Professional Documents
Culture Documents
2 (Fall)
155
CI therapy studies that have reported first by Mott and Sherrington (1895)
positive results for improving motor and subsequently replicated (Lassek,
deficit after stroke. Its use is therefore 1953; Twitchell, 1954). It formed one
beginning to spread. of the major pillars underlying Sher-
CI therapy basically involves the rington’s formulation of the reflexo-
use of operant training techniques in logical position (Sherrington, 1910),
a rehabilitation context. The origin of which became one of the dominant
the therapy is described in publica- positions in neurology for the first
tions from this laboratory, but it is 70 years of the 20th century. Howev-
not well recognized or understood er, we showed that there were two
and is therefore often overlooked, the behavioral techniques that could in-
main reason probably being that duce a monkey to make use of a
there is little familiarity with behavior single deafferented forelimb.
analysis in the fields associated with One technique was training of the
neurorehabilitation. deafferented extremity. At first a
The theoretical roots of CI therapy discrete-trial avoidance conditioning
emerged from principles developed procedure was used. The monkey had
during graduate work at Columbia to make a simple flexion of the
University with Fred Keller and W. deafferented limb at the sound of a
Schoenfeld. The initial laboratory buzzer (Knapp, Taub, & Berman,
work that led to CI therapy began 1959, 1963) or click (Taub & Berman,
in the Department of Experimental 1963, 1968) to avoid an electric
Neurology in a research institute at shock. When the research shifted to
the Jewish Chronic Disease Center in the IBR, shaping was used. It proved
Brooklyn, New York. Monkeys re- to be a particularly effective means of
ceived a surgical abolition of somatic improving the motor deficit of the
sensation from one or both forelimbs, deafferented extremity. When dis-
and then were given training based, crete-trial procedures were used,
in part, on operant learning princi- transfer of limb use from the condi-
ples. Work continued at the Institute tioning chamber to the colony envi-
for Behavioral Research (IBR) in ronment was never observed (Taub &
Silver Spring, Maryland. The Chair- Berman, 1963, 1968; Taub, Ellman,
man of the Board of IBR was Joseph & Berman, 1966; Taub, Goldberg, &
V. Brady, who played a leading role Taub, 1975; Taub, Williams, Barro,
in founding behavioral pharmacolo- & Steiner, 1978). However, when
gy. The translation of CI therapy manual shaping with food reward
from monkeys to humans was stim- was employed in subsequent experi-
ulated by Brady’s example. CI ther- ments, there was a substantial im-
apy can be viewed as a type of provement of movement in the life
behavioral neurorehabilitation. situation as well (Taub, 1977). The
actions shaped included pointing at
DEAFFERENTATION visual targets (Taub et al., 1975) and ;
IN MONKEYS prehension in juveniles deafferented
on day of birth (Taub, Perrella, &
When somatic sensation is abol- Barro, 1973) and prenatally (Taub,
ished from a single forelimb in Perrella, Miller, & Barro, 1975) that
monkeys by the serial section of all had never exhibited prehension pre-
sensory roots of spinal nerves inner- viously. In both cases, the manual
vating that extremity, the monkey shaping-with-food-reward procedure
never again uses the deafferented produced an almost complete rever-
limb. This is the case even though sal of the motor disability, which
the motor outflow over the ventral progressed from total absence of the
roots of spinal nerves is left intact. target behavior to very good (al-
This was a classic observation, made though not normal) behavior. In the
using the deafferented limb during was never restrained, the ability to
the period before spontaneous recov- use the deafferented limb continued
ery of function had taken place, they to develop as the animals matured
would not learn that the limb could until it was similar to the extensive
not be used. LNU of the affected (though impaired) use of a limb when
extremity should therefore not devel- they were given limb deafferentation
op. In addition, the intact limb was as adults. This, then, constitutes a
restrained for the same period so that second line of evidence that supports
the animals could not receive rein- the LNU formulation.
forcement for use of that extremity
alone. In conformity with the predic- Translation of the LNU Model from
tion, which without the LNU formu- Deafferentation in Monkeys to CNS
lation would have been counterintu- Injury in Humans
itive, the animals were able to use
their deafferented extremity in the The results of the experiments de-
free situation after the restraint was scribed above show that simple behav-
removed 3 months after surgery, and ior-analytic techniques employed in
this was permanent, persisting for the discrete-trial or shaping contexts re-
rest of the animals’ lives. sulted in the conversion of a useless
Suggestive evidence in support of deafferented upper extremity to a limb
LNU was also obtained during deaf- that could be used extensively. Later, it
ferentation experiments carried out became apparent that this could be
< prenatally (Taub et al., 1973, 1975). viewed as a rapid and substantial
Life in the physically restrictive uter- rehabilitation of movement (although
ine environment imposes major con- that term was not usually applied to
straints on the ability to use the primates at the time). Thus, it appeared
forelimbs for such purposes as alter- possible that the same two techniques
ing body orientation to adjust for might be appropriately used to reha-
shifts in maternal position. Although bilitate motor disability in humans. An
use of the fetal limbs is not prevented implication of the concept of LNU as
entirely, their movement is restricted the outcome of the punishments and
in utero, thereby functioning like a rewards that result from attempted use
sling or a padded mitt in a human CI of an impaired extremity is that it
therapy experiment (to be discussed should, in principle, operate after any
below). Four animals were studied CNS injury when the initial effect is to
who had received unilateral forelimb temporarily abolish movement, re-
deafferentation by an intrauterine gardless of the injury’s location or
approach during the prenatal period; extent. There was also no a priori
three when two-thirds the way reason to suppose that it would not
through gestation and one when operate in humans as well as monkeys.
two-fifths the way through gestation. Specifically, stroke often leaves pa-
These animals exhibited functional tients with an apparently permanent
use of the deafferented extremity loss of function in an upper extremity,
from the first day of extrauterine life, although the limb is not paralyzed. In
in contrast to animals deafferented addition, the motor impairment is
after maturity that did not use the preponderantly unilateral. These fac-
affected extremity unless given train- tors are similar to those that pertain
ing of the deafferented arm or after unilateral forelimb deafferenta-
restraint of the intact arm. At birth, tion in monkeys. Therefore, it seemed
the prenatally deafferented animals reasonable to formulate a protocol
all used that limb for postural sup- that simply transferred the behavior-
port during ‘‘sprawling’’ and for analytic techniques used for overcom-
pushing into a sitting position. Sub- ing LNU of a deafferented limb
sequently, although the intact limb in monkeys to humans who had
stroke victim could not be modified if the training conditions are ar-
in the chronic phase, no matter what ranged appropriately. Recently, the
technique was employed. This view field has begun to accept that the
still has considerable force. Even word constraint is meant to include
today, after 25 years of research and training. The use of this term is
clinical practice, a substantial per- consistent with Timberlake’s analysis
centage of the chronic patients who (1993) of reinforcement as constitut-
come to the UAB CI therapy clinic ing ‘‘constraint of a functional causal
for treatment have been told by their system comprised of multiple interre-
physicians and therapists that there is lated causal sequences, complex link-
nothing that can be done to improve ages between causes and effects and a
their motor deficit. set of initial conditions’’ (p. 105).
rehabilitation is not really being contract. The diary is kept for the
accomplished. When CI therapy re- part of the day spent outside the
search was begun, there were no laboratory and is reviewed in detail
methods or tests being used to assess each morning with the therapist.
how or whether a patient was using a Daily administration of the MAL.
stroke-affected extremity in the life The MAL collects information about
situation. However, a behavior-ana- use of the more affected extremity in
lytic approach made it intuitively 30 important activities of daily living
obvious that a primary goal of (ADL) in all major domains of
rehabilitation treatment had to be everyday life. The daily repetition of
the development of methods to in- this detailed report, which is probed
duce use of the more affected arm in and verified in a number of ways,
the life situation and then monitoring serves to keep the patient’s attention
that use. This was the case indepen- on the use of the more affected
dent of considerations relating to extremity outside the laboratory or
LNU, although these certainly rein- clinic.
forced the need for real-world mon- Problem solving. During adminis-
itoring of behavior; hence, the devel- tration of the MAL, the therapist
opment of the motor activity log helps patients analyze, circumvent, or
(MAL; (Taub et al., 1993; Uswatte, overcome any barriers to using the
Taub, Morris, Light, & Thompson, more impaired arm in the life situa-
2006; Uswatte, Taub, Morris, Vig- tion. For example, if the patient is
nolo, & McCulloch, 2005). The MAL concerned about spilling liquid from
results have been confirmed by accel- a glass, the therapist may suggest
erometry data from transducers worn filling the glass only half way. If
on both arms for 3 days before and patients use the less affected arm for
3 days after the end of treatment manipulating eating utensils in a
(Uswatte, Miltner, et al., 1997, 1998; restaurant because they are embar-
Uswatte, Spraggins, Walker, Cal- rassed by dropping food from a
houn, & Taub, 1997). The TP con- utensil onto a table, the therapist
sists of the following techniques. may suggest not going to a restaurant
Behavioral contract. At the outset during the course of the treatment.
of treatment, the therapist negotiates Home skill assignments. During
a contract with the patient (and treatment, subjects are asked to carry
separately with the caregiver, if one out at home five difficult (for them)
is available) in which agreement is ADL tasks and five easy tasks using
reached that the patient will use his the more affected arm, selected daily
or her more impaired extremity as from a list of approximately 200 (e.g.,
much as possible outside the labora- brush teeth, wash hands, use TV
tory. Specific activities during which remote). In addition, patients are
the patient will practice using the asked to spend 15 to 30 min at home
more impaired extremity are dis- on a daily basis repetitively perform-
cussed, agreed on, and written down. ing with their more affected arm
At the end of this process, the specific upper extremity tasks that
negotiated document is signed by are similar to those performed in the
the patient (or caregiver), the thera- laboratory or clinic. The tasks are
pist, and a witness to emphasize chosen for practice to improve the
the character of the document as most significant movement deficits.
a contract. Subjects check off the ADL activities
Daily home diary. During treat- and exercises carried out on a form
ment, the patients catalogue on a provided to them each day.
daily diary form how much they have Weekly telephone contacts with
used the more affected arm for the patients. For the first month after
activities specified in the behavioral the end of treatment, the MAL is
administered by phone, and problem resting hand splint and a sling (Taub
solving is carried out. et al., 1993). However, this level of
Posttreatment practice. Toward the restraint was found to be unneces-
end of treatment, an individualized sary, and currently a mitt with a
posttreatment home practice program heavily padded palmar surface is
of approximately 100 tasks is developed employed. It prevents the use of the
and given to the patients. They are fingers and hand for a target of 90%
encouraged to perform two or three of waking hours and gives as good
tasks for 10 min daily after the treat- results as the resting hand splint and
ment period, but to continually focus sling arrangement. This was, and still
attention on using the more affected is, generally considered to be the
arm in ADL whenever possible. signature if not the differential com-
In most physical rehabilitation ponent of CI therapy. This is unfor-
regimens, there is a passive element; tunate, because there is evidence that
the patient is responsible for carrying less affected limb restraint is not
out the therapist’s instructions pri- necessary or even important for
marily or only during treatment producing a maximal treatment effect
sessions. A major difference in CI (Taub et al., 1999; Uswatte, Taub,
therapy is the involvement of the Morris, Barman, & Crago, 2006).
patient as an active participant in all However, although less affected limb
requirements of the therapy, not only restraint is not necessary with adult
during the treatment sessions but also humans, it is important for monkeys
at home during the treatment period and young children (pediatric CI
and for the first month after labora- therapy), who have less capacity for
tory therapy has been completed (and self-suppression of behavior and de-
afterward, although this is not mon- ferral of reinforcement. Even in adult
itored). The TP makes patients re- humans, when restraint of the less
sponsible for adhering to the require- affected arm is used, it may make
ments of the therapy, and therefore in some contribution to promoting a
effect they become responsible for long-lasting increase in use of the
their own improvement. more affected arm in the home. This
The TP is the main way in which CI is a clinical opinion not based on a
therapy differs from other rehabilita- controlled study, but it is thought to
tion procedures. Its critical importance be a sufficiently real possibility that
in producing a large treatment effect use of the restraining mitt during the
was recently demonstrated (Gauthier treatment period is still retained in
et al., 2008). Twenty subjects were the UAB laboratory clinic.
given the full CI therapy protocol
including the TP. A second group CI THERAPY IN OTHER
received the same treatment in the LABORATORIES
laboratory, but none of the TP tech- In the UAB laboratory, over 400
niques were administered. Both groups patients with stroke have been given
showed a significant increase in the one variant or another of CI therapy
amount of spontaneous use of the and all but three of these patients
more affected arm in the life situation, have demonstrated substantial im-
but the improvement of the CI therapy provement in motor ability. There
TP group was approximately 2.5 times have also been over 300 papers from
as great as the improvement recorded
other laboratories on adult and
for the non-TP group.
pediatric CI therapy published to
date. To our knowledge all but two
Less Affected Limb Restraint of the studies have reported positive
In initial experiments, limb re- results. In particular, CI therapy was
straint was achieved using a rigid the subject of a multisite randomized
controlled trial (Wolf et al., 2006), mately 50% of the amount of use of
the gold standard of proof of efficacy the more affected arm they had
in medical fields. The results were before stroke from an initial level of
positive. approximately 10%. This is a five
With respect to magnitude of the times difference and a substantial
treatment effect, this laboratory’s improvement, but it is not a cure.
results have been replicated with There is still considerable room for
patients with chronic stroke in pub- further improvement. CI therapy can
lished studies from four laboratories also produce a large treatment effect
in which therapists were trained in (although not as large) in patients
this laboratory and monitored twice with more severe motor deficits than
yearly (Dettmers et al., 2005; Kunkel those in the mild or moderate deficit
et al., 1999; Miltner, Bauder, Som- category treated in most CI therapy
mer, Dettmers, & Taub, 1999; Sterr studies, including patients with ini-
et al., 2002). Some of the other papers tially plegic hands (see below).
report outcomes as large as those
obtained in this and related labora- APPLICATIONS OF CI THERAPY
tories; however, many studies report
The LNU formulation predicts
results that are significant but only
that any substantial damage to the
one half to one third as large as those
CNS may lead to LNU. Thus, CI
obtained here. The likely reasons for
therapy, which initially had been
this disparity are twofold: (a) There
found to be helpful in overcoming
was incomplete or complete lack of
LNU in stroke patients with mild or
use of the procedures of the transfer
moderate motor deficits, should be
package, which, although reported in
applicable to motor limitations more
the papers from this laboratory, had
severe than those originally worked
been largely ignored. As noted above,
with, to deficits other than motor
we have replicated the reduced treat-
impairment of the upper extremity,
ment effect obtained by others by
and to other types of neurological
duplicating everything that is nor-
conditions.
mally done in treatment here except
implementation of the TP (Gauthier
et al., 2008). (b) A protocol with Lower Functioning Patients
attenuated intensity (tasks or move- Most of the patients treated at the
ments per unit time) was used, such UAB laboratory could be character-
as in a study by van der Lee, ized as having deficits that were mild
Beckerman, Lankhorst, and Bouter or moderate, defined as having the
(1999). ability to extend 20u at the wrist and
The techniques of the TP have 10u at each of the fingers. Experi-
often been used separately by indi- ments have also been carried out with
vidual therapists, but rarely system- patients with moderate and moder-
atically and never combined together ately severe deficits (Taub et al.,
in an attempt to make patients’ 1999). Their treatment change was
compliance with the protocol outside somewhat less than for higher func-
the laboratory critical so that they tioning patients (e.g., increases of
become responsible for their own approximately 400% and 350% for
improvement. Even when the behav- patients with moderate and moder-
ioral techniques of the TP and ately severe deficits, respectively,
intensive training are used, CI thera- compared to approximately 500%
py does not constitute a ‘‘cure’’ for for patients with mild or moderate
the motor deficit following stroke. deficits), but the treatment changes
On a group basis, patients in studies were nevertheless very large. Most
from this laboratory with mild or recently, work has been carried out
moderate deficits regain approxi- with patients with useless, plegic
hands that were initially fisted. Con- (e.g., overground walking, treadmill
ventional physical rehabilitation pro- walking with and without a partial
cedures, including some from neuro- body weight support harness, sit-to-
developmental treatment (NDT) stand, lie-to-sit, step climbing, walk-
and functional electrical stimulation ing over obstacles, various balance
(FES) were used to maintain the and support exercises) for at first 6
fingers in a sufficiently extended and and then 3 hr per day with inter-
aligned position so that CI therapy spersed rest intervals as needed over
training procedures could be carried 3 weeks and 0.5 hr per day devoted
out. At the end of treatment, the to TP procedures. Task performance
patients exhibited a 186% improve- is shaped as in the upper extremity
ment in the real-world use of the protocol. Training is enhanced th-
more affected arm. This arm had rough the use of force feedback (limb
been converted into a useful ‘‘helper’’ load monitor) and limb displacement
in the life situation (e.g., keeping a (joint angle/electric goniometer) feed-
piece of paper in place while writing back devices. No restraining device is
with the less affected hand, holding a placed on the less affected leg. The
toothpaste tube while unscrewing the lower extremity procedure is consid-
cap, bearing body weight for bed ered to be a form of CI therapy
mobility). because of the use of the TP, the
We estimate that CI therapy is strong massed practice and shaping
applicable to at least 50% of the element, and because the reinforce-
chronic stroke population with motor ment of adaptive patterns of ambu-
deficit, perhaps more. This is a very lation over maladaptive patterns in
large group of individuals; an esti- our training procedure constitutes a
mated 4,000,000 people in this coun- significant general form of constraint.
try have had strokes in previous Control data were provided by a
years, and in addition, there are more general fitness control group that
than 3,000,000 people who have had received the same battery of lower
had traumatic brain injuries. Very extremity tests as the treatment sub-
few of the more than 50% of these jects. The ES of the change in real-
individuals with persisting motor world performance due to the treat-
deficit are given any rehabilitation ment was very large, but not quite as
treatment. Thus, CI therapy could large as for the upper extremity. The
potentially improve the quality of life improved lower extremity use was
and increase the independence of a retained without any decrement for
large number of currently untreated the 2 years that were tested.
persons with brain damage.
Conditions Other Than Stroke
Lower Extremity
The CI therapy protocol has been
An obvious target for transfer of applied with success, as noted at the
the CI therapy techniques developed beginning of the article, to traumatic
for the upper extremity was the more brain injury (Shaw, Morris, Uswatte,
affected lower extremity of stroke McKay, & Taub, 2003), upper and
patients. The 38 chronic stroke pa- lower extremity in multiple sclerosis
tients treated to date have had a wide (Mark, Taub, Bashir, et al., 2008;
range of disability extending from Mark, Taub, Uswatte, et al., 2008),
being close to nonambulatory to cerebral palsy and pediatric motor
having moderately impaired coordi- disorders of neurological origin
nation (Taub et al., 1999). The across the full range of age from
treatment (lower extremity CI thera- 1 year old through the teenage years
py) consists of massed or repetitive (Taub, Griffin, et al., 2006; Taub
practice of lower extremity tasks et al., 2007, 2011; Taub, Ramey,
DeLuca, & Echols, 2004), focal hand good result with CIMT. Groups of
dystonia in musicians (Candia et al., three patients and a therapist partic-
1999, 2002), and, though not a motor ipated in a language card game
disorder, phantom limb pain after (Pulvermüller, 1990; Pulvermüller &
amputation (Weiss, Miltner, Adler, Schonle, 1993). The exercise resem-
Bruckner, & Taub, 1999). bles the child’s card game ‘‘Go Fish.’’
Aphasia. The application of CI A participant asks one of the other
therapy that is probably of greatest players if they have in their hand a
interest from a behavior-analytic card with a specific pictured object to
point of view is to aphasia, especially match one in their own. If they do, the
the work being done currently. Apha- requester can meld those cards. Par-
sia arises as a consequence of focal ticipants win the game if they meld
brain damage, often in association each of the cards they were dealt so
with stroke. There is as much LNU that none are left. The difficulty of the
after stroke associated with the ver- required request by each patient is
bal behavior of aphasics as there is progressively increased in small steps
with motor deficit. Because of halting (i.e., shaped) along several dimen-
and slow verbal production and sions: number of words in the request
incomplete understanding, speech be- (or response to it), number of formu-
comes very effortful and often em- las of politeness, precision of patient’s
barrassing. The person compensates card description (animal, pet, dog),
by greatly reducing attempts to speak complexity of card depiction (dog,
or remaining silent entirely and by two dogs, one red and one blue dog),
using gestures and other nonverbal and grammatical correctness.
means of communication. In addi- CIAT I patients in the initial RCT
tion, when there is difficulty in improved much more than patients
understanding speech, many aphasics who received conventional aphasia
with receptive problems (Wernicke’s therapy. This study has since been
aphasia, fluent aphasia) simply tune replicated (Bhogal, Teasell, & Spee-
out. Thus, the demonstration that chley, 2003; Kirmess & Maher, 2010;
motor deficits are modifiable in Maher et al., 2006; Meinzer et al.,
chronic stroke raised the possibility 2004, 2007). Following a positive
that verbal impairment could also be evaluation of a committee appointed
rehabilitated by an appropriate mod- by the American Speech and Hearing
ification of the CI therapy protocol. Association (Raymer et al., 2008),
The LNU formulation predicted that CIAT I is now beginning to spread.
this was a strong possibility. In The results of the CIAT I protocol
the first studies (Pulvermüller et al., have been positive; however, the
2001; Taub, 2002), aphasic patients intervention was only an incomplete
with chronic stroke who had previ- translation of CIMT. CIMT pro-
ously received extensive conventional duced an improvement of approxi-
speech therapy and had reached an mately 500% in real-world use of the
apparent maximum in recovery of more affected extremity of chronic
language function were induced to stroke patients with mild to moderate
talk and improve their verbal skills motor deficit in one experiment
for 3 hr each weekday over a 2-week (Taub, Uswatte, King, et al., 2006).
period. The intervention was termed Other experiments from this labora-
constraint-induced aphasia therapy tory have reported treatment effects
(CIAT I). The constraint was im- of similar size. Aphasic patients given
posed by the contingencies of rein- CIAT I showed an improvement of
forcement in the shaping paradigm 30% in real-world verbal behavior.
that was used; there was no physical This is a large treatment effect
restraint, although as noted, physical compared to conventional speech
restraint is not necessary to obtain a language therapies, but it is very
small compared to the results pro- therapy achieves its therapeutic ef-
duced by CIMT. Consequently, to fect. Another important mechanism
determine whether this large differ- relates to the fact that CI therapy
ence was the result of an incomplete produces large plastic changes in the
translation of the CI therapy proto- structure and function of the brain.
col employed in the UAB laboratory Starting in the 1980s, Merzenich
with motor deficits to the treatment and collaborators showed in mon-
of language impairment, the initial keys that a decrease or increase in the
aphasia treatment protocol (CIAT I) amount of use of a body part or a
was modified to more closely resem- sensory function decreased or in-
ble the CIMT protocol. creased the size of the brain region
In the restructured and enhanced that represented that function (e.g.,
protocol (CIAT II), use of behavior- Jenkins, Merzenich, Ochs, Allard, &
analytic procedures was increased and Guic-Robles, 1990; Merzenich et al.,
emphasized. Revisions involved addi- 1983). This phenomenon was origi-
tion of new exercises, including the final nally termed cortical reorganization
exercise, considered to be the most and is now called brain plasticity
important, in which everyday verbal or neuroplasticity. In the 1990s,
interactions are simulated and mod- Taub and collaborators in Germany
eled. In addition a TP parallel to that showed that neuroplastic cortical
used in CIMT was introduced, there reorganization occurred in humans,
was increased emphasis on the shaping and that it had functional signifi-
of responses, and the primary caregiver cance in that it could affect move-
was trained as an alternate therapist so ment, behavior, and the quality of
that the training begun in the labora- sensory experience (e.g., Elbert, Pan-
tory could be continued at home, both tev, Wienbruch, Rockstroh, &Taub,
during and after formal training. 1995; Flor et al., 1995).
To date, only four patients have A substantial number of studies
been treated with the new protocol. have now shown that CI therapy
However, their results have far exceed- produces a large neuroplastic cortical
ed those obtained with CIAT I and are reorganization in humans with stroke-
comparable to the results obtained with related paresis of an upper limb. This
CIMT. With CIAT I, as noted, there was first demonstrated by Nudo,
was a 30% improvement in real-world Wise, SiFuentes, and Milliken (1996)
verbal behavior; for the recent patients, in an animal model of CI therapy.
the mean improvement was 537%, Subsequently, Liepert, Bauder, Milt-
which is approximately 18 times greater ner, Taub, and Weiller (2000) used
than for CIAT I and roughly equiva- focal transcranial magnetic stimula-
lent to the treatment effect for CIMT. tion (TMS) to map the area of the
Of additional interest is the fact that at motor cortex that controls an impor-
6-month follow-up, the patients tant muscle of the hand (abductor
showed no loss in retention; instead, pollicis brevis) in 15 patients with a
the verbal behavior scores increased chronic upper extremity hemiparesis
substantially to a 643% improvement (mean chronicity 5 6 years) before
over pretreatment scores. This increase and after CI therapy. We first repli-
appears to be attributable to the cated the clinical result that CI
continuation of training by the care- therapy produces a very large increase
givers in the real-world environment. in patients’ amount of arm use in the
home over a 2-week treatment period.
Over the same interval, the cortical
CI THERAPY AND
region from which electromyography
BRAIN PLASTICITY
responses of the abductor pollicis
As noted, overcoming LNU is one brevis muscle could be elicited by
of the mechanisms by which CI TMS was greatly increased, and both
the clinical effect and the alteration in that after CI therapy there was a large
brain function persisted for the increase in the activation of the
6 months tested. CI therapy had led usually weakly active healthy, ipsilat-
to an increase in the excitability and eral hemisphere with more affected
recruitment of a large number of hand movement in confirmation of
neurons in the innervation of move- the findings of Kopp et al. (1999). In
ments of the more affected limb addition, Wittenberg et al. (2003)
adjacent to those originally involved found in a positron emission tomog-
in control of the extremity prior to raphy study that before CI therapy
treatment. The effect was sufficiently there was a larger activation in
large that it represented a return to multiple areas of the brain with more
normal size of the motor output area affected arm movement than in
of the abductor pollicis brevis muscle healthy control subjects. This exces-
on the infarcted side of the brain, sive activation diminished after CI
although it was the size of excitable therapy. The preliminary interpreta-
cortical area that had become normal, tion of this result is that less effort is
not its function; the affected hand, required to produce movements after
though much improved after CI CI therapy than before treatment.
therapy, was not normal in function. Since these initial studies, there
In a third study, Kopp et al. (1999) have been approximately 20 other
carried out dipole modeling of steady- studies that have demonstrated an
state movement-related cortical po- alteration in brain function associat-
tentials (EEG) of patients before and ed with a CI therapy-induced im-
after CI therapy. We found that provement in movement after CNS
3 months after treatment the undam- damage. By providing a physiological
aged motor cortex ipsilateral to the basis for the treatment effect reported
affected arm, which normally controls for CI therapy, these results have
movements of the contralateral (less tended to increase confidence in the
affected) arm, had been recruited to clinical results.
generate movements of the affected The studies described to this point
arm. This effect was not in evidence show that alterations in afferent
immediately after treatment and was input can alter the function and
presumably due to the sustained organization of specific brain regions,
increase in more affected arm use in but until recently there was no
the life situation produced by CI evidence that environmental stimuli
therapy over the 3-month follow-up could measurably alter brain struc-
period. This experimental evidence tures in adult humans. It has now
that CI therapy is associated with been shown that seasoned taxi drivers
substantial changes in brain activity have significantly expanded hippo-
has been confirmed by convergent campi (Maguire et al., 2000), jugglers
data from two other neurophysiolog- acquire significantly increased tem-
ical studies that used two additional poral lobe density (Draganski et al.,
techniques in association with the 2004), and thalamic density signifi-
administration of CI therapy. Bauder, cantly declines after limb amputation
Sommer, Taub, and Miltner (1999) (Draganski et al., 2006). Moreover,
showed that there is a large increase in in an animal model of stroke, CI
the amplitude of the late components therapy combined with exercise re-
of the Bereitschaftspotential (a move- duced brain tissue loss associated
ment-related cortical potential) after with stroke (DeBow, Davies, Clarke,
CI therapy, suggesting that an en- & Colbourne, 2003). Accordingly,
hanced neuronal excitability is in- structural imaging studies became a
duced in the damaged hemisphere; logical next step toward understand-
this is consistent with the results of ing whether there are anatomical
Liepert et al. (2000). We also found changes following the administration
Figure 3. Cortical surface-rendered images of changes in gray matter. Gray matter increases
displayed on a standard brain for (A) participants who received the CI therapy transfer package
and (B) those who did not. Surface rendering was performed with a depth of 20 mm. Bar values
indicate t statistics ranging from 2.2 to 6.7.
after CNS damage. The talk was in a modified forms that are more readily
department of physical medicine and reimbursed by insurance. In addition,
rehabilitation. I described the work as I understand it, schools of physical
with primates in the context of a and occupational therapy are begin-
possible translation to humans, al- ning to teach CI therapy and at least
though the latter was implied and not some of the principles of behavior
specifically stated. The chairman of analysis. Behavior analysis has thus
the department, who was a promi- begun to make an appearance on the
nent clinician and rehabilitation in- stage of neurorehabilitation.
vestigator, sat quietly through my talk
but with a frigid expression. After I REFERENCES
was finished, he said with progressive-
ly increasing volume, ‘‘Are you trying Azrin, N. H., & Holz, W. C. (1966). Punish-
ment. In W. K. Honig (Ed.), Operant
to tell me that you have a behavioral behavior: Areas of research and application
intervention that you think will im- (pp. 380–447). New York, NY: Appleton-
prove the symptoms of a neurological Century-Crofts.
lesion?’’ I saw that I was on danger- Bauder, H., Sommer, M., Taub, E., & Miltner,
ous ground and so I said, ‘‘Well, no’’; W. H. R. (1999). Effect of CI therapy on
movement-related brain potentials [Ab-
but that, of course, was what I had stract]. Psychophysiology, 36 (Suppl. 1), S31.
been implying for the past hour. I Bhogal, S. K., Teasell, R., & Speechley, M.
paused for a while and then said as (2003). Intensity of aphasia therapy, impact
mildly as I could, ‘‘But after all, isn’t on recovery. Stroke, 34, 987–993.
that what physical therapy is?’’ That Candia, V., Elbert, T., Altenmüller, E., Rau,
H., Schäfer, T., & Taub, E. (1999). Con-
was a mistake. He sputtered a few straint-induced movement therapy for focal
words, which I didn’t catch, while his hand dystonia in musicians. Lancet, 353, 42.
face quite literally began purpling. Candia, V., Schafer, T., Taub, E., Rau, H.,
However, to his credit, after the first Altenmüller, E., Rockstroh, B., et al. (2002).
Sensory motor retuning: A behavioral
experiment was completed and the treatment for focal hand dystonia of pia-
report appeared in print, he changed nists and guitarists. Archives of Physical
his opinion and became a strong and Medicine and Rehabilitation, 83, 1342–1348.
valuable supporter. This pretty well Catania, A. C. (1998). Learning (4th ed.).
sums up the way in which the Upper Saddle River, NJ: Prentice Hall.
Cohen, J. (1988). Statistical power analysis for
rehabilitation community has reacted the behavioral sciences (2nd ed.). Hillsdale,
to CI therapy. At first, there was a NJ: Erlbaum.
strong bias against a treatment based DeBow, S. B., Davies, M. L. A., Clarke, H. L.,
on behavior analysis; few members of & Colbourne, F. (2003). Constraint-induced
the rehabilitation community had any movement therapy and rehabilitation exer-
cises lessen motor deficits and volume of
familiarity with behavior analysis or brain injury after striatal hemorrhagic
had even heard of it. However, as the stroke in rats. Stroke, 34, 1021–1026.
evidence began to mount and at- Dettmers, C., Teske, U., Hamzei, F., Uswatte,
tempts at replication were successful, G., Taub, E., & Weiller, C. (2005). Distrib-
uted form of constraint-induced movement
attitudes began to change. With the therapy improves functional outcome and
success of the multisite RCT cited quality of life after stroke. Archives of
above, and the fact that CI therapy Physical Medicine and Rehabilitation, 86,
produced substantial plastic structur- 204–209.
al changes in the brain, the case had Draganski, B., Gaser, C., Busch, V.,
Schuierer, G., Bogdahn, U., & May, A.
essentially been made. Use of CI (2004). Changes in grey matter induced by
therapy is still not by any means training. Nature, 427, 311–312.
universal, probably in part because of Draganski, B., Moser, T., Lummel, N.,
insurance reimbursement problems Gänssbauer, S., Bogdahn, U., Haas, F.,
due to the duration and therefore et al. (2006). Decrease of thalamic gray
matter following limb amputation. Neuro-
expense of the treatment. However, image, 31, 951–957.
even with that, the treatment is Elbert, T., Pantev, C., Wienbruch, C., Rock-
beginning to spread, especially in stroh, B., & Taub, E. (1995). Increased
ment of mixed transcortical aphasia. Cog- Taub, E., Goldberg, I. A., & Taub, P. B.
nition, 48, 139–161. (1975). Deafferentation in monkeys: Point-
Raymer, A., Beeson, P., Holland, A., Kendall, ing at a target without visual feedback.
D., Maher, L., Martin, N., et al. (2008). Experimental Neurology, 46, 178–186.
Translational research in aphasia: From Taub, E., Griffin, A., Gammons, K., Nick, J.,
neuroscience to neurorehabilitation. Journal Uswatte, G., & Law, C. R. (2006). CI
of Speech, Hearing, & Language Research, therapy for young children with congenital
5(1), S259–S275. hemiparesis. Atlanta, GA: Society for Neu-
Shaw, S. E., Morris, D. M., Uswatte, G., roscience.
McKay, S. B., & Taub, E. (2003). Adher- Taub, E., Griffin, A., Nick, J., Gammons, K.,
ence to constraint induced movement ther- Uswatte, G., & Law, C. R. (2007). Pediatric
apy procedures and outcome for persons CI therapy for stroke-induced hemiparesis
with traumatic brain injury. Journal of in young children. Developmental Neurore-
Neurologic Physical Therapy, 27(4), 180. habilitation, 10, 1–16.
Sherrington, C. S. (1910). Remarks on the Taub, E., Griffin, A., Uswatte, G., Gammons,
reflex mechanism of the step. Brain, 33, K., Nick, J., & Law, C. R. (2011).
1–25. Treatment of congenital hemiparesis with
Skinner, B. F. (1938). The behavior of organ- pediatric constraint-induced movement
isms. New York, NY: Appleton-Century- therapy Journal of Child Neurology, 26,
Crofts. 1163–1173.
Skinner, B. F. (1968). The technology of Taub, E., Heitman, R., & Barro, G. (1977).
teaching. New York, NY: Appleton-Century- Alertness and level of activity and purposive
Crofts. movement following deafferentation in the
Sterr, A., Elbert, T., Berthold, I., Kölbel, S., adult and developing monkey. In H. P.
Rockstroh, B., & Taub, E. (2002). Longer Zeigler & B. M. Wenzel (Eds.), Tonic
versus shorter daily constraint-induced functions of sensory systems (Vol. 290, pp.
movement therapy of chronic hemiparesis: 348–365). Ann, NY: Academic Sciences.
An exploratory study. Archives of Physical Taub, E., Miller, N. E., Novack, T., Cook,
Medicine and Rehabilitation, 83, 1374–1377. E. W., III., Fleming, W. C., Nepomuceno,
Taub, E. (1977). Movement in nonhuman C. S., et al. (1993). Technique to improve
primates deprived of somatosensory feed- chronic motor deficit after stroke. Archives
back. Exercise and sports science reviews of Physical Medicine and Rehabilitation, 74,
(Vol. 4, pp. 335–374). Santa Barbara, CA: 347–354.
Journal Publishing Affiliates. Taub, E., Perrella, P. N., & Barro, G. (1973).
Taub, E. (1980). Somatosensory deafferenta- Behavioral development following forelimb
tion research with monkeys: Implications deafferentation on day of birth in monkeys
for rehabilitation medicine. In L. P. Ince with and without blinding. Science, 181,
(Ed.), Behavioral psychology in rehabilitation 959–960.
medicine: Clinical applications (pp. 371– Taub, E., Perrella, P. N., Miller, D., & Barro,
401). New York, NY: Williams & Wilkins. G. (1975). Diminution of early environmen-
Taub, E. (2002). CI therapy: A new rehabil- tal control through perinatal and prenatal
itation technique for aphasia and motor somatosensory deafferentation. Biological
disability after neurological injury. Klinik Psychiatry, 10, 609–626.
und Forschung, 8, 48–49. Taub, E., Ramey, S. L., DeLuca, S., & Echols,
Taub, E., & Berman, A. J. (1963). Avoidance E. (2004). Efficacy of constraint-induced
conditioning in the absence of relevant (CI) movement therapy for children with
proprioceptive and exteroceptive feedback. cerebral palsy with asymmetric motor im-
Journal of Comparative and Physiological pairment. Pediatrics, 113, 305–312.
Psychology, 56, 1012–1016. Taub, E., & Uswatte, G. (2009). Constraint-
Taub, E., & Berman, A. J. (1968). Movement induced movement therapy: A paradigm for
and learning in the absence of sensory translating advances in behavioral neurosci-
feedback. In S. J. Freedman (Ed.), The ence into rehabilitation treatments. In G.
neuropsychology of spatially oriented behav- Berntson & J. Cacioppo (Eds.), Handbook
ior (pp. 173–192). Homewood, IL: Dorsey. of neuroscience for the behavioral sciences
Taub, E., Crago, J. E., Burgio, L. D., (Vol. 2, pp. 1296–1319). Hoboken, NJ:
Groomes, T. E., Cook, E. W., DeLuca, S. Wiley.
C., et al. (1994). An operant approach to Taub, E., Uswatte, G., King, D. K., Morris,
rehabilitation medicine: Overcoming D., Crago, J., & Chatterjee, A. (2006). A
learned nonuse by shaping. Journal of the placebo controlled trial of constraint-in-
Experimental Analysis of Behavior, 61, duced movement therapy for upper extrem-
281–293. ity after stroke. Stroke, 37, 1045–1049.
Taub, E., Ellman, S. J., & Berman, A. J. Taub, E., Uswatte, G., Mark, V., & Morris,
(1966). Deafferentation in monkeys: Effect D. (2006). The learned nonuse phenome-
on conditioned grasp response. Science, 151, non: Implications for rehabilitation. Europa
593–594. Medicophysica, 42(2), 241–255.
Taub, E., Uswatte, G., & Pidikiti, R. (1999). Wittenberg, G. F., Chen, R., Ishii, K.,
Constraint-induced movement therapy: A Bushara, K. O., Taub, E., Gerber, L. H.,
new family of techniques with broad appli- et al. (2003). Constraint- induced therapy in
cation to physical rehabilitation—a clinical stroke: Magnetic-stimulation motor maps
review. Journal of Rehabilitation Research and cerebral activation. Neurorehabilitation
and Development, 36, 237–251. and Neural Repair, 17, 48–57.
Taub, E., Williams, E., Barro, G., & Steiner, Wolf, S., Winstein, C., Miller, J., Taub, E.,
S. S. (1978). Comparison of the perfor- Uswatte, G., Morris, D., et al. (2006). Effect
mance of deafferented and intact monkeys of constraint-induced movement therapy on
on continuous and fixed ratio schedules of upper extremity function 3–9 months after
reinforcement. Experimental Neurology, 58, stroke: The EXCITE randomized clinical
1–13. trial. Journal of the American Medical
Timberlake, W. (1993). Behavior systems and Association, 296, 2095–2104.
reinforcement: An integrative approach.
Journal of the Experimental Analysis of
Behavior, 60, 105–128. APPENDIX A
Twitchell, T. E. (1954). Sensory factors in
purposive movement. Journal of Neurophys- Steps Involved in the Shaping
iology, 17, 239–254.
Uswatte, G., Miltner, W. H. R., Varma, M.,
Progression from Total Absence of the
Moran, S., Sharma, V., Foo, B., et al. Target Behavior to Thumb-Forefinger
(1998). Accelerometry: An objective ap- Grasp of a Food Object in Juvenile
proach to real-world outcome measurement Monkeys Deafferented Prenatally or
in physical rehabilitation. Paper presented at on Day of Birth
the inaugural meeting of the Program in
Cognitive Rehabilitation of the James S. The steps in shaping were as
McDonnell Foundation, St. Louis, MO. follows:
Uswatte, G., Miltner, W., Walker, H., Sprag-
gins, S., Moran, S., Calhoun, J., et al. 1. Showing the juvenile a desirable
(1997). Accelerometers in rehabilitation: food object (e.g. small apple cube,
Objective measurement of extremity use at peanuts) and reinforcing any move-
home [Abstract]. Rehabilitation Psychology,
42, 139.
ment of the arm, whether in the correct
Uswatte, G., Spraggins, S., Walker, H., direction or not, by food in the mouth.
Calhoun, J., & Taub, E. (1997). Validity 2. Requiring arm movements of
and reliability of accelerometry as an progressively greater excursion and
objective measure of upper extremity use more accurate direction for place-
at home [Abstract]. Archives of Physical
Medicine and Rehabilitation, 78, 896.
ment of food in the mouth.
Uswatte, G., Taub, E., Morris, D., Barman, 3. Requiring that the food object
J., & Crago, J. (2006). Contribution of the be touched for food to be placed in
shaping and restraint components of con- hand so that it could be returned by
straint-induced movement therapy to treat- the animal to its mouth.
ment outcome. NeuroRehabilitation, 21(2),
147–156. 4. Requiring that fingers be opened
Uswatte, G., Taub, E., Morris, D., Light, K., & so that hand could be baited with a
Thompson, P. (2006). The motor activity food object; wrist supported by
Log-28: Assessing daily use of the hemiparetic experimenter at end of arm trajecto-
arm after stroke. Neurology, 67, 1189–1194. ry; fingers opened, first by passive
Uswatte, G., Taub, E., Morris, D., Vignolo,
M., & McCulloch, K. (2005). Reliability
manipulation by experimenter and
and validity of the upper-extremity motor subsequently with progressively more
activity Log-14 for measuring real-world active finger extension required.
arm use. Stroke, 36, 2493–2496. 5. Grasping of food object by the
van der Lee, J., Beckerman, H., Lankhorst, animal at end of arm trajectory with
G., & Bouter, L. (1999). Constraint-induced
movement therapy [Letter to the Editor]. no support of wrist.
Archives of Physical Medicine and Rehabil- 6. Picking food object up from
itation, 80, 1606. experimenter’s palm, which was mold-
Weiss, T., Miltner, W. H. R., Adler, T., ed and moved to make prehension
Bruckner, L., & Taub, E. (1999). Decrease easier; any type of grasp permitted.
in phantom limb pain associated with
prosthesis-induced increased use of an 7. Picking food object up from a
amputation stump in humans. Neuroscience flat wooden board. Lateral thumb-
Letters, 272, 131–134. forefinger grasp (a monkey’s normal
mode of prehension) developed spon- ence among tasks that have similar
taneously over sessions, as did ap- potential for producing specific im-
proaching the food object from provements.
above rather than accomplishing the Shaping tasks should be modeled
grasp while the ulnar surface of wrist for the patient and encouragement
and lower forearm were supported by and coaching (verbal prompts) pro-
the board. vided liberally.
8. Placement of food objects (ap- The level of difficulty of the
proximately 1-cm3 apple cubes) in shaping task should be slightly be-
shallow (0.5 mm) wells on a Klüver yond what the patient can accom-
board (a board with multiple wells plish easily (e.g., encouraging him or
from which monkeys extract pieces of her to do a little better than the
food) to promote more accurate previous performance).
thumb-finger approximation. In the shaping progression, mov-
9. Placement of apple cubes on a ing to the next higher level of
Klüver board with deeper (1 cm) difficulty should be carried out
wells to promote pincer grasp (ap- when the patient has reached a
proximation of the palmar tips of the relative plateau with regard to
thumb and forefinger). performance. For the present pur-
10. Use of smaller food objects, poses, when a patient has performed
first peanuts, then raisins. five trials in a row with no improve-
ment evidenced in their score, the
The terminal behavior achieved next level of difficulty should be
was retrieval of raisins from wells on attempted. If subjects are permitted
the first attempt by pincer grasp on to achieve greater mastery, they
approximately 50% of trials. Some- frequently have a tendency to be-
times, after two or more attempts come ‘‘locked in’’ at that level.
failed, the monkeys would move the Subsequently, improvement be-
food object out of a well with the comes more difficult to achieve.
forefinger so that it could be grasped (This is a guideline only. If the
on the flat surface between wells. patient is ‘‘on a roll,’’ progressing
rapidly, he or she should be shifted
to the next performance difficulty
APPENDIX B level as rapidly as the trainer feels
Shaping Guidelines the performance will keep improv-
ing at a maximal level).
Shaping is a training method in The shaping task is made progres-
which a motor or behavioral objec- sively more difficult only as the
tive is approached in small steps by patient improves in performance.
successive approximations, or a task Any of the shaping progression
is gradually made more difficult in parameters can be changed to in-
accordance with a subject’s motor crease the difficulty of the task (e.g.,
capabilities. The following guidelines time, number of repetitions, height,
employed in the UAB laboratory placement, etc.).
should be followed when using shap- When increasing the level of diffi-
ing for inducing recovery of motor culty of an activity, the shaping pro-
function. gression parameters selected should
Specific shaping tasks should be relate to the subject’s movement prob-
selected for patients by considering lems (i.e., in the flipping dominoes
(a) specific joint movements that task, if the subject’s most significant
exhibit the most pronounced deficits, deficits are in thumb and forefinger
(b) the joint movements that trainers dexterity, the task progression should
believe have the greatest potential for involve using, depending on the nature
improvement, and (c) patient prefer- of the deficit, either larger or smaller
dominoes. If the subject’s most signif- Also, note any placement changes on
icant movement deficits are at the the data sheet when a shaping task is
shoulder, the task progression should made more difficult.
involve moving the dominoes farther To quantify a shaping task, only
away). one shaping progression parameter
Shaping tasks are made more difficult can be allowed to vary. For example,
when it is clear that, for the most part, on an elbow extension task, there
the patient will be able to accomplish the would be three parameters: time,
task, though with effort. number of repetitions, and distance.
Positive reinforcement or reward The time and number of the repeti-
should be provided visually (i.e., tions can be held constant and the
keeping the shaping data form in distance can be slowly increased until
plain view of the patient so that he or the subject can no longer perform a
she can see performance history and specified number of extensions in a
‘‘personal best’’; task performance given period of time (e.g., 10 exten-
becomes like an arcade game). Task sions in 30 s). Alternatively, distance
performance information should can be held constant (e.g., 10 in.) and
also be given verbally at frequent the subject would be encouraged to
intervals. progressively increase the number of
An important function of the repetitions in a set period of time
trainer is to act as a cheerleader, (e.g., 30 s). For a given task, more
continuously encouraging the subject than one parameter should not be
on a moment-to-moment basis to varied at the same time (e.g., both
keep improving the performance.
distance and number of repetitions).
Performance regressions are never
If the trainer feels that the subject
punished and are usually ignored.
would benefit from varying a second
If a patient is experiencing exces-
parameter, that is permissible. How-
sive difficulty with a task, a simpler
task involving similar movements can ever, it should be understood that
be substituted. this training now must be quantified
Rest intervals should be allowed as a new entity on separate data
during each shaping session. The rest sheets.
period is usually the same length as
the trial period, although longer Example of Shaping Tasks:
intervals are sometimes needed to Flipping Dominoes
prevent fatigue.
Trainers should rate the perfor- Activity description:
mance of each shaping task trial Approximately 25 dominoes are
using the quality-of-movement scale placed in front of the subject. The
attached. subject is asked to reach forward and
The results of each shaping task flip the dominoes using either fore-
trial, including quality-of-movement arm pronation or supination. The
rating, should be recorded on the correct movement can be best isolat-
shaping data form. ed by asking the subject to rest his or
Encouragement and quality-of- her forearm on the table during the
movement recordings should be given task.
to the subject verbally on at least 50%
of the trials. Potential shaping progression:
Placement of equipment used in Placing the dominoes farther away to chal-
shaping tasks should be recorded on lenge elbow extension.
the shaping data form so that the Using larger or smaller dominoes to challenge
task can be duplicated. Adhesive wrist and finger control.
markers on the task performance Place dominoes on a box to challenge shoulder
table can be used for this purpose. flexion.
Authors Queries
Journal: The Behavior Analyst
Paper: bhan-35-02-03
Title: The Behavior-Analytic Origins of Constraint-Induced Movement
Therapy: An Example of Behavioral Neurorehabilitation
Dear Author
During the preparation of your manuscript for publication, the questions listed
below have arisen. Please attend to these matters and return this form with your
proof. Many thanks for your assistance