.g., movement time,
amplitude), making possible an essentially limitless number of combinations of
specific throwing movements, each of which contains the same relative timing.
193


Locomotion represents another class of movements that could be considered to
be controlled by a GMP. However, the research by Shapiro and colleagues
suggests that, in fact, there are at least two separate GMPs for gait, each with
unique relative timings—one for walking and another for running. Note, however,
that we can speed up and slow down either the walking or running gait
selectively without having to abandon the GMP (see Focus on Research 5.4).
Note that the relative timing actually produced by a performer can be thought of
as a kind of fingerprint unique to a particular movement class. This pattern can
be used to identify which of several motor programs has been executed
(Schneider & Schmidt, 1995; Young & Schmidt, 1990). Focus on Application 5.2
provides more examples of how our GMPs reflect other kinds of biological
“fingerprints.”
The relative timing of a person’s motor program for typing a name or password
               has the potential to serve as verification of identity.
194


Focus on Research 5.4
Relative Timing in Locomotion
Shapiro and collaborators (1981) studied the shifts in relative timing in
locomotion. They filmed people on a treadmill at speeds ranging from 3
to 12 kph and measured the durations of various phases of the step cycle
as the movement speed increased. The step cycle can be separated into
four parts, as shown in figure 5.10 a. For the right leg, the interval
between the heel strike at the left until the leg has finished yielding
(flexing) under the body's load is termed extension phase 2 (or E2), and
the interval from maximum flexion until toe-off is E3; together, E2 and E3
make up the stance phase. The interval from toe-off until maximum knee
flexion is termed the flexion phase (F), and the interval from maximum
flexion to heel strike is E1; together F and E1 make up the swing phase.
Figure 5.10 One cycle of gait can be divided into four parts, representing
   the swing and stance portions (a) . These four parts of the step cycle
 occupy relatively consistent relative timing within walking and running
                  speeds, but change between gaits (b) .
195


Reprinted by permission from Shapiro et al. 1981.
The locomotion data shown in Figure 5.10b are expressed as the
proportions of the step cycle occupied by each of the four phases; the
duration of each phase is divided by the total step-cycle time. When the
treadmill speed ranged from 3 to 6 kph, all subjects walked, each with a
particular pattern of relative timing. About half of the step cycle was
occupied by E3; about 10% of it was occupied by F and E2 each, with
about 28% occupied by E1. Notice that as speed increased from 3 to 6
kph, there was almost no shift in the relative timing for any of the parts of
the step cycle. When the speed was increased to 8 kph, however, where
now all subjects were running, we see that the relative-timing pattern was
completely different. Now E1 had the largest percentage of the step cycle
(32%), and E2 had the smallest (15%). E3, which had the largest
proportion of the step cycle in walking, was now intermediate, at about
28%. But notice that as the running speed increased from 8 to 12 kph,
there again was a tendency for these proportions to remain nearly
invariant.
The interpretation is that there are two GMPs operating here—one for
walking and one for running. Each has its own pattern of relative timing
and is quite different from the other. When the treadmill speed increases
for walking, the parameter values are changed, which speeds up the
movement with the same program while maintaining the relative timing.
At about 7 kph, a critical speed is reached, and the subject abruptly shifts
to a running program; the relative timing of this activity is maintained
nearly perfectly as running speed is increased further.
196


Exploring Further
1. Suppose that Shapiro and colleagues had found that a single GMP
   controlled both the walking and running gaits. How would the graph
   in figure 5.10b have appeared if the findings had supported this
   alternative hypothetical result?
2. Another gait that humans sometimes use is skipping. Given the
   findings of Shapiro and colleagues, how would the various parts of
   the step cycle appear during skipping slowly versus quickly?
Focus on Application 5.2
Relative-Timing Fingerprints
Identity fraud has represented a major threat to security for years. Forging
someone’s signature on a check and hacking into an account with
someone’s password are just two methods used by fraudsters to get
illegal access to wealth and information. However, the invariant features
of one’s GMP provide an important tool to combat the problem.
A person’s signature is usually considered unique and different from
anyone else’s signature. Forging the spatial characteristics of a signature
is not a very difficult task. All the forger needs to do is obtain the target
signature, compare the illegal and legal signatures, and continue to
practice by making improvements on the imperfections until a realistic
forgery is very difficult to distinguish from the original. A password that
is typed into an account on a computer is even easier to forge if the
fraudster knows the characters to enter. All that is needed is to enter the
correct sequence, and access is granted. However, relative timing is the
missing ingredient in both cases of fraudulent information.
Suppose, for example, that when you signed your name, the spatial and
temporal recordings of each of the various loops and cursives in
producing the letters, as well as the timing of your “t” crossings and “i”
197


dottings and so on, were compared to a data bank in which a large
number of your previous signatures had been stored. According to GMP
theory, the invariant characteristics of your signature would be repeated
regardless of the tool you used to sign your name (e.g., familiar or
unfamiliar pen), the surface on which you wrote it (e.g., paper or digital
tablet), or the size of the writing. Moreover, the fraudster who had access
only to the spatial characteristics of your signature would be completely
at a loss to replicate its relative timing.
It is possible that typing your password also has a relative-timing
characteristic that is uniquely yours, especially for those such as the
authors of this book who are not trained typists. We each have our own
unique style of typing—which letters are typically contacted with which
fingers, how long each key is held down (dwell time), and the transition
times that usually occur between particular letters. Once again, a data
bank of previous executions of our passwords would give rise to a range
of overall timings of these dwell and transition times, from which a
relative-timing “profile” could be derived and to which the fraudster
would not have access.
Fortunately for us, these methods of using digital knowledge of our GMPs
are now a reality. A Google search of terms such as “keystroke
dynamics” reveals a large number of articles about the theory and
technology underlying this security advance, as well as information about
a growing number of security firms that are developing the industry. In
many ways you can think of your signature and passwords as relative-
timing “fingerprints” that are unique to you.
Parameters Added to the GMP
In the previous section we discussed some of the features of movement that
remain the same from one time to the next—the invariant features of the GMP.
According to the theory, surface features need to be specified each time a
movement is performed. That is, the GMP needs to be parameterized before it
can be executed. What are some of these parameters?
198


Movement Time
Both the Armstrong study (Focus on Research 5.3) and the study by Shapiro and
collaborators (Focus on Research 5.4) provided strong evidence that overall
movement time could be varied without affecting the relative timing of the GMP.
In Armstrong`s study, the subject who accidentally sped up the movement pattern
still retained that same timing of reversals in the movement. And the subjects in
the study by Shapiro and colleagues could vary speeds of walking and running
without disrupting the relative timing of the step cycle. This also agrees with the
common experience that we seem to have no trouble speeding up and slowing
down a given movement, such as throwing a ball at various speeds, or writing
more slowly or more quickly. These findings indicate that, when movement time
is changed, the new movement preserves the essential temporal-pattern features
of the old movement. Therefore, both movements are represented by a common
underlying temporal (and sequential) pattern that can be run off at different
speeds. Therefore, overall movement time is a parameter of the GMP.
Movement Amplitude
The amplitude of movements can also be modulated easily in a way that is much
like varying the time. For example, you can write your signature either on a check
or five times larger on a blackboard, and in each case the signature is clearly
“yours” (Lashley, 1942; Merton, 1972). Making this size change seems almost
trivially easy.
The handwriting phenomenon was studied more formally by Hollerbach (1978),
who had subjects write the word “hell” in different sizes. He measured the
accelerations of the pen (or, alternatively, the forces delivered to the pen) during
the production of the words. These accelerations are graphed in figure 5.11.
When the trace moves upward, this indicates acceleration (force) away from the
body; a downward trace indicates acceleration toward the body. Of course, when
the word is written larger, the overall magnitude of the accelerations produced
must be larger, seen as the uniformly larger amplitudes for the larger word. But
what is of most interest is that the temporal patterns of acceleration over time are
almost identical for the two words, with the accelerations having similar
modulations in upward and downward fluctuations.
199


   Figure 5.11 Acceleration-time tracings of two instances of writing the word
 “hell,” once in small script (red) and again in larger script (blue). Although the
 amplitudes (which are proportional to the forces exerted on the pen) for the two
traces are markedly different, the temporal organization of the patterning remains
                       nearly the same in the two instances.
Adapted by permission from Hollerbach 1978.
This leads to an observation similar to the one just made about movement time. It
is easy to increase the amplitude of the movements by uniformly increasing the
accelerations (forces) that are applied, while preserving their temporal
patterning. Therefore, the same word is written with different amplitudes can be
based on a common underlying structure that can be run off with scaled forces
that govern the entire movement in order to produce different actions of overall
different sizes. Therefore, overall amplitude of force is a parameter of the GMP.
Effectors
A performer can also modulate a movement by using a different limb – and,
hence, different muscles – to produce the action. In the signature example, writing
on a blackboard involves very different muscles and joints than writing on a
check. In blackboard writing, the fingers are mainly fixed, and the writing is done
with the muscles controlling the elbow and the shoulder. In check writing the
elbow and the shoulder are mainly fixed, and the writing is done with the muscles
controlling the fingers. Yet the writing patterns produced are essentially the same.
This indicates that a given pattern can be produced even when the effectors – and
the muscles that drive them – are different.
These phenomena were studied by Raibert (1977), who wrote the sentence “Able
200


was I ere I saw Elba” (a palindrome, spelled the same way backward as
forward) with different effectors (i.e., limbs). In figure 5.12, line a shows his
writing with the right (dominant) hand, line b with the right arm with the wrist
immobilized, and line c with the left hand. These patterns are very similar. Even
more remarkable is that line d was written with the pen gripped in the teeth, and
line e used the pen taped to the foot! There are obvious similarities among the
writing styles, and it seems clear that the same person wrote each of them, yet the
effector system was completely different for each.
Figure 5.12 Five samples of writing a palindrome by the same subject, using (A)
  the dominant hand, (B) the dominant hand with the wrist immobilized, (C) the
nondominant hand, (D) with pen gripped by the teeth, and (E) with the pen taped
                            between toes of the foot.
Reprinted by permission from Raibert 1977.
This all indicates that changing the limb and effector system can preserve the
essential features of the movement pattern relatively easily. Therefore, the
selection of effectors can be thought of as a kind of “parameter” that must be
selected prior to action. There is some underlying temporal structure common to
these actions, which can be run off with different effector systems while using the
same GMP.
Summary of GMP Concepts
Some of these elements of the theory of GMPs can be summarized as follows:
A GMP underlies a class of movements and is structured in memory with a
rigidly defined temporal organization.
This structure is characterized by its relative timing, which can be measured
201


by a set of ratios among the durations of various events in the movement.
Variations in movement time, movement amplitude, and the limb used
represent the movement's surface structure, achieved by executing the action
with different parameters, whereas relative timing represents its deep,
fundamental structure.
Even though a movement may be carried out with different surface features
(e.g., duration, amplitude), the relative timing remains invariant.
Whereas surface features are very easy to alter by parameter adjustment, the
deeper relative-timing structure is very difficult to alter.
A particularly good way to understand the invariant features of a GMP with
certain added parameters is to consider movement as you would the various
components of a stereo system (see Focus on Application 5.3).
We started this section on GMPs by expressing dissatisfaction with the simple
motor program views as developed earlier in the chapter. Two issues were
considered to be especially problematic: the storage problem and the novelty
problem. The GMP theory provides solutions to both of these problems. For the
storage problem, the theory holds that an infinite number of movements can be
produced by a single GMP, so only one program needs to be stored for each class
of movement rather than an infinite number. And for the novelty problem, the
theory suggests that a second memory representation, a schema, is the theoretical
structure responsible for supplying parameters needed at the time of movement
execution. Note that, by using a parameter not used before, the person can
produce a novel action. Much more will be said about schema development in
later chapters. But, for now, think of the schema as a mechanism responsible for
selecting the parameters for the chosen GMP.
Focus on Application 5.3
The Stereo System Analogy
A good analogy for GMPs involves the standard phonograph/stereo
system, in which a turntable sends signals from a record into an amplifier,
whose output is delivered to speakers. In this analogy, illustrated in the
202


top portion of figure 5.13, the phonograph record itself is the GMP, and
the speakers are the muscles and limbs. (Does anyone remember what a
record is?) The record has all of the features of programs, such as
information about the order of events (the guitar comes before the
harmonica), the temporal structure among the events (i.e., the rhythm, or
relative timing), and the relative amplitudes of the sounds (the first
drumbeat is twice as loud as the second). This information is stored on
the record, just as GMP theory says that the analogous information is
stored in the program. Also, there are many different records to choose
from, just as humans have many motor programs to choose from (e.g.,
throwing, jumping), each stored with different kinds of information.
Figure 5.13 Illustration of the stereo system analogy.
Notice, though, that the record’s output is not fixed (lower portion of
figure 5.13): The speed of output can be changed if the speed of the
turntable is increased. Yet relative timing (rhythm) is preserved even
though the speed of the music is increased. You can change the amplitude
of the output by raising the volume uniformly; this increases the
amplitudes of all the features of the sounds. Also, you have a choice of
which effectors to use: You can switch the output from a set of speakers
in the den to a second set of speakers in the living room; this is analogous
to hammering either with the left hand or with the right, or with a different
hammer, still using the same pattern. Perhaps if you think of the theory of
203


GMPs in concrete terms like a stereo system, you can understand most of
the important features of the theory more easily. For example, when
subjects in the study by Shapiro and colleagues switched from walking to
running (Focus on Research 5.4), they first had to remove the walking
“record” and replace it with a running “record.” Then they had to
parameterize it, like setting the volume, speed, and speaker controls. This
analogy of the GMP and its parameters to the characteristics of a stereo
system sometimes helps people to understand the basic idea.
Summary
In very brief actions, there is no time for the system to process feedback about
errors and to correct them. The mechanism that controls this type of behavior is
open loop, called the “motor program.” This chapter is about motor programming
activities. Considerable evidence supports the motor program idea: (a) Reaction
time is longer for more complex movements; (b) complex movements can be
elicited in their complete form by certain stimuli; (c) animals deprived of
feedback information by deafferentation are capable of strong, relatively
effective movements; (d) some cyclical movements in animals are controlled by
inherited central pattern generators; and (e) a limb’s electrical muscle activity
patterns are unaffected for 100 to 120 ms when the limb is blocked by a
mechanical perturbation.

Even though the motor program is responsible for the major events in the
movement pattern, there is considerable interaction with sensory processes, such
as the organization of various reflex processes to generate rapid corrections,
making the movement flexible in the face of changing environmental demands.
Finally, motor programs are thought to be generalized to account for a class of
actions (such as throwing), and parameters must be supplied to define the way in
which the pattern is to be executed (such as throwing either rapidly or slowly).
The schema concept and how a schema is acquired with experience is discussed
extensively in chapter 10.

0 comments:

Posting Komentar