Neuroscience Letters, 16 (1980) 67--70 © El~vier/North-Holland Scientific Publishers Ltd.

6~

~ ~ R R E ~ T E S O F ~ A C Q U I S m O N OF mGH- AND LOW-

M.D.~ RUG(] and P ~ . ~ N ~ D e ~ ~ of P s y e h o ~ , University of St. Andrews, St. Andrews and Departmen~ of Psychology, Univerdty of York, York (United Kingdom)

( ~ d ~ b e r 20th, 1979) (Accepted ~ 14th, 1979)

SUMMARY

EEG was recorde:t from left and right temporal areas while ~abjects learn~ high-~ low~y words. ~etry in power in the alpha band, in the fcsma of less power ~om the left hemisphere, was obser~ ~} only during the

O ~ m • • m," r rmgn ~ wordS. This condition also gave ~ bilaterally to more ~ power. The difference between conditions in levels of inter-hemi- splun~e coherence pl~licted, in both frequency bands, relativ~ levels of recall of the two word types. It is suggested ~hat the alpha and theta bands of the EF_~ may reflect different ~ of information processing ir~ these tasks.

The rationale for ~ study is ~ on the observations that (i) word:: rated high in imageability am recalled, more efficiently than comparable words rated low on ~ dimension [7] and (fi) high imagery words appear to be more bilaterally represented inthe orain than those of low imagery [6 ]. In v i e w o f ~ ~ o b ~ n t h a t imagery mediated learning of linguistic material~is d e p e n d e n t ~ ,the i n t ~ of the ~ t temporal lobe [ 5 ] it was h ~ ~ ~ one ~ o n forage above mentioned recail advantage of ~ ~ words ~ ~ ~ ~ more extensive bilateral processing that they initiate when subjects attemp~ to learn them. This hypothesi~ was tested using electroencephldographic (EEG) techniques by recording bilateral tern- pond EEG while subjects learnt words which varied on the dimension of ~ , Acomflderable body of evidence [2,3] suggests that the EEG alpha baud csn be ~ to provide evidence about the relative contributions of ~ h ~ ~ r ~ ~ 0 ~ cognitive activities and is based on tlhe assump- tion that relatively less alpha ~ctivity ~ from cortical areas involved in the 0 on. ln the present study, the parameters of power ~ interhemis~ ~eren~ in the alp~ and thet~ bands were investigated, ~ coherence ~ e beingutilised to provide an index of

68 •

the extent to which activity in each frequency band from the two hemi- spheres was correlated in the different experi'm~ntal conditions.

Ten right-handed males served D sttbJ ects- Stimuli consisted of two lists, each ot wordsi Zn and arranfp~ vertically in a random order. One list c o ~ o f W o ~ ~ I high in imagery (HI) and the other of low-imagery (LI) words. The two lists were e q u ~ l for frequency of occurrence in the Engiish language. of frequency were obtained from Thomdike a u d ~ ~ [9] ~ o f from Paivio e ta l . [8] . The mean imsgmry r a t i n p were 6 ~ snd3.14 for the HI and LI words respectively. •

Ag/AgCI electr,xles were placed on scalp sites Ts andT~ of the 10--20 system [ 4] and referred to an electrode at Fz. The two chsnnels of EEG (~/~ amplitude cut~off points I Hz and 35 Hz) were monitored on paper and recorded on an FM tape-recorder prior to offline analysis.

Following the application of the electrodes the subject was seated in front of a screen on which could be mounted the lists of stimulus words. S u ~ were required to memorise and subsequently to recall each list of words, the order of presentation of which was counterbalmuced over subjects. Prior to the performance of the experimental ~ k s subject ' E ~ were recorded during a 2.Sdmin period ofrelaxatton with eyes open. Subsequently s u b ~ t s were p r e s e n ~ with a list of words with hmt~ctions to memorise the lint in any way they chose. After 2.5-rain they were required to count b a e M ~ in 3s for 30 sec (thus eliminating any contribution to ~ of s h ~ r m memory) and then to write down as ~ y Of the words as possible. The same procedure was then followed with the second list.

Spectral analysis of subjects' EEGs was performed using the Sande.Tukey FFT algorithm [1 ]. This was done on 12 artefact-free 2.564ec epochs (100-Hz sampling rate) obtained from the EEG zecozded in each condition. Mean power and coherence ~ues in the alpha (7~81--12.11 Hz) and theta (3.91--7.42Hz) bands were computed for ~of the ~ conditions, no- task, learning HI words and l~ing LI words.

Analysis of subjects' performances revealed a highly significant effect of imagery (P <0 .001) i n d i c a ~ that subjects m~flled m o ~ M words(Means: HI = 17,0, LI - 12.1), The effects of the experimental tasks on p0wer we:e evaluated using as an index the amount of change m power m each hemi-

T A B L E I

Mean alpha power change scores in each condi t ion

HI LI

L. hem. R. hem. L. hem. R. hem.

M 9.30 16.12 15.07 13.05 SD 25.92 30 .36 23.64 36.00

TABLE H

Mean theta power change scores in-ear l condit ion

69

H I ¸ LI

L . hem- R. hem. L. hem. R. hem.

M --9.98 --20.01 7.81 1.24 SD 29.82 20.58 16.33 19.03

sphere from the no-task to task conditions using the formula (no-~ask p o w e r - task power/no-task power + task power) X 100. Mean alpha change scores are shown in Table I. These were analysed using a 2-way ANOVA. An effect was found for i m a ~ w (P < 0,05) and the interactior of hemispheres and imagery (P~ 0.05). Multiple comparisons using the T,~key test [10] revealed that these effects were caused by the right hemisphere having, in the HI condition, a significantly smaller change score than the left hemisphere, a n d a smaller score than either hemisphere during the l e ~ i n g of the LI words. Thus, relatively greater alpha supp~ssion in the left hemisphere o c c ~ only while subjects learnt the H! words, this asymmetry bein~ absent during the learning of the LI words. Table II shows the mean chmlge scores for theta power. These scores were significantly lower while subjects learnt the HI words (P <~ 0,025), indicating that relatively more theta was present during the l e ~ ~ of these words than those of LI.

No significant differences between HI and LI words were found with respect to the coherence data, which are shown in Table III. To assess the extent to which changes in correlated EEG activity between the hemispheres were associated with differential performance on the two memory tasks the difference in coherence levels during the learning of the HI and LI words w~,~ correlated, for each frequency band, with the proportion, as a function of total recall, of the number of LI words successfully recalled (this ia:ter m~a- sure allows an assessment of the relative difference in recall of the two word types independently of overall levels of recall). This correlation c Jefficient w u +0.602 (P < 0.05) for the alpha band and -0 .672 (P .~ 0.05) for the theta band. Thus, subjects in whom the difference in the recall o~ HI and LI

TABLE HI

Mean leveb of coherence (z transform of square root of the original levelg) in eac~ condition

No tj~sk HI LI

Alpha M SD

Theta M SD

0,834 0.974, O. 82 0.286 :).364 0.440 1.259 1.482 1.348 0.399 0.426 0.268

70

words was small tended to have large d i f f ~ in interbemisplumc coher- ence in the alpha band and small ~ i i n ~ t h e ~ band. ~-.

The h y p o ~ that HI..words are mo~e ']ikelFto en~qp thece reb r~ hemi- spheres b g a t m ~ y is not supported by this study, in that the analysis of task- dependent asymmetries the learn- ir~ of the LI words that occurred, this pedmps r e ~ the f ae t t h ~ subjects found these words more difficult to learn. It is ~ interest, however, to note that the HI words gave rise bilaterally to more power in the theta bsnd; an effect which may be of relevsnee to the more ~ t ~ ~ o f ~ words.~

Abo of mtermt m the ~ ~ : t ~ ~ ~ ~ and theta mter- hemisphe~ coherence ~ e e n ~ t w o prediclmd, in iffe t ways, the relative recaU o f the ~ o t y p e s o f wonb~ ~ ! i ~ c e o f this finding in relation ~to i n ~ ¢ . ~ ~ d ~ mnmnonic pro-

may be ~ ~ , ~ ~ n S m i t ~ d ~ t h a g ~ in • ~ • 4. 4,

memory, performance are differently , t s i g n i f i e ~ t l y s m o c i ~ with.two different m e ~ a r ~ o f ~ t y in the E E G . ~ o f t h e hemispheres, i.e., the coherence levels o ~ m the two ~bads, However, these ~,ffects require further investipt /on p ~ ~ a a / ~ t o i n ~ their meanir, g wi~ respect to the pomi'ble p e y © h o ~ . ~ ~ ~ may reflect.

~ v e of the difficulties in ~ ~ of ~ese results they pro- vide strong ev ince in .upport ~ ~ ~ ~ ~ j ~ e n t frequency ~ of the EEG may independently ~ procewet u n d ~ ~ complex informa- tion proce~___ing. It wouldbe of ~ k ~ t o determinewhether the effects observed in the alpha and theta bands in t h e p t ~ e a t ~ ~ reflect ,t~e same or different aspects of the many mnemonic and linguistic p r ~ that make up the experimental tasks used in this study.

REFERENCES

1 BlOomfield, P., Fourier Analysis o f ~ Series: An lntroduet/on, WHey, New York, 1976.

2 Butler, 8,R. and Glass, A, F~G ~ t e s o f ~ dominanee. In A.R. Riuen and R,F. Thompson(Kds.), A d ~ in Psyehobio~.-VoL 11~ Amdemie Prom, New York, 1977. ~

3 Donehin, ~ El' Kutas, M. and M ~ y , G., ~ e a l indices of hemkphede utilisation. In S. Harnad, R.W. Doty, L. Goldstein, J. Jaynes and G. Krauthsmer (Eds.) Lateralmation in the Nervous System, Academic Press, London, 1977.

4 Jasper,H,H., The ten twenty e ~ ; ~ of ~ i n ~ f l m m l f ~ t i o n , Eleetr0- enceph, olin. Neurophysioi., 10 (1958) 371--375.

5 Jones~otman, M. and Milner, B,, Right temporal lobe contributions to inn~e~mii- ated verbal learning, Neuropsyehol~ 16 (19"/3) 61--71.~ ~

6 Marcel, T. ~nd Pattemo~, K,, Word ~ t i o n and production : reeiprocity in clinical and norms] studies. In J. Requin (Bd.), A t t e n ~ , a n d P e r f o ~ VII, L EHbaum, New Jersey, 1977.

7 Paivio, A,, Mental imagery in associative l e ~ smi me~, lho 'eho~. Rev., 76 (1969) 241--263. ~

8 ~ngfutuem

9 Bureau of

10 ~don, 1972.