- Email: [email protected]
Electroencephalographic spectral power and lateralized motor activities
Electroencephalographv and clinical NeurophvvioloRv, 1985, 6 0 : 2 2 8 - 2 3 6
228
Elsevier Scientific Publishers Ireland, Lid.
Experimental Section E L E C T R O E N C E P H A L O G R A P H I C S P E C T R A L P O W E R AND L A T E R A L I Z E D M O T O R ACTIVITIES A. A U T R E T , L. AUVERT. F. L A F F O N T and P. L A R M A N D E
Clinique Neurologique, C.H.U. Bretonneau, 37000 Tours (France) (Accepted for publication: September 18, 1984)
The numerous studies bearing on the variations in spectral power of the EEG during tasks implying activity of the cerebral hemispheres implicitly assume that to any given task corresponds a particular cerebral electrical state which is localised and can be recorded on the scalp. The methodological precautions needed in the execution of the tasks, the recording conditions, the derivations chosen, the parameters studied and the statistical analysis were the subject of a general review by Gevins and Schaffer (1980). The results in the literature concerning lateralised modifications of the a rhythm by the hemispheric activities are, however, contradictory. Several teams have failed to bring out any diminution of the a rhythm in relation to the regions concerned (Giannitrapani 1971; Dolce and Waldeier 1974; Gevins et al. 1979a, b, c; Grabow et al. 1979; Rugg and Dickens 1982). On the other hand, other teams have shown an asymmetry of the o~ rhythm (Doyle et al. 1974; Beaumont et al. 1978; Furst 1978; Rebert and Low 1978; Ornstein et al. 1980). From here a controversy has developed (Davidson and Ehrlichman 1980) in which Gevins et al. (1980) have suggested that the sensorimotor components of the tasks carried out could considerably influence the asymmetries observed. We have tried to show more clearly to what extent lateral movements of the hand or the eyes modify the spectral power in the symmetrical regions of the brain. Material and Method
Sixteen males between 20 and 30 years age, all right-handed, were the subjects of this study. Dur-
ing recording each sat in a comfortable armchair, in a half-darkened, partially sound-proof room. Sixteen recording sequences were carried out, each lasting 3 rain. Orders were communicated verbally between sequences. The eye movements were checked by electro-oculogram. Rest sequences alternated with motor sequences. Four sequences were carried out at rest with eyes closed (EC) and 4 with eyes open (EO). Each type of hand motor sequence was carried out EC, first to the right, then to the left: the tonic movements (TM) consisted in pressing continuously against a dynamometer, the phasic movements (PM) in bending and straightening the fingers at a rate of approximately 0 . 5 - 1 / s e c , and the sequential movements (SM) in striking the arm of the chair alternately with the palm, the side of the hand and the fist. The oculomotor sequences consisted in gazing laterally (LG) with EO to the right then to the left. The 16 sequences were set up following a latin square design to eliminate possible effects related to the rank of execution. Eight symmetrical bipolar EEG derivations were recorded simultaneously. The interelectrode distance was strictly 6 cm: derivation 1 rolando-frontal. between C3 and an anterior point towards the nasion: derivation 2 rolando-parietal between C3 and a point lying towards O1; derivation 3 occipito-parietal, between O1 and a point in the direction of C3; and derivation 4 temporal, between T3 and a point lying towards the external canthus ( - 3 dB bandwidth: 0.5 Hz 2.5 kHz). Sampling was carried out in real time at a frequency of 250 Hz by an A / D converter coupled to a Minc 11/03 computer and stored on disc. Before being sampled the signals were filtered by anti-aliasing filters whose cut-off frequency was
0013-4649/85/$03.30 ~'9 1985 Elsevier Scientific Publishers Ireland, Ltd.
EEG POWER A N D MOTOR ACTIVITIES
set to 30 Hz with an attenuation of 24 dB/octave. Processing was carried out later. Successive elementary spectra of 2.56 sec were calculated, each period of analysis being filtered by temporal multiplication through a Hanning window. An intermediate spectrum was furnished by the average of 4 elementary spectra. Intermediate spectra whose intensity was superior to 200 ~tV2/Hz in the 0-1 Hz band were rejected in order to avoid artefacts. For each sequence a maximum of 12 intermediate spectra served to establish a definitive spectrum. The frequency bands retained were 0 : 3 . 5 7 Hz; a: 7.4 12.3 Hz; fl~: 12.9 14.8 Hz.
Statistical analysis (1) The interhemispheric differences were studied first. At rest for each condition 4 sequences were available. An analysis of variance (ANOVA) (Winer 1962) of intermediate spectra in 3 dimensions (subject-sequence-hemisphere) enabled us to clarify in which derivations (1, 2, 3, 4) a significant difference between hemispheres appeared. For each motor activity only one sequence was available; an A N O V A in 2 dimensions (subject-hemisphere) permitted us to specify the derivations (1, 2, 3, 4) in which a significant difference between hemispheres appeared. When there was a significant interhemispheric difference, the side on which spectral intensities predominated was given by a simple comparison of the means of the definitive spectra. (2) Next, for each derivation, the power at the time of right motor activity was compared to that at the time of left motor activity. To appreciate also the modification in intensities between the corresponding rest condition (EC for the manual sequences and EO for the oculomotor sequences) and those of the motor activities, an A N O V A in 2 dimensions (subject-sequence) was used which included 4 sequences: right motor, left motor and the 2 adjacent corresponding rest sequences (EC for the manual sequences and EO for the oculomotor sequences). When a significant inter-sequence difference appeared we compared by a t test the definitive spectrum intensities at the time of right with those at the time of left motor activity.
229
Results
(1) Interhemispheric differences (A) Rest sequences. The F values of the comparison between hemispheres extracted from the 3-way A N O V A (subject hemisphere, 4 rest sequences are shown in the two top lines of Table 1). In the EC rest situations a right superiority ( P < 0.05) of the spectral intensities was found in 2 derivations (1 and 3) for O, in 3 derivations (1, 2 and 4) for a and in 2 derivations (1 and 3) for/~1. In the EO situations it is remarkable that the same superiorities to the right were observed. Furthermore, right superiorities appeared for 0 in derivations (2 and 4) and for B1 in derivation 4. In this study the sequence-hemisphere interactions were only significant at P < 0.05 in derivation 4 for the B~ rhythm in the EC rest condition and in derivations 1 and 4 for B1 rhythm in the EO rest condition. (B) To gain a better idea of this right EEG spectral dominance at rest we performed a special stud)' for the a rhythm EC. Analysis of each interhemispheric difference in the a spectrum in the EC rest condition showed for the 240 differences examined (15 subjects, 4 derivations, 4 sequences) a significant predominance towards the right in 98 cases and towards the left in 33 cases. There was no significantly different distribution of right and left predominances among the derivations (X ~ test, P < 0.05). Each individual could have a maximum of 16 significant differences (4 derivations, 4 sequences) ( P < 0.05) towards the right or towards the left. For our population the median of the significant differences in favour of the right was 6.5 and that of the significant differences in favour of the left was 1. (C) Motor sequences. The F values of the comparison of intermediate spectra between hemispheres extracted from the 2-way A N O V A (subject-hemisphere) are shown in the lower lines of Table I. For each motor sequence, the location of the significant interhemispheric differences are shown in Fig. 1. Considering that for the 4 motor sequences to the same side and for the'. 3 rhythms studied (0, a, ill) a maximum of 48 possible
230
A. A U T R E T
I
I
I
¢'1
*
I
I
Z
.
.
7d r--
.
,--:
.
.
o
.
~0
¢q
Z
~*
-Z~R**
•
.
G*
~.
~c~Z
o;.
,-d Z ¢,i Z
~*
~*
¢5 Z
,~
~.
.
~.
,'-q Z ,~ Z
.
,
.
,~**
~,.
~f*
,'-i Z
g,
F2
"~dZ .< >-
© z .<
~:.
~Z
,-,-: z
rv *
Gr, ""I. * O', * ¢'5 *
E
g
> m
.
.
Z
Z - *
=
N II
e-
z
$
N
~D ",D
.~.
~ , ._~
-
~
..~_~ ._ V
2.
e~
,A Q
.;
C~ V
<
[-.
,,
ET AL
EEG POWER AND MOTOR ACTIVITIES significant differences was o b t a i n a b l e , it is interesting to note that when the m o v e m e n t was to the right 28 a s y m m e t r i e s in favour of the right were o b s e r v e d a n d none in favour of the left, whereas when the m o v e m e n t was to the left only 21 a s y m metries were observed in favour of the right while 4 a s y m m e t r i e s were in favour of the left. Only the left lateral gaze was c a p a b l e of inverting this right electrical d o m i n a n c e . A m o n g the h a n d movements, the left T M s seemed most able to diminish the right electrical d o m i n a n c e while the left PMs seemed to a c c e n t u a t e it. These m o d i f i c a t i o n s concerned in d e s c e n d i n g o r d e r the ,8 l, o~ and 0 r h y t h m s and affected all 4 derivations.
HANDMOVEV~ENTS/~ TONIC
PHASIC
SEQUENTIAL
LATERALEAZE
Fig. 1. Differences between hemispheres (P<0.05) during lateralized motor sequences. Variance analysis (subjects, hemispheres) 0, o~ or ,81 are located for each derivation on the side where they predominate.
231 It should be noted, however, that in this analysis the s u b j e c t - h e m i s p h e r e interaction is non-significant ( P < 0.05) in 29 out of the 96 cases.
(I1) Variation in spectral intensities J~r each deri~,ation according to the sequences of rest and of right and [eft motor actitriO' The results of the A N O V A of i n t e r m e d i a t e spectra in two d i m e n s i o n s (subject, sequence) are shown in T a b l e II. Eighty-two significant ( P < 0.05) intersequence differences were observed on analysis of the variance. F o r these, c o m p a r i s o n of the spectral intensities at the time of right activity with lhose at the time of left activity led to the o b s e r v a t i o n of 32 significant differences P < 0.05 (t test),. The latter m a y be g r o u p e d into two types of variation. Eighteen concern symmetrical derivations and go in the same direction: i.e., when the spectral intensities d u r i n g right tonic m o v e m e n t s are comp a r e d with those during left tonic m o v e m e n t s the intensity was greater on both sides for 0 in derivation 3, for c~ in d e r i v a t i o n s 2, 3 a n d 4 and for ,81 in d e r i v a t i o n 3: when the right lateral gaze was comp a r e d with the left, the power was less for 0 in d e r i v a t i o n 3. for c~ in derivations 2 a n d 3 and for /31 in derivation 3. T h e other differences were a s y m m e t r i c a l : their locations are shown in Fig. 2. All these 14 differences c o r r e s p o n d to a d i m i n u t i o n in p o w e r during the c o n t r a l a t e r a l m o v e m e n t in c o m p a r i s o n with the ipsilateral m o v e m e n t or, which comes to the s a m e thing, to an increase during ipsilateral movements. Thus in Fig. 2 these differences are marked ' +" on the right hemisphere, meaning that the power is greater at this level when the m o v e m e n t s are executed to the right, that is to say, during ipsilateral movements. T h e y are m a r k e d " - ' on the left hemisphere, m e a n i n g that the power is less at this level when the m o v e m e n t is executed to the right (or, which comes to the same thing, greater on the left h e m i s p h e r e when the m o v e m e n t is executed to the left, i.e., d u r i n g ipsilateral movements). Further, it is n o t e w o r t h y (1) that the 0, ~x a n d / 3 j r h y t h m s are equally affected; (2) that phasic m o v e m e n t s induce only a weak effect and that lateral gaze is the most powerful.
232
A. A U T R E T ET A L
.
~8
e.,
.-
r-
©
.
Z
cj
~
-
*
.
.
Kw
~ z
I
Z
Z
Z
*
~E
>~ © z~
,-: Z
e.i Z
~t *
.-: Z
,,
• ~
d
Z
,o
....~
>
~
,4*
© ;,--
~
.
~.
c; ~
- ~n
;¢
.~~ r-...
•.2.
~
V
._
.~_.
,...
t-
e-
~
m
g
g
~o
~ =
8 ,j e-,
~
.-8
z.
EEG POWER A N D MOTOR ACTIVITIES
233
.
.
.
I
Z
.
I
Z
.
.
.
I
Z
*
i
.
",~ -~
¢',1 ~
~
Z
¢'h
~
¢,,I -*
.
I
.
i
¢',1 -*
e,',
-*
I
Z
.*
.
.
~ Z
*
.
/ z
- *
.
~* .
.
I
* ra
Z
.
I
Z
Z H
0
V
o. V e.,..,
~ o [..-
.~ e~
g ~.~
,:I
v
234
A. A,L T R E T El" AI.,
O, 12 times out of 27 for a and only 7 times out of
HAND MOVEMENTS
29 for ill. Conversely, the intensity, during motor sequences is greater than at rest 8 times for fi~, twice for ~ and not at all for 0.
Discussion
Fig. 2. Comparison of power intensities ( P < 0.05) between right and left motor sequences on each lead. Asymmetrical variations. " + ' means spectral intensity during right motor sequences > spectral intensity during left motor sequences. " - " means spectral intensity during right motor sequences < spectral intensity during left motor sequences.
Table III shows, for the 82 significant interfrequency differences, the various positions of the intensities of the two rest sequences in comparison with those of the motor ipsi- and contralateral sequences. Thus at rest the power is greater than that of the motor sequences 25 times out of 26 for T A B L E II1 Classification of power intensities of motor (M) ipsi- (i) and contra- (c) lateral sequences and corresponding rest (Ri,2) sequences. Results are drawn from the 82 significant ( P < 0.05) inter-sequence differences. Rhythm
0 B~
Classifications M,,,, < RI, 2
M c < R1. 2 RI, 2 < Mix < M,
Other classification
25 12 7
0 9 6
1 4 8
0 2 8
The methodological precautions suggested by Gevins and Schaffer (1980) have been followed: comparison of identical but symmetrical sensorimotor tasks, verification of their execution, results presented separately by derivation and non-consideration of the ($ and fie rhythms, which could be perturbed by artefacts of movement or of the myogram. However, like previous authors dealing with A N O V A and EEG spectral analysis, we made the assumption of a gaussian distribution of the intensities during each of our experimental conditions. The choice of an c~ risk of 5% in the ANOVA should lead in the null hypothesis and (supposing that all the variables were independent) to a maximum number of significant differences due to chance, respectively: of 1 in each of the A N O V A in 3 dimensions (12 interhemisphere comparisons at rest EC and EO), 5 in the A N O V A in 2 dimensions (96 comparisons between hemispheres during motor sequences of both sides) and 5 in the A N O V A in 2 dimensions (96 comparisons between right an left motor activities). Our results show a much greater number of significant differences than predicted as due to chance and their coherence must be emphasized. The parameter studied was the spectral power itself. In the statistical analyses the inter-subject variance was important and the interactions were frequently significant. This could perhaps have been reduced by choosing only 'small a ' subjects (Elevenon et al. 1982). It is not known whether the use of certain functions of the spectral power is of interest, for the use of the logarithm of the power by Giannitrapani (1971) and of the normalized power by Rugg and Dickens (1982) did not allow detection of clearly significant asymmetries. The use of bipolar derivations allowed us to present the variation of the EEG, per se, but not to speculate about the electrical activity located
t:I~!G POWER A N D MOTOR ACTIVITIES
beneath either electrode. Furthermore, changes in power spectra may be due to phase changes at either electrode. Our work shows, in right-handed men, a right spectral dominance at rest EO and EC which, however, does not appear in all derivations for all the rhythms under study. Our results agree with those of Grabow et al. (1979) who found in righthanded women a right parietal c~ dominance, and differ from those of Rugg and Dickens (1982) and from those of Giannitrapani (1971). Neither the former, at the parietal level, nor the latter, at the central level, showed any asymmetry of the a rhythm. It will be noted from our results that this right spectral dominance is not a very powerful biological phenomenon since in comparing the values of each spectrum we find 3 right dominances for 1 left dominance and that for a maximum possible of 16 asymmetries per subject the median is situated at 6.5 asymmetries for the right and 1 for the left. Further, none of the 3 rhythms studied, nor any of the 4 derivations, seems more lateralized than the others. The relation between the 0, a and /3~ rhythms and contralateral motor activity was studied in two ways. First, we noted during the right movements a greater number of right spectral dominances than during the left movements, certain of the latter inducing a left spectral dominance. Everything occurred as though the right movements increased and the left movements reduced or inverted the rest asymmetry and as though in this respect the 0, a and//~ rhythms played an equivalent role. Secondly, all the significant asymmetrical modifications observed at each derivation during the lateralized motor activities revealed a spectral diminution during the contralateral movement, i.e., also an increase during the ipsilateral movement. In our opinion, a relationship is thereby clearly demonstrated between motor activities depending principally on one hemisphere and a diminution in the corresponding 0, a and fl~ spectral power. This corroborates the remarks of Gevins et al. (1980) on the importance of sensorimotor events in the creation of EEG asymmetries. The powerful activation effect of the lateralization of the gaze noted during psychological protocols (Kinsbourne 1974) finds here an EEG concomitant.
235
One observes a spectral power at rest greater than that of the motor sequences, almost constant fort O, occurring in nearly half of the cases for and more irregular for /?l- This is the quantified measure of the EEG activation observed visually during motor activities. Thus, this quantitative EEG work on righthanded subjects in the 0, c, and fll spectral bands shows a clear right spectral dominance at rest and confirms the unproven intuitive notions of a diminution in the 0, ~ and also fi~ intensities during contralateral movements (and in particular during lateral gaze).
Summa~ The powers of the 0 (3.5 7 Hz), c~ (7.4 -12.3 Hz) and /?l (12.9-14.8 Hz) rhythms were analysed in symmetrical derivations. Sixteen right-handed young male adults were observed: at rest, with eyes closed (EC) (4 sequences) and eyes open (EO) (4 sequences); during right then left tonic alternate or sequential movements of the hand and while gazing to the right then to the left. At rest there exists a clear and significant spectral dominance to the right which does not appear in all the rhythms or in all the derivations. As far as the c~ rhythm EC is concerned, individual analysis of the sequences shows that of those which are significantly lateralized, three-quarters are lateralized towards the right. Right motor activity exaggerates, left motor activity' diminishes, this right electrical dominance. Lateral gaze is the most powerful activity in this respect. At each derivation, comparison of intensities during right and left activities shows that the contralateral movements diminish the spectral power in the 3 bands studied. This effect seems to be obtained most frequently at the centro-parietal level. The spectral power at rest compared with that during motor activity is higher constantly for 0 rhythm, in almost half the cases for the c~ and occasionally for the ill.
236
Resum6 Puissance spectrale de l'blectroencbphalogramme et activitb motrice latbralis~e La p u i s s a n c e des r y t h m e s thEta ( 3 , 5 - 7 Hz), a l p h a ( 7 , 4 - 1 2 , 3 Hz) et beta~ ( 1 2 , 9 - 1 4 , 8 Hz) fur a n a l y s 6 e au n i v e a u de 4 d d r i v a t i o n s s y m d t r i q u e s c h e z 16 h o m m e s j e u n e s , droitiers. Seize s e q u e n c e s o n t 6t6 enregistr6es c h e z c h a q u e sujet: r e p o s y e u x ferm6s (4 s e q u e n c e s ) et y e u x o u v e r t s (4 s e q u e n c e s ) , d u r a n t des m o u v e m e n t s d ' u n e m a i n les y e u x ferm6s s u c c e s s i v e m e n t t o n i q u e s , altern6s, p u i s s d q u e n t i e l s et lors du m a i n t i e n du r e g a r d lat6ral les yeux o u v e r t s . Ces 4 s 6 q u e n c e s m o r t i c e s 6taient rdalis6es d ' a b o r d 'a d r o i t e puis/~ gauche. A u r e p o s il existe une d o m i n a n c e s p e c t r a l e sign i f i c a t i v e m e n t p l u s i m p o r t a n t e vers la d r o i t e qui n ' a p p a r a i t c e p e n d a n t pas p o u r t o u s l e s r y t h m e s , ni d a n s toutes les s6quences. En ce qui c o n c e r n e le r y t h m e a l p h a au r e p o s y e u x fermes, l ' a n a l y s e de c h a q u e s 6 q u e n c e m o n t r e q u e 3 / 4 des s 6 q u e n c e s s i g n i f i c a t i v e m e n t latdralis6es le s o n t vers la droite. L ' a c t i v i t 6 m o t r i c e d r o i t e a u g m e n t e et l ' a c t i v i t e gauche diminue cette dominance 61ectrique c6r6brale droite. Le r e g a r d lat6ral vers la d r o i t e est la p l u s p u i s s a n t e activit6 h cet 6gard. A u n i v e a u de c h a q u e d 6 r i v a t i o n , la c o m p a r a i son des p u i s s a n c e s s p e c t r a l e s d u r a n t les activit6s d r o i t e s et g a u c h e s m o n t r e q u e le m o u v e m e n t c o n tro-lat6ral d i m i n u e l'activit6 s p e c t r a l e d a n s les 3 b a n d e s 6tudi6es. C e t effet s e m b l e 6tre o b t e n u le plus frdquemment au n i v e a u c e n t r o - p a r i d t a l . L ' a c t i v i t 6 s p e c t r a l e au r e p o s c o m p a r 6 e ~. celle des activitEs m o t r i c e s est p l u s g r a n d e c o n s t a m m e n t p o u r le r y t h m e theta, d a n s la m o i t i 6 des cas p o u r le r y t h m e a l p h a et s e u l e m e n t o c c a s i o n n e l l e m e n t p o u r le r y t h m e b e t a t.
References Beaumont, J.G., Mayes, A.R. and Rugg, M.D. Asymmetry in
EEG alpha coherence and power: effect of task and sex. Electroenceph. clin. Neurophysiol., 1978, 45: 393-401. Davidson, R.J. and Ehrlichman, H. Lateralized cognitive processes and the electroencephalogram. Science, 1980, 207: 1005-1006.
A. AUTRET ET AL. Dolce, G. and Waldeier, H. Spectral and multivariate analysis of EEG changes during mental activity in man. Electroenceph. clin. Neurophysiol., 1974, 36:577 584. Doyle, J.C., Ornstein, R. and Galin, D. Lateral specialization of cognitive mode. lI. EEG frequency analysis. Psychophysiology, 1974, 11:567 578. Etevenon, P., Samson-Dollfus, D., Kemali, D. and Perris, (7. In quest of the best spectral EEG parameters towards a multicentric consensus for measuring normality, psychopathology and psychotropic drug effects. Adv. Biol. Psychiat., 1982, 9:28 38. Furst, C.J. EEG alpha asymmetry and visuospatial performance. Nature (Lond.), 1978, 260:254 255. Gevins, A.S. and Schaffer, R.E. A critical review of electroencephalographic (EEG) correlates of higher cortical functions. C.R.C. Crit. Rev. Bioengng, 1980, 4: 113-164. Gevins, A.S., Zeitlin, G.M., Doyle, J.C., Yingling, C.D., Schaffer, R.E., Callaway, E. and Yeager, C.L Electroencephalogram correlates of higher cortical functions. Science, 1979a, 203: 665-668. Gevins, A.S., Zeitlin, G.M., Yingling, C.D., Doyle, J.C., Dedon, M.F., Schaffer, R.E., Roumasset, J.T. and Yeager, C.L. EEG patterns during 'cognitive' tasks. I. Methodology and analysis of complex behaviors. Electroenceph. olin. Neurophysiol., 1979b, 47:693 703. Gevins, A.S., Zeitlin, G.M., Doyle, J.C., Schaffer. R.E. and Callaway, E. EEG patterns during ~cognitive" tasks. 11. Analysis of controlled tasks. Electroenceph. din. Neurophysiol., 1979c, 47: 704-710. Gevins. A.S., Doyle, J.C., Schaffer, R.E.. Callaway, E. and Yeager, C. Lateralized cognitive processes and the electroencephalogram. Science. 1980, 207:1006 1007. Giannitrapani, D. Scanning mechanism and the EEG. Electroenceph, clin. Neurophysiol., 1971, 30:139 146. Grabow, J.D., Aronson, A.E., Greene, K.L. and Offord, K.P. A comparison of EEG activity in the left and right cerebral hemispheres by power-spectrum analysis during language and non-language tasks. Electroenceph. clm. Neurophysiol., 1979, 47: 460-472. Kinsbourne, M. Direction of gaze and distribution of cerebral thought processes. Neuropsychologia, 1974, 12:279 281. Ornstein, R., Johnstone, J., Herron, J. and Swencionis, C. Differential right hemisphere engagement in visuospatial tasks. Neuropsychologia, 1980, 18: 49-64. Rebert, C.S. and Low, D.W. Differential hemispheric activation during complex visuomotor performance. Electroenceph. clin. Neurophysiol., 1978, 44: 724-734. Rugg, M.D. and Dickens, A.M.J. Dissociation of alpha and theta activity as a function of verbal and visuospatial tasks. Electroenceph. clin. Neurophysiol., 1982, 53: 201-207. Winer, B.J. (Ed.). Factorial Experiments -- Computational Procedures and Numerical Examples. Statistical Principles in Experimental Design. McGraw-Hill, New York, 1962: 431-505.
228
Elsevier Scientific Publishers Ireland, Lid.
Experimental Section E L E C T R O E N C E P H A L O G R A P H I C S P E C T R A L P O W E R AND L A T E R A L I Z E D M O T O R ACTIVITIES A. A U T R E T , L. AUVERT. F. L A F F O N T and P. L A R M A N D E
Clinique Neurologique, C.H.U. Bretonneau, 37000 Tours (France) (Accepted for publication: September 18, 1984)
The numerous studies bearing on the variations in spectral power of the EEG during tasks implying activity of the cerebral hemispheres implicitly assume that to any given task corresponds a particular cerebral electrical state which is localised and can be recorded on the scalp. The methodological precautions needed in the execution of the tasks, the recording conditions, the derivations chosen, the parameters studied and the statistical analysis were the subject of a general review by Gevins and Schaffer (1980). The results in the literature concerning lateralised modifications of the a rhythm by the hemispheric activities are, however, contradictory. Several teams have failed to bring out any diminution of the a rhythm in relation to the regions concerned (Giannitrapani 1971; Dolce and Waldeier 1974; Gevins et al. 1979a, b, c; Grabow et al. 1979; Rugg and Dickens 1982). On the other hand, other teams have shown an asymmetry of the o~ rhythm (Doyle et al. 1974; Beaumont et al. 1978; Furst 1978; Rebert and Low 1978; Ornstein et al. 1980). From here a controversy has developed (Davidson and Ehrlichman 1980) in which Gevins et al. (1980) have suggested that the sensorimotor components of the tasks carried out could considerably influence the asymmetries observed. We have tried to show more clearly to what extent lateral movements of the hand or the eyes modify the spectral power in the symmetrical regions of the brain. Material and Method
Sixteen males between 20 and 30 years age, all right-handed, were the subjects of this study. Dur-
ing recording each sat in a comfortable armchair, in a half-darkened, partially sound-proof room. Sixteen recording sequences were carried out, each lasting 3 rain. Orders were communicated verbally between sequences. The eye movements were checked by electro-oculogram. Rest sequences alternated with motor sequences. Four sequences were carried out at rest with eyes closed (EC) and 4 with eyes open (EO). Each type of hand motor sequence was carried out EC, first to the right, then to the left: the tonic movements (TM) consisted in pressing continuously against a dynamometer, the phasic movements (PM) in bending and straightening the fingers at a rate of approximately 0 . 5 - 1 / s e c , and the sequential movements (SM) in striking the arm of the chair alternately with the palm, the side of the hand and the fist. The oculomotor sequences consisted in gazing laterally (LG) with EO to the right then to the left. The 16 sequences were set up following a latin square design to eliminate possible effects related to the rank of execution. Eight symmetrical bipolar EEG derivations were recorded simultaneously. The interelectrode distance was strictly 6 cm: derivation 1 rolando-frontal. between C3 and an anterior point towards the nasion: derivation 2 rolando-parietal between C3 and a point lying towards O1; derivation 3 occipito-parietal, between O1 and a point in the direction of C3; and derivation 4 temporal, between T3 and a point lying towards the external canthus ( - 3 dB bandwidth: 0.5 Hz 2.5 kHz). Sampling was carried out in real time at a frequency of 250 Hz by an A / D converter coupled to a Minc 11/03 computer and stored on disc. Before being sampled the signals were filtered by anti-aliasing filters whose cut-off frequency was
0013-4649/85/$03.30 ~'9 1985 Elsevier Scientific Publishers Ireland, Ltd.
EEG POWER A N D MOTOR ACTIVITIES
set to 30 Hz with an attenuation of 24 dB/octave. Processing was carried out later. Successive elementary spectra of 2.56 sec were calculated, each period of analysis being filtered by temporal multiplication through a Hanning window. An intermediate spectrum was furnished by the average of 4 elementary spectra. Intermediate spectra whose intensity was superior to 200 ~tV2/Hz in the 0-1 Hz band were rejected in order to avoid artefacts. For each sequence a maximum of 12 intermediate spectra served to establish a definitive spectrum. The frequency bands retained were 0 : 3 . 5 7 Hz; a: 7.4 12.3 Hz; fl~: 12.9 14.8 Hz.
Statistical analysis (1) The interhemispheric differences were studied first. At rest for each condition 4 sequences were available. An analysis of variance (ANOVA) (Winer 1962) of intermediate spectra in 3 dimensions (subject-sequence-hemisphere) enabled us to clarify in which derivations (1, 2, 3, 4) a significant difference between hemispheres appeared. For each motor activity only one sequence was available; an A N O V A in 2 dimensions (subject-hemisphere) permitted us to specify the derivations (1, 2, 3, 4) in which a significant difference between hemispheres appeared. When there was a significant interhemispheric difference, the side on which spectral intensities predominated was given by a simple comparison of the means of the definitive spectra. (2) Next, for each derivation, the power at the time of right motor activity was compared to that at the time of left motor activity. To appreciate also the modification in intensities between the corresponding rest condition (EC for the manual sequences and EO for the oculomotor sequences) and those of the motor activities, an A N O V A in 2 dimensions (subject-sequence) was used which included 4 sequences: right motor, left motor and the 2 adjacent corresponding rest sequences (EC for the manual sequences and EO for the oculomotor sequences). When a significant inter-sequence difference appeared we compared by a t test the definitive spectrum intensities at the time of right with those at the time of left motor activity.
229
Results
(1) Interhemispheric differences (A) Rest sequences. The F values of the comparison between hemispheres extracted from the 3-way A N O V A (subject hemisphere, 4 rest sequences are shown in the two top lines of Table 1). In the EC rest situations a right superiority ( P < 0.05) of the spectral intensities was found in 2 derivations (1 and 3) for O, in 3 derivations (1, 2 and 4) for a and in 2 derivations (1 and 3) for/~1. In the EO situations it is remarkable that the same superiorities to the right were observed. Furthermore, right superiorities appeared for 0 in derivations (2 and 4) and for B1 in derivation 4. In this study the sequence-hemisphere interactions were only significant at P < 0.05 in derivation 4 for the B~ rhythm in the EC rest condition and in derivations 1 and 4 for B1 rhythm in the EO rest condition. (B) To gain a better idea of this right EEG spectral dominance at rest we performed a special stud)' for the a rhythm EC. Analysis of each interhemispheric difference in the a spectrum in the EC rest condition showed for the 240 differences examined (15 subjects, 4 derivations, 4 sequences) a significant predominance towards the right in 98 cases and towards the left in 33 cases. There was no significantly different distribution of right and left predominances among the derivations (X ~ test, P < 0.05). Each individual could have a maximum of 16 significant differences (4 derivations, 4 sequences) ( P < 0.05) towards the right or towards the left. For our population the median of the significant differences in favour of the right was 6.5 and that of the significant differences in favour of the left was 1. (C) Motor sequences. The F values of the comparison of intermediate spectra between hemispheres extracted from the 2-way A N O V A (subject-hemisphere) are shown in the lower lines of Table I. For each motor sequence, the location of the significant interhemispheric differences are shown in Fig. 1. Considering that for the 4 motor sequences to the same side and for the'. 3 rhythms studied (0, a, ill) a maximum of 48 possible
230
A. A U T R E T
I
I
I
¢'1
*
I
I
Z
.
.
7d r--
.
,--:
.
.
o
.
~0
¢q
Z
~*
-Z~R**
•
.
G*
~.
~c~Z
o;.
,-d Z ¢,i Z
~*
~*
¢5 Z
,~
~.
.
~.
,'-q Z ,~ Z
.
,
.
,~**
~,.
~f*
,'-i Z
g,
F2
"~dZ .< >-
© z .<
~:.
~Z
,-,-: z
rv *
Gr, ""I. * O', * ¢'5 *
E
g
> m
.
.
Z
Z - *
=
N II
e-
z
$
N
~D ",D
.~.
~ , ._~
-
~
..~_~ ._ V
2.
e~
,A Q
.;
C~ V
<
[-.
,,
ET AL
EEG POWER AND MOTOR ACTIVITIES significant differences was o b t a i n a b l e , it is interesting to note that when the m o v e m e n t was to the right 28 a s y m m e t r i e s in favour of the right were o b s e r v e d a n d none in favour of the left, whereas when the m o v e m e n t was to the left only 21 a s y m metries were observed in favour of the right while 4 a s y m m e t r i e s were in favour of the left. Only the left lateral gaze was c a p a b l e of inverting this right electrical d o m i n a n c e . A m o n g the h a n d movements, the left T M s seemed most able to diminish the right electrical d o m i n a n c e while the left PMs seemed to a c c e n t u a t e it. These m o d i f i c a t i o n s concerned in d e s c e n d i n g o r d e r the ,8 l, o~ and 0 r h y t h m s and affected all 4 derivations.
HANDMOVEV~ENTS/~ TONIC
PHASIC
SEQUENTIAL
LATERALEAZE
Fig. 1. Differences between hemispheres (P<0.05) during lateralized motor sequences. Variance analysis (subjects, hemispheres) 0, o~ or ,81 are located for each derivation on the side where they predominate.
231 It should be noted, however, that in this analysis the s u b j e c t - h e m i s p h e r e interaction is non-significant ( P < 0.05) in 29 out of the 96 cases.
(I1) Variation in spectral intensities J~r each deri~,ation according to the sequences of rest and of right and [eft motor actitriO' The results of the A N O V A of i n t e r m e d i a t e spectra in two d i m e n s i o n s (subject, sequence) are shown in T a b l e II. Eighty-two significant ( P < 0.05) intersequence differences were observed on analysis of the variance. F o r these, c o m p a r i s o n of the spectral intensities at the time of right activity with lhose at the time of left activity led to the o b s e r v a t i o n of 32 significant differences P < 0.05 (t test),. The latter m a y be g r o u p e d into two types of variation. Eighteen concern symmetrical derivations and go in the same direction: i.e., when the spectral intensities d u r i n g right tonic m o v e m e n t s are comp a r e d with those during left tonic m o v e m e n t s the intensity was greater on both sides for 0 in derivation 3, for c~ in d e r i v a t i o n s 2, 3 a n d 4 and for ,81 in d e r i v a t i o n 3: when the right lateral gaze was comp a r e d with the left, the power was less for 0 in d e r i v a t i o n 3. for c~ in derivations 2 a n d 3 and for /31 in derivation 3. T h e other differences were a s y m m e t r i c a l : their locations are shown in Fig. 2. All these 14 differences c o r r e s p o n d to a d i m i n u t i o n in p o w e r during the c o n t r a l a t e r a l m o v e m e n t in c o m p a r i s o n with the ipsilateral m o v e m e n t or, which comes to the s a m e thing, to an increase during ipsilateral movements. Thus in Fig. 2 these differences are marked ' +" on the right hemisphere, meaning that the power is greater at this level when the m o v e m e n t s are executed to the right, that is to say, during ipsilateral movements. T h e y are m a r k e d " - ' on the left hemisphere, m e a n i n g that the power is less at this level when the m o v e m e n t is executed to the right (or, which comes to the same thing, greater on the left h e m i s p h e r e when the m o v e m e n t is executed to the left, i.e., d u r i n g ipsilateral movements). Further, it is n o t e w o r t h y (1) that the 0, ~x a n d / 3 j r h y t h m s are equally affected; (2) that phasic m o v e m e n t s induce only a weak effect and that lateral gaze is the most powerful.
232
A. A U T R E T ET A L
.
~8
e.,
.-
r-
©
.
Z
cj
~
-
*
.
.
Kw
~ z
I
Z
Z
Z
*
~E
>~ © z~
,-: Z
e.i Z
~t *
.-: Z
,,
• ~
d
Z
,o
....~
>
~
,4*
© ;,--
~
.
~.
c; ~
- ~n
;¢
.~~ r-...
•.2.
~
V
._
.~_.
,...
t-
e-
~
m
g
g
~o
~ =
8 ,j e-,
~
.-8
z.
EEG POWER A N D MOTOR ACTIVITIES
233
.
.
.
I
Z
.
I
Z
.
.
.
I
Z
*
i
.
",~ -~
¢',1 ~
~
Z
¢'h
~
¢,,I -*
.
I
.
i
¢',1 -*
e,',
-*
I
Z
.*
.
.
~ Z
*
.
/ z
- *
.
~* .
.
I
* ra
Z
.
I
Z
Z H
0
V
o. V e.,..,
~ o [..-
.~ e~
g ~.~
,:I
v
234
A. A,L T R E T El" AI.,
O, 12 times out of 27 for a and only 7 times out of
HAND MOVEMENTS
29 for ill. Conversely, the intensity, during motor sequences is greater than at rest 8 times for fi~, twice for ~ and not at all for 0.
Discussion
Fig. 2. Comparison of power intensities ( P < 0.05) between right and left motor sequences on each lead. Asymmetrical variations. " + ' means spectral intensity during right motor sequences > spectral intensity during left motor sequences. " - " means spectral intensity during right motor sequences < spectral intensity during left motor sequences.
Table III shows, for the 82 significant interfrequency differences, the various positions of the intensities of the two rest sequences in comparison with those of the motor ipsi- and contralateral sequences. Thus at rest the power is greater than that of the motor sequences 25 times out of 26 for T A B L E II1 Classification of power intensities of motor (M) ipsi- (i) and contra- (c) lateral sequences and corresponding rest (Ri,2) sequences. Results are drawn from the 82 significant ( P < 0.05) inter-sequence differences. Rhythm
0 B~
Classifications M,,,, < RI, 2
M c < R1. 2 RI, 2 < Mix < M,
Other classification
25 12 7
0 9 6
1 4 8
0 2 8
The methodological precautions suggested by Gevins and Schaffer (1980) have been followed: comparison of identical but symmetrical sensorimotor tasks, verification of their execution, results presented separately by derivation and non-consideration of the ($ and fie rhythms, which could be perturbed by artefacts of movement or of the myogram. However, like previous authors dealing with A N O V A and EEG spectral analysis, we made the assumption of a gaussian distribution of the intensities during each of our experimental conditions. The choice of an c~ risk of 5% in the ANOVA should lead in the null hypothesis and (supposing that all the variables were independent) to a maximum number of significant differences due to chance, respectively: of 1 in each of the A N O V A in 3 dimensions (12 interhemisphere comparisons at rest EC and EO), 5 in the A N O V A in 2 dimensions (96 comparisons between hemispheres during motor sequences of both sides) and 5 in the A N O V A in 2 dimensions (96 comparisons between right an left motor activities). Our results show a much greater number of significant differences than predicted as due to chance and their coherence must be emphasized. The parameter studied was the spectral power itself. In the statistical analyses the inter-subject variance was important and the interactions were frequently significant. This could perhaps have been reduced by choosing only 'small a ' subjects (Elevenon et al. 1982). It is not known whether the use of certain functions of the spectral power is of interest, for the use of the logarithm of the power by Giannitrapani (1971) and of the normalized power by Rugg and Dickens (1982) did not allow detection of clearly significant asymmetries. The use of bipolar derivations allowed us to present the variation of the EEG, per se, but not to speculate about the electrical activity located
t:I~!G POWER A N D MOTOR ACTIVITIES
beneath either electrode. Furthermore, changes in power spectra may be due to phase changes at either electrode. Our work shows, in right-handed men, a right spectral dominance at rest EO and EC which, however, does not appear in all derivations for all the rhythms under study. Our results agree with those of Grabow et al. (1979) who found in righthanded women a right parietal c~ dominance, and differ from those of Rugg and Dickens (1982) and from those of Giannitrapani (1971). Neither the former, at the parietal level, nor the latter, at the central level, showed any asymmetry of the a rhythm. It will be noted from our results that this right spectral dominance is not a very powerful biological phenomenon since in comparing the values of each spectrum we find 3 right dominances for 1 left dominance and that for a maximum possible of 16 asymmetries per subject the median is situated at 6.5 asymmetries for the right and 1 for the left. Further, none of the 3 rhythms studied, nor any of the 4 derivations, seems more lateralized than the others. The relation between the 0, a and /3~ rhythms and contralateral motor activity was studied in two ways. First, we noted during the right movements a greater number of right spectral dominances than during the left movements, certain of the latter inducing a left spectral dominance. Everything occurred as though the right movements increased and the left movements reduced or inverted the rest asymmetry and as though in this respect the 0, a and//~ rhythms played an equivalent role. Secondly, all the significant asymmetrical modifications observed at each derivation during the lateralized motor activities revealed a spectral diminution during the contralateral movement, i.e., also an increase during the ipsilateral movement. In our opinion, a relationship is thereby clearly demonstrated between motor activities depending principally on one hemisphere and a diminution in the corresponding 0, a and fl~ spectral power. This corroborates the remarks of Gevins et al. (1980) on the importance of sensorimotor events in the creation of EEG asymmetries. The powerful activation effect of the lateralization of the gaze noted during psychological protocols (Kinsbourne 1974) finds here an EEG concomitant.
235
One observes a spectral power at rest greater than that of the motor sequences, almost constant fort O, occurring in nearly half of the cases for and more irregular for /?l- This is the quantified measure of the EEG activation observed visually during motor activities. Thus, this quantitative EEG work on righthanded subjects in the 0, c, and fll spectral bands shows a clear right spectral dominance at rest and confirms the unproven intuitive notions of a diminution in the 0, ~ and also fi~ intensities during contralateral movements (and in particular during lateral gaze).
Summa~ The powers of the 0 (3.5 7 Hz), c~ (7.4 -12.3 Hz) and /?l (12.9-14.8 Hz) rhythms were analysed in symmetrical derivations. Sixteen right-handed young male adults were observed: at rest, with eyes closed (EC) (4 sequences) and eyes open (EO) (4 sequences); during right then left tonic alternate or sequential movements of the hand and while gazing to the right then to the left. At rest there exists a clear and significant spectral dominance to the right which does not appear in all the rhythms or in all the derivations. As far as the c~ rhythm EC is concerned, individual analysis of the sequences shows that of those which are significantly lateralized, three-quarters are lateralized towards the right. Right motor activity exaggerates, left motor activity' diminishes, this right electrical dominance. Lateral gaze is the most powerful activity in this respect. At each derivation, comparison of intensities during right and left activities shows that the contralateral movements diminish the spectral power in the 3 bands studied. This effect seems to be obtained most frequently at the centro-parietal level. The spectral power at rest compared with that during motor activity is higher constantly for 0 rhythm, in almost half the cases for the c~ and occasionally for the ill.
236
Resum6 Puissance spectrale de l'blectroencbphalogramme et activitb motrice latbralis~e La p u i s s a n c e des r y t h m e s thEta ( 3 , 5 - 7 Hz), a l p h a ( 7 , 4 - 1 2 , 3 Hz) et beta~ ( 1 2 , 9 - 1 4 , 8 Hz) fur a n a l y s 6 e au n i v e a u de 4 d d r i v a t i o n s s y m d t r i q u e s c h e z 16 h o m m e s j e u n e s , droitiers. Seize s e q u e n c e s o n t 6t6 enregistr6es c h e z c h a q u e sujet: r e p o s y e u x ferm6s (4 s e q u e n c e s ) et y e u x o u v e r t s (4 s e q u e n c e s ) , d u r a n t des m o u v e m e n t s d ' u n e m a i n les y e u x ferm6s s u c c e s s i v e m e n t t o n i q u e s , altern6s, p u i s s d q u e n t i e l s et lors du m a i n t i e n du r e g a r d lat6ral les yeux o u v e r t s . Ces 4 s 6 q u e n c e s m o r t i c e s 6taient rdalis6es d ' a b o r d 'a d r o i t e puis/~ gauche. A u r e p o s il existe une d o m i n a n c e s p e c t r a l e sign i f i c a t i v e m e n t p l u s i m p o r t a n t e vers la d r o i t e qui n ' a p p a r a i t c e p e n d a n t pas p o u r t o u s l e s r y t h m e s , ni d a n s toutes les s6quences. En ce qui c o n c e r n e le r y t h m e a l p h a au r e p o s y e u x fermes, l ' a n a l y s e de c h a q u e s 6 q u e n c e m o n t r e q u e 3 / 4 des s 6 q u e n c e s s i g n i f i c a t i v e m e n t latdralis6es le s o n t vers la droite. L ' a c t i v i t 6 m o t r i c e d r o i t e a u g m e n t e et l ' a c t i v i t e gauche diminue cette dominance 61ectrique c6r6brale droite. Le r e g a r d lat6ral vers la d r o i t e est la p l u s p u i s s a n t e activit6 h cet 6gard. A u n i v e a u de c h a q u e d 6 r i v a t i o n , la c o m p a r a i son des p u i s s a n c e s s p e c t r a l e s d u r a n t les activit6s d r o i t e s et g a u c h e s m o n t r e q u e le m o u v e m e n t c o n tro-lat6ral d i m i n u e l'activit6 s p e c t r a l e d a n s les 3 b a n d e s 6tudi6es. C e t effet s e m b l e 6tre o b t e n u le plus frdquemment au n i v e a u c e n t r o - p a r i d t a l . L ' a c t i v i t 6 s p e c t r a l e au r e p o s c o m p a r 6 e ~. celle des activitEs m o t r i c e s est p l u s g r a n d e c o n s t a m m e n t p o u r le r y t h m e theta, d a n s la m o i t i 6 des cas p o u r le r y t h m e a l p h a et s e u l e m e n t o c c a s i o n n e l l e m e n t p o u r le r y t h m e b e t a t.
References Beaumont, J.G., Mayes, A.R. and Rugg, M.D. Asymmetry in
EEG alpha coherence and power: effect of task and sex. Electroenceph. clin. Neurophysiol., 1978, 45: 393-401. Davidson, R.J. and Ehrlichman, H. Lateralized cognitive processes and the electroencephalogram. Science, 1980, 207: 1005-1006.
A. AUTRET ET AL. Dolce, G. and Waldeier, H. Spectral and multivariate analysis of EEG changes during mental activity in man. Electroenceph. clin. Neurophysiol., 1974, 36:577 584. Doyle, J.C., Ornstein, R. and Galin, D. Lateral specialization of cognitive mode. lI. EEG frequency analysis. Psychophysiology, 1974, 11:567 578. Etevenon, P., Samson-Dollfus, D., Kemali, D. and Perris, (7. In quest of the best spectral EEG parameters towards a multicentric consensus for measuring normality, psychopathology and psychotropic drug effects. Adv. Biol. Psychiat., 1982, 9:28 38. Furst, C.J. EEG alpha asymmetry and visuospatial performance. Nature (Lond.), 1978, 260:254 255. Gevins, A.S. and Schaffer, R.E. A critical review of electroencephalographic (EEG) correlates of higher cortical functions. C.R.C. Crit. Rev. Bioengng, 1980, 4: 113-164. Gevins, A.S., Zeitlin, G.M., Doyle, J.C., Yingling, C.D., Schaffer, R.E., Callaway, E. and Yeager, C.L Electroencephalogram correlates of higher cortical functions. Science, 1979a, 203: 665-668. Gevins, A.S., Zeitlin, G.M., Yingling, C.D., Doyle, J.C., Dedon, M.F., Schaffer, R.E., Roumasset, J.T. and Yeager, C.L. EEG patterns during 'cognitive' tasks. I. Methodology and analysis of complex behaviors. Electroenceph. olin. Neurophysiol., 1979b, 47:693 703. Gevins, A.S., Zeitlin, G.M., Doyle, J.C., Schaffer. R.E. and Callaway, E. EEG patterns during ~cognitive" tasks. 11. Analysis of controlled tasks. Electroenceph. din. Neurophysiol., 1979c, 47: 704-710. Gevins. A.S., Doyle, J.C., Schaffer, R.E.. Callaway, E. and Yeager, C. Lateralized cognitive processes and the electroencephalogram. Science. 1980, 207:1006 1007. Giannitrapani, D. Scanning mechanism and the EEG. Electroenceph, clin. Neurophysiol., 1971, 30:139 146. Grabow, J.D., Aronson, A.E., Greene, K.L. and Offord, K.P. A comparison of EEG activity in the left and right cerebral hemispheres by power-spectrum analysis during language and non-language tasks. Electroenceph. clm. Neurophysiol., 1979, 47: 460-472. Kinsbourne, M. Direction of gaze and distribution of cerebral thought processes. Neuropsychologia, 1974, 12:279 281. Ornstein, R., Johnstone, J., Herron, J. and Swencionis, C. Differential right hemisphere engagement in visuospatial tasks. Neuropsychologia, 1980, 18: 49-64. Rebert, C.S. and Low, D.W. Differential hemispheric activation during complex visuomotor performance. Electroenceph. clin. Neurophysiol., 1978, 44: 724-734. Rugg, M.D. and Dickens, A.M.J. Dissociation of alpha and theta activity as a function of verbal and visuospatial tasks. Electroenceph. clin. Neurophysiol., 1982, 53: 201-207. Winer, B.J. (Ed.). Factorial Experiments -- Computational Procedures and Numerical Examples. Statistical Principles in Experimental Design. McGraw-Hill, New York, 1962: 431-505.