Electroencephalography and clinical Neurophysiology, 81 ( 1991 ) 353- 358
353
© 1991 Elsevier Scientific Publishers Ireland, Ltd. 0924-980X/91/$03.50 ADONIS 0924980X9100095N
E L M O C O 90250
Long latency postural responses are functionally modified by cognitive set D.J. B e c k l e y a, B . R . B l o e m a,b, M . P . R e m l e r a, R . A . C . R o o s b a n d J . G . V a n D i j k b " Department of Neurology, School of Medicine, Unit,ersity of California, Dat,is, CA (U.S.A.), and b Department of Neurology, University Hospital, Leiden (The Netherlands) (Accepted for publication: 25 March 1991)
Summary We examined how cognitive set influences the long latency components of normal postural responses in the legs. We disturbed the postural stability of standing human subjects with sudden toe-up ankle rotations. To influence the subjects' cognitive set, we varied the rotation amplitude either predictably (serial 4 ° versus serial 10°) or unpredictably (random mixture of 4 ° and 10°). The subjects' responses to these ankle rotations were assessed from the EMG activity of the tibialis anterior, the medial gastrocnemius, and the vastus lateralis muscles of the left leg. The results indicate that, when the rotation amplitude is predictable, only the amplitude of the long latency (LL) response in tibialis anterior and vastus lateralis varied directly with perturbation size. Furthermore, when the rotation amplitude is unpredictable, the central nervous system selects a default amplitude for the LL response in the tibialis anterior. When normal subjects are exposed to 2 perturbation amplitudes which include the potential risk of falling, the default LL response in tibialis anterior appropriately anticipates the larger amplitude perturbation rather than the smaller or an intermediate one.
Key words: Long-latency; EMG; Cognitive set; Posture; Motor control; Balance
In standing subjects, toe-up rotational perturbations of a tiltable support platform elicit short latency (SL) and medium latency (ML) responses in the stretched gastrocnemius muscle and long latency (LL) responses in the shortened tibialis anterior and vastus lateralis muscles (Nashner 1976; Dichgans and Diener 1987). Studies in normal subjects (Nashner 1976; Allum and Biidingen 1979; Nashner and Cordo 1981; Allum 1983; Diener et al. 1983, 1984b, 1988) and neurological patients (Diener et al. 1984a,c, 1985; Scholz et al. 1987; Bloem et al. 1990; Beckley et al. 1991) indicate that LL responses contribute to postural stabilization, whereas ML responses further destabilize the subject who is being propelled backward by the rotational perturbation. Since LL responses are important in postural control, it would be of interest to determine how these responses are regulated during conditions in which the exact requirements of force and amplitude cannot be reliably predicted. Studies of upper limb trajectory control have shown that the nervous system prepares default responses under conditions of uncertainty (Cordo 1987; Hening et al. 1988). Such default reCorrespondence to: Dennis J. Beckley, M.D., Veterans Administration Medical Center - 127, 150 Muir Road, Martinez, CA 94553
(U.S.A.). Tel.: (415) 229 0328.
sponses appear to be midway between those needed for large and small movements. In an analogous manner, Horak and colleagues (1989) used translational perturbations of a supporting platform to study this effect of "central set" on lower extremity postural reflexes by varying perturbation amplitudes either predictably or unpredictably. Similar to the results obtained in upper limb trajectory studies, the size of the posturally stabilizing response during unpredictable perturbations tended towards a default value corresponding to a medium size perturbation. However, a medium size default postural response might be insufficient to prevent a fall in reaction to unexpectedly large perturbations. Therefore, it would seem more appropriate if the default postural response anticipated large rather than intermediate or small size perturbations in order to prevent falls during unpredictable conditions. It is possible that Horak et al. (1989) arrived at their conclusion of a medium size default response because their protocol of translational perturbations did not entail the risk of failing. Therefore, we studied the effect of cognitive set on the default postural response in a paradigm of rotational perturbations which contained the potential risk of falling. Specifically, we tested postural reflexes in normal healthy subjects in response to serial predictable small amplitude (4°) and large amplitude (10 °) toe-up rotational perturbations
354 and compared them to responses obtained during a random mixture of large and small amplitude perturbations. In the random condition, in which loss of balance could result from making the wrong preparation, we predicted that the size of the default LL response in the stabilizing tibialis anterior and vastus lateralis muscles would be altered by cognitive set to be consistently large rather than small or intermediate in amplitude. Preliminary results of this study have been published in abstract form (Beckley and Remler 19891.
Subjects and methods Ten healthy adults (age range 25-41 years, mean 34.3; 5 females and 5 males) were selected for testing after having given informed consent, as approved by the Institutional Review Board at the Martinez Veteran's Administration Medical Center. All subjects were normal according to a detailed neurological examination. Surface E M G recordings, with 9 mm diameter disk electrodes 3 cm apart, were obtained from 3 muscles in the left leg (tibialis anterior (TA), medial gastrocnemius (MG), and vastus lateralis (VL)). The subjects stood with their feet 10 cm apart on a movable forceplate platform (NeuroCom) and were told to fixatc their gaze on a spot at eye level directly in front. Initial and inter-trial vertical stance were monitored by visual inspection of the subject and live oscilloscopic display of T A and M G activity. No pretest information regarding perturbation type, amplitude, or sequence was provided. Stimuli consisted of 3 different protocols, each of 20 trials of forceplate toe-up ramp movements of preselected amplitudes and variable predictability: (a) serial predictable: 4 ° amplitude (stimulus duration = 80 msec); (b) serial predictable: 10° amplitude (stimulus duration = 200 msec); (c) random mixture of 4°/10 ° amplitude, all at a constant velocity of 50°/sec (50 msec to peak). The 3 protocols produced 4 different testing conditions for each subject: 4 ° predictable (4P); 4 ° random (4R); 10° predictable (10P) and 10 ° random (10R). These two perturbation amplitudes were selected after an earlier pilot study had shown that 4 ° appears to be the minimum amplitude that yields consistent postural reflexes, whereas 10 ° is just at the upper tolerance range of stability for most normal subjects. Most subjects had occasional loss of balance during the 10° perturbation, but never during a 4 ° perturbation and all reported initial feelings as if they might fall during the large perturbation. The sequence of the 3 sets was randomly varied between subjects to reduce a potential confounding effect of ordering. The interval between consecutive trials varied from 10 to 20 sec.
D.J. BECKLEY ET At.. Raw E M G data were pre-amplified, bandpass filtered (10 H z - 1 kHz), full-wave rectified and integrated over a time constant of 4.5 msec and stored to computer disk for off-line analysis (RC Electronics). The E M G sweep was triggered by an electrical pulse simultaneous with the onset of platform movement. The total post-stimulus sample period was 500 msec and the data sampling rate was 2 kHz (bin width 500/.,sec). Optimal gain settings for each subject and control for possible crosstalk were obtained by having each subject maximally contract each muscle prior to actual testing. Gains were kept constant for each subject throughout the experiment. Onset latencies for St. and Mr. responses m the medial gastrocnemius and LL responses in the tibialis anterior (LL (TA)) and vastus lateralis (LL (VL)) were measured relative to the falling edge of the trigger pulse, which coincided with the onset of forceplatc movement. The long latency innervation sequence was calculated by subtracting the LL (TA) onset latency from the LL (VL) onset latency. Normal subjects have a distal to proximal innervation pattern that results in a positive LL innervation sequence (Nashner 19771. Onset criteria R~r the various E M G responses were determined by reference windows as follows: 30-60 msec post trigger (SL), 70-120 msec post trigger (ML), and 100-200 msec post lrigger (LL (TA) and LL (VL)) and by voltage thresholds set at twice the pre-stirnulus noise level. Mean amplitudes of the SL response Cover its complete measured duration) and of the ML response and both LL responses (over the first 75 msec) were calculated by computer algorithm. Mean amplitudes of SL, ML and both LL responses were corrected for background activity by subtracting the first 50 msec of E M G activity prior to the onset of platform movement. The resultant value (amplitude score) was then converted to a standardized Z score to permit comparison of E M G responses between subjects over the range of testing conditions (Hansen et al. 1988)i
Statistical analysis Statistical analysis consisted of 2 factor repeated measures A N O V A comparing the E M G response mean amplitude Z scores and latencies for all 10 subjects for the 4 testing conditions (4P, 4R, 10P, and 10R) (SYS"FAT v4.1). This was followed by post hoc multiple comparisons with linear contrast and appropriate Bonferroni adjustments if the univariate F statistic had a significant P value of < 0.05: Univariate A N O V A was chosen over M A N O V A because of the greater reduction in power when using the latter method with small sample sizes (Wilkinson 19901. Greenhouse-Geisser corrections were not applied to the degrees of freedom because the F ratio is not significantly positively biased
355
LL R E F L E X E S A R E M O D I F I E D BY C O G N I T I V E SET
Ch,4 TRIGGER
J"l
I I
I
Ch.2 MG EMG
I I I
Ch.1 TA EMG
.75
II I
o 15
tsL t.L
I
1 I
Ch3. VL EMG
I
I
-150
tLLTA
I
I
I I I I I
0
~
I
I
.35 ~
tLLVL I
150 50 msec/dlv.
I
'
~
I 0 VOLTS
I
I
300
Fig. 1. Average of 20 trials in the 4 ° predictable condition for a normal subject. Vertical arrows indicate onset of the SL and M L responses in the medial gastrocnemius muscle (MG) and the LL responses in the tibialis anterior (TA) and vastus lateralis (VL) muscles. The dotted reference line indicates the onset of forceplate movement relative to which response latencies are measured. Amplifications are x 5000 for channels 1 and 2 and × 2000 for channel 3. Ch = channel.
with only 2 levels of each repeated factor (Keppel 1982).
Results
E M G response latencies
A representative sample of recorded E M G activity is shown in Fig. 1. Onset latencies for all responses, as well as the LL innervation sequence, are within the normal ranges reported by others (Diener et al. 1983; Scholz et al. 1987) (Table I). There were no significant differences for the various response latencies or the LL innervation sequence when comparing the 4 testing conditions (4P, 4R, 10P, and 10R) with 2 factorial repeated measures ANOVA. The number of missing responses included 13% for the SL response, 4% for the ML response and none for the LL responses. E M G response mean amplitudes
We found that the mean amplitude of the LL (TA) response varied directly with perturbation size in the
predictable test conditions (10P > 4P; P < 0.05). A more notable finding was that the size of the LL (TA) response for the 4R condition was considerably increased compared to the 4P condition ( P < 0.005). This is shown as a significant interaction effect ( P < 0.05) between perturbation amplitude and cognitive set and is displayed graphically in Fig. 2A. Post hoc multiple comparisons showed that the mean LL (TA) Z scores for the 4P condition were significantly less than the mean LL (TA) Z scores for the 4R, the 10P, and the 10R conditions ( P < 0.005) and that there was no significant difference between the 4R, 10P, or 10R conditions. The LL (TA) response in the 4R condition is upregulated to a size appropriate to anticipate a maximal size perturbation, as in condition 10P. Furthermore, confidence intervals show no overlap between the mean of condition 4P and the means of conditions 4R, 10P and 10R at the 99% confidence limits (Fig. 3). In contrast, neither the SL nor the ML responses displayed significant mean amplitude differences across the 4 testing conditions. The LL (VL) response revealed a significant main effect for perturbation ampli-
TABLE I Mean latencies of short (SL), medium (ML) and long latency (LL) responses for all subjects (n = 10) for each test condition (4P, 4 ° predictable amplitude; 4R, 4 ° random amplitude; 10P, I0 o predictable amplitude; 10R, 10 ° random amplitude). Data are expressed as mean_+ S.D. Mean latencies (msec)
SL ML LL (TA) LL (VL) LL IS
Test conditions 4P
4R
10P
10R
42.9_+ 3.3 86.6_+ 6.4 128.6_+ 10.1 144.6 + 13.4 16.0-t- 13.7
42.2_+ 3.0 84.4_+ 3.0 127.3 -+ 11.3 143.0 _+ 10.8 16.0_+ 12.6
42.3_+ 4.4 86.4_+ 4.4 127.2-+ 10.4 146.6 -+ 11.0 20.0-+ 8.7
42.7_+ 3.6 87.2_+ 4.3 128.1 + 10.7 147.3 -+ 10.4 19.4_+ 11.2
LL (TA), tibialis anterior long latency response; LL (VL), vastus tateralis long latency response; LL IS, long latency innervation sequence [LL (VL) - LL (TA)].
356
D.J. B E C K L b Y E T AL.
ing role in this paradigm, and possibly also in daily life (Allure 1983; Diener et al. 1984a,b). Response latencies and the distal-proximal innervation sequence were not influenced by either perturbation amplitude or cognitive set, confirming previous o b ~ r v a t i o n s (Diener et al. 1984b; Horak et at. 19891• This is in keeping with the suggestion that latencies arc programmed from minimal sensory information and separately from response amplitudes (Diener et al. 1988). Reversal of the distal-proximal innervation sequence or co-contraction of distal agonist-antagonist muscles has been reported in normals perturbed while standing on narrowed support surface configurations (Horak and Nashncr 19861 or while pre-leaning backward during toe-up tilts (Diener et al. 1983). Unlike those paradigms, it is probable that even our 10° rotational perturbations did not create the degree of postural instability that would require our subjects to switch from an "ankle strategy" to a "hip strategy" or "cocontraction strategy/" The lack of a default LL (VL) response was unexpected since it is in contrast to what is predicted by the "postural synergy hypothesis" (Nashner and McCollum 1985), The lack of influence of cognitive set on proxi-real muscle synergists is consistent with the theory that set-dependent postural regulation is first directed toward distal stabilizing muscles (Horak et al. 19891. Unlike other studies which used a translational paradigm (Diener et al. 1988; Horak et al. 1989), we did find a main effect in the proximal muscle response for stimulus amplitude, possibly because of the more destabilizing nature of our rotational paradigm. Indeed, responses to rotational perturbations differ in many ways from translational perturbations (Nashner 1976; Diener et al. 1983: Nardone et al. 1990). In
0 r.~
0.6
N ~
~
+
10R
10P
o.o
-o,4
10° ¢'4P
-o,s ~.I
TOE-UP FORCEPLAI~ i
, i
AMPLITUDE
TEST CONDITIONS 0 0
(,'3 N
o.s
P~.
o,~ o.o
~ ~
-o.~
m
0
-0.¢
r,z,.1
:.5
"F
-0,6
FORCEPLATE AMPLITUDE SERt~
P,A N ~ M
TEST CONDITIONS Fig. 2. A: c o m p a r i s o n of the g r a n d m e a n a n d s t a n d a r d e r r o r of the m e a n for tibialis a n t e r i o r (TA) LL Z scores for all subjects for each of the 2 levels of b o t h i n d e p e n d e n t v a r i a b l e s - - f o r c e p l a t e a m p l i t u d e (4 ° and 10°) and d e g r e e of predictability ( " S e r i a l " or "Random*'). T h e r e is a significant i n t e r a c t i o n effect ( F (1, 9 ) = 10.36, P < 0.05) for the 4 ° r a n d o m condition (4R), i n d i c a t i n g that this r e s p o n s e is strongly i n f l u e n c e d by the d e g r e e of p e r t u r b a t i o n predictability. The difference in m e a n a m p l i t u d e b e t w e e n the p r e d i c t a b l e and unpredictable 10° p e r t u r b a t i o n is not significant. B: c o m p a r i s o n of the g r a n d m e a n and s t a n d a r d e r r o r of the m e a n for vastus lateralis (VL) LL Z scores for all subjects for each of the 2 levels of both i n d e p e n d e n t v a r i a b l e s - - f o r c e p l a t e a m p l i t u d e (4 ° and 10°) and d e g r e e of predictability ( " S e r i a l " or " R a n d o m " ) . T h e r e is a significant main effect for a m p l i t u d e only ( F (1, 9 ) = 13.90, P < 0.01) bu! no interaction effect ( F (1, 9 1 = 1 . 2 4 , P < 0 . 3 0 1 . Thus, while the g r a p h suggests a t r e n d for condition 4 R to be larger t h a n 4P. t h e r e is no clear e v i d e n c e t h a t cognitive set modifies the L L r e s p o n s e in the proximal vastus lateralis muscle.
tude only (10P > 4P. P < 0.01) and no significant interaction effect ( P < 0.30) (Fig. 2B).
Discussion
CONFIDENCE INTERVALS ABOUT THE
uJ O O if) N u.l ¢3
E
5r~
353
© 1991 Elsevier Scientific Publishers Ireland, Ltd. 0924-980X/91/$03.50 ADONIS 0924980X9100095N
E L M O C O 90250
Long latency postural responses are functionally modified by cognitive set D.J. B e c k l e y a, B . R . B l o e m a,b, M . P . R e m l e r a, R . A . C . R o o s b a n d J . G . V a n D i j k b " Department of Neurology, School of Medicine, Unit,ersity of California, Dat,is, CA (U.S.A.), and b Department of Neurology, University Hospital, Leiden (The Netherlands) (Accepted for publication: 25 March 1991)
Summary We examined how cognitive set influences the long latency components of normal postural responses in the legs. We disturbed the postural stability of standing human subjects with sudden toe-up ankle rotations. To influence the subjects' cognitive set, we varied the rotation amplitude either predictably (serial 4 ° versus serial 10°) or unpredictably (random mixture of 4 ° and 10°). The subjects' responses to these ankle rotations were assessed from the EMG activity of the tibialis anterior, the medial gastrocnemius, and the vastus lateralis muscles of the left leg. The results indicate that, when the rotation amplitude is predictable, only the amplitude of the long latency (LL) response in tibialis anterior and vastus lateralis varied directly with perturbation size. Furthermore, when the rotation amplitude is unpredictable, the central nervous system selects a default amplitude for the LL response in the tibialis anterior. When normal subjects are exposed to 2 perturbation amplitudes which include the potential risk of falling, the default LL response in tibialis anterior appropriately anticipates the larger amplitude perturbation rather than the smaller or an intermediate one.
Key words: Long-latency; EMG; Cognitive set; Posture; Motor control; Balance
In standing subjects, toe-up rotational perturbations of a tiltable support platform elicit short latency (SL) and medium latency (ML) responses in the stretched gastrocnemius muscle and long latency (LL) responses in the shortened tibialis anterior and vastus lateralis muscles (Nashner 1976; Dichgans and Diener 1987). Studies in normal subjects (Nashner 1976; Allum and Biidingen 1979; Nashner and Cordo 1981; Allum 1983; Diener et al. 1983, 1984b, 1988) and neurological patients (Diener et al. 1984a,c, 1985; Scholz et al. 1987; Bloem et al. 1990; Beckley et al. 1991) indicate that LL responses contribute to postural stabilization, whereas ML responses further destabilize the subject who is being propelled backward by the rotational perturbation. Since LL responses are important in postural control, it would be of interest to determine how these responses are regulated during conditions in which the exact requirements of force and amplitude cannot be reliably predicted. Studies of upper limb trajectory control have shown that the nervous system prepares default responses under conditions of uncertainty (Cordo 1987; Hening et al. 1988). Such default reCorrespondence to: Dennis J. Beckley, M.D., Veterans Administration Medical Center - 127, 150 Muir Road, Martinez, CA 94553
(U.S.A.). Tel.: (415) 229 0328.
sponses appear to be midway between those needed for large and small movements. In an analogous manner, Horak and colleagues (1989) used translational perturbations of a supporting platform to study this effect of "central set" on lower extremity postural reflexes by varying perturbation amplitudes either predictably or unpredictably. Similar to the results obtained in upper limb trajectory studies, the size of the posturally stabilizing response during unpredictable perturbations tended towards a default value corresponding to a medium size perturbation. However, a medium size default postural response might be insufficient to prevent a fall in reaction to unexpectedly large perturbations. Therefore, it would seem more appropriate if the default postural response anticipated large rather than intermediate or small size perturbations in order to prevent falls during unpredictable conditions. It is possible that Horak et al. (1989) arrived at their conclusion of a medium size default response because their protocol of translational perturbations did not entail the risk of failing. Therefore, we studied the effect of cognitive set on the default postural response in a paradigm of rotational perturbations which contained the potential risk of falling. Specifically, we tested postural reflexes in normal healthy subjects in response to serial predictable small amplitude (4°) and large amplitude (10 °) toe-up rotational perturbations
354 and compared them to responses obtained during a random mixture of large and small amplitude perturbations. In the random condition, in which loss of balance could result from making the wrong preparation, we predicted that the size of the default LL response in the stabilizing tibialis anterior and vastus lateralis muscles would be altered by cognitive set to be consistently large rather than small or intermediate in amplitude. Preliminary results of this study have been published in abstract form (Beckley and Remler 19891.
Subjects and methods Ten healthy adults (age range 25-41 years, mean 34.3; 5 females and 5 males) were selected for testing after having given informed consent, as approved by the Institutional Review Board at the Martinez Veteran's Administration Medical Center. All subjects were normal according to a detailed neurological examination. Surface E M G recordings, with 9 mm diameter disk electrodes 3 cm apart, were obtained from 3 muscles in the left leg (tibialis anterior (TA), medial gastrocnemius (MG), and vastus lateralis (VL)). The subjects stood with their feet 10 cm apart on a movable forceplate platform (NeuroCom) and were told to fixatc their gaze on a spot at eye level directly in front. Initial and inter-trial vertical stance were monitored by visual inspection of the subject and live oscilloscopic display of T A and M G activity. No pretest information regarding perturbation type, amplitude, or sequence was provided. Stimuli consisted of 3 different protocols, each of 20 trials of forceplate toe-up ramp movements of preselected amplitudes and variable predictability: (a) serial predictable: 4 ° amplitude (stimulus duration = 80 msec); (b) serial predictable: 10° amplitude (stimulus duration = 200 msec); (c) random mixture of 4°/10 ° amplitude, all at a constant velocity of 50°/sec (50 msec to peak). The 3 protocols produced 4 different testing conditions for each subject: 4 ° predictable (4P); 4 ° random (4R); 10° predictable (10P) and 10 ° random (10R). These two perturbation amplitudes were selected after an earlier pilot study had shown that 4 ° appears to be the minimum amplitude that yields consistent postural reflexes, whereas 10 ° is just at the upper tolerance range of stability for most normal subjects. Most subjects had occasional loss of balance during the 10° perturbation, but never during a 4 ° perturbation and all reported initial feelings as if they might fall during the large perturbation. The sequence of the 3 sets was randomly varied between subjects to reduce a potential confounding effect of ordering. The interval between consecutive trials varied from 10 to 20 sec.
D.J. BECKLEY ET At.. Raw E M G data were pre-amplified, bandpass filtered (10 H z - 1 kHz), full-wave rectified and integrated over a time constant of 4.5 msec and stored to computer disk for off-line analysis (RC Electronics). The E M G sweep was triggered by an electrical pulse simultaneous with the onset of platform movement. The total post-stimulus sample period was 500 msec and the data sampling rate was 2 kHz (bin width 500/.,sec). Optimal gain settings for each subject and control for possible crosstalk were obtained by having each subject maximally contract each muscle prior to actual testing. Gains were kept constant for each subject throughout the experiment. Onset latencies for St. and Mr. responses m the medial gastrocnemius and LL responses in the tibialis anterior (LL (TA)) and vastus lateralis (LL (VL)) were measured relative to the falling edge of the trigger pulse, which coincided with the onset of forceplatc movement. The long latency innervation sequence was calculated by subtracting the LL (TA) onset latency from the LL (VL) onset latency. Normal subjects have a distal to proximal innervation pattern that results in a positive LL innervation sequence (Nashner 19771. Onset criteria R~r the various E M G responses were determined by reference windows as follows: 30-60 msec post trigger (SL), 70-120 msec post trigger (ML), and 100-200 msec post lrigger (LL (TA) and LL (VL)) and by voltage thresholds set at twice the pre-stirnulus noise level. Mean amplitudes of the SL response Cover its complete measured duration) and of the ML response and both LL responses (over the first 75 msec) were calculated by computer algorithm. Mean amplitudes of SL, ML and both LL responses were corrected for background activity by subtracting the first 50 msec of E M G activity prior to the onset of platform movement. The resultant value (amplitude score) was then converted to a standardized Z score to permit comparison of E M G responses between subjects over the range of testing conditions (Hansen et al. 1988)i
Statistical analysis Statistical analysis consisted of 2 factor repeated measures A N O V A comparing the E M G response mean amplitude Z scores and latencies for all 10 subjects for the 4 testing conditions (4P, 4R, 10P, and 10R) (SYS"FAT v4.1). This was followed by post hoc multiple comparisons with linear contrast and appropriate Bonferroni adjustments if the univariate F statistic had a significant P value of < 0.05: Univariate A N O V A was chosen over M A N O V A because of the greater reduction in power when using the latter method with small sample sizes (Wilkinson 19901. Greenhouse-Geisser corrections were not applied to the degrees of freedom because the F ratio is not significantly positively biased
355
LL R E F L E X E S A R E M O D I F I E D BY C O G N I T I V E SET
Ch,4 TRIGGER
J"l
I I
I
Ch.2 MG EMG
I I I
Ch.1 TA EMG
.75
II I
o 15
tsL t.L
I
1 I
Ch3. VL EMG
I
I
-150
tLLTA
I
I
I I I I I
0
~
I
I
.35 ~
tLLVL I
150 50 msec/dlv.
I
'
~
I 0 VOLTS
I
I
300
Fig. 1. Average of 20 trials in the 4 ° predictable condition for a normal subject. Vertical arrows indicate onset of the SL and M L responses in the medial gastrocnemius muscle (MG) and the LL responses in the tibialis anterior (TA) and vastus lateralis (VL) muscles. The dotted reference line indicates the onset of forceplate movement relative to which response latencies are measured. Amplifications are x 5000 for channels 1 and 2 and × 2000 for channel 3. Ch = channel.
with only 2 levels of each repeated factor (Keppel 1982).
Results
E M G response latencies
A representative sample of recorded E M G activity is shown in Fig. 1. Onset latencies for all responses, as well as the LL innervation sequence, are within the normal ranges reported by others (Diener et al. 1983; Scholz et al. 1987) (Table I). There were no significant differences for the various response latencies or the LL innervation sequence when comparing the 4 testing conditions (4P, 4R, 10P, and 10R) with 2 factorial repeated measures ANOVA. The number of missing responses included 13% for the SL response, 4% for the ML response and none for the LL responses. E M G response mean amplitudes
We found that the mean amplitude of the LL (TA) response varied directly with perturbation size in the
predictable test conditions (10P > 4P; P < 0.05). A more notable finding was that the size of the LL (TA) response for the 4R condition was considerably increased compared to the 4P condition ( P < 0.005). This is shown as a significant interaction effect ( P < 0.05) between perturbation amplitude and cognitive set and is displayed graphically in Fig. 2A. Post hoc multiple comparisons showed that the mean LL (TA) Z scores for the 4P condition were significantly less than the mean LL (TA) Z scores for the 4R, the 10P, and the 10R conditions ( P < 0.005) and that there was no significant difference between the 4R, 10P, or 10R conditions. The LL (TA) response in the 4R condition is upregulated to a size appropriate to anticipate a maximal size perturbation, as in condition 10P. Furthermore, confidence intervals show no overlap between the mean of condition 4P and the means of conditions 4R, 10P and 10R at the 99% confidence limits (Fig. 3). In contrast, neither the SL nor the ML responses displayed significant mean amplitude differences across the 4 testing conditions. The LL (VL) response revealed a significant main effect for perturbation ampli-
TABLE I Mean latencies of short (SL), medium (ML) and long latency (LL) responses for all subjects (n = 10) for each test condition (4P, 4 ° predictable amplitude; 4R, 4 ° random amplitude; 10P, I0 o predictable amplitude; 10R, 10 ° random amplitude). Data are expressed as mean_+ S.D. Mean latencies (msec)
SL ML LL (TA) LL (VL) LL IS
Test conditions 4P
4R
10P
10R
42.9_+ 3.3 86.6_+ 6.4 128.6_+ 10.1 144.6 + 13.4 16.0-t- 13.7
42.2_+ 3.0 84.4_+ 3.0 127.3 -+ 11.3 143.0 _+ 10.8 16.0_+ 12.6
42.3_+ 4.4 86.4_+ 4.4 127.2-+ 10.4 146.6 -+ 11.0 20.0-+ 8.7
42.7_+ 3.6 87.2_+ 4.3 128.1 + 10.7 147.3 -+ 10.4 19.4_+ 11.2
LL (TA), tibialis anterior long latency response; LL (VL), vastus tateralis long latency response; LL IS, long latency innervation sequence [LL (VL) - LL (TA)].
356
D.J. B E C K L b Y E T AL.
ing role in this paradigm, and possibly also in daily life (Allure 1983; Diener et al. 1984a,b). Response latencies and the distal-proximal innervation sequence were not influenced by either perturbation amplitude or cognitive set, confirming previous o b ~ r v a t i o n s (Diener et al. 1984b; Horak et at. 19891• This is in keeping with the suggestion that latencies arc programmed from minimal sensory information and separately from response amplitudes (Diener et al. 1988). Reversal of the distal-proximal innervation sequence or co-contraction of distal agonist-antagonist muscles has been reported in normals perturbed while standing on narrowed support surface configurations (Horak and Nashncr 19861 or while pre-leaning backward during toe-up tilts (Diener et al. 1983). Unlike those paradigms, it is probable that even our 10° rotational perturbations did not create the degree of postural instability that would require our subjects to switch from an "ankle strategy" to a "hip strategy" or "cocontraction strategy/" The lack of a default LL (VL) response was unexpected since it is in contrast to what is predicted by the "postural synergy hypothesis" (Nashner and McCollum 1985), The lack of influence of cognitive set on proxi-real muscle synergists is consistent with the theory that set-dependent postural regulation is first directed toward distal stabilizing muscles (Horak et al. 19891. Unlike other studies which used a translational paradigm (Diener et al. 1988; Horak et al. 1989), we did find a main effect in the proximal muscle response for stimulus amplitude, possibly because of the more destabilizing nature of our rotational paradigm. Indeed, responses to rotational perturbations differ in many ways from translational perturbations (Nashner 1976; Diener et al. 1983: Nardone et al. 1990). In
0 r.~
0.6
N ~
~
+
10R
10P
o.o
-o,4
10° ¢'4P
-o,s ~.I
TOE-UP FORCEPLAI~ i
, i
AMPLITUDE
TEST CONDITIONS 0 0
(,'3 N
o.s
P~.
o,~ o.o
~ ~
-o.~
m
0
-0.¢
r,z,.1
:.5
"F
-0,6
FORCEPLATE AMPLITUDE SERt~
P,A N ~ M
TEST CONDITIONS Fig. 2. A: c o m p a r i s o n of the g r a n d m e a n a n d s t a n d a r d e r r o r of the m e a n for tibialis a n t e r i o r (TA) LL Z scores for all subjects for each of the 2 levels of b o t h i n d e p e n d e n t v a r i a b l e s - - f o r c e p l a t e a m p l i t u d e (4 ° and 10°) and d e g r e e of predictability ( " S e r i a l " or "Random*'). T h e r e is a significant i n t e r a c t i o n effect ( F (1, 9 ) = 10.36, P < 0.05) for the 4 ° r a n d o m condition (4R), i n d i c a t i n g that this r e s p o n s e is strongly i n f l u e n c e d by the d e g r e e of p e r t u r b a t i o n predictability. The difference in m e a n a m p l i t u d e b e t w e e n the p r e d i c t a b l e and unpredictable 10° p e r t u r b a t i o n is not significant. B: c o m p a r i s o n of the g r a n d m e a n and s t a n d a r d e r r o r of the m e a n for vastus lateralis (VL) LL Z scores for all subjects for each of the 2 levels of both i n d e p e n d e n t v a r i a b l e s - - f o r c e p l a t e a m p l i t u d e (4 ° and 10°) and d e g r e e of predictability ( " S e r i a l " or " R a n d o m " ) . T h e r e is a significant main effect for a m p l i t u d e only ( F (1, 9 ) = 13.90, P < 0.01) bu! no interaction effect ( F (1, 9 1 = 1 . 2 4 , P < 0 . 3 0 1 . Thus, while the g r a p h suggests a t r e n d for condition 4 R to be larger t h a n 4P. t h e r e is no clear e v i d e n c e t h a t cognitive set modifies the L L r e s p o n s e in the proximal vastus lateralis muscle.
tude only (10P > 4P. P < 0.01) and no significant interaction effect ( P < 0.30) (Fig. 2B).
Discussion
CONFIDENCE INTERVALS ABOUT THE
uJ O O if) N u.l ¢3
E
5r~
