the behavior-analytic origins of constraint-induced movement therapy… · the behavior-analytic...

25
The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation Edward Taub University of Alabama at Birmingham Constraint-induced (CI) therapy is a term given to a family of efficacious neurorehabilitation treatments including to date: upper extremity CI movement therapy, lower extremity CI movement therapy, pediatric CI therapy, and CI aphasia therapy. The purpose of this article is to outline the behavior analysis origins of CI therapy and the ways in which its procedures incorporate behavior analysis methods and principles. The intervention is founded on the concept of learned nonuse, a mechanism now empirically demonstrated to exist, which occurs after many different types of damage to the central nervous system (CNS). It results from the dramatic alteration of the contingencies of reinforcement that results from substantial CNS damage and leads to a greater deficit than is warranted by the actual damage sustained. CI therapy produces a countervailing alteration in the contingencies of reinforcement. The intervention has been used successfully to substantially improve motor deficits after stroke, traumatic brain injury, spinal cord injury, multiple sclerosis, with cerebral palsy in a pediatric population, and for language impairment in poststroke aphasia. The protocol of CI therapy consists primarily of standard behavior-analytic methods. It produces a marked plastic brain change that is correlated with its therapeutic effect, and therefore provides an example of the way in which behavior change can contribute to a profound remodeling of the brain. CI therapy may be viewed as an example of behavioral neurorehabilitation. Key words: CI therapy, CI movement therapy, CI aphasia therapy, stroke, central nervous system injury, neurorehabilitation, behavior analysis Constraint-induced movement the- rapy (CIMT) is a family of neuror- ehabilitation treatments developed at the University of Alabama at Bir- mingham (UAB). It involves the application of behavior-analytic tech- niques to the improvement of deficits that result from different types of substantial damage to the central nervous system (CNS), such as stroke, traumatic brain injury, spinal cord injury, multiple sclerosis, cere- bral palsy, and other pediatric motor disorders (summarized in Taub & Uswatte, 2009; Taub, Uswatte, & Pidikiti, 1999). The deficits treated are mainly motor in nature but also include verbal behavior in aphasia and phantom limb pain after limb amputation. The first application of CI therapy was to motor deficit after stroke (Taub et al., 1993), and this continues to be the most frequent application. Its efficacy has been demonstrated by a multisite random- ized controlled trial (RCT; Wolf et al., 2006), which is rare for the rehabil- itation field, and multiple single-site RCTs. There are now well over 300 The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:15 155 Cust # MS 12-105 This research was supported by Grant HD34273 from the National Institutes of Health, Grants W98 0410 and B2490T from the U.S. Department of Veterans Affairs, Grant RG 4221-A-201 from the Multiple Sclerosis Society, Grants 0365163B, 0815065E, and 0715450B from the American Heart Association Southeast Affiliate, and Grant 97-41 from the James S. McDonnell Foundation. I thank the following collabora- tors: Gitendra Uswatte, Neal E. Miller, Victor Mark, David Morris, Jean E. Crago, Angi Griffin, Mary M. Bowman, Staci Bishop- McKay, Danna Kay King, Sonya Pearson, Camille Bryson, Michelle Spear, Adriana Delgado, Francilla Allen, Christy Bussey, Margaret Johnson, Leslie Harper, Jamie Wade, Edwin W. Cook, III, and Louis D. Burgio. I also thank Gitendra Uswatte and Edgar E. Coons for critical and insightful readings of this manuscript. Correspondence concerning this article should be addressed to Edward Taub, Uni- versity of Alabama at Birmingham, 1530 3rd Ave. S, CPM 712, Birmingham, Alabama 35294 (e-mail: [email protected]). The Behavior Analyst 2012, 35, 155–178 No. 2 (Fall) 155

Upload: phungdiep

Post on 13-Apr-2018

218 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

The Behavior-Analytic Origins of Constraint-InducedMovement Therapy: An Example of

Behavioral Neurorehabilitation

Edward TaubUniversity of Alabama at Birmingham

Constraint-induced (CI) therapy is a term given to a family of efficacious neurorehabilitationtreatments including to date: upper extremity CI movement therapy, lower extremity CImovement therapy, pediatric CI therapy, and CI aphasia therapy. The purpose of this article isto outline the behavior analysis origins of CI therapy and the ways in which its proceduresincorporate behavior analysis methods and principles. The intervention is founded on theconcept of learned nonuse, a mechanism now empirically demonstrated to exist, which occursafter many different types of damage to the central nervous system (CNS). It results from thedramatic alteration of the contingencies of reinforcement that results from substantial CNSdamage and leads to a greater deficit than is warranted by the actual damage sustained. CItherapy produces a countervailing alteration in the contingencies of reinforcement. Theintervention has been used successfully to substantially improve motor deficits after stroke,traumatic brain injury, spinal cord injury, multiple sclerosis, with cerebral palsy in a pediatricpopulation, and for language impairment in poststroke aphasia. The protocol of CI therapyconsists primarily of standard behavior-analytic methods. It produces a marked plastic brainchange that is correlated with its therapeutic effect, and therefore provides an example of theway in which behavior change can contribute to a profound remodeling of the brain. CI therapymay be viewed as an example of behavioral neurorehabilitation.

Key words: CI therapy, CI movement therapy, CI aphasia therapy, stroke, central nervoussystem injury, neurorehabilitation, behavior analysis

Constraint-induced movement the-rapy (CIMT) is a family of neuror-ehabilitation treatments developed at

the University of Alabama at Bir-mingham (UAB). It involves theapplication of behavior-analytic tech-niques to the improvement of deficitsthat result from different types ofsubstantial damage to the centralnervous system (CNS), such asstroke, traumatic brain injury, spinalcord injury, multiple sclerosis, cere-bral palsy, and other pediatric motordisorders (summarized in Taub &Uswatte, 2009; Taub, Uswatte, &Pidikiti, 1999). The deficits treatedare mainly motor in nature but alsoinclude verbal behavior in aphasiaand phantom limb pain after limbamputation. The first application ofCI therapy was to motor deficit afterstroke (Taub et al., 1993), and thiscontinues to be the most frequentapplication. Its efficacy has beendemonstrated by a multisite random-ized controlled trial (RCT; Wolf et al.,2006), which is rare for the rehabil-itation field, and multiple single-siteRCTs. There are now well over 300

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:15 155 Cust # MS 12-105

This research was supported by GrantHD34273 from the National Institutes ofHealth, Grants W98 0410 and B2490T fromthe U.S. Department of Veterans Affairs,Grant RG 4221-A-201 from the MultipleSclerosis Society, Grants 0365163B,0815065E, and 0715450B from the AmericanHeart Association Southeast Affiliate, andGrant 97-41 from the James S. McDonnellFoundation. I thank the following collabora-tors: Gitendra Uswatte, Neal E. Miller, VictorMark, David Morris, Jean E. Crago, AngiGriffin, Mary M. Bowman, Staci Bishop-McKay, Danna Kay King, Sonya Pearson,Camille Bryson, Michelle Spear, AdrianaDelgado, Francilla Allen, Christy Bussey,Margaret Johnson, Leslie Harper, JamieWade, Edwin W. Cook, III, and Louis D.Burgio. I also thank Gitendra Uswatte andEdgar E. Coons for critical and insightfulreadings of this manuscript.

Correspondence concerning this articleshould be addressed to Edward Taub, Uni-versity of Alabama at Birmingham, 1530 3rdAve. S, CPM 712, Birmingham, Alabama35294 (e-mail: [email protected]).

The Behavior Analyst 2012, 35, 155–178 No. 2 (Fall)

155

Page 2: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

CI therapy studies that have reportedpositive results for improving motordeficit after stroke. Its use is thereforebeginning to spread.

CI therapy basically involves theuse of operant training techniques ina rehabilitation context. The origin ofthe therapy is described in publica-tions from this laboratory, but it isnot well recognized or understoodand is therefore often overlooked, themain reason probably being thatthere is little familiarity with behavioranalysis in the fields associated withneurorehabilitation.

The theoretical roots of CI therapyemerged from principles developedduring graduate work at ColumbiaUniversity with Fred Keller and W.Schoenfeld. The initial laboratorywork that led to CI therapy beganin the Department of ExperimentalNeurology in a research institute atthe Jewish Chronic Disease Center inBrooklyn, New York. Monkeys re-ceived a surgical abolition of somaticsensation from one or both forelimbs,and then were given training based,in part, on operant learning princi-ples. Work continued at the Institutefor Behavioral Research (IBR) inSilver Spring, Maryland. The Chair-man of the Board of IBR was JosephV. Brady, who played a leading rolein founding behavioral pharmacolo-gy. The translation of CI therapyfrom monkeys to humans was stim-ulated by Brady’s example. CI ther-apy can be viewed as a type ofbehavioral neurorehabilitation.

DEAFFERENTATIONIN MONKEYS

When somatic sensation is abol-ished from a single forelimb inmonkeys by the serial section of allsensory roots of spinal nerves inner-vating that extremity, the monkeynever again uses the deafferentedlimb. This is the case even thoughthe motor outflow over the ventralroots of spinal nerves is left intact.This was a classic observation, made

first by Mott and Sherrington (1895)and subsequently replicated (Lassek,1953; Twitchell, 1954). It formed oneof the major pillars underlying Sher-rington’s formulation of the reflexo-logical position (Sherrington, 1910),which became one of the dominantpositions in neurology for the first70 years of the 20th century. Howev-er, we showed that there were twobehavioral techniques that could in-duce a monkey to make use of asingle deafferented forelimb.

One technique was training of thedeafferented extremity. At first adiscrete-trial avoidance conditioningprocedure was used. The monkey hadto make a simple flexion of thedeafferented limb at the sound of abuzzer (Knapp, Taub, & Berman,1959, 1963) or click (Taub & Berman,1963, 1968) to avoid an electricshock. When the research shifted tothe IBR, shaping was used. It provedto be a particularly effective means ofimproving the motor deficit of thedeafferented extremity. When dis-crete-trial procedures were used,transfer of limb use from the condi-tioning chamber to the colony envi-ronment was never observed (Taub &Berman, 1963, 1968; Taub, Ellman,& Berman, 1966; Taub, Goldberg, &Taub, 1975; Taub, Williams, Barro,& Steiner, 1978). However, whenmanual shaping with food rewardwas employed in subsequent experi-ments, there was a substantial im-provement of movement in the lifesituation as well (Taub, 1977). Theactions shaped included pointing atvisual targets (Taub et al., ;1975) andprehension in juveniles deafferentedon day of birth (Taub, Perrella, &Barro, 1973) and prenatally (Taub,Perrella, Miller, & Barro, 1975) thathad never exhibited prehension pre-viously. In both cases, the manualshaping-with-food-reward procedureproduced an almost complete rever-sal of the motor disability, whichprogressed from total absence of thetarget behavior to very good (al-though not normal) behavior. In the

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:26 156 Cust # MS 12-105

156 EDWARD TAUB

Page 3: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

case of thumb–forefinger prehension,this took approximately 30 half-hoursessions. The steps involved in theshaping progression are described inAppendix A.

Another technique resulting in useof the deafferented limb was restraintof the intact limb while the deaf-ferented limb was left free (Knappet al., 1959, 1963; Taub & Berman,1968). This rendered the animalvirtually helpless. However, withinseveral hours of the imposition ofrestraint, the animal began to use thedeafferented extremity extensively. Ifthe restraint was left in place for1 week, then on removal of therestraint the animal continued touse the limb when the intact limbwas not restrained, and that use waspermanent. The movements were notnormal; they were clumsy becausesomatic sensation had been abol-ished, but they were extensive andeffective (Taub, 1977, 1980).

Thus, both the training and shap-ing conditions and the situation inwhich the intact limb was restrainedinduced the monkeys to make purpos-ive use of the deafferented extremity.In trying to understand whether thesetwo experimental manipulations in-volved a common mechanism, it wasnoted that in the unrestricted colonyenvironment the monkeys were free touse the intact limb to accomplishobjectives including those normallycarried out by both forelimbs inconcert, rather than attempting tocoordinate use of the intact forelimbwith an impaired extremity. However,both the restraint and the trainingsituation reversed the contingencies ofreinforcement. Either the monkeysused the deafferented limb or theywere punished: In the training situa-tions, they were either subjected to anoxious electric shock or could notobtain food or water reinforcementwhen 22 hr hungry or thirsty; in therestraint situation, they were renderedvirtually helpless. Consequently, themonkey used the deafferented limb.

This set of results seemed toresolve the enigma posed by theabsence of purposive movement bya deafferented limb posed originallyby the Mott and Sherrington exper-iment (1895): Why didn’t the mon-keys use a single deafferented limb?Sherrington’s reasonable answer hadbeen that extremity deafferentationinterrupted the afferent limb of spinalreflexes, and it was this that abol-ished use of the extremity eventhough motor innervation remainedintact. Hence the idea emerged thatspinal reflexes were the basic buildingblocks from which behavior waselaborated, which was the fundamen-tal tenet of Sherringtonian reflexolo-gy. This was a pervasive view fordecades, whose influence, as theexemplar of the ‘‘peripheralist posi-tion,’’ extended into a number ofbehaviorist systems. For example,the second half of the first chapterof Skinner’s (1938) The Behavior ofOrganisms is devoted to Sherring-ton’s laws of the reflex. However, thetwo simple behavioral techniquesnoted above (and later control exper-iments) showed that this formulationcould not be correct. What thencould account for the absence ofpurposive movement after unilateralforelimb deafferentation? The need toaddress that salient question led tothe formulation of the concept oflearned nonuse.

LEARNED NONUSE

Several converging lines of evi-dence suggested that the nonuse of asingle deafferented forelimb in mon-keys is a learning phenomenon thatinvolves a conditioned suppression ofmovement that was termed learnednonuse (LNU). The restraint andtraining techniques appeared to beeffective because they overcame LNU(Taub, 1977, 1980; Taub, Uswatte,Mark, & Morris, 2006).

Substantial neurological injury usu-ally leads to a depression of CNSexcitability and a consequent reduction

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:26 157 Cust # MS 12-105

ORIGINS OF CI THERAPY 157

Page 4: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

or even elimination of the motor orsensory function with which the affect-ed CNS area is associated. Subsequent-ly, a spontaneous recovery of CNSexcitability takes place, but this processcan require considerable time in bothnonhuman and human primates(Taub, 1980, Taub, Heitman, & Barro,1977). Thus, immediately after surgicaldeafferentation of a single forelimb,monkeys cannot use that extremity.Efforts to use it often lead to painfuland otherwise aversive consequences,such as uncoordination and falling,loss of food objects, and in general,failure of any activity attempted withthe deafferented limb. Many learningexperiments have demonstrated thatpunishment has the effect of suppress-ing the behavior that precedes it (Azrin& Holz, 1966; Catania, 1998; Estes,1944). The monkeys, meanwhile, getalong reasonably well in the laboratoryenvironment on three limbs and aretherefore reinforced for this pattern ofless effective compensatory behaviorthat, as a result, is strengthened. Thus,the response tendency to not use theaffected limb persists and, consequent-ly, monkeys never learn that the limbhad become potentially useful severalmonths after surgery. The mechanism

by which LNU develops is depictedschematically in Figure 1.

When the movements of the intactlimb are restricted beginning severalmonths or longer after unilateral deaf-ferentation, the situation is changeddramatically. Animals either must usethe deafferented limb or cannot withany degree of efficiency feed them-selves, locomote, or carry out largeportions of their daily activities. Thisnew constraint on behavior increasesthe tendency to use the deafferentedlimb, thereby overcoming LNU. More-over, current ongoing conditions, suchas the relative inefficiency of theaffected upper extremity comparedwith the unaffected forelimb, continueto affect the contingencies of reinforce-ment associated with use of the affectedextremity. If the movement-restrictiondevice is removed a short while afterthe early display of purposive move-ment, the newly learned use of thedeafferented limb will have acquiredlittle strength and is quickly over-whelmed by the well-learned tendencynot to use the limb. However, if themovement-restriction device is left onfor several days or longer, use of thedeafferented limb acquires strengthand, then when the device is removed,

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:26 158 Cust # MS 12-105

Figure 1. Schematic model for the development of learned nonuse.

158 EDWARD TAUB

Page 5: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

can compete successfully with thestrongly overlearned nonuse of thatlimb. The counterconditioning of LNUis depicted schematically in Figure 2.

The training situations described inthe previous section, just like therestriction of the intact limb, placemajor constraints on the animals’behavior. In the discrete-trial trainingsituation, if the monkeys do notperform the required response withthe deafferented limb, they are eitherpunished by such aversive conse-quences as falling to the deafferentedside, or do not receive food pellets orfluid when hungry or thirsty, respec-tively. Similarly, during shaping, re-ward is contingent on making pro-gressively improved movements withthe deafferented limb. The monkeyscannot get by using just the intactforelimb as they can in the colonyenvironment. These new sets ofconditions, just like the movement-restriction device, constrain the ani-mals to use their deafferented limbsto avoid punishment or obtain re-ward and thereby induce the use ofthe deafferented limb. As a result,LNU is overcome.

As noted, use of the deafferentedlimb does not transfer from thediscrete-trial situation to the colonyenvironment. This lack of transfer

may be due to the restriction oftraining in the conditioning paradigmto a few specific movements within anarrow context, with the result thatarm use is not generalized to a varietyof movements or situations. Theshaping situation, however, is moreflexible and free-form; there is freedomfor the animal to use many differenttypes of movement and movementstrategies to attempt to achieve behav-ioral objectives that are differentiallyreinforced. Therefore, what is learnedin the shaping situation transfers tothe colony environment and evengeneralizes to movement categoriesother than those trained. The shapingprocess appears to provide a bridgefrom the training to the life situation.

Direct Test of the LNU Hypothesis

An experiment was carried out totest the LNU formulation directly(Taub, 1977, 1980). Movement of aunilaterally deafferented forelimbwas prevented immediately after sur-gery with a restraining device inseveral animals so that they couldnot attempt to use that extremity fora period of 3 months. Restraint wasbegun while the animals were stillunder anesthesia. The reasoning wasthat by preventing animals from

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:27 159 Cust # MS 12-105

Figure 2. Schematic model of mechanism for overcoming learned nonuse.

ORIGINS OF CI THERAPY 159

Page 6: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

using the deafferented limb duringthe period before spontaneous recov-ery of function had taken place, theywould not learn that the limb couldnot be used. LNU of the affectedextremity should therefore not devel-op. In addition, the intact limb wasrestrained for the same period so thatthe animals could not receive rein-forcement for use of that extremityalone. In conformity with the predic-tion, which without the LNU formu-lation would have been counterintu-itive, the animals were able to usetheir deafferented extremity in thefree situation after the restraint wasremoved 3 months after surgery, andthis was permanent, persisting for therest of the animals’ lives.

Suggestive evidence in support ofLNU was also obtained during deaf-ferentation experiments carried outprenatally< (Taub et al., 1973, 1975).Life in the physically restrictive uter-ine environment imposes major con-straints on the ability to use theforelimbs for such purposes as alter-ing body orientation to adjust forshifts in maternal position. Althoughuse of the fetal limbs is not preventedentirely, their movement is restrictedin utero, thereby functioning like asling or a padded mitt in a human CItherapy experiment (to be discussedbelow). Four animals were studiedwho had received unilateral forelimbdeafferentation by an intrauterineapproach during the prenatal period;three when two-thirds the waythrough gestation and one whentwo-fifths the way through gestation.These animals exhibited functionaluse of the deafferented extremityfrom the first day of extrauterine life,in contrast to animals deafferentedafter maturity that did not use theaffected extremity unless given train-ing of the deafferented arm orrestraint of the intact arm. At birth,the prenatally deafferented animalsall used that limb for postural sup-port during ‘‘sprawling’’ and forpushing into a sitting position. Sub-sequently, although the intact limb

was never restrained, the ability touse the deafferented limb continuedto develop as the animals matureduntil it was similar to the extensive(though impaired) use of a limb whenthey were given limb deafferentationas adults. This, then, constitutes asecond line of evidence that supportsthe LNU formulation.

Translation of the LNU Model fromDeafferentation in Monkeys to CNSInjury in Humans

The results of the experiments de-scribed above show that simple behav-ior-analytic techniques employed indiscrete-trial or shaping contexts re-sulted in the conversion of a uselessdeafferented upper extremity to a limbthat could be used extensively. Later, itbecame apparent that this could beviewed as a rapid and substantialrehabilitation of movement (althoughthat term was not usually applied toprimates at the time). Thus, it appearedpossible that the same two techniquesmight be appropriately used to reha-bilitate motor disability in humans. Animplication of the concept of LNU asthe outcome of the punishments andrewards that result from attempted useof an impaired extremity is that itshould, in principle, operate after anyCNS injury when the initial effect is totemporarily abolish movement, re-gardless of the injury’s location orextent. There was also no a priorireason to suppose that it would notoperate in humans as well as monkeys.Specifically, stroke often leaves pa-tients with an apparently permanentloss of function in an upper extremity,although the limb is not paralyzed. Inaddition, the motor impairment ispreponderantly unilateral. These fac-tors are similar to those that pertainafter unilateral forelimb deafferenta-tion in monkeys. Therefore, it seemedreasonable to formulate a protocolthat simply transferred the behavior-analytic techniques used for overcom-ing LNU of a deafferented limbin monkeys to humans who had

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:27 160 Cust # MS 12-105

160 EDWARD TAUB

Page 7: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

experienced a cerebrovascular acci-dent (Taub, 1980; Taub, Uswatte,Mark, et al., 2006).

At the time that the LNU mecha-nism was proposed (Taub, 1977;Taub & Berman, 1968) and itstransfer to humans suggested (Taub,1980), there had been little attempt totranslate basic research findings inneuroscience to pathological condi-tions in humans, and none as far as Iknow in neurorehabilitation. Howev-er, given a belief in the generality ofthe laws of behavior for all mammals,and the example of Brady’s pioneer-ing use of operant conditioningtechniques to evaluate the effect ofpharmacological agents on behaviorin animals and subsequent use of theresultant data as a basis for estimat-ing the possible effects of thoseagents on humans, the translation ofwhat was to be called CI therapyfrom monkeys to humans seemedstraightforward. The nervous systemsmight be different, but the principlesof behavior were the same.

A series of studies was then carriedout starting in 1986 (Taub et al.,1993) in which chronic stroke pa-tients were trained in the laboratory,initially for 6 hr per day with an hourof interpolated rest for 10 consecutiveweekdays; later it was found that 3 hrper day of training for 10 consecutiveweekdays was equally effective (Taubet al., 1999). In addition, the lessaffected arm was restrained for arequested 90% of waking hours (i.e.,both in the laboratory and at home).A timer was inserted into the re-straining device that was activated bycontact with the hand so there was anobjective record of the amount oftime the patient was compliant withthe instruction to wear the restraintoutside the laboratory.

The primary training techniqueemployed was shaping (Taub et al.,1994), which had been so successfulwith deafferented monkeys. In addi-tion, a set of behavioral techniquestermed the transfer package (TP;described below) was also employed

to promote transfer of the improve-ments in motor ability achieved in thelaboratory to the life situation. Allsubjects in the studies were at least1 year poststroke (M 5 4.4 years).The subjects in the early studies hadupper extremity motor deficits thatcould be characterized as mild ormoderate, which actually involves asubstantial deficit compared to nor-mal motor function. Two of the earlystudies employed placebo controlgroups (Taub et al., 1993; Taub,Uswatte, King, et al., 2006). Theresults showed that the treatmentproduced very large improvementsin the patients’ ability to use the moreaffected arm, just as in the case of thedeafferented monkeys.

To date, several hundred subjectswith mild or moderate stroke symp-toms have been treated with CI therapyin the UAB laboratory. The magnitudeof the treatment change can be evalu-ated with effect size (ES) statistics. Byconvention, a d9 statistic of 0.47 isconsidered large (Cohen, 1988). TheES for the treatment change on ourmeasure of actual use of the moreaffected arm in the life situationranged from a d9 of 2.1 to 4.0 (M 53.3), depending on the experiment. Ina key experiment (Taub, Uswatte,King, et al., 2006), the amount ofreal-world spontaneous arm use com-pared to before stroke increased from9% prior to treatment to 52% aftertreatment, a more than five timesimprovement. Similar results havebeen obtained in other experimentsfrom this laboratory.

The deafferented monkeys in theexperiments in which the CI therapyrehabilitation techniques were devel-oped were all in the chronic phase,more than 1 year after their surgicalprocedures. It therefore seemed thatthese techniques should work wellwith patients in the chronic phase ifthe translation of the approach wasefficacious at all. However, the gen-eral, essentially axiomatic belief inthe rehabilitation field at the timewas that the impaired movement of a

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:27 161 Cust # MS 12-105

ORIGINS OF CI THERAPY 161

Page 8: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

stroke victim could not be modifiedin the chronic phase, no matter whattechnique was employed. This viewstill has considerable force. Eventoday, after 25 years of research andclinical practice, a substantial per-centage of the chronic patients whocome to the UAB CI therapy clinicfor treatment have been told by theirphysicians and therapists that there isnothing that can be done to improvetheir motor deficit.

Naming the Treatment

The movement-restriction andtraining situations share a commonfeature: They both are powerfulmeans of inducing use of the moreaffected arm. One procedure physi-cally restrains the less affected arm sothat the individual, whether monkeyor human, must use the more affectedextremity to avoid being renderedhelpless and thereby subject to mul-tiple sources of punishment. Theother method, training, induces useof the more affected arm by alteringthe contingencies of reinforcementso that it must be used in order toobtain reinforcement or, in monkeys,to avoid punishment. Thus, bothprocedures constitute constraints thatpromote use of the more affected armby a major alteration of the contin-gencies of reinforcement. Althoughthe name is accurate, the use of theterm constraint in the title of thetherapy has turned out to be confus-ing. The rehabilitation field wasnot used to thinking of training asimposing a constraint on behavior.Instead, the large majority of profes-sionals interpreted the focal word inthe name of the therapy as being analternate way of saying ‘‘restraint,’’so that the general impression arosethat restraint of the less affected armwas the central and most importantfeature of the therapy. As indicatedbelow, that is very far from beingtrue; physical restraint of the lessaffected arm can be dispersed withentirely in achieving a maximal result

if the training conditions are ar-ranged appropriately. Recently, thefield has begun to accept that theword constraint is meant to includetraining. The use of this term isconsistent with Timberlake’s analysis(1993) of reinforcement as constitut-ing ‘‘constraint of a functional causalsystem comprised of multiple interre-lated causal sequences, complex link-ages between causes and effects and aset of initial conditions’’ (p. 105).

COMPONENTS OF CI THERAPY

As noted, the CI therapy protocolincorporates a number of proceduresthat are commonly used in behavior-al approaches to modifying behavior.First, shaping is used in the labora-tory so that movement of the moreimpaired extremity is brought tomore closely approximate that ofindividuals who have not suffered aneurological injury. (The manualused for training therapists in theprocedures of CI therapy includesshaping plans for 52 training tasks;nine are given in Appendix B.) Inaddition, and perhaps most impor-tant, a set of behavioral techniques(the TP) is used to increase thefrequency with which the more af-fected extremity is used spontaneous-ly in the performance of activities inthe real-world environment. The ob-jective of the TP is to effect transferor generalization of gains that aremade in the laboratory to the lifesituation and to then make themhabitual. For the interested reader,further description of the methodsemployed in the CI therapy protocolmay be found in the followingpapers: Taub et al. (1994); Morrisand Taub (2008); Taub, Uswatte,Mark, et al. (2006). An extended casestudy is attached to Morris and Taub(2008) as an =appendix.

Training of the More AffectedExtremity: Shaping

A standard approach to shaping isemployed in which a behavioral

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:27 162 Cust # MS 12-105

162 EDWARD TAUB

Page 9: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

objective is approached in small stepsby successive approximations (Mor-gan, 1974; Panyan, 1980; Skinner,1938, 1968). A task is made moredifficult in accordance with a pa-tient’s progressively increasing motorcapabilities; alternatively, the require-ment for speed of performance isincrementally increased (Taub et al.,1994). A battery of over 100 shapingtasks has been developed with apreliminary written shaping plan foreach. (See Appendix B for an enu-meration of the principles used inshaping and a description of sampleshaping tasks and generic shapingplans for each.) Each subject’s pro-gram is individualized by selectingapproximately 12 tasks from thelarger battery and creating new oneswhen it seems that they would beadvantageous for that individual.The selection of tasks for each persondepends on (a) specific joint move-ments that exhibit the most promi-nent deficit; (b) joint movements thatproject staff believe have the greatestpotential for improvement; and (c)the subject’s preference among tasksthat have a similar potential forproducing specific improvements.Prior to treatment, a patient’s func-tional movement capacity and thenature of his or her impairment aredetermined on an individual jointmovement basis during a systematicevaluation and are recorded on astandardized form.

In developing the battery, care wastaken to select tasks that can bebroken down into subtasks and canbe objectively measured even whensmall improvements are noticeable.Each activity is usually practiced fora set of 10 trials (30 s each) andexplicit, immediate feedback is pro-vided regarding the subject’s perfor-mance on each trial. When the levelof difficulty of a shaping task isincreased, the parameter selected forchange relates to the participant’smovement problems, as determinedin the course of training by thetherapist. For example, if the partic-

ipant’s most significant movementdeficits are with thumb and fingerdexterity and an object-flipping taskis used, the difficulty of the taskwould be increased by making theobject progressively smaller if themovement problem is in thumbadduction and finger flexion (i.e.,making a pincer grasp); in contrast,if the movement problem involvesthumb abduction and finger exten-sion (i.e., releasing a pincer grasp),the difficulty of the task would beincreased by making the object pro-gressively larger. As another exam-ple, if there is a significant deficit inelbow extension and a pointing orreaching task is used, the shapingprogression might involve placing thetarget object at increasing distancesfrom the participant.

In the shaping progression, theamount of task-difficulty increase issuch that it is likely that the partic-ipant will at each step be able toaccomplish the task, although witheffort. This incremental increase indifficulty often makes it possible toachieve a given objective that mightnot be attainable if several largeincrements in motor performancewere required. Coaching is providedliberally throughout all shaping pro-cedures, including the usual tech-niques of cuing and prompting.Modeling is also employed as needed.Verbal reinforcement is providedenthusiastically at frequent intervals(e.g., ‘‘that’s excellent,’’ ‘‘first class,’’‘‘keep trying’’). Criticism is nevermade; poor performance is generallyignored and further efforts at im-provement are encouraged.

The Transfer Package

One of the overriding goals of CItherapy is to achieve transfer oftherapeutic gains made in the re-search or clinical setting to theparticipant’s real-world environment.It could almost be said that if whatpatients learn in the clinic is notgeneralized to the life situation, then

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:27 163 Cust # MS 12-105

ORIGINS OF CI THERAPY 163

Page 10: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

rehabilitation is not really beingaccomplished. When CI therapy re-search was begun, there were nomethods or tests being used to assesshow or whether a patient was using astroke-affected extremity in the lifesituation. However, a behavior-ana-lytic approach made it intuitivelyobvious that a primary goal ofrehabilitation treatment had to bethe development of methods to in-duce use of the more affected arm inthe life situation and then monitoringthat use. This was the case indepen-dent of considerations relating toLNU, although these certainly rein-forced the need for real-world mon-itoring of behavior; hence, the devel-opment of the motor activity log(MAL; (Taub et al., 1993; Uswatte,Taub, Morris, Light, & Thompson,2006; Uswatte, Taub, Morris, Vig-nolo, & McCulloch, 2005). The MALresults have been confirmed by accel-erometry data from transducers wornon both arms for 3 days before and3 days after the end of treatment(Uswatte, Miltner, et al., 1997, 1998;Uswatte, Spraggins, Walker, Cal-houn, & Taub, 1997). The TP con-sists of the following techniques.

Behavioral contract. At the outsetof treatment, the therapist negotiatesa contract with the patient (andseparately with the caregiver, if oneis available) in which agreement isreached that the patient will use hisor her more impaired extremity asmuch as possible outside the labora-tory. Specific activities during whichthe patient will practice using themore impaired extremity are dis-cussed, agreed on, and written down.At the end of this process, thenegotiated document is signed bythe patient (or caregiver), the thera-pist, and a witness to emphasizethe character of the document asa contract.

Daily home diary. During treat-ment, the patients catalogue on adaily diary form how much they haveused the more affected arm for theactivities specified in the behavioral

contract. The diary is kept for thepart of the day spent outside thelaboratory and is reviewed in detaileach morning with the therapist.

Daily administration of the MAL.The MAL collects information aboutuse of the more affected extremity in30 important activities of daily living(ADL) in all major domains ofeveryday life. The daily repetition ofthis detailed report, which is probedand verified in a number of ways,serves to keep the patient’s attentionon the use of the more affectedextremity outside the laboratory orclinic.

Problem solving. During adminis-tration of the MAL, the therapisthelps patients analyze, circumvent, orovercome any barriers to using themore impaired arm in the life situa-tion. For example, if the patient isconcerned about spilling liquid froma glass, the therapist may suggestfilling the glass only half way. Ifpatients use the less affected arm formanipulating eating utensils in arestaurant because they are embar-rassed by dropping food from autensil onto a table, the therapistmay suggest not going to a restaurantduring the course of the treatment.

Home skill assignments. Duringtreatment, subjects are asked to carryout at home five difficult (for them)ADL tasks and five easy tasks usingthe more affected arm, selected dailyfrom a list of approximately 200 (e.g.,brush teeth, wash hands, use TVremote). In addition, patients areasked to spend 15 to 30 min at homeon a daily basis repetitively perform-ing with their more affected armspecific upper extremity tasks thatare similar to those performed in thelaboratory or clinic. The tasks arechosen for practice to improve themost significant movement deficits.Subjects check off the ADL activitiesand exercises carried out on a formprovided to them each day.

Weekly telephone contacts withpatients. For the first month afterthe end of treatment, the MAL is

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:27 164 Cust # MS 12-105

164 EDWARD TAUB

Page 11: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

administered by phone, and problemsolving is carried out.

Posttreatment practice. Toward theend of treatment, an individualizedposttreatment home practice programof approximately 100 tasks is developedand given to the patients. They areencouraged to perform two or threetasks for 10 min daily after the treat-ment period, but to continually focusattention on using the more affectedarm in ADL whenever possible.

In most physical rehabilitationregimens, there is a passive element;the patient is responsible for carryingout the therapist’s instructions pri-marily or only during treatmentsessions. A major difference in CItherapy is the involvement of thepatient as an active participant in allrequirements of the therapy, not onlyduring the treatment sessions but alsoat home during the treatment periodand for the first month after labora-tory therapy has been completed (andafterward, although this is not mon-itored). The TP makes patients re-sponsible for adhering to the require-ments of the therapy, and therefore ineffect they become responsible fortheir own improvement.

The TP is the main way in which CItherapy differs from other rehabilita-tion procedures. Its critical importancein producing a large treatment effectwas recently demonstrated (Gauthieret al., 2008). Twenty subjects weregiven the full CI therapy protocolincluding the TP. A second groupreceived the same treatment in thelaboratory, but none of the TP tech-niques were administered. Both groupsshowed a significant increase in theamount of spontaneous use of themore affected arm in the life situation,but the improvement of the CI therapyTP group was approximately 2.5 timesas great as the improvement recordedfor the non-TP group.

Less Affected Limb Restraint

In initial experiments, limb re-straint was achieved using a rigid

resting hand splint and a sling (Taubet al., 1993). However, this level ofrestraint was found to be unneces-sary, and currently a mitt with aheavily padded palmar surface isemployed. It prevents the use of thefingers and hand for a target of 90%of waking hours and gives as goodresults as the resting hand splint andsling arrangement. This was, and stillis, generally considered to be thesignature if not the differential com-ponent of CI therapy. This is unfor-tunate, because there is evidence thatless affected limb restraint is notnecessary or even important forproducing a maximal treatment effect(Taub et al., 1999; Uswatte, Taub,Morris, Barman, & Crago, 2006).However, although less affected limbrestraint is not necessary with adulthumans, it is important for monkeysand young children (pediatric CItherapy), who have less capacity forself-suppression of behavior and de-ferral of reinforcement. Even in adulthumans, when restraint of the lessaffected arm is used, it may makesome contribution to promoting along-lasting increase in use of themore affected arm in the home. Thisis a clinical opinion not based on acontrolled study, but it is thought tobe a sufficiently real possibility thatuse of the restraining mitt during thetreatment period is still retained inthe UAB laboratory clinic.

CI THERAPY IN OTHERLABORATORIES

In the UAB laboratory, over 400patients with stroke have been givenone variant or another of CI therapyand all but three of these patientshave demonstrated substantial im-provement in motor ability. Therehave also been over 300 papers fromother laboratories on adult andpediatric CI therapy published todate. To our knowledge all but twoof the studies have reported positiveresults. In particular, CI therapy wasthe subject of a multisite randomized

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:28 165 Cust # MS 12-105

ORIGINS OF CI THERAPY 165

Page 12: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

controlled trial (Wolf et al., 2006),the gold standard of proof of efficacyin medical fields. The results werepositive.

With respect to magnitude of thetreatment effect, this laboratory’sresults have been replicated withpatients with chronic stroke in pub-lished studies from four laboratoriesin which therapists were trained inthis laboratory and monitored twiceyearly (Dettmers et al., 2005; Kunkelet al., 1999; Miltner, Bauder, Som-mer, Dettmers, & Taub, 1999; Sterret al., 2002). Some of the other papersreport outcomes as large as thoseobtained in this and related labora-tories; however, many studies reportresults that are significant but onlyone half to one third as large as thoseobtained here. The likely reasons forthis disparity are twofold: (a) Therewas incomplete or complete lack ofuse of the procedures of the transferpackage, which, although reported inthe papers from this laboratory, hadbeen largely ignored. As noted above,we have replicated the reduced treat-ment effect obtained by others byduplicating everything that is nor-mally done in treatment here exceptimplementation of the TP (Gauthieret al., 2008). (b) A protocol withattenuated intensity (tasks or move-ments per unit time) was used, suchas in a study by van der Lee,Beckerman, Lankhorst, and Bouter(1999).

The techniques of the TP haveoften been used separately by indi-vidual therapists, but rarely system-atically and never combined togetherin an attempt to make patients’compliance with the protocol outsidethe laboratory critical so that theybecome responsible for their ownimprovement. Even when the behav-ioral techniques of the TP andintensive training are used, CI thera-py does not constitute a ‘‘cure’’ forthe motor deficit following stroke.On a group basis, patients in studiesfrom this laboratory with mild ormoderate deficits regain approxi-

mately 50% of the amount of use ofthe more affected arm they hadbefore stroke from an initial level ofapproximately 10%. This is a fivetimes difference and a substantialimprovement, but it is not a cure.There is still considerable room forfurther improvement. CI therapy canalso produce a large treatment effect(although not as large) in patientswith more severe motor deficits thanthose in the mild or moderate deficitcategory treated in most CI therapystudies, including patients with ini-tially plegic hands (see below).

APPLICATIONS OF CI THERAPY

The LNU formulation predictsthat any substantial damage to theCNS may lead to LNU. Thus, CItherapy, which initially had beenfound to be helpful in overcomingLNU in stroke patients with mild ormoderate motor deficits, should beapplicable to motor limitations moresevere than those originally workedwith, to deficits other than motorimpairment of the upper extremity,and to other types of neurologicalconditions.

Lower Functioning Patients

Most of the patients treated at theUAB laboratory could be character-ized as having deficits that were mildor moderate, defined as having theability to extend 20u at the wrist and10u at each of the fingers. Experi-ments have also been carried out withpatients with moderate and moder-ately severe deficits (Taub et al.,1999). Their treatment change wassomewhat less than for higher func-tioning patients (e.g., increases ofapproximately 400% and 350% forpatients with moderate and moder-ately severe deficits, respectively,compared to approximately 500%for patients with mild or moderatedeficits), but the treatment changeswere nevertheless very large. Mostrecently, work has been carried outwith patients with useless, plegic

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:28 166 Cust # MS 12-105

166 EDWARD TAUB

Page 13: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

hands that were initially fisted. Con-ventional physical rehabilitation pro-cedures, including some from neuro-developmental treatment (NDT)and functional electrical stimulation(FES) were used to maintain thefingers in a sufficiently extended andaligned position so that CI therapytraining procedures could be carriedout. At the end of treatment, thepatients exhibited a 186% improve-ment in the real-world use of themore affected arm. This arm hadbeen converted into a useful ‘‘helper’’in the life situation (e.g., keeping apiece of paper in place while writingwith the less affected hand, holding atoothpaste tube while unscrewing thecap, bearing body weight for bedmobility).

We estimate that CI therapy isapplicable to at least 50% of thechronic stroke population with motordeficit, perhaps more. This is a verylarge group of individuals; an esti-mated 4,000,000 people in this coun-try have had strokes in previousyears, and in addition, there are morethan 3,000,000 people who have hadhad traumatic brain injuries. Veryfew of the more than 50% of theseindividuals with persisting motordeficit are given any rehabilitationtreatment. Thus, CI therapy couldpotentially improve the quality of lifeand increase the independence of alarge number of currently untreatedpersons with brain damage.

Lower Extremity

An obvious target for transfer ofthe CI therapy techniques developedfor the upper extremity was the moreaffected lower extremity of strokepatients. The 38 chronic stroke pa-tients treated to date have had a widerange of disability extending frombeing close to nonambulatory tohaving moderately impaired coordi-nation (Taub et al., 1999). Thetreatment (lower extremity CI thera-py) consists of massed or repetitivepractice of lower extremity tasks

(e.g., overground walking, treadmillwalking with and without a partialbody weight support harness, sit-to-stand, lie-to-sit, step climbing, walk-ing over obstacles, various balanceand support exercises) for at first 6and then 3 hr per day with inter-spersed rest intervals as needed over3 weeks and 0.5 hr per day devotedto TP procedures. Task performanceis shaped as in the upper extremityprotocol. Training is enhanced th-rough the use of force feedback (limbload monitor) and limb displacement(joint angle/electric goniometer) feed-back devices. No restraining device isplaced on the less affected leg. Thelower extremity procedure is consid-ered to be a form of CI therapybecause of the use of the TP, thestrong massed practice and shapingelement, and because the reinforce-ment of adaptive patterns of ambu-lation over maladaptive patterns inour training procedure constitutes asignificant general form of constraint.Control data were provided by ageneral fitness control group thatreceived the same battery of lowerextremity tests as the treatment sub-jects. The ES of the change in real-world performance due to the treat-ment was very large, but not quite aslarge as for the upper extremity. Theimproved lower extremity use wasretained without any decrement forthe 2 years that were tested.

Conditions Other Than Stroke

The CI therapy protocol has beenapplied with success, as noted at thebeginning of the article, to traumaticbrain injury (Shaw, Morris, Uswatte,McKay, & Taub, 2003), upper andlower extremity in multiple sclerosis(Mark, Taub, Bashir, et al., 2008;Mark, Taub, Uswatte, et al., 2008),cerebral palsy and pediatric motordisorders of neurological originacross the full range of age from1 year old through the teenage years(Taub, Griffin, et al., 2006; Taubet al., 2007, 2011; Taub, Ramey,

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:28 167 Cust # MS 12-105

ORIGINS OF CI THERAPY 167

Page 14: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

DeLuca, & Echols, 2004), focal handdystonia in musicians (Candia et al.,1999, 2002), and, though not a motordisorder, phantom limb pain afteramputation (Weiss, Miltner, Adler,Bruckner, & Taub, 1999).

Aphasia. The application of CItherapy that is probably of greatestinterest from a behavior-analyticpoint of view is to aphasia, especiallythe work being done currently. Apha-sia arises as a consequence of focalbrain damage, often in associationwith stroke. There is as much LNUafter stroke associated with the ver-bal behavior of aphasics as there iswith motor deficit. Because of haltingand slow verbal production andincomplete understanding, speech be-comes very effortful and often em-barrassing. The person compensatesby greatly reducing attempts to speakor remaining silent entirely and byusing gestures and other nonverbalmeans of communication. In addi-tion, when there is difficulty inunderstanding speech, many aphasicswith receptive problems (Wernicke’saphasia, fluent aphasia) simply tuneout. Thus, the demonstration thatmotor deficits are modifiable inchronic stroke raised the possibilitythat verbal impairment could also berehabilitated by an appropriate mod-ification of the CI therapy protocol.The LNU formulation predicted thatthis was a strong possibility. Inthe first studies (Pulvermuller et al.,2001; Taub, 2002), aphasic patientswith chronic stroke who had previ-ously received extensive conventionalspeech therapy and had reached anapparent maximum in recovery oflanguage function were induced totalk and improve their verbal skillsfor 3 hr each weekday over a 2-weekperiod. The intervention was termedconstraint-induced aphasia therapy(CIAT I). The constraint was im-posed by the contingencies of rein-forcement in the shaping paradigmthat was used; there was no physicalrestraint, although as noted, physicalrestraint is not necessary to obtain a

good result with CIMT. Groups ofthree patients and a therapist partic-ipated in a language card game(Pulvermuller, 1990; Pulvermuller &Schonle, 1993). The exercise resem-bles the child’s card game ‘‘Go Fish.’’A participant asks one of the otherplayers if they have in their hand acard with a specific pictured object tomatch one in their own. If they do, therequester can meld those cards. Par-ticipants win the game if they meldeach of the cards they were dealt sothat none are left. The difficulty of therequired request by each patient isprogressively increased in small steps(i.e., shaped) along several dimen-sions: number of words in the request(or response to it), number of formu-las of politeness, precision of patient’scard description (animal, pet, dog),complexity of card depiction (dog,two dogs, one red and one blue dog),and grammatical correctness.

CIAT I patients in the initial RCTimproved much more than patientswho received conventional aphasiatherapy. This study has since beenreplicated (Bhogal, Teasell, & Spee-chley, 2003; Kirmess & Maher, 2010;Maher et al., 2006; Meinzer et al.,2004, 2007). Following a positiveevaluation of a committee appointedby the American Speech and HearingAssociation (Raymer et al., 2008),CIAT I is now beginning to spread.The results of the CIAT I protocolhave been positive; however, theintervention was only an incompletetranslation of CIMT. CIMT pro-duced an improvement of approxi-mately 500% in real-world use of themore affected extremity of chronicstroke patients with mild to moderatemotor deficit in one experiment(Taub, Uswatte, King, et al., 2006).Other experiments from this labora-tory have reported treatment effectsof similar size. Aphasic patients givenCIAT I showed an improvement of30% in real-world verbal behavior.This is a large treatment effectcompared to conventional speechlanguage therapies, but it is very

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:28 168 Cust # MS 12-105

168 EDWARD TAUB

Page 15: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

small compared to the results pro-duced by CIMT. Consequently, todetermine whether this large differ-ence was the result of an incompletetranslation of the CI therapy proto-col employed in the UAB laboratorywith motor deficits to the treatmentof language impairment, the initialaphasia treatment protocol (CIAT I)was modified to more closely resem-ble the CIMT protocol.

In the restructured and enhancedprotocol (CIAT II), use of behavior-analytic procedures was increased andemphasized. Revisions involved addi-tion of new exercises, including the finalexercise, considered to be the mostimportant, in which everyday verbalinteractions are simulated and mod-eled. In addition a TP parallel to thatused in CIMT was introduced, therewas increased emphasis on the shapingof responses, and the primary caregiverwas trained as an alternate therapist sothat the training begun in the labora-tory could be continued at home, bothduring and after formal training.

To date, only four patients havebeen treated with the new protocol.However, their results have far exceed-ed those obtained with CIAT I and arecomparable to the results obtained withCIMT. With CIAT I, as noted, therewas a 30% improvement in real-worldverbal behavior; for the recent patients,the mean improvement was 537%,which is approximately 18 times greaterthan for CIAT I and roughly equiva-lent to the treatment effect for CIMT.Of additional interest is the fact that at6-month follow-up, the patientsshowed no loss in retention; instead,the verbal behavior scores increasedsubstantially to a 643% improvementover pretreatment scores. This increaseappears to be attributable to thecontinuation of training by the care-givers in the real-world environment.

CI THERAPY ANDBRAIN PLASTICITY

As noted, overcoming LNU is oneof the mechanisms by which CI

therapy achieves its therapeutic ef-fect. Another important mechanismrelates to the fact that CI therapyproduces large plastic changes in thestructure and function of the brain.

Starting in the 1980s, Merzenichand collaborators showed in mon-keys that a decrease or increase in theamount of use of a body part or asensory function decreased or in-creased the size of the brain regionthat represented that function (e.g.,Jenkins, Merzenich, Ochs, Allard, &Guic-Robles, 1990; Merzenich et al.,1983). This phenomenon was origi-nally termed cortical reorganizationand is now called brain plasticityor neuroplasticity. In the 1990s,Taub and collaborators in Germanyshowed that neuroplastic corticalreorganization occurred in humans,and that it had functional signifi-cance in that it could affect move-ment, behavior, and the quality ofsensory experience (e.g., Elbert, Pan-tev, Wienbruch, Rockstroh, &Taub,1995; Flor et al., 1995).

A substantial number of studieshave now shown that CI therapyproduces a large neuroplastic corticalreorganization in humans with stroke-related paresis of an upper limb. Thiswas first demonstrated by Nudo,Wise, SiFuentes, and Milliken (1996)in an animal model of CI therapy.Subsequently, Liepert, Bauder, Milt-ner, Taub, and Weiller (2000) usedfocal transcranial magnetic stimula-tion (TMS) to map the area of themotor cortex that controls an impor-tant muscle of the hand (abductorpollicis brevis) in 15 patients with achronic upper extremity hemiparesis(mean chronicity 5 6 years) beforeand after CI therapy. We first repli-cated the clinical result that CItherapy produces a very large increasein patients’ amount of arm use in thehome over a 2-week treatment period.Over the same interval, the corticalregion from which electromyographyresponses of the abductor pollicisbrevis muscle could be elicited byTMS was greatly increased, and both

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:28 169 Cust # MS 12-105

ORIGINS OF CI THERAPY 169

Page 16: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

the clinical effect and the alteration inbrain function persisted for the6 months tested. CI therapy had ledto an increase in the excitability andrecruitment of a large number ofneurons in the innervation of move-ments of the more affected limbadjacent to those originally involvedin control of the extremity prior totreatment. The effect was sufficientlylarge that it represented a return tonormal size of the motor output areaof the abductor pollicis brevis muscleon the infarcted side of the brain,although it was the size of excitablecortical area that had become normal,not its function; the affected hand,though much improved after CItherapy, was not normal in function.In a third study, Kopp et al. (1999)carried out dipole modeling of steady-state movement-related cortical po-tentials (EEG) of patients before andafter CI therapy. We found that3 months after treatment the undam-aged motor cortex ipsilateral to theaffected arm, which normally controlsmovements of the contralateral (lessaffected) arm, had been recruited togenerate movements of the affectedarm. This effect was not in evidenceimmediately after treatment and waspresumably due to the sustainedincrease in more affected arm use inthe life situation produced by CItherapy over the 3-month follow-upperiod. This experimental evidencethat CI therapy is associated withsubstantial changes in brain activityhas been confirmed by convergentdata from two other neurophysiolog-ical studies that used two additionaltechniques in association with theadministration of CI therapy. Bauder,Sommer, Taub, and Miltner (1999)showed that there is a large increase inthe amplitude of the late componentsof the Bereitschaftspotential (a move-ment-related cortical potential) afterCI therapy, suggesting that an en-hanced neuronal excitability is in-duced in the damaged hemisphere;this is consistent with the results ofLiepert et al. (2000). We also found

that after CI therapy there was a largeincrease in the activation of theusually weakly active healthy, ipsilat-eral hemisphere with more affectedhand movement in confirmation ofthe findings of Kopp et al. (1999). Inaddition, Wittenberg et al. (2003)found in a positron emission tomog-raphy study that before CI therapythere was a larger activation inmultiple areas of the brain with moreaffected arm movement than inhealthy control subjects. This exces-sive activation diminished after CItherapy. The preliminary interpreta-tion of this result is that less effort isrequired to produce movements afterCI therapy than before treatment.

Since these initial studies, therehave been approximately 20 otherstudies that have demonstrated analteration in brain function associat-ed with a CI therapy-induced im-provement in movement after CNSdamage. By providing a physiologicalbasis for the treatment effect reportedfor CI therapy, these results havetended to increase confidence in theclinical results.

The studies described to this pointshow that alterations in afferentinput can alter the function andorganization of specific brain regions,but until recently there was noevidence that environmental stimulicould measurably alter brain struc-tures in adult humans. It has nowbeen shown that seasoned taxi drivershave significantly expanded hippo-campi (Maguire et al., 2000), jugglersacquire significantly increased tem-poral lobe density (Draganski et al.,2004), and thalamic density signifi-cantly declines after limb amputation(Draganski et al., 2006). Moreover,in an animal model of stroke, CItherapy combined with exercise re-duced brain tissue loss associatedwith stroke (DeBow, Davies, Clarke,& Colbourne, 2003). Accordingly,structural imaging studies became alogical next step toward understand-ing whether there are anatomicalchanges following the administration

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:28 170 Cust # MS 12-105

170 EDWARD TAUB

Page 17: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

of CI therapy in humans and whetherthese are correlated with clinicalimprovements. Moreover, anatomi-cal studies that make use of structuralMRI have advantages over fMRIstudies, including the fact that notask is carried out during scanning sothat there is no need to exerciseexperimental control over the topog-raphy and force of task-related move-ments.

Longitudinal (pre- vs. posttreat-ment) voxel-based morphometry wasperformed on subjects enrolled in ourstudy of the contribution made by theTP to CI therapy outcome (Gauthieret al., 2008). It was found thatstructural brain changes paralleledchanges in amount of use of theimpaired extremity for activities ofdaily living. Groups receiving the TPshowed profuse increases in graymatter tissue in sensorimotor areason both sides of the brain (Figure 3)as well as in bilateral hippocampi. Incontrast, the groups that did notreceive the TP showed relatively smallimprovements in real-world arm use

and failed to demonstrate gray matterincreases.

The research just reviewed makes itclear that behavior and sensoryexperience are involved in a funda-mental feedback loop that keepsremodeling the nervous system. Therelation of the nervous system andbehavior is not a one-way street; thenervous system is involved in aprocess of continual plastic changethroughout the life span based onfeedback from the environment andfrom a person’s own behavior. Inaddition, CI therapy appears to har-ness this life-long plasticity of thenervous system to produce an im-provement in movement and languageafter damage to the nervous system.

CONCLUSION

It might be a fitting end to thisstory of my journey into the worldof clinical treatment to recount theevents associated with a grandrounds I gave just before attemptingthe translation of the research withdeafferented monkeys to humans

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:28 171 Cust # MS 12-105

Figure 3. Cortical surface-rendered images of changes in gray matter. Gray matter increasesdisplayed on a standard brain for (A) participants who received the CI therapy transfer packageand (B) those who did not. Surface rendering was performed with a depth of 20 mm. Bar valuesindicate t statistics ranging from 2.2 to 6.7.

ORIGINS OF CI THERAPY 171

Page 18: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

after CNS damage. The talk was in adepartment of physical medicine andrehabilitation. I described the workwith primates in the context of apossible translation to humans, al-though the latter was implied and notspecifically stated. The chairman ofthe department, who was a promi-nent clinician and rehabilitation in-vestigator, sat quietly through my talkbut with a frigid expression. After Iwas finished, he said with progressive-ly increasing volume, ‘‘Are you tryingto tell me that you have a behavioralintervention that you think will im-prove the symptoms of a neurologicallesion?’’ I saw that I was on danger-ous ground and so I said, ‘‘Well, no’’;but that, of course, was what I hadbeen implying for the past hour. Ipaused for a while and then said asmildly as I could, ‘‘But after all, isn’tthat what physical therapy is?’’ Thatwas a mistake. He sputtered a fewwords, which I didn’t catch, while hisface quite literally began purpling.However, to his credit, after the firstexperiment was completed and thereport appeared in print, he changedhis opinion and became a strong andvaluable supporter. This pretty wellsums up the way in which therehabilitation community has reactedto CI therapy. At first, there was astrong bias against a treatment basedon behavior analysis; few members ofthe rehabilitation community had anyfamiliarity with behavior analysis orhad even heard of it. However, as theevidence began to mount and at-tempts at replication were successful,attitudes began to change. With thesuccess of the multisite RCT citedabove, and the fact that CI therapyproduced substantial plastic structur-al changes in the brain, the case hadessentially been made. Use of CItherapy is still not by any meansuniversal, probably in part because ofinsurance reimbursement problemsdue to the duration and thereforeexpense of the treatment. However,even with that, the treatment isbeginning to spread, especially in

modified forms that are more readilyreimbursed by insurance. In addition,as I understand it, schools of physicaland occupational therapy are begin-ning to teach CI therapy and at leastsome of the principles of behavioranalysis. Behavior analysis has thusbegun to make an appearance on thestage of neurorehabilitation.

REFERENCES

Azrin, N. H., & Holz, W. C. (1966). Punish-ment. In W. K. Honig (Ed.), Operantbehavior: Areas of research and application(pp. 380–447). New York, NY: Appleton-Century-Crofts.

Bauder, H., Sommer, M., Taub, E., & Miltner,W. H. R. (1999). Effect of CI therapy onmovement-related brain potentials [Ab-stract]. Psychophysiology, 36 (Suppl. 1), S31.

Bhogal, S. K., Teasell, R., & Speechley, M.(2003). Intensity of aphasia therapy, impacton recovery. Stroke, 34, 987–993.

Candia, V., Elbert, T., Altenmuller, E., Rau,H., Schafer, T., & Taub, E. (1999). Con-straint-induced movement therapy for focalhand dystonia in musicians. Lancet, 353, 42.

Candia, V., Schafer, T., Taub, E., Rau, H.,Altenmuller, E., Rockstroh, B., et al. (2002).Sensory motor retuning: A behavioraltreatment for focal hand dystonia of pia-nists and guitarists. Archives of PhysicalMedicine and Rehabilitation, 83, 1342–1348.

Catania, A. C. (1998). Learning (4th ed.).Upper Saddle River, NJ: Prentice Hall.

Cohen, J. (1988). Statistical power analysis forthe behavioral sciences (2nd ed.). Hillsdale,NJ: Erlbaum.

DeBow, S. B., Davies, M. L. A., Clarke, H. L.,& Colbourne, F. (2003). Constraint-inducedmovement therapy and rehabilitation exer-cises lessen motor deficits and volume ofbrain injury after striatal hemorrhagicstroke in rats. Stroke, 34, 1021–1026.

Dettmers, C., Teske, U., Hamzei, F., Uswatte,G., Taub, E., & Weiller, C. (2005). Distrib-uted form of constraint-induced movementtherapy improves functional outcome andquality of life after stroke. Archives ofPhysical Medicine and Rehabilitation, 86,204–209.

Draganski, B., Gaser, C., Busch, V.,Schuierer, G., Bogdahn, U., & May, A.(2004). Changes in grey matter induced bytraining. Nature, 427, 311–312.

Draganski, B., Moser, T., Lummel, N.,Ganssbauer, S., Bogdahn, U., Haas, F.,et al. (2006). Decrease of thalamic graymatter following limb amputation. Neuro-image, 31, 951–957.

Elbert, T., Pantev, C., Wienbruch, C., Rock-stroh, B., & Taub, E. (1995). Increased

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:30 172 Cust # MS 12-105

172 EDWARD TAUB

Page 19: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

cortical representation of the fingers of theleft hand in string players. Science, 270,305–307.

Estes, W. K. (1944). An experimental studyof punishment. Psychological Monographs,57(Serial No. 263).

Flor, H., Elbert, T., Knecht, S., Wienbruch,C., Pantev, C., Birbaumer, N., et al. (1995).Phantom limb pain as a perceptual correlateof massive reorganization in upper limbamputees. Nature, 375, 482–484.

Gauthier, L. V., Taub, E., Perkins, C.,Ortmann, M., Mark, V. W., & Uswatte,G. (2008). Remodeling the brain: Plasticstructural brain changes produced by dif-ferent motor therapies after stroke. Stroke,39, 1520–1525.

Jenkins, W. M., Merzenich, M. M., Ochs,M. T., Allard, T., & Guic-Robles, E. (1990).Functional reorganization of primary so-matosensory cortex in adult owl monkeysafter behaviorally controlled tactile stimu-lation. Journal of Neurophysiology, 63,82–104.

Kirmess, M., & Maher, L. (2010). ConstraintInduced language therapy in early aphasiarehabilitation. Aphasiology, 24, 725–736.

Knapp, H. D., Taub, E., & Berman, A. J.(1959). Conditioned response followingdeafferentation in the monkey. Transactionsof the American Neurological Association,84, 185–187.

Knapp, H. D., Taub, E., & Berman, A. J.(1963). Movements in monkeys with deaf-ferented limbs. Experimental Neurology, 7,305–315.

Kopp, B., Kunkel, A., Muhlnickel, W.,Villringer, K., Taub, E., & Flor, H. (1999).Plasticity in the motor ststem related totherapy-induced improvement of movementafter stroke. NeuroReport, 10, 807–810.

Kunkel, A., Kopp, B., Muller, G., Villringer,K., Villringer, A., Taub, E., et al. (1999).Constraint-induced movement therapy formotor recovery in stroke patients. Archivesof Physical Medicine and Rehabilitation, 80,624–628.

Lassek, A. M. (1953). Inactivation of volun-tary motor function following rhizotomy.Journal of Neuropathology and ExperimentalNeurology, 3, 83–87.

Liepert, J., Bauder, H., Miltner, W. H. R.,Taub, E., & Weiller, C. (2000). Treatment-induced cortical reorganization after strokein humans. Stroke, 31, 1210–1216.

Maguire, E. A., Gadian, D. G., Johnsrude,I. S., Good, C. D., Ashburner, J., Frack-owiak, R. S. J., et al. (2000). Navigation-related structural change in the hippocampiof taxi drivers. Proceedings of the NationalAcademy of Sciences, USA, 97, 4398–4403.

Maher, L. M., Kendall, D., Swearengin, J. A.,Rodriguez, A., Leon, S. A., Pingel, K., et al.(2006). A pilot study of use-dependentlearning in the context of constraint inducedlanguage therapy. Journal of the Interna-

tional Neuropsychological Society, 12,843–852.

Mark, V., Taub, E., Bashir, K., Uswatte, G.,Delgado, A., Bowman, M., et al. (2008).Constraint-induced movement therapy canimprove hemiparetic progressive multiplesclerosis. Multiple Sclerosis, 14(7), 992–994.

Mark, V. W., Taub, E., Uswatte, G., Bashir,K., Bryson, C., Bowman, M., et al. (2008).Impaired gait in progressive multiple scle-rosis improves with constraint-inducedmovement therapy [Abstract]. Archives ofPhysical Medicine and Rehabilitation,89(10), E5.

Meinzer, M., Elbert, T., Barthel, G., Djundja,D., Taub, E., & Rockstroh, B. (2007).Extending the constraint-induced move-ment therapy (CIMT) approach to cognitivefunctions: Constraint-induced aphasia ther-apy (CIAT) of chronic aphasia. Neuro-Rehabilitation, 22, 311–318.

Meinzer, M., Elbert, T., Wienburch, C.,Djundja, D., Barthel, G., & Rockstroh, B.(2004). Intensive language training enhancesbrain plasticity in chronic aphasia. BMCBiology, 2, 20–29.

Merzenich, M. M., Kaas, J. H., Wall, J.,Nelson, R. J., Sur, M., & Felleman, D.(1983). Topographic reorganization of so-matosensory cortical areas 3b and 1 in adultmonkeys following restricted deafferenta-tion. Neuroscience, 8, 33–55.

Miltner, W. H. R., Bauder, H., Sommer, M.,Dettmers, C., & Taub, E. (1999). Effects ofconstraint-induced movement therapy onchronic stroke patients: A replication.Stroke, 30, 586–592.

Morgan, W. G. (1974). The shaping game: Ateaching technique. Behavior Therapy, 5,271–272.

Mott, F. W., & Sherrington, C. S. (1895).Experiments upon the influence of sensorynerves upon movement and nutrition of thelimbs. Proceedings of the Royal Society ofLondon, 57, 481–488.

Nudo, R. J., Wise, B. M., SiFuentes, F., &Milliken, G. W. (1996). Neural substratesfor the effects of rehabilitative training onmotor recovery after ischemic infarct. Sci-ence, 272, 1791–1794.

Panyan, M. V. (1980). How to use shaping.Lawrence, KS: H & H Enterprises.

Pulvermuller, F. (1990). Aphasische kommuni-kation. Grundfragen ihrer analyse und ther-apie [Communication in aphasics. Basicquestions and therapeutic approaches]. Tue-bingen, Germany: Gunter Narr Verlag.

Pulvermuller, F., Neininger, B., Elbert, T.,Mohr, B., Rockstroh, B., Kobbel, P., et al.(2001). Constraint-Induced therapy ofchronic aphasia following stroke. Stroke,32, 1621–1626.

Pulvermuller, F., & Schonle, P. (1993). Be-havioral and neuronal changes during treat-

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:30 173 Cust # MS 12-105

ORIGINS OF CI THERAPY 173

Page 20: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

ment of mixed transcortical aphasia. Cog-nition, 48, 139–161.

Raymer, A., Beeson, P., Holland, A., Kendall,D., Maher, L., Martin, N., et al. (2008).Translational research in aphasia: Fromneuroscience to neurorehabilitation. Journalof Speech, Hearing, & Language Research,5(1), S259–S275.

Shaw, S. E., Morris, D. M., Uswatte, G.,McKay, S. B., & Taub, E. (2003). Adher-ence to constraint induced movement ther-apy procedures and outcome for personswith traumatic brain injury. Journal ofNeurologic Physical Therapy, 27(4), 180.

Sherrington, C. S. (1910). Remarks on thereflex mechanism of the step. Brain, 33,1–25.

Skinner, B. F. (1938). The behavior of organ-isms. New York, NY: Appleton-Century-Crofts.

Skinner, B. F. (1968). The technology ofteaching. New York, NY: Appleton-Century-Crofts.

Sterr, A., Elbert, T., Berthold, I., Kolbel, S.,Rockstroh, B., & Taub, E. (2002). Longerversus shorter daily constraint-inducedmovement therapy of chronic hemiparesis:An exploratory study. Archives of PhysicalMedicine and Rehabilitation, 83, 1374–1377.

Taub, E. (1977). Movement in nonhumanprimates deprived of somatosensory feed-back. Exercise and sports science reviews(Vol. 4, pp. 335–374). Santa Barbara, CA:Journal Publishing Affiliates.

Taub, E. (1980). Somatosensory deafferenta-tion research with monkeys: Implicationsfor rehabilitation medicine. In L. P. Ince(Ed.), Behavioral psychology in rehabilitationmedicine: Clinical applications (pp. 371–401). New York, NY: Williams & Wilkins.

Taub, E. (2002). CI therapy: A new rehabil-itation technique for aphasia and motordisability after neurological injury. Klinikund Forschung, 8, 48–49.

Taub, E., & Berman, A. J. (1963). Avoidanceconditioning in the absence of relevantproprioceptive and exteroceptive feedback.Journal of Comparative and PhysiologicalPsychology, 56, 1012–1016.

Taub, E., & Berman, A. J. (1968). Movementand learning in the absence of sensoryfeedback. In S. J. Freedman (Ed.), Theneuropsychology of spatially oriented behav-ior (pp. 173–192). Homewood, IL: Dorsey.

Taub, E., Crago, J. E., Burgio, L. D.,Groomes, T. E., Cook, E. W., DeLuca, S.C., et al. (1994). An operant approach torehabilitation medicine: Overcominglearned nonuse by shaping. Journal of theExperimental Analysis of Behavior, 61,281–293.

Taub, E., Ellman, S. J., & Berman, A. J.(1966). Deafferentation in monkeys: Effecton conditioned grasp response. Science, 151,593–594.

Taub, E., Goldberg, I. A., & Taub, P. B.(1975). Deafferentation in monkeys: Point-ing at a target without visual feedback.Experimental Neurology, 46, 178–186.

Taub, E., Griffin, A., Gammons, K., Nick, J.,Uswatte, G., & Law, C. R. (2006). CItherapy for young children with congenitalhemiparesis. Atlanta, GA: Society for Neu-roscience.

Taub, E., Griffin, A., Nick, J., Gammons, K.,Uswatte, G., & Law, C. R. (2007). PediatricCI therapy for stroke-induced hemiparesisin young children. Developmental Neurore-habilitation, 10, 1–16.

Taub, E., Griffin, A., Uswatte, G., Gammons,K., Nick, J., & Law, C. R. (2011).Treatment of congenital hemiparesis withpediatric constraint-induced movementtherapy Journal of Child Neurology, 26,1163–1173.

Taub, E., Heitman, R., & Barro, G. (1977).Alertness and level of activity and purposivemovement following deafferentation in theadult and developing monkey. In H. P.Zeigler & B. M. Wenzel (Eds.), Tonicfunctions of sensory systems (Vol. 290, pp.348–365). Ann, NY: Academic Sciences.

Taub, E., Miller, N. E., Novack, T., Cook,E. W., III., Fleming, W. C., Nepomuceno,C. S., et al. (1993). Technique to improvechronic motor deficit after stroke. Archivesof Physical Medicine and Rehabilitation, 74,347–354.

Taub, E., Perrella, P. N., & Barro, G. (1973).Behavioral development following forelimbdeafferentation on day of birth in monkeyswith and without blinding. Science, 181,959–960.

Taub, E., Perrella, P. N., Miller, D., & Barro,G. (1975). Diminution of early environmen-tal control through perinatal and prenatalsomatosensory deafferentation. BiologicalPsychiatry, 10, 609–626.

Taub, E., Ramey, S. L., DeLuca, S., & Echols,E. (2004). Efficacy of constraint-induced(CI) movement therapy for children withcerebral palsy with asymmetric motor im-pairment. Pediatrics, 113, 305–312.

Taub, E., & Uswatte, G. (2009). Constraint-induced movement therapy: A paradigm fortranslating advances in behavioral neurosci-ence into rehabilitation treatments. In G.Berntson & J. Cacioppo (Eds.), Handbookof neuroscience for the behavioral sciences(Vol. 2, pp. 1296–1319). Hoboken, NJ:Wiley.

Taub, E., Uswatte, G., King, D. K., Morris,D., Crago, J., & Chatterjee, A. (2006). Aplacebo controlled trial of constraint-in-duced movement therapy for upper extrem-ity after stroke. Stroke, 37, 1045–1049.

Taub, E., Uswatte, G., Mark, V., & Morris,D. (2006). The learned nonuse phenome-non: Implications for rehabilitation. EuropaMedicophysica, 42(2), 241–255.

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:30 174 Cust # MS 12-105

174 EDWARD TAUB

Page 21: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

Taub, E., Uswatte, G., & Pidikiti, R. (1999).Constraint-induced movement therapy: Anew family of techniques with broad appli-cation to physical rehabilitation—a clinicalreview. Journal of Rehabilitation Researchand Development, 36, 237–251.

Taub, E., Williams, E., Barro, G., & Steiner,S. S. (1978). Comparison of the perfor-mance of deafferented and intact monkeyson continuous and fixed ratio schedules ofreinforcement. Experimental Neurology, 58,1–13.

Timberlake, W. (1993). Behavior systems andreinforcement: An integrative approach.Journal of the Experimental Analysis ofBehavior, 60, 105–128.

Twitchell, T. E. (1954). Sensory factors inpurposive movement. Journal of Neurophys-iology, 17, 239–254.

Uswatte, G., Miltner, W. H. R., Varma, M.,Moran, S., Sharma, V., Foo, B., et al.(1998). Accelerometry: An objective ap-proach to real-world outcome measurementin physical rehabilitation. Paper presented atthe inaugural meeting of the Program inCognitive Rehabilitation of the James S.McDonnell Foundation, St. Louis, MO.

Uswatte, G., Miltner, W., Walker, H., Sprag-gins, S., Moran, S., Calhoun, J., et al.(1997). Accelerometers in rehabilitation:Objective measurement of extremity use athome [Abstract]. Rehabilitation Psychology,42, 139.

Uswatte, G., Spraggins, S., Walker, H.,Calhoun, J., & Taub, E. (1997). Validityand reliability of accelerometry as anobjective measure of upper extremity useat home [Abstract]. Archives of PhysicalMedicine and Rehabilitation, 78, 896.

Uswatte, G., Taub, E., Morris, D., Barman,J., & Crago, J. (2006). Contribution of theshaping and restraint components of con-straint-induced movement therapy to treat-ment outcome. NeuroRehabilitation, 21(2),147–156.

Uswatte, G., Taub, E., Morris, D., Light, K., &Thompson, P. (2006). The motor activityLog-28: Assessing daily use of the hemipareticarm after stroke. Neurology, 67, 1189–1194.

Uswatte, G., Taub, E., Morris, D., Vignolo,M., & McCulloch, K. (2005). Reliabilityand validity of the upper-extremity motoractivity Log-14 for measuring real-worldarm use. Stroke, 36, 2493–2496.

van der Lee, J., Beckerman, H., Lankhorst,G., & Bouter, L. (1999). Constraint-inducedmovement therapy [Letter to the Editor].Archives of Physical Medicine and Rehabil-itation, 80, 1606.

Weiss, T., Miltner, W. H. R., Adler, T.,Bruckner, L., & Taub, E. (1999). Decreasein phantom limb pain associated withprosthesis-induced increased use of anamputation stump in humans. NeuroscienceLetters, 272, 131–134.

Wittenberg, G. F., Chen, R., Ishii, K.,Bushara, K. O., Taub, E., Gerber, L. H.,et al. (2003). Constraint- induced therapy instroke: Magnetic-stimulation motor mapsand cerebral activation. Neurorehabilitationand Neural Repair, 17, 48–57.

Wolf, S., Winstein, C., Miller, J., Taub, E.,Uswatte, G., Morris, D., et al. (2006). Effectof constraint-induced movement therapy onupper extremity function 3–9 months afterstroke: The EXCITE randomized clinicaltrial. Journal of the American MedicalAssociation, 296, 2095–2104.

APPENDIX A

Steps Involved in the ShapingProgression from Total Absence of theTarget Behavior to Thumb-ForefingerGrasp of a Food Object in JuvenileMonkeys Deafferented Prenatally oron Day of Birth

The steps in shaping were asfollows:

1. Showing the juvenile a desirablefood object (e.g. small apple cube,peanuts) and reinforcing any move-ment of the arm, whether in the correctdirection or not, by food in the mouth.

2. Requiring arm movements ofprogressively greater excursion andmore accurate direction for place-ment of food in the mouth.

3. Requiring that the food objectbe touched for food to be placed inhand so that it could be returned bythe animal to its mouth.

4. Requiring that fingers be openedso that hand could be baited with afood object; wrist supported byexperimenter at end of arm trajecto-ry; fingers opened, first by passivemanipulation by experimenter andsubsequently with progressively moreactive finger extension required.

5. Grasping of food object by theanimal at end of arm trajectory withno support of wrist.

6. Picking food object up fromexperimenter’s palm, which was mold-ed and moved to make prehensioneasier; any type of grasp permitted.

7. Picking food object up from aflat wooden board. Lateral thumb-forefinger grasp (a monkey’s normal

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:30 175 Cust # MS 12-105

ORIGINS OF CI THERAPY 175

Page 22: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

mode of prehension) developed spon-taneously over sessions, as did ap-proaching the food object fromabove rather than accomplishing thegrasp while the ulnar surface of wristand lower forearm were supported bythe board.

8. Placement of food objects (ap-proximately 1-cm3 apple cubes) inshallow (0.5 mm) wells on a Kluverboard (a board with multiple wellsfrom which monkeys extract pieces offood) to promote more accuratethumb-finger approximation.

9. Placement of apple cubes on aKluver board with deeper (1 cm)wells to promote pincer grasp (ap-proximation of the palmar tips of thethumb and forefinger).

10. Use of smaller food objects,first peanuts, then raisins.

The terminal behavior achievedwas retrieval of raisins from wells onthe first attempt by pincer grasp onapproximately 50% of trials. Some-times, after two or more attemptsfailed, the monkeys would move thefood object out of a well with theforefinger so that it could be graspedon the flat surface between wells.

APPENDIX B

Shaping Guidelines

Shaping is a training method inwhich a motor or behavioral objec-tive is approached in small steps bysuccessive approximations, or a taskis gradually made more difficult inaccordance with a subject’s motorcapabilities. The following guidelinesemployed in the UAB laboratoryshould be followed when using shap-ing for inducing recovery of motorfunction.

Specific shaping tasks should beselected for patients by considering(a) specific joint movements thatexhibit the most pronounced deficits,(b) the joint movements that trainersbelieve have the greatest potential forimprovement, and (c) patient prefer-

ence among tasks that have similarpotential for producing specific im-provements.

Shaping tasks should be modeledfor the patient and encouragementand coaching (verbal prompts) pro-vided liberally.

The level of difficulty of theshaping task should be slightly be-yond what the patient can accom-plish easily (e.g., encouraging him orher to do a little better than theprevious performance).

In the shaping progression, mov-ing to the next higher level ofdifficulty should be carried outwhen the patient has reached arelative plateau with regard toperformance. For the present pur-poses, when a patient has performedfive trials in a row with no improve-ment evidenced in their score, thenext level of difficulty should beattempted. If subjects are permittedto achieve greater mastery, theyfrequently have a tendency to be-come ‘‘locked in’’ at that level.Subsequently, improvement be-comes more difficult to achieve.(This is a guideline only. If thepatient is ‘‘on a roll,’’ progressingrapidly, he or she should be shiftedto the next performance difficultylevel as rapidly as the trainer feelsthe performance will keep improv-ing at a maximal level).

The shaping task is made progres-sively more difficult only as thepatient improves in performance.

Any of the shaping progressionparameters can be changed to in-crease the difficulty of the task (e.g.,time, number of repetitions, height,placement, etc.).

When increasing the level of diffi-culty of an activity, the shaping pro-gression parameters selected shouldrelate to the subject’s movement prob-lems (i.e., in the flipping dominoestask, if the subject’s most significantdeficits are in thumb and forefingerdexterity, the task progression shouldinvolve using, depending on the natureof the deficit, either larger or smaller

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:30 176 Cust # MS 12-105

176 EDWARD TAUB

Page 23: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

dominoes. If the subject’s most signif-icant movement deficits are at theshoulder, the task progression shouldinvolve moving the dominoes fartheraway).

Shaping tasks are made more difficultwhen it is clear that, for the most part,the patient will be able to accomplish thetask, though with effort.

Positive reinforcement or rewardshould be provided visually (i.e.,keeping the shaping data form inplain view of the patient so that he orshe can see performance history and‘‘personal best’’; task performancebecomes like an arcade game). Taskperformance information shouldalso be given verbally at frequentintervals.

An important function of thetrainer is to act as a cheerleader,continuously encouraging the subjecton a moment-to-moment basis tokeep improving the performance.

Performance regressions are neverpunished and are usually ignored.

If a patient is experiencing exces-sive difficulty with a task, a simplertask involving similar movements canbe substituted.

Rest intervals should be allowedduring each shaping session. The restperiod is usually the same length asthe trial period, although longerintervals are sometimes needed toprevent fatigue.

Trainers should rate the perfor-mance of each shaping task trialusing the quality-of-movement scaleattached.

The results of each shaping tasktrial, including quality-of-movementrating, should be recorded on theshaping data form.

Encouragement and quality-of-movement recordings should be givento the subject verbally on at least 50%of the trials.

Placement of equipment used inshaping tasks should be recorded onthe shaping data form so that thetask can be duplicated. Adhesivemarkers on the task performancetable can be used for this purpose.

Also, note any placement changes onthe data sheet when a shaping task ismade more difficult.

To quantify a shaping task, onlyone shaping progression parametercan be allowed to vary. For example,on an elbow extension task, therewould be three parameters: time,number of repetitions, and distance.The time and number of the repeti-tions can be held constant and thedistance can be slowly increased untilthe subject can no longer perform aspecified number of extensions in agiven period of time (e.g., 10 exten-sions in 30 s). Alternatively, distancecan be held constant (e.g., 10 in.) andthe subject would be encouraged toprogressively increase the number ofrepetitions in a set period of time(e.g., 30 s). For a given task, morethan one parameter should not bevaried at the same time (e.g., bothdistance and number of repetitions).If the trainer feels that the subjectwould benefit from varying a secondparameter, that is permissible. How-ever, it should be understood thatthis training now must be quantifiedas a new entity on separate datasheets.

Example of Shaping Tasks:Flipping Dominoes

Activity description:

Approximately 25 dominoes areplaced in front of the subject. Thesubject is asked to reach forward andflip the dominoes using either fore-arm pronation or supination. Thecorrect movement can be best isolat-ed by asking the subject to rest his orher forearm on the table during thetask.

Potential shaping progression:

Placing the dominoes farther away to chal-lenge elbow extension.

Using larger or smaller dominoes to challengewrist and finger control.

Place dominoes on a box to challenge shoulderflexion.

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:30 177 Cust # MS 12-105

ORIGINS OF CI THERAPY 177

Page 24: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

Potential feedback variables:

Number of dominoes flipped in a set period oftime.

Time required to flip a set number ofdominoes.

Movements emphasized:

Lateral pincer grasp.

Wrist extension.

Forearm supination or pronation (depending ondirection of flip).

Shoulder flexion (if placed on a box).

Example of Shaping Tasks:TurningMagazine Pages

Activity description:

Place a magazine on the table. Ask thesubject to turn the pages. Have the subjectconcentrate on turning pages by either pro-nating or supinating.

Potential shaping progression:

The position of the magazine can bechanged (moved farther away from thesubject) to challenge elbow extension.

Increase the amount of time for thesubject to turn the pages or increase thenumber of pages that the subject mustturn to challenge the subject’s endurance.

Potential feedback variables:

Number of pages turned in a set amount of time.

Time required to turn a set number of pages.

Movements emphasized:

Forearm supination.

Forearm pronation.

Pincer or lateral pincer grasp.

Shoulder internal and external rotation.

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:31 178 Cust # MS 12-105

178 EDWARD TAUB

Page 25: The Behavior-Analytic Origins of Constraint-Induced Movement Therapy… · The Behavior-Analytic Origins of Constraint-Induced Movement Therapy: An Example of Behavioral Neurorehabilitation

Authors Queries

Journal: The Behavior AnalystPaper: bhan-35-02-03Title: The Behavior-Analytic Origins of Constraint-Induced MovementTherapy: An Example of Behavioral Neurorehabilitation

Dear AuthorDuring the preparation of your manuscript for publication, the questions listedbelow have arisen. Please attend to these matters and return this form with yourproof. Many thanks for your assistance

QueryReference

Query Remarks

1 Which Taub et al. (1975)ref? List all authors.

2 Which Taub et al. (1975)ref? List all authors.

3 Add Morris and Taub(2008) to ref list.

The Behavior Analyst bhan-35-02-03.3d 7/9/12 13:45:31 179 Cust # MS 12-105

ORIGINS OF CI THERAPY 179