application of motor learning principles to complex...

Journal of Motor Behavior, 2007, Vol, 39. No, 1, 40-48Copyright © 2007 Heldref Publications

Application of Motor Learning Principlesto Complex Surgical Tasks: Searchingfor the Optimal Practice Schedule

Ryan BrydgesDepartment of Surgery

Heather CarnahanDepartment of SurgeryDepartment of Occupational Scienceand Occupational Therapy

University of Toronto, Ontario, Canada

David BacksteinAdam DubrowskiDepartment of SurgeryUniversity of Toronto, Ontario, Canada

ABSTRACT. Practice of complex tasks can be scheduled in sever-al ways: as whole-task practice or as practice of the individual skillscomposing the task in either a blocked or a random order. Theauthors used those 3 schedules to study 18 participjuits' learning ofan orthopedic surgical task. They assessed leitming by obtaitiitigexpert evaluation of pcifomiance and objective kinctnatic measuresbefore, immediately after, and I week after practice (transfer test).During acqtiisition. the blocked group showed supetior pertbrmancefor sitiiple skills hut not for more complex skills. For the expert-hased measures of pertbrmance. all groups improved from pretest toposttest and remained constant front posttest to transfer. Measures ofthe tlnal product showed that [he whole-practice group's outcomeswere significantly hetter than tho,se of the random group on transfer.All groups showed hetter efficiency of motions in the posttest thanin the pretest. Those measures were also poorer on the transfer testthan on the posttest. The present evidence does not supptm the con-textual interference effect—hypothetically, because of the inherentcognitive effbrt effect associated with some of the component skills.TTie authors recommend that surgical tasks composed of several dis-crete skills be practiced as a whole. The results of this study demon-strate the importance of critically appraising basic theories in appliedetiv iron men ts.

Key words: contextual interference, motor learning, practiceschedule, skill acquisition

T heoretical advances in the principles governitig tnotorskill acquisition are often used to guide interested edu-

cators and health professionals in applied settings to devel-op theory-ba.sed educational and rehabilitative goals. Con-versely, careful implementation of the theoretical conceptsin the applied world can serve as testing grounds that canlead to etiipirical validation of those concepts and to thegeneratioti of interesting theoretical questions.

One applied field in vi'hich theoretical motor leamingprinciples can be tested is the training of technical surgical

skills. The ability to adequately prepare new surgicaltrainees by using the traditional Halstedian apprenticeshipmodel of sutgical education—coined see one. do one, teachone—has been recently questioned. Demands on new surgi-cal trainees are increasing because they are required to learnabout more diseases and procedutes in the same or fewerworking hours. Learning opportunities are further limitedby economic constraints in health care and by a greaterrespect for the rights of patients. It is becoming necessaryfor instructors to exploit the educational value of everylearning experience in an eftbrt to reduce the initial portionof the hypothetical leaming curve for the technical aspectsof the surgical craft (Hall. Ellis, & Hamdorf. 2003).

In an effort to standardize and optimize the learning expe-rience, investigators have shown an increased interest in therole of surgical skills laboratories in the teaching of at leastsome of the tnost basic surgical skills (Aggarwal. Hatice, &Darzi. 2004; Aggarwal. Moorthy. & Dar^i. 2004). That envi-rontnent provides an opportunity for trainees to practicetechnical tasks and skills repeatedly until they achieve pro-ficiency. There is evidence that low-fidelity bench top mod-els are as effective as high-fidelity models (Grober ct al..2004): the former tiiodels improve the cost-effectiveness ofskills laboratoties and tiiaintain the learning objectives.However, the application of learning principles to the devel-opment of pedagogically sound and cost-effective practiceschedules has been largely unexplored. That is, to date,effotts in surgical training have been focused on developingappropriate models for practice (Grober et al.) and methods

Correspondence address: Adam Dubrowski. University ofToronto. Surgical Skills Centre at Mount Sinai Hospital. 600 Uni-ver.siry Avenue. Level 2-Room 250. Toronto. Ontario. M5G 1X5,Canada. E-mail address: [email protected]

40

Motor Learning Principles in Surgery

of evaluating learning (Martin et al.. 1997; Perkins. Starkes,Lee, & Hutchison. 2002). How those technical skills andtasks are best practiced, though, has not been well addressedwithin the literature.

ln teaching complex surgical tasks composed of severaltechnical skills, practice may be arranged either as wholepractice—that is, the entire task is taught in its serial order—or as part practice—that is, the task is divided into its fun-damental movement segments or skills {Dubrowski.. Back-stein, Abughaduma, Leidl, & Carnahan, 2005). Basictechnical skills are thus the building blocks of tasks and eanbe objectively assessed in isolation (Bann, Khan, & Darzi,2003: Carnahan. 1993; Datta et al., 2002). Pioneers in taskanalysis, Naylor and Briggs (1963) hypothesized that atask's complexity and its organization detennine how itshould be practiced. Task complexity has been defined as thenumber of movement segments (Magill, 2000), whereas taskorganization refers to the temporal relationship between thecomposite movement segments. According to that theoreti-cal paradigm, optimal learning of a high-complexity, low-organization task will occur under paii-practice conditions.Park. Wilde, and Shea (2004) described that method as onein which complex movement sequences (akin to technicalsurgical tasks) are decompo.sed into smaller, more manage-able parts (akin to technical surgical skills) that one can laterrecombine to create a consolidated sequence. Another viewis that improvement in the performance of sequential motortasks can be regarded as the ability to perform the smallersubsequences together so that the transitions between themovement elements ultimately disappear (Hansen. Trem-blay. & Elliott. 2005), In that view, whole practice wouldseem to be the preferred practice regime. In practical terms,however, part-practice, when conducted on surgically rele-vant bench models, is the most cost-effective choice becausethe tiainees can share equipment, space, and the attention ofthe instructor.

When arranging a part-practice schedule, the instructormay teach individual technical skills in either a blocked ora random order. Comparison of those two practice sched-ules is extensive in motor learning research. Research inpart-practice schedules has historically been focused onmotor skills such as barrier knock-down (Shea & Morgan,1979) and timing (Wulf & Lee, 1993) tasks. What differen-tiates blocked and random schedules is the amount of con-textual interference (Battig, 1966): In the blocked schedule,contextual interference is low (Lee. Wishart, Cunningham.& Carnahan. 1997). whereas in the random schedule, con-textual interference is high because that practice schedulerequires that all skills be intermixed, creating a sessionwithout a definite pattern. The results of a majority of stud-ies have demonstrated that although random practice isdetrimental to performance during the initial practice ses-sions, it leads to better performance on retention and trans-fer tests than does blocked practice of the same skills (Mag-ill & Hall. 1990). The contextual interference effect is notalways present when studied within the realm of complex

real-world tasks, however (Hebert. Landin, & Solmon,19%). Wulf and Shea (2002) explained that the tasks com-monly used to elicit the contextual interference effect ai'erelatively simple, placing low demands on attention andmemory and often requiring a small number of practice tri-als before performance achieves a steady level. Our uniqueaim in the present study was to address the issue of whethera complex surgical task can be learned efficiently by usingthe basic tenets of part-whole learning and contextual inter-ference research.

In this study, we examined the differences in the acquisi-tion of a bicortical bone-plating task imposed by differentpractice schedules. That surgical task ean be defined as a ser-ial muUisegmental task. We subsequently divided the taskinto tlve fundamental technical skills. We isolated those skillsfor use in the part-practice conditions (Table 1). Althougheach skill is independent of the others, the order in whichthey must be conducted is crucial in the operating room.

Our purpose in the present study was to investigate part-versus-whole training in teaching the complex task of ortho-pedic bone plating. We used a transfer test to a more realis-tic model after a retention period to investigate the effect ofpractice conditions on the leaming of the bone-plating task.We assumed that the transfer test would serve not only as atrue indicator of actual learning (Schmidt & Bjork, 1992) butalso as a good index of transfer of learning. Demonstratingtransfer of learning from models with low fidelity (low dif-ficulty) to human tissue (high difficulty) is highly .sought inthe surgical education domain. Based on earlier research(Dubrowski et al., 2005), our experimental hypothesis wasthat the blocked group would show superior performanceduring the acquisition phase, but that random and whole-practice schedules would lead to better transfer performancebecause those conditions contain a higher level of contextu-al interference.

MethodParticipants

The University of Toronto Research Ethics Boardapproved this research, and we obtained informed consentfrom 18 postgraduute Ist-year surgical residents and 8 3rd-year undergraduate medical students (clerks), We randomlyassigned participants to one of three groups; all orthopedicresidents (n = 6) and clerks were equally distributed amongeach group.

Apparatus

We used anatomically correct foam cortical shell models(Pacitlc Research Laboratories. Sawbones, Vashon. WA) inthe shape of a radius bone during the pre-, and posttests aswell as during acquisition (Cristofolini & Viceconti. 2(XX);S/Jvek & Gealer. 1991; Szivek. Weng, & Karpman. 1990).

One of the five discrete skills required drilling actiotis(Table 1). Participants used a pneumatic drill (Hall Series 4,Model 5067. Zimmer Inc.. Warsaw, IN) to perform thattask. The other four skills composing the bone-plating task

January2007, Vol.39, No. 1 41

R. Brydges, H. Carnahan, D, Backstein, & A. Dubrowski

TABLE 1. Description of Separate SkillsPerformed During the Practice Phaseof the Experiment

Skill Description

Reduction and The (obliquely) fractured radius boneappiication of ends are opposed in an anatomicalplate position and held wilh a bone clamp.

Artificial soft tissues surrounded thebone. The bone plate is applied to thattemporary fixation.

Drilling This skill requires the appropriate useof a tissue guard and a power drill. Theparticipant drilled tlve holes toaccommodate the tlve screws requiredfor the fixation. Each hole was drilledindependently and with precisedirection. The surgeon had to apply anappropriate tiniount of pressure topenetrate the bone while avoidingplunging.

Depth The performer uses a depth gauge tomeasurement accurately measure each hole. The

application of that device requiresdexterity and the ability to feel tbe farcortex without actually seeing it(because of tbe sott tissue envelope).

Bone tapping The participant performs this skill toproduce threads for the screw. Thetechnique requires repnuluction of tbeidentical angle used for the drillingprocedure.

Insertion of This skill requires the reproduction ofscrews tbe drilled hole's angle ami tbe

application of enougb torque to enabletbe performer (o in,'̂ crt all five screwstightly without stripping tbe threads.

Note. A panel of three orthopedic surjicons decomposed the entirebone-plating lask inio five functinnally iniiiviJiial p,syL-honiotorskills.

were performed with the Zimmer ECT intern;il fracture fix-ation, small fragtiient plates and screws set {Zimmer Inc.).Participants used a cadaver specimen (whole arm) for thetransfer test.

Procedure

Before the experimental session, all participants viewed ashort instructional video in which an expeti orthopedic sur-geon slowly demonstrated the entire bone-plating task whileexplaining the cotiiponent skills. The detnon strati on waserror free. Then, all participants performed a pretest by com-pleting the entire task, which consisted of proper bonereduction and plate application and the placement of fivescrews into the artificial bone. Next, they were tandomlyassigned to one of the three experimental groups. Each

group performed all five skills involved in the bone-platingtask (Table 1) in an assigned order. Whole training consi.stedof teaching all tWe component parts of the skill during eachpractice bout. Part training consisted of teaching each of thefive component parts of the task separately. Furthermore.participants performed part training in either a blocked orrandom fashion. In blocked training, we taught one compo-nent tlve times before moving to the second component. Inrandom training, we taught each component once, but theorder of components was presented in a random fashion.Thus, the contextual interference (e.g.. Shea & Morgan,1979) varied between the two part-training methods.

We held the number of practice repetitions lor each skiliconstant for all participants in all groups. One criticism of ourmethod is [hat the number of trials was small. However, weintentionally selected that number to represent the most real-istic practice schedule within the current junior residencyeducational program.

After the acquisition phase, participants performed aposttest on the same artificial radius. Following a I-weekretention period, participants returned to complete a trans-fer test on an artificially fractured cadaver radius (an expertsurgeon cut the radius by using a pneumatic saw; HallSeries 4 oscillator). Thus, we assessed participants" petfor-mances three times: before acquisition (pretest); 5-minpostacquisition (posttest); and, after a 1-week rest period,on cadaveric tissue (transfer test). During all three tests, itwas not feasible to measure the five individual surgicalskills separately. The measurements derived from those per-formances therefore represented the entire procedure.

Measurement

We measured technical performance during acquisitionwith the Imperial College (London) Surgical AssessmentDevice (ICSAD) motion analysis system, an objectivemethod of quantifying the movement process of surgicalskills (Aggarwal. Moorthy. et al.. 2004), The ICSAD tnon-itored hand motion characteristics by tracking tbe positionsof magnetic markers placed on the dorsum of each pattici-pant's hands (Datta, Mackay. Mandalia. & Durzi. 2000). Weused a commercially available motion-tracking system (Iso-trak II. Polhcmus. VT) to track the movement of the partic-ipatits' instrutnentcd hands in three-dimensional spacecoordinates. The sampling frequency of the system was 20Hz. Using the position data, custom software (Imperial Col-lege, London) enabled us to derive two motion paranieteisonline: (a) number of hand movements and (b) total time ontask. Movement duration specifically has been consideredthe most sensitive measure of surgical competency onmicroscopic (Starkes. Payk, & Hodges. 1998) and laparo-scopic (Perkins et al., 2002) suturing tasks.

We also captured the performances on all three tests onvideotape for subsequent offline analysis by two expertorthopedic surgeons; the experts used three separate meth-ods for evaluation of the tapes (Dath et al.. 2004). The firstwas a modified version of a six-question, five-anchor-points

42 Journal of Motor Behavior


global rating scale (GRS) developed to measure generaloperative performance (Martin et al.. 1997). The second wasa 15-item checklist of detailed, operation-specific proce-dures identified by a panel of three orthopedic surgeons asnecessary to perfonn the operative task effeetively. Thosethree surgeons were not otherwise involved in the study. Thethird expert-based evaluation method was a final productanalysis, similar to the one developed by Szalay, MacRae,Regehr, and Reznick (2000) for cnd-to-side anastomosis.

As in the case of the duration of the acquisition phase, weselected all of those scoring systems because they are cur-rently used during formative evaluation of surgical compe-tency and would therefore likely enhance the applicabilityof our findings to the surgical education domain.

Statistical Analysis

Because of logistical constraints associated with expertrater availability, we applied only the computer-basedassessment method during the acquisition phase of theexperiment. We subjected number of movements (NM) andtotal time (TTime) to separate 3 (group: whole, random,blocked) x 5 (trial) mixed-design analyses of variance(ANOVAs) with repeated measures on the last variable. Weanalyzed the data for each of the five practiced skills in sep-arate ANOVAs: (a) fracture reduction and application ofplate, (b) drilling, (c) depth measurement, (d) bone tapping,and (e) insertion of screws.

We assessed the normality of the GRSs, the task-specificchecklists, final product analysis, and ICSAD measures(NM and TTime) with a number of Shapiro-Wilk tests; thetests showed that all measures except the final productanalysis were normally distributed (p < .05). On tbe basis ofthat result, we conducted appropriate statistical tests toexamine whether the three groups were equivalent beforeany manipulations. For that purpose, we subjected thescores on the pretest to a one-way ANOVA and to aKruskal-Wallis nonparametric test for final product analy-sis to ensure that all groups were equivalent before theacquisition phase. Where groups were found to differ atpretest, we calculated difference scores between the rawscores on the pre- and posttests and the pre- and transfertests: we used the difference .scores as an index of learning(Dubrowski etal,, 2005).

The tests showed group equivalence for the checklist,F(2, 23) = 0.72, p = .50; GRS, FU, 23) = 1.26. p = .30; NM,F(2. 23) = 0.63, p = .54; and TTime, F(2, 23) = 0.43, p =.66. However, there was a significant difference for the finalproduct analysis, x'̂ '*- N = 3) = 7.77, p < .05. Post hocanalysis (Mann-Whitney U test) revealed that the randomgroup (M = 4.50. SE = 0.25) had a significantly higher .scorethan did both the whole group (M = 3.33. SE = 0.38; U =13.5. p < .05) and the blocked group (M = 3.63, SE = 0.23;U = 12.5, p < .05) on pretest. Despite our best attempts atbalancing the experimental groups on the basis of level oftraining, the pretest differences on that measure may beattributed to inadequate randomizafion or high variability in

the final product analysis measure. We subsequently calcu-lated difference scores for that dependent measure and usedthem for further group comparisons.

We analyzed all scores for the task-specific checklist,GRS. and the two kinematic variables (NM and TTime) infour separate repeated measures ANOVAs. Each ANOVAmodel consisted of a between-participants variable. 3(group; whole, random, blocked), and a within-participantsvariable. 3 (test phase; pretest, posttest, and transfer test),with repeated measures on the last variable. For allANOVAs, we further analyzed effects significant at /? < .05by using Tukey's honestly significant difference post hocmethod for comparison of means. We analyzed the differ-ence scores calculated for the final product analysis with theKruskal-Wallis test to investigate main effects, and we usedthe Mann-Whitney U test to perform post hoc analysis.

Results

Acquisition Phase

Computer-Based Assessment

Table 2 illustrates the group and trial main effects for thetwo computer-based assessment variables measured duringacquisition. No group main effects were found for thereduction and application of plate, depth measurement,bone-tapping, or insertion of screws skills. For the drillingskill, post hoc analysis on the group main effect indicatedthat the blocked group performed fewer movements (i.e.,more efficient execution) and required less time to completethe skill than did the other groups, and that the whole andrandom groups did not differ. The depth measurement andbone-tapping skills did not show any significant trial maineffects for either of the two measures. For the insertion ofscrew skill, there was no trial main effect for the NM vari-able. However, the time to complete the skill did decreasebetween Trial 1 and Trial 5 (y7 < .05). There was similarly astatistically significant difference for both computer-basedmeasures (indicating improvement) between Trial I andTrial 5 for both the reduction and application of plate andthe drilling skills. No Group x Trial interactions were foundduring the acquisition phase.

Testing Phase

Computer-Based Measures

The analyses of the number of movements revealed amain effect only for test, F(l. 23) = ll.l\. p < .01 (FigurelA). All groups improved their performance from pretest toposttest. That measure was also sensitive to changes in per-formance during the 1-week rest period. There was anincrease in the number of movements from posttest to trans-fer test (Figure lA).

Analyses of TTime similarly revealed a main effect onlyfor test, F(l. 23) = 41.40, p < ,001 (see Figure IB); allgroups improved their performance from pretest to posttest.That measure was also sensitive to changes in performanceduring the I-week rest period. There was an increase in

January 2007, Vol, 39, No, 1 43

R, Brydges, H. Carnahan, D, Backstein, & A, Dubrowski

TABLE 2.

Main effect

GroupTrial

GroupTrial

GroupTrial

GroupTrial

GroupTrial

Acquisition Findings for

I

F

1.4014.74

9.625,65

2.513.38

O.U0.56

1.980,88

Each Individua

•Jo, of movements

4f

Reduction

2,23!.23

2.231.23

r

Psychomotor Skill

F

and application of plate

.27 0,94<,001 15.93

Drilling

< ,001<,05

Depth measurement

2,23 .101,23 .08

2,231,23

Bone tapping

.89

.46

Insertion of screws

2,23 ,161.23 ,36

3.5113.51

3.161.60

0.350.06

0.165.58

Time taken (s)

df

2.231. 12

2.231.23

2.231.23

2,231.23

2.231.23

P

.40<.OOI

<.O5< ,001

.06

.22

,71,82

.85<.O5

TTime from posttest to transfer test (Figure IB), It is notclear whether those increases in number of movements andtotal time on tasit were a result of skill degradation orwhether they were a function of the more realistic modelused for the transfer test.

Expert-Ba.ied Measures

Analysis of the GRS revealed a main effect for test. F(l.23) = 33.31, p < .00! {see Figure IC); all three groupsimproved their performance from the pretest to both theposttest and the transfer test, with no statistically significantdifferences between the post- and transfer tests. The analy-ses of the practice-specific checklists similarly revealed amain effect for test. F{1. 23) = 33.55, p < ,001 (Figure ID).All groups showed improvements in scores from the pretestto both the posttest and the transfer test. There were no inter-actions between the experimental groups and test phase foreither the checklist or the GRS.

Because there was a significant difference between therandom group and the two other groups on pretest for thefmal product analysis, we used individual difference scoresfor that measure. The change in the final product analysisfrom pretest to posttest did not yield statistically significantdifferences across the three experimental groups. X'*^- ^ ~3) = 4.39, p = .11. However, the change from pretest totransfer test was significantly different. ^-(2. N =3) =5.91,/) < .05. Here, ihe whole group demonstrated greaterimprovements on the final product than did the random

group, and the improvements demonstrated by the blockedgroup were equivalent to those of both the whole and therandom groups (Figure IE).

Discussion

In this study, we focused on investigating the most appro-priate practice schedule for an orthopedic surgical task.That issue is of utmost importance lor individuals whoteach surgical technical skills, who are always searching forways to enhance the teaching capabilities of staff surgeonsand the learning opportunities of residents in laboratory-based teaching venues to improve patient care delivery andsafety (Dubrowski et al.. 2005; Hamstra & Dubrowski.2005), Our aim in the current investigation was lo deter-mine the most beneficial learning paradigm for novice sur-geons practicing a complex bone-plating task, with the ulti-mate goal of optimizing the learning experience in thelaboratory setting. We used two well-known theoreticalprinciples from the motor learning literature to achieve thatgoal, namely, ihe part-whole practice paradigm and contex-tual interference.

Acquisition Phase

Our results in the acquisition phase showed contextualinterference for only a sub.set of skills: the relative com-plexity of each of the five skills composing the bone-piatingtask can explain that finding. Drilling and depth measure-

44 Journai of Motor Behavior


I d Whole Random Blocked

(A)

600-

Pretest Pastiest Transfer test Pretest Posttest Transfer tesi

(C)

Tiansler lesl

Whole Random Blocked

FIGURE 1. Significant effects for ali computer- and expert-based measures. The computer-based measures. (A) number of move-ments and (B) time on task, are both charted as a function of test. The experl-based ineasures. (C) global rating scale and (D) check-list score, are charted across test. The difference score between the pretest and ibe transfer test, which was calculated for tbe (E) finalproduct analysis (HPA Diff. Seore). is charted across group, *p < ,001. *V < •'̂ " • ***/̂ < •^^^

ment showed specific improvements in perronnance as afunction of practice for the blocked group over those of therandom and whole groups, whereas blocked practice did notdifferentially improve the reduction and application of theplate, bone-tapping, and insertion of screws skills. One

plausible explanation for the differential effect is the inher-ent cognitive effort involved in producing a paiticular skill.One theoretical explanation of the contextual interferenceeffect suggests that the high cognitive effort associated withpracticing under random conditions promotes later retention

January 2007, Vol. 39, No. 1 45

R. Brydges, H. Carnahan, D, Backstein, & A. Dubrowski

and transfer of the practiced task (Lee & Simon. 2004:Schmidt & Lee. 2005). However. Albaret and Thon (1998)put foi-ward the proposition that as the complexity of thepracticed skill increases, participants are forced to use inten-sive cognitive processing and recurrent calls on long-termmemory, regardless of the practice conditions. For example.the information in a participant's working memory couldundergo decay because of the processing of feedbackreceived while completing each complex submovement,forcing him or her to reconstruct the action plan on the sub-sequent trial even if the skill does not change (Albaret &Thon). Thus, the intraskill interference created by the inher-ent complexity of the skill may obscure the contextual inter-ference effect because the increase in cognitive load couldpossibly override the benefits of the inlerskill variation pro-duced by a random practice schedule. Moreover, the combi-nation of random practice and effectively complex sub-movciiients could overload the system, reducing the amountlearned within that schedule (Wulf & Shea. 2002).

Therefore, we first suggest that the component skills ofthe bone-plating task may be categorized as simple or com-plex. For instance, the drilling and depth measurementskills may be categorized as simple because the movementsrequire a single tool, only one to two degrees of freedom,and minimal coordination between the joints of the upperlimb. Those tasks are inherently similar to the simple motortasks (e.g.. key press sequences) commonly employed inprominent contextual interference and part-whole practiceexperiments (Hansen et al., 2005: Shea & Morgan. 1979),On the other hand, the skills of reduction and application ofplate, bone-tapping, and insertion of screws are complexbecause tbey require the use of multiple tools, temporalcoordination of upper limb joints, and movements throughseveral degrees of freedom. Our post hoc categorizationsupports Albaret and Thon's (1998) contention that skillcomplexity interferes (interacts or masks) with the contex-tual interference effect. One limitation of the present designis that we decomposed the task of bone-plating to five skills(movement segments). We based our selection of thatdecomposition of the task on consultations with practicingorthopedic surgeons responsible for teaching the task in theoperating room to novice trainees and on the physicalarrangements used in laboratory-based training, which aredictated mainly by equipment and space limitations. There-fore, the task decomposition was functional. We recognizethat, with further decomposition of the complex skiiis intomore refined and simpler skills, the results of the acquisi-tion phase could possibly be different and more in line withthe contextual interference effect.

Testing Phase

The present results suggest that the combined effect ofthe intraskiil difficulties within the three complex skills ledto an increase in the cognitive load experienced by partici-pants in all three experimental groups. Our results supportthe notion that an increase in contextual interference facili-

tated retention in all groups, as evaluated by the computer-based asse.ssment and by two of the expert-based assess-ments. That fmding differs from the results reported byDubrowski et al. (2005). who conducted a similar studywith medical students. In their study, the whole and randomgroups performed better than did the blocked group on aretention test (i.e.. the same model was used across tests).indeed, nonsignificant trends in the expert-based transfertest data suggested similar results, with the whole and ran-dom groups scoring better than the blocked group on theGRS and checklist measures (Table 3).

An alternative explanation of the transfer test findingsmakes use of tenets from the optimal lramewt)rk proposed byGuadagnoli and Lee (2004). In their challenge-point frame-work, they proposed that every motor task can possess twotypes of difficulties: nominal and functional. The nominaltask difficulty is the complexity of the task under optimalconditions, and the functional difficulty of the task is the dif-ficulty modulated by external factors. For example, pertbrm-ing an orthopedic bone-plating task entails a certain amountof nominal difficulty. Performing the same task on an inani-mate model in a laboratory environment results in a func-tional difficulty lower than the nominal difficulty because ofthe less stressful environment (Hauge. Wanzek, & Godellas,2001; Lingard. Reznick, Espin. Regehr, & DeVito, 2002).Performing the task on a cadaveric model may result in ahigher functional difficulty, and performing the task in theoperating room will further increase the level of functionaldifficulty. One of the premises of Guadagnoli and Lee'sframework is that to evaluate learning objectively, one mustadjust the functional difficulty of the task to the trainees' cur-rent performance level. In the present study, we trained theparticipants during the acquisition phase by using a low-fidelity model of a fractured radius bone. Guadagnoli andLee would describe that task as having low-to-moderatefunctional difficulty because the model would present a rela-tively small challenge to the tested population. The level offunctional task difficulty was significantly increased, howev-er, on the transfer test because the use of the cadaveric armintroduced a number of new challenges, including a new

TABLE 3. Results of Transfer Test GroupComparisons

Dependent variable

Number of movementsTime takenGlobal rating scaleChecklist

F

0.310.311.942.39

Result

4f

2. 232.232.232,23

Nose. The final product analysis (difference score) was %-(2.3t = 5,91,;)<.05,

P

.74

.93,17.11

N^



.sense of realism, wet surfaces, visual obstruction of the oper-ating field by tissue surrounding the bone, denser bone tissue,and variability in the size and shape of the bone. Accordingto the challenge-point framework, the amount of potentiallyavailable infonnation in a task increases exponentially fornovice performers with small increases in the functional dif-ficulty. Thus,, we propose that because our acquisition phasedid not advance the capabilities of our perfonners far enoughbeyond the level of novice, their performance on the transfertest would be quite poor because of the high intemal com-plexity of the task. That result was confirmed—all partici-pants performed at a degraded level on the transfer testregardless of the practice schedule they followed.

Measurement Systems

There was no conformity between the process-oriented(e.g., computer-based measures,, GRS. and task-specificchecklist) and the outcome-oriented (e.g.. final productanalysis) dependent measures. Although the process-orientedmeasures showed that the leaming of all groups was indistin-guishable when assessed on the transfer test, it remains sig-nificant that the outcome-oriented measure was sensitive indetecting group differences following the retention period.That is. the improvements in the fmal product analyses frompretest, through posttest, to transfer test showed that thewhole-practice group significantly improved their outcome incomparison with that of the random group. That outcomemeasure is. however, potentially limited by a ceiling effectbecause the scale of the final product analysis ranged from 0to 5. The random group obtained an improvement score ofonly O.I 1 because the group averaged similar values on boththe pre- and transfer tests. On the basis of that observation,we feel that in future studies in which final product analysisis used as a dependent variable, researchers should considera larger and more sensitive scale. Also, the significant groupdifference at outcome would have found greater support inthis study had we used another computer-based outcome-ori-ented measure to evaluate the plated bone. To account for thatlimitation, in future studies investigators should find con-comitant computer-based outcome-oriented measures for theapplied task being investigated.

Practical Implications

Despite the aforementioned limitations, our results sug-gest that whole practice is the most beneficial schedule forthe complex bone-plating task. When placed on the contex-tual interference continuum, that schedule would likelycause medium levels of contextual interference, whichappear to be sufficient for improving performance.

On the basis of our results, it appears that bone-plating isa task of high organization and high complexity (Naylor &Briggs. 1963). That finding suggests that the compositeskills must be practiced in whole order—that is. underwhole practice conditions—probably because the transitionbetween skills creates a significant change in the kinematiccharacteristics of each component. Wendeioth, Puttemans,

Vangheluwe, and Swinnen (2003), who studied the effec-tiveness of pait-versus-whole training on learning a biman-ual task, provided support for that view. They concludedthat whole-practice conditions enhance the learning of tasksinvolving a high degree of interlimb coordination, such asathletic or musical tasks. Moreover. Hansen et al, (2005)discussed a phenomenon termed the one-target advantage,showing that when individuals string together subcompo-nents of a task to perform a complete movement series, thekinematic characteristics of each skill often change becauseof planning or initiation of adjacent movement elements.Both of those examples suggest that the order of learning ofsubmovements (here, termed technical skills) of a complexmotor task is important to the overall performance of thetask. Evidence is mounting in support of practice schedulesihat require the entire movement sequence to be performedduring acquisition, especially for real-world tasks that areinherently complex. Furthermore, the results indicate theapplicability of Guadagnoli and Lee's (2004) challenge-point framework in the surgical domain. The results of thepresent study have specifically shown the importance ofmatching the practice schedule to the performer, environ-ment, and task. In general, surgical educators can exploitthis framework to determine how practice should be set upfor efficient learning by the novice trainee.

Theoretical Contributions

From a theoretical perspective, the debate regarding thetransferability of principles derived from studies using simpleskills to complex skill learning is also of significant interest.Wulf and Shea (2(X)2) speculated that one feature of motorleaming that is difficult for experimenters to simulate by usingsimple laboratory skills is that the leamer's attention may bedirected to several task elements. For example, the skills prac-ticed in this study consisted of large compound movementswhose subcomponents differed in many characteristics andoften involved a total movement time of 10-30 s during justone trial. Most motor-leaming principles have been developedon the basis of less complex movements with relatively fewerdegrees of freedom and vastly shorter movement times (on themillisecond scale). We believe that the use of existing theoriesand principles in complex environments plays a significantrole in knowledge translation from theoretical to applied tlelds,and vice versa. In particular, principles from the challenge-point framework (Guadagnoli & Lee. 2004) have been usefulin understanding the changes in motor behavior caused bymanipulation of practice schedules in the applied surgical set-ting. There appears to be a necessary requirement forresearchers interested in motor leaming to apply long-standingtheories in more naturalistic settings while preserving the sci-entific methods of the past (Wulf & Shea).

REFERENCES

Aggarwal. R.. Hance. J.. & Darzi. A, (2004), Surgical educationand training in the new millenniuni. Surgical Endoscopy. 18,1409-1410,

January 2007, Vol. 39, No. 1 47

R. Brydges, H. Carnahan, D. Backstein, & A, Dubrowski

Aggarwal, R,. Moorthy. K.. & Darzi, A. (2004), Laparoscopicskills training and assessment. Briti.'ih Journal of Surgery. 91.1549-1558,

Albaret. J. M.. & Thon, B, (1998). Ditferenlial eftects of task com-plexity on contextual interference in a drawing lask. Acta P.s\-clwlogicci. 100. 9-24,

Bann. S, D.. Khun. M. S,. & Darzi. A. (2003), Measurement ofsurgical dexterity using molion analysis of simple bench skills.World Journal of Surgery. 27. 390-394.

Baiiig. W, F. (1966). Hacilitation and interference. In E, A,Bilodeau (Ed.). Acquisition of skill {pp. 134-146), New York:Academic Press.

Carnahan. H. (1993), The role of three-dimensional movementanalysis in the assessment of motor expertise. In J. Starkes & F.Allard (Eds.), Cognitive issues in motor expertise (pp. 35-53).Amsterdam: Elsevier.

Cristofolini, L,, & Vieecontl. M. (2000). Mechanical validation ofwhole bone composite tibia models. Journal of Biomechanics,33. 279-288.

Dalh. D,, Regehr. G.. Bircb. D,. Schlachta, C . Poulin. E.. Mamazza.J., et al, (2004). Toward reliable operative assessment: The relia-bility and feasibility of videotaped assessment of laparoscopictecbnical skills. Surgical Endoscopy. 18. 1800-1804,

Datta. V,. Mackay, S,. Mandalia, M.. & Darzi, A, (2000). The useof electromagnetic motion tracking analysis to obiectively mea-sure open surgical skitl in tbe laboratory-based model. Journalof the American College of Surgeons, 193. 479-485,

Datta, V,. Mandalia. M.. Mackay. S,. Chang. A.. Cheshir, N,, &Darzi, A, (2002), Relationship between skill and outcome in thelaboratory-based model. Surgery. 131, 318-323.

Dubrowski. A.. Backstein. D,. Abughaduma. R,. Leidl, D.. & Car-nahan. H, (2(X)5). The intluence of practice schedules in thelearning of a complex bone-philing surgical task. AmericanJournal of Surgery. 190. 359-363.

Grober, E. D.. Hamstra. S, J,, Wanzel. K. R.. Reznick, R, K.. Mat-sumoto. E. D.. Sidhu, R, S., et al. (20(M), The educationalimpact of bench model tldelity on the acquisition of technicalskill: The use of clinically relevant outcome measures. Annal.tof Surgery. 240. 374-381,

Guadagnoli, M. A,, & Lee. T. D, (2004). Challenge point: A frame-work for conceptualizing tbe etTects for viirious practice condi-tions in motor learning. Journal of Motor Behavior. 36. 212-224,

Hall. J, C . Ellis. C, & Hamdorf. J, (2003), Surgeons and cogni-tive processes. British Journal of Surgery. 90. 10-16,

Hamstra, S, J., & Dubrowski, A. (2005). Effective training andassessment of surgical skills, and the correlates of performance.Surgical Innovation. 12. IX-ll.

Hansen, S,. Tremblay. L.. & Elliott. D, (2005), Part and wholepractice: Chunking and online control in the acquisition of aserial motor task. Research Quarterly for Exercise and Sport.76. 60-66.

Hauge. L. S.. Wanzek. J. A,. & Godellas. C, (2001). The reliabilityof an instrument for identifying and quantifying surgeons" teach-ing in tbe operating room. American Journal of Sur^er\\ 181.333-337.

Hebert, E, P.. Landin, D., & Solmon, M. A, (1996). Practice sched-ule effects on the performance and learning of low- and high-skilled students: An applied study. Research Quarterly for Exer-cise and Sport. 67. 52-58.

Lee. T, D.. & Simon, D, A, (2004), Contextual interference. In A.M. Williams (Ed.). Skill iwquisition in sport: Research, theoryand practice (pp. 29-44). London: Routledge.

Lee. T. D., Wishari. L, R.. Cunningham. S,, & Camahan, H.(1997), Modeled timing information during random practiceeliminates the contextual interference effect. Research Quaner-iy for Exercise and Sport. 68. UX)-IO5.

Lingard, L.. Reznick. R,. Espin. S,. Regehr. G,. & DeVito, I,(2002). Team communications in the operating room: Talk pat-terns, sites of tension, and implications for novices. AcademicMedicine. 77. 232-237.

Magill. R, A. (20(X)), Motor learning: Concepts and applications.New York: McGraw-Hill,

Magill. R, A,, & Hall, K, G, (1990), A review of the contextualinterference effects in motor skill acquisition. Human Move-ment Science, 9. 241-289,

Martin. J. A,. Regehr. G,. Reznick. R.. MacRae. H,. Murnaghan,J,. Hutchison. C et al, (1997), Objective structured assessmentof technical skill (OSATS) for surgical residents. British Jour-nal of Surgery. 84. 273-278.

Naylor. J. C. & Briggs. G, E. (1963), Effects of task complexity andtask organization on the relative efficiency of part and whole train-ing methods. Journal of Experimental Psvchologv. 65. 217-244.

Park. J. H.. Wilde. H.. & Shea. C. H. (2(K)4), Part-whole practiceof movement sequences. Journal of Motor Behavior. 36. 51-61,

Perkins. N.. Starkes. J, L.. Lee. T D.. & Hutchison. C, (2002).Learning to use minimal access surgical instruments and 2-dimensional remote visual feedback: How difficull is the taskfor novices? Advances in Health Science.^ Education: Theoryand Practice, 7, 117-131.

Schmidt, R. A,, & Bjork, R, A. (1992), New conceptualization ofpractice: Common principles in three research paradigms sug-gest important new concepts for practice. Psychological Sci-ence. 3. 207-217.

Schmidt. R, A.. & Lee. T. D. (2005), Motor control anil learning: Abehavioral empha.m (4lh ed.). Champaign. IL: Human Kinetics.

Shea. J, B,. & Morgan. R. L. (1979), Contextual interference onthe acquisition, retention, and iransfer of a motor skill. Journalof Experimental Psychology. 5. 179-187.

Starkes, J. L., Payk. L. & Hodges, N. J. (1998), Developing a stan-dardized test for the assessment of suturing skill in novicemicrosurgeons. Mf(7™7/r,(,'crv. 18. 19-22.

Szalay. D,. MacRae. H.. Regehr. G.. & Reznick. R, (2(K)()), Usingoperative outcome to assess technical skitl, American Journal ofSurgery, 180. 234-237.

Szivek. J, A,. & Gealer. R, L, (1991). Comparison of the deforma-tion response of synthetic and cadaveric femora during simulated.one-legged slance. Journal of Applied Bioniaterial.\. 2. 277-280,

Szivek, J. A.. Weng, M,. & Karpman, R. (1990). Variability in thetorsional and bending response of a commercially availablecomposite femur. Journal of Applied Biomaterials. I. 183-186,

Wenderoth. N., Puttemans. V.. Vangheluwe. S., & Swinnen. S. P,(2003). Bimanual training reduces spatial interference. Journalof Motor Behavior, 35, 296-308,

Wulf. G.. & Lee. T D, (1993), Contextual interference in move-ments of tbe same class: Differential effects on program andparameter learning. Journal of Motor Behavior. 25, 254-263,

Wulf. G,. & Shea. C. H, (2002), Principles derived from the studyof simple skills do not generalize to complex skill learning,P.sychonomic Bulletin and Review, 9, 185-211.

Submitted December 12, 2005Revised June 13, 20()6

Second revision July 18, 2006


application of motor learning principles to complex...

Documents