RESEARCH ARTICLE

Sheng Li Æ Charles T. Leonard

The effect of enslaving on perception of finger forces

Received: 6 April 2005 / Accepted: 8 December 2005 / Published online: 18 January 2006� Springer-Verlag 2006

Abstract The primary purpose was to examine the effectof enslaving on finger force perception during isometricfinger force production using an ipsilateral force-matching paradigm. Fourteen subjects were instructedto produce varying levels of reference forces [10, 20, 30,and 40% maximal voluntary contraction (MVC)] forceusing one finger (index, I or little, L) and to reproducethese forces using the same finger (homo-finger tasks, I/Iand L/L) or a different finger (hetero-finger tasks, I/Land L/I). Forces of all fingers were recorded. Duringhomo-finger tasks, no differences were found in forcemagnitude or relative level of force (expressed as aproportion of MVC). The index finger matching forcemagnitudes were greater than the little finger referenceforce magnitudes, with significantly lower levels of rel-ative force during L/I tasks; while the little fingermatching forces underestimated the index finger refer-ence forces with significantly higher levels of relativeforce during I/L tasks. The difference in the matchingand reference forces by the instructed finger(s), i.e.,matching error, was larger in hetero-finger tasks than inhomo-finger tasks, particularly at high reference forcelevels (30, 40% MVC). When forces of all fingers wereconsidered, enslaving (uninstructed finger forces) sig-nificantly minimized matching errors of the total forceduring both I/L and L/I hetero-finger tasks, especially athigh reference force levels. Our results show that there isa tendency to match the absolute magnitude of the totalforce during ipsilateral finger force-matching tasks. Thistendency is likely related to enslaving effects. Our resultsprovide evidence that all (instructed and uninstructed)finger forces are sensed, thus resulting in perception ofthe absolute magnitude of total finger force.

Keywords Force perception Æ Enslaving Æ Forcematching Æ Finger

Introduction

Many skilled manual activities (e.g., writing, usingchopsticks) require precise control and interactions ofindividual finger forces. The inability to precisely controlindividual finger forces and the interdependence of fin-ger force production has been reported extensively. Forexample, when a finger produces force, other unin-structed fingers within the hand also produce forces. Theterm ‘‘enslaving’’ is used to explain the phenomenon ofaccompanying force production by uninstructed fingers.Enslaving has been observed at various levels of volun-tary force production (Kilbreath and Gandevia 1994;Zatsiorsky et al. 1998, 2000; Li et al. 1998a, b, 2003;Kilbreath et al. 2002; Slobounov et al. 2002). Duringsingle-finger tasks, enslaving increases with the level offorce production of the instructed finger (Slobounovet al. 2002). This effect is more evident during ring andlittle finger tasks than during index finger tasks (Zatsi-orsky et al. 1998).

Peripheral mechanisms, such as shared finger flexormuscles and passive connections among tendons, canpartially account for enslaving. A series of studies,however, have provided evidence that enslaving is lar-gely mediated by the central nervous system (CNS;Danion et al. 2000, 2003; Li et al. 2000a, b, 2004; Latashet al. 2002). The effect has been ascribed to the inabilityof the CNS to precisely partition and direct motorcommands to the motor neuron pool for the desiredfinger, due to the phenomena of convergence anddivergence in the human primary motor cortex (Schieberand Hibbard 1993; Schieber 2001).

Precise force control requires an accurate perceptionof exerted forces. Contralateral paradigms, i.e., referenceforces exerted by one limb being matched by matchingforces generated by the contralateral limb, are usuallyused to investigate perception of isometric force (force

S. Li (&) Æ C. T. LeonardMotor Control Laboratory, School of Physical Therapyand Rehabilitation Science, The University of Montana,Missoula, MT 59812, USAE-mail: sheng.li@umontana.eduTel.: +1-406-2434428Fax: +1-406-2432795

Exp Brain Res (2006) 172: 301–309DOI 10.1007/s00221-005-0332-3

matching) and of heaviness associated with lengtheningand shortening contractions (weight estimation). In suchparadigms, force perception appears to be sensitive tothe maximal force-generating capacities of muscles.When force of a weak muscle(s) (e.g., finger flexors) ismatched by force of a stronger muscle (e.g., elbowflexor), the elbow flexor generates a larger absolutematching force. When forces are each expressed as aproportion of maximal voluntary contraction (MVC)force, the relative forces are more closely estimated(Jones 2003), i.e., force is perceived relative to themaximal force-generating capability of the muscle. Thishas led to the idea that muscular forces are perceivedcentrally based on a ‘‘sense of effort’’ (Gandevia andMcCloskey 1977a, b; Gandevia 1987), leading to thesame relative effort (% MVC) of non-homologousmuscles during contralateral matching tasks.

In contralateral matching tasks involving homolo-gous muscles, when the maximum force-generatingcapacity is decreased by alterations in the state of thematching muscles [e.g., length change (Cafarelli andBigland-Ritchie 1979); fatigue (Jones and Hunter 1983;Carson et al. 2002); or partial curarization (Gandeviaand McCloskey 1977a, b)], the subject perceives, as aresult of the alteration, that the intensity of the referencecontraction in the contralateral homologous muscles hasincreased. Forces are still generated relative to themaximal forces that could be obtained by the matchingmuscle in the altered state.

Finger force perception can also be altered by theactivity of adjacent, functionally related digits. For in-stance, the perceived heaviness of a weight lifted by onedigit increases when an adjacent digit lifts a weightsimultaneously. The magnitude of this error increasesprogressively with the size of the concurrently liftedweight by an adjacent digit, but not by activity of aremote muscle group (e.g., ankle dorsiflexors; Kilbreathand Gandevia 1991).

The purpose of the present study was to investigatethe role of enslaving on finger force perception by usingan ipsilateral force-matching paradigm. Subjects wereasked to produce a reference force using one finger (in-dex, I or little, L) and to reproduce the same amount offorce using the same finger (homo-finger tasks) or adifferent finger (hetero-finger tasks). One might expectthat during hetero-finger tasks, forces would be per-ceived relative to the differences in the maximal force-generating capabilities between index and little fingers.However, the inability to precisely partition centralmotor commands, which is manifested by enslaving,might conceivably alter force perception (cf. Kilbreathand Gandevia 1991, 1992). The little finger has largerenslaving effects than the index finger (Zatsiorsky et al.1998), compensating for differences in maximal force-generating capacities during hetero-finger tasks. There-fore, we hypothesized that enslaving would lead toperception of the absolute magnitude of the total force,instead of the instructed finger force, thereby alteringfinger force generation during hetero-finger tasks. Spe-

cifically, we predicted that force matching would be lessaccurate during hetero-finger tasks than during homo-finger tasks, but enslaving effects would minimize totalforce-matching errors during hetero-finger tasks.

Methods

Fourteen healthy volunteers (four males, ten females;27.1±10.1 years; range 23–51 years) participated in theexperiments. All subjects were right-handed accordingto their preferential use of the right hand duringwriting and eating. All subjects gave informed consentaccording to the procedures approved by The Institu-tional Review Board of the University of Montanaand all procedures were consistent with The HelsinkiDeclaration.

Apparatus

During testing, the subject was seated facing a testingtable. The two upper limbs were symmetrical with re-spect to the body midline with the upper arms atapproximately 45� of abduction in the frontal plane and45� of flexion in the sagittal plane, elbow joints atapproximately 135� of flexion. The right hand and fin-gers were positioned and stabilized into a suspensiondevice. In the suspension device, four unidirectionalpiezoelectric force sensors (208C02, PCB PiezotronicsInc., Depew, NY, USA) were each connected in serieswith wire cables that were suspended by swivel attach-ments from slots in the top plate of the inverted U-shaped frame. The rubber-coated loops, located at thebottom of each wire, were placed in the middle of thedistal phalanges. A hand fixation device was located atthe bottom of the frame and used to stabilize the palm ofthe hand and to ensure a constant hand configurationthroughout the experiment (the wrist was fixed atapproximately 20� of extension and the fingers wereapproximately 20� of flexion at the metacarpophalan-geal joints). The left forearm and hand rested on thetesting table. The suspension system has been repeatedlyused before (Danion et al. 2000; Latash et al. 2002;Shinohara et al. 2003; see Fig. 1 of Latash et al. 2002 fora schematic illustration).

In the described configuration, finger forces wereproduced by flexing the distal interphalangeal joints ofthe fingers. Four force sensors were used to measureindividual finger forces. The force sensors had a range of444.8 N. Analog output signals from the sensors wereconnected to separate signal conditioners (model484B11, PCB Piezotronics Inc.). The signal conditionersoperated in a direct-current-coupled mode, utilizing thesensor’s discharge time constant, as established by thebuilt-in microelectronic circuits within the sensors. Assuch, the time constant of the sensor was ‡500 s. Thesystem involved approximately 1% error over the typicalepoch of recording of a constant signal. Force signals

302

were sampled at 1,000 Hz using a 16-bit analog-to-dig-ital converter (PCI-6229, National Instruments, Austin,TX, USA). The resolution of the system was 2.715 mN/bit. A PC desktop equipped with customized LabVIEWsoftware (National Instruments, Austin, TX, USA) was

used for data acquisition and processing. A Matlabprogram was written for data analysis.

Force-matching tasks

As illustrated in Fig. 1, the subjects were instructed togenerate a reference force pre-determined by a visualtarget (a: reference force generation) using an instructedfinger (3 s), to relax during the middle 3-s period (b:force relaxation), and then to reproduce the same forceas in the first period using the same or another instructedfinger and maintain this force to the end of a trial (4 s; c:force matching). Force sensor data were displayed visu-ally on a computer monitor positioned in front of thesubject. On-line visual display of reference forces al-lowed subjects to monitor the instructed finger forceduring the first 3-s reference force generation. Duringmatching force periods, visual feedback was not pro-vided to subjects (dotted line in Fig. 1). During forcerelaxation, traces of all finger forces were displayed toensure force relaxation.

At the beginning of the experiment, subjects wereasked to produce MVCs using the index finger and thelittle finger. The highest peak value from three trialswas considered as the MVC force. Based on individualMVCs, four target levels were created: 10, 20, 30, 40%MVC of the index (I) and little (L) fingers, respec-tively.

0 1 2 3 4 5 6 7 8 9 10

0

4

8

12

16

20

Time (sec)

Force (N)

a cb

Fig. 1 Illustration of force-matching tasks by a representativehomo-finger task (I20I). The subject was instructed to produce areference force pre-determined by a visual target (20% MVC) usingthe index finger (a: reference force production), to relax for 3 s (b:force relaxation), and to reproduce the same force using the samefinger (c: matching force production). Visual feedback wasprovided for the subject during a and b, but not c. Mean forceswere calculated from 2.0 to 2.5 s for the reference force and from7.5 to 8.0 s for the matching force

Table 1 Mean and standard error (SE) of the reference and matching forces for different matching conditions. FREF, FMATCH: only forceof the instructed finger was considered;

PFREF,

PFMATCH: the total force of all fingers was considered. I/L indicates the reference force

by the index (I) finger and the matching force by the little (L) finger. Similar coding for I/I, L/L, and L/I. The imbedded number representsthe relative level of the reference force (% MVC) [e.g., I20L: reference force of the index finger (I) at 20% MVC, matching force by thelittle finger (L)]. Note that the sum of the forces across all of the fingers was lower than the force applied by one of the fingers, i.e., negativeenslaving, in some tasks such as I10I FREF and FMATCH; I10L FREF and I20L FREF. The weight of the resting fingers was offset to zero atthe beginning of the experiment. The negative enslaving occurred when the uninstructed fingers were lifted off the loops

I/I L/L

I10I I20I I30I I40I L10L L20L L30L L40L

FREF 4.96 9.50 14.27 18.81 2.56 4.77 7.04 9.17SE 0.19 0.38 0.54 0.71 0.19 0.37 0.55 0.70FMATCH 5.81 10.58 14.79 19.80 3.24 6.26 7.29 9.94SE 0.57 0.95 0.95 1.17 0.34 0.70 0.60 0.91P

FREF 4.81 9.65 14.62 19.55 2.88 5.33 8.41 11.35SE 0.19 0.36 0.59 0.80 0.20 0.37 0.54 0.79P

FMATCH 5.68 10.78 15.17 20.71 3.60 7.12 8.67 12.39SE 0.57 0.94 0.99 1.36 0.69 1.47 1.50 1.45

I/L L/I

I10L I20L I30L I40L L10I L20I L30I L40I

FREF 4.90 9.54 14.28 18.57 2.33 4.40 6.48 8.66SE 0.20 0.37 0.51 0.76 0.24 0.47 0.64 0.90FMATCH 6.73 10.45 12.40 14.75 4.58 8.30 8.81 12.50SE 0.61 1.09 1.07 0.95 1.22 1.62 1.43 2.07P

FREF 4.83 9.49 14.61 19.59 2.78 5.35 8.35 11.10SE 0.20 0.37 0.54 0.79 0.36 0.86 1.33 1.39P

FMATCH 7.30 12.63 16.10 20.80 4.61 8.62 9.21 13.06SE 0.69 1.47 1.50 1.45 1.26 1.82 1.58 2.24

303

Four matching conditions were tested:

1. I/L: Reference force (FREF) by I and matching force(FMATCH) by L at four target levels: I10L, I20L, I30L,and I40L, respectively; similarly,

2. I/I: FREF by I and FMATCH by I—I10I, I20I, I30I,I40I;

3. L/L: FREF by L and FMATCH by L—L10L, L20L,L30L, L40L;

4. L/I: FREF by L and FMATCH by I—L10I, L20I, L30I,L40I.

Note that the first three conditions were tested in all14 subjects (n=14), while the fourth condition was tes-ted in eight subjects (n=8). I/I and L/L tasks werehomo-finger tasks in which the same finger producedboth the reference and matching forces, while hetero-finger tasks (I/L and L/I tasks) denoted the referenceproduced by one finger was matched by a matchingforce of another finger.

In each trial, the subjects were explicitly instructed tomatch the magnitude of reference force, i.e., match theforce rather than the level of effort. Approximately 8–10practice trials were allowed for each subject. The intervalbetween two consecutive trials was approximately 20 s.The order of matching conditions was randomized. Eachmatching condition was conducted in a block of fourtrials. Subjects received no feedback regarding force-matching accuracy during the experiment.

Data analysis

During off-line analysis, to standardize data analysis, weselected a 0.5-s period of force production for bothreference and matching forces when forces were moststable. The mean force value from 2.0 to 2.5 s of thereference force was measured as FREF, while the mean

force value from 7.5 to 8.0 s of the matching force wascalculated as FMATCH.

Absolute and relative forces of the instructed fingerswere used to compare reference and matching forces.Absolute forces were recorded finger forces (in newtons),while relative forces were normalized forces with respectto individual MVCs.

Force-matching errors between the instructed fingerswere measured by constant errors (CEs). The followingequation was used: CE = FMATCH - FREF. The indexof CE indicates the direction of matching error. A po-sitive CE indicates an overestimation of FREF duringforce-matching period while a negative CE reflects anunderestimation.

To account for contributions of uninstructed fingerforces (enslaving) to force-matching errors, totalmatching errors were calculated using a similar equa-tion: total CEs (

PCE) = total matching force

(P

FMATCH) - total reference force (P

FREF). Parame-ters for each matching condition were averaged fromfour repeated trials for a given matching condition.

Statistics

Statistical analyses included computing the differences inforce matching between homo- and hetero-finger tasksand the effect of enslaving on force matching duringhetero-finger tasks. Repeated measures ANOVAs wereused with factors FORCE (four levels; 10, 20, 30, 40%MVC), MATCH (two levels; reference and matching) tocompare absolute and relative forces. For comparisonsbetween homo- and hetero-finger tasks, factors FORCEand COND (matching conditions, two levels; homo vs.hetero) were used to compare CEs. ANOVAs were per-formed with factors ENSL (enslaving, two levels; yes orno) and FORCE to assess the effect of enslaving on each

0 1 2 3 4 5 6 7 8 9 100

4

8

12

16

20 Index

Little Total

Ring Middle

Time (sec)

Force (N)Fig. 2 A representative hetero-finger force-matching task(I20L). Note that the matchingforce of the instructed littlefinger is less than the referenceforce of the instructed indexfinger. However, the total forceduring little finger forceproduction is roughly equal tothe total force during indexfinger force production due tothe large contribution ofuninstructed finger forces(enslaving)

304

hetero-finger (I/L, L/I) tasks separately. Whenever nec-essary, post hoc Tukey’s honest significant difference testswere utilized. Data from eight subjects who completed L/I tasks were used (n=8) when comparisons involved L/Itasks. The level of significance was set at p £ 0.05.

Results

For all subjects, the mean index finger MVC (47.4 N)was larger than the mean little finger MVC (23.4 N;p<0.001). When forces by uninstructed fingers (enslav-ing) were included, MVC was 53.8 N for the index fingertask, and 40.8 N for the little finger task (p=0.002).Table 1 summarizes means and standard errors of thereference and matching forces during different matchingconditions.

Homo-finger tasks exhibited different force-matchingbehaviors from hetero-finger tasks (Fig. 1 vs. Fig. 2). Asshown by a representative homo-finger task (Fig. 1,I20I), the instructed finger matching force approximatedits reference force generated by the same (index) finger.In contrast, when subjects were asked to match thisreference force using the little finger (hetero-finger task;Fig. 2, I20L), the little finger matching force (thickdotted line) underestimated the index finger referenceforce. However, the total forces generated by the littlefinger matching task (little finger force + enslavingforces) approximated total index finger reference forces

(index finger force + enslaving forces). As also evidentin Fig. 2, force by the uninstructed ring finger (thindotted line) contributed largely to the increase in thetotal matching force. Further analysis, therefore, fo-cused on the differences between homo- and hetero-fin-ger tasks and the role of uninstructed finger forces(enslaving) on matching behaviors in hetero-finger tasks.

Homo- vs. hetero-finger matching tasks

During homo-finger matching (I/I and L/L) tasks, thematching forces usually approximated the referenceforces by the same instructed finger (Fig. 3a, c, filleddiamonds). No significant differences in magnitudeswere found between the reference forces (FREF) and thematching forces (FMATCH) during I/I tasks. Findingswere the same for L/L tasks with one exception. A two-way ANOVA (n=14) showed significant effects ofMATCH (F[1,13]=6.85, p=0.020), FORCE(F[3,39]=116.47, p<0.001) and a significant interactionMATCH · FORCE (F[3,39]=4.16, p=0.001). Post hoctests revealed that FMATCH was significantly greater thanFREF at 20% MVC reference level (p=0.020). Duringhomo-finger matching tasks, the MVC was the same forthe reference and matching forces. As expected, there-fore, comparisons of relative forces yielded the samepattern of results to those of absolute forces (Fig. 3b, d,filled diamonds).

0

5

10

15

20

25

0 5 10 15 20 250

10

20

3040

50

60

70

0 10 20 30 40 50

I/II/L

0

2

4

68

10

12

14

0 2 4 6 8 10

L/LL/I

0

10

20

30

40

50

0 10 20 30 40 50

L/LL/I

I/II/L

Reference force (N) Reference force (%MVC)

Matching force (N) Matching force (%MVC)

A B

DC

Matching force (N) Matching force (%MVC)

Fig. 3 Absolute and relative forces of the instructed fingersaveraged across trials and subjects. Relative force was defined asa percentage of individual maximal voluntary contraction (MVC)force. Both absolute and relative forces of the matching force areextremely close to that of the reference force during homo-finger

tasks (I/I and L/L filled diamonds). During hetero-finger tasks, thelittle finger force underestimated the index finger reference force (a)and had higher levels of effort (relative force, b); in contrast, theindex finger force overestimated the little finger reference force (c),but with lower levels of effort (relative force, d)

305

Figure 3 depicts differences in absolute and relativereference and matching forces of the instructed fingersbetween homo- and hetero-finger tasks. When the littlefinger was instructed to match a reference force by theindex finger (I/L tasks, Fig. 3a, open squares), the littlefinger matching force was greater than the referenceforce at low reference levels (10% MVC), but less atthe higher reference levels (30 and 40% MVC). A two-way ANOVA (n=14) showed a significant effect ofFORCE (F[3,39]=209.55, p<0.001) and a significantinteraction MATCH · FORCE (F[3,39]=21.89,p<0.001). Post hoc tests showed that FMATCH in I/Ltasks was significantly higher than FREF at 10% MVC,but significantly smaller than FREF at 30 and 40%MVC (p<0.001); FREF and FMATCH were not signifi-cantly different at 20% MVC. Surprisingly, the relativematching force of the little finger was significantlyhigher (by about 20%) than the relative reference forceof the index finger at all tested force levels (Fig. 3b,open squares). A two-way ANOVA (n=14) revealedsignificant effects of MATCH (F[1,13]=57.52, p<0.001)and FORCE (F[3,39]=165.11, p<0.001). No significantinteraction was found.

In contrast, when the index finger was instructed togenerate a force to match the reference force producedby the little finger (L/I, Fig. 3c, open squares), thematching force was greater than the reference force at alltested force levels. A two-way ANOVA (n=8) showedsignificant effects of MATCH (F[1,7]=6.95, p=0.034)and FORCE (F[3,21]=46.22, p<0.001) with no signifi-cant interaction. The relative forces, however, showed aslightly different pattern (Fig. 3d, squares). A two-wayANOVA (n=8) revealed a significant effect of FORCE(F[3,21]=205.82, p<0.001) and a significant interactionFORCE · MATCH (F[3,21]=18.40, p<0.001). Post hoctests showed that the relative FMATCH (index finger) was

-6

-4

-2

0

2

4

6

10 20 30 40FREF

I/LI/I

Constant error (N)

-6

-4

-2

0

2

4

6

10 20 30 40

L/IL/L

FREF

A

B

Fig. 4 Constant errors (CEs) of force-matching tasks for theinstructed reference and matching fingers. Means and standarderrors are presented. CEs are small for homo-finger (a, I/I and b, L/L) tasks at all levels of the reference force (FREF). In contrast, CEsare negative at high levels of FREF during I/L tasks (a), indicatingan underestimation; CEs are constantly larger during L/I tasks (b),indicating an overestimation

0

5

10

15

20

25

I10L I20L I30L I40L L10I L20I L30I L40I

Force (N)

FREF

FMATCH

ÂFREF ÂFMATCH

I/LL/I

Fig. 5 Effect of enslaving on reference and matching forcemagnitudes during hetero-finger tasks (I/L squares; L/I triangles).When enslaving is considered, differences in the total forces aresmaller during reference and matching force production. Theenslaving effect is more obvious for the little finger especially athigh levels of force production. FREF, reference force of the

instructed finger; FMATCH, matching force of the instructed finger;PFREF, sum of all finger forces during reference force production;PFMATCH, sum of all finger forces during matching force

production. Matching conditions, e.g., I20L: reference force by20% MVC of the index finger, matching force by the little finger

306

not significantly different from the relative FREF (littlefinger) at low levels (10, 20% MVC), but was signifi-cantly lower at high FREF levels (30, 40% MVC;p<0.001).

Accuracy of force matching, as measured by CE, wasalso different between homo- and hetero-finger tasks.During the homo-finger matching (I/I and L/L) tasks,no significant differences in CE between the two homo-finger tasks were found, regardless of reference forcelevels. Hetero-finger matching (I/L and L/I) tasks re-sulted in a different CE profile. As compared to CE in I/Itasks (Fig. 4a), CE in I/L tasks was significantly differ-ent at 10, 30, and 40% MVC. CE was negative at 30 and40% MVC, indicating underestimation of the referenceforces (Fig. 4a). A two-way ANOVA (n=14) showedsignificant effects of COND (F[1,13]=6.87, p=0.021),FORCE (F[3,39]=12.42, p<0.001) and a significantinteraction COND · FORCE (F[3,39]=19.78,p<0.001).

Constant error was significantly larger in L/I tasksthan in L/L tasks at all tested force levels (Fig. 4b). Atwo-way ANOVA (n=8) showed significant effects ofCOND (F[3,21]=3.21, p=0.044) and FORCE(F[1,7]=6.05, p=0.043). No significant interaction wasfound.

Both hetero-finger tasks (I/L and L/I) showed similarmatching errors at the 10% MVC reference force level.CE were larger in L/I tasks than in I/L tasks at the 20%MVC level. There was an overestimation of matchingforces (positive CE) for L/I tasks and an underestima-tion of matching forces (negative CE) for I/L tasks athigh reference force levels (30, 40% MVC). A two-wayANOVA (n=8) show main effects of COND(F[1,7]=8.57, p=0.022), and FORCE (F[3,21]=4.73,p=0.011) and a significant interaction COND· FORCE (F[3,21]=13.02, p<0.001).

Enslaving effects on force-matching errors during het-ero-finger tasks

As reported previously, uninstructed finger forces(enslaving) played an important role in hetero-fingerforce-matching tasks (Fig. 2). Figure 5 (see also Table 1)further demonstrates that in hetero-finger tasks,enslaving forces lessened the differences in total forceproduction between two fingers during hetero-fingermatching conditions (I/L and L/I), particularly at highforce levels (30, 40% MVC). This effect, however, wasnot observed in homo-finger matching tasks.

Separate ANOVAs were performed to assess theenslaving effect on each hetero-finger tasks. For I/Ltasks (Fig. 6), a two-way ANOVA (n=8) showed maineffects of ENSL (F[1,7]=15.90, p=0.005), FORCE(F[3,21]=3.17, p=0.045) and a significant interactionFORCE · ENSL (F[3,21]=5.55, p=0.005). Post hoctests (p<0.005 for all tested force levels) indicated thatenslaving led to a significant increase in CEs at 20%MVC (

PCE vs. CE 3.14 N vs. 0.73 N), a significant

change (from negative to positive CE) at 30% MVC(P

CE vs. CE 1.71 N vs. �1.80 N) and a significantlybetter matching accuracy at 40% MVC (

PCE vs. CE

0.65 N vs. �4.15 N), while there was no significant dif-ference at 10% MVC (

PCE vs. CE 2.60 N vs. 1.53 N).

Similarly, there was a significant change in CEs athigher reference force levels in L/I tasks. A two-wayANOVA (n=8) showed a significant effect of ENSL(F[1,7]=9.07, p=0.019), and a significant interactionENSL · FORCE (F[3,21]=4.96, p=0.009). Accordingto post hoc tests, enslaving resulted in a significantreduction in matching errors at 30 and 40% MVC(P

CE vs. CE 0.87, 1.95 N vs. 2.33, 3.84 N; p<0.009),but no significant changes at 10 and 20% MVC (

PCE

vs. CE 1.83, 3.27 N vs. 2.25, 3.90 N).

Discussion

Results from the present experiments included: (1)Homo-finger (I/I, L/L) tasks showed better matchingperformance than hetero-finger (I/L, L/I) matchingtasks, as manifested by significantly larger matchingerrors between the instructed finger forces during hetero-finger tasks. (2) Enslaving effects significantly minimizedmatching errors between total forces generated duringhetero-finger tasks especially at high reference forcelevels (30, 40% MVC) during hetero-finger tasks. Thisresulted in a tendency to match the absolute magnitudeof total force output.

Our results provide evidence supporting the hypoth-esis that enslaving leads to perception of the absolutemagnitude of the total force, instead of only the in-structed finger, thus altering generation of isometricfinger forces during hetero-finger tasks. Differences inenslaving effects during index and little finger tasksappeared to compensate for differences in maximal

-5

-4

-3

-2

-1

0

1

2

3

4

5

I10L

I20L

I30L

I40L

L10I

L20I

L30I

L40I

Force (N)

CEÂ CE

Fig. 6 Effect of enslaving on constant errors (CEs). CEs (CE andPCE) are differences between the corresponding matching and

reference forces shown in Fig. 5. Note that enslaving minimizesCEs for different matching tasks particularly at high reference forcelevels (30, 40% MVC)

307

force-generating capabilities of these two fingers duringhetero-finger tasks (Figs. 5, 6). The index finger had alarger MVC, but demonstrated smaller enslaving par-ticularly at the low force levels. In contrast, the littlefinger had a smaller MVC with larger enslaving effects(cf. Zatsiorsky et al. 2000). When the little finger at-tempted to produce forces with a higher level of effort(% MVC), this increased not only its force output, butalso enslaving forces from uninstructed fingers (cf. Slo-bounov et al. 2002). This indicates that, during hetero-finger matching tasks, other uninstructed fingers con-tribute more to the total force during attempted higherlevels of little finger force production, resulting in thetendency to match the absolute magnitude of total force(see Figs. 5, 6).

In addition, cognitive factors might play a role inhetero-finger force matching, particularly relating toeach individual’s expectations concerning the relativeforce-generating capacities of the index and little fingers.For example, subjects might purposefully increase theintensity of little finger force production (our data indi-cated about 20%) in an attempt to match the referenceforce generated by the stronger index finger during I/Ltasks at all tested force levels (Fig. 3b). An increase in thelittle finger force concomitantly increased enslaving ef-fects contributing to the total matching force, thus min-imizing matching errors of the total force. The oppositeeffect was observed during L/I tasks (Fig. 3d).

The enslaving effects reported here are consistentwith earlier studies that used different methodologies.Kilbreath and Gandevia (1991, 1992) demonstrated thatwhen subjects lifted weights simultaneously with twoinstructed digits but were asked to only estimate theheaviness of the weight on one of the two digits by liftinga weight on the contralateral side, perceived heavinesswas greater than when the weight was lifted only by onedigit. The authors proposed that the CNS is unable topartition precisely the destination of motor commandsto functionally related ‘‘muscles,’’ such that motorcommands for the target digit increase when two digitsvoluntarily lift weights simultaneously, resulting in in-creased perceived heaviness. Our results of enslavingeffects demonstrate that forces of instructed and unin-structed (i.e., enslaving) fingers are perceived within theCNS, and that perception of enslaving influences sub-sequent force-matching performance.

In addition to centrally mediated mechanisms,peripheral afferent signals may modify central motorcommands and play an important role in finger forcematching. Finger flexors, sharing common tendons, of-ten act as a single muscle. There is some anatomicalcompartmentalization of motor units within the flexors.The index finger component has the greatest amount ofmotor unit compartmentalization (Kilbreath and Gan-devia 1994). In addition, the flexor digitorum profundusexhibits ‘‘reflex partitioning,’’ i.e., each of its motoneu-ron sub-pools has a distinct pattern of cutaneomuscularreflexes (Kilbreath et al. 1997). Cutaneomuscular re-flexes of the index finger are facilitatory, leading to an

increased perceived heaviness after index finger anes-thesia, while the corresponding effect is inhibitory in thelittle finger (Kilbreath et al. 1997). Conceivably, whenthe reference force generated by the index is matched bythe little finger force (I/L tasks), due to inhibitory cu-taneomuscular reflexes in the little finger, motor com-mands could be increased, resulting in a higher percentMVC in the little finger.

The results further contribute to our understanding ofthe role of the CNS in precise finger control duringfunctional tasks. Enslaving affects finger force perceptionand subsequent force-matching performance. This sug-gests that all finger forces (instructed and uninstructedfingers) are perceived within the CNS and that all fingersare indeed integrated into a meaningful synergy for de-sired functional purposes. For instance, during hand-writing tasks, forces generated by each finger changecontinuously and precise control of force and moment isrequired (Latash et al. 2003). Perception of forces fromall fingers, therefore, becomes extremely important forpen stabilization and movement during such a task.Precise regulation of individual finger forces thus providegrasp stability during manipulative functions of the hand(Johansson et al. 1999; Pataky et al. 2004a, b).

The tendency to match the absolute magnitude of thetotal force during ipsilateral hetero-finger tasks seems tobe a departure from previous findings using contralateralmatching paradigm that forces are perceived relatively tomaximal force-generating capabilities (finger force vs.elbow flexion force, see Sect. ‘‘Introduction’’). Theseapparent disparate findings, however, are perhaps notcontradictory, but rather indicate the possibility that theCNS adopts different strategies for different functionalpurposes. Conceivably, it is more practical for the CNS tomaintain relative rather than absolute efficacy in muscleforce production during contralateral force matching.This imparts an advantage when the CNS needs to adaptto changes/differences in the contralateral muscles, e.g.,contractile changes in the contralateral homologousmuscle after fatigue (Carson et al. 2002) or maximalforce-generating capacities between contralateral non-homologous muscles (Jones 2003). This behavior ofmaintaining relative efficacy might also be beneficial forcoordinated movement involving non-homologous mus-cles of two limbs. Taken together, it seems that the CNSadopts different strategies during perception of fingerforces within the hand and in the contralateral hand.Selection of different strategies required for differentfunctional tasks suggests that the CNS is able to choose astrategy which imparts most precision and success.

To conclude, during ipsilateral hetero-finger force-matching tasks, subjects showed a tendency to match theabsolute magnitude of the total force during single-fin-ger force production tasks. This tendency is likely re-lated to the forces by uninstructed fingers (enslaving).Enslaving affects finger force perception and subsequentforce-matching performance. These effects reflect therole of the CNS in precise finger control during func-tional tasks.

308

Acknowledgments This work was supported in part by startupfunds provided to S. Li from The University of Montana and TheMJ Murdock Charitable Trust Foundation. The authors aregrateful to anonymous reviewers for their constructive comments,to Professor M.L. Latash for his valuable comments on an earlierversion of the manuscript, and to Professor E.R. Ikeda for helpfuldiscussions.

References

Cafarelli E, Bigland-Ritchie B (1979) Sensation of static force inmuscles of different length. Exp Neurol 65:511

Carson RG, Riek S, Shahbazpour N (2002) Central and peripheralmediation of human force sensation following eccentric orconcentric contractions. J Physiol 539:913–925

Danion F, Latash ML, Li ZM, Zatsiorsky VM (2000) The effect offatigue on multifinger co-ordination in force production tasksin humans. J Physiol 523(Pt 2):523–532

Danion F, Latash M, Li S (2003) Finger interactions studied withtranscranial magnetic stimulation during multi-finger forceproduction tasks. Clin Neurophysiol 114:1445–1455

Gandevia SC (1987) Roles for perceived voluntary motor com-mands in motor control. Trends Neurosci 10:81–85

Gandevia SC, McCloskey DI (1977a) Changes in motor com-mands, as shown by changes in perceived heaviness, duringpartial curarization and peripheral anaesthesia in man. JPhysiol 272:673–689

Gandevia SC, McCloskey DI (1977b) Effects of related sensoryinputs on motor performances in man studied through changesin perceived heaviness. J Physiol 272:653–672

Johansson RS, Backlin JL, Burstedt MK (1999) Control of graspstability during pronation and supination movements. ExpBrain Res 128:20–30

Jones LA (2003) Perceptual constancy and the perceived magnitudeof muscle forces. Exp Brain Res 151:197–203

Jones LA, Hunter IW (1983) Effect of fatigue on force sensation.Exp Neurol 81:640–650

Kilbreath SL, Gandevia SC (1991) Independent digit control:failure to partition perceived heaviness of weights lifted bydigits of the human hand. J Physiol (Lond) 442:585–599

Kilbreath SL, Gandevia SC (1992) Independent control of thedigits: changes in perceived heaviness over a wide range offorce. Exp Brain Res 91:539–542

Kilbreath SL, Gandevia SC (1994) Limited independent flexion ofthe thumb and fingers in human subjects. J Physiol 479:487–497

Kilbreath SL, Refshauge K, Gandevia SC (1997) Differentialcontrol of the digits of the human hand: evidence from digitalanaesthesia and weight matching. Exp Brain Res 117:507–511

Kilbreath SL, Gorman RB, Raymond J, Gandevia SC (2002)Distribution of the forces produced by motor unit activity in thehuman flexor digitorum profundus. J Physiol 543:289–296

Latash ML, Li S, Danion F, Zatsiorsky VM (2002) Centralmechanisms of finger interaction during one- and two-handforce production at distal and proximal phalanges. Brain Res924:198–208

LatashML,Danion F, Scholz JF, Zatsiorsky VM, SchonerG (2003)Approaches to analysis of handwriting as a task of coordinating aredundant motor system. HumMov Sci 22:153–171

Li ZM, Latash ML, Newell KM, Zatsiorsky VM (1998a) Motorredundancy during maximal voluntary contraction in four-fin-ger tasks. Exp Brain Res 122:71–78

Li ZM, Latash ML, Zatsiorsky VM (1998b) Force sharing amongfingers as a model of the redundancy problem. Exp Brain Res119:276–286

Li S, Danion F, Latash ML, Li ZM, Zatsiorsky VM (2000a)Characteristics of finger force production during one and two-hand tasks. Hum Mov Sci 19:897–924

Li S, Danion F, Latash ML, Li ZM, Zatsiorsky VM (2000b) Fingercoordination in multi-finger force production tasks involvingfingers of the right hand and/or fingers of the left hand. J ApplBiomech 16:379–391

Li S, Latash ML, Zatsiorsky VM (2003) Finger interaction duringmulti-finger tasks involving finger addition and removal. ExpBrain Res 150:230–236

Li S, Latash ML, Zatsiorsky VM (2004) Effects of motor imageryon finger force responses to transcranial magnetic stimulation.Cogn Brain Res 20:273–280

Pataky TC, Latash ML, Zatsiorsky VM (2004a) Prehension syn-ergies during nonvertical grasping. I experimental observations.Biol Cybern 91:148–158

Pataky TC, Latash ML, Zatsiorsky VM (2004b) Prehension syn-ergies during nonvertical grasping. II: modeling and optimiza-tion. Biol Cybern 91:231–242

Schieber MH (2001) Constraints on somatotopic organization inthe primary motor cortex. J Neurophysiol 86:2125–2143

Schieber MH, Hibbard LS (1993) How somatotopic is the motorcortex hand area? Science 261:489–492

Shinohara M, Li S, Kang N, Zatsiorsky VM, Latash ML (2003)Effects of age and gender on finger coordination in MVC andsubmaximal force-matching tasks. J Appl Physiol 94:259–270

Slobounov S, Johnston J, Chiang H, Ray W (2002) The role of sub-maximal force production in the enslaving phenomenon. BrainRes 954:212–219

Zatsiorsky VM, Li ZM, Latash ML (1998) Coordinated forceproduction in multi-finger tasks: finger interaction and neuralnetwork modeling. Biol Cybern 79:139–150

Zatsiorsky VM, Li ZM, Latash ML (2000) Enslaving effects inmulti-finger force production. Exp Brain Res 131:187–195

309