Gamma-band activity, as examined by Pulvermueller et al. (1994) is not necessarily a genuine and independent phenomenon but may be an epiphenomenon of activity changes in lower frequency bands. Recent experiments have shown a strong contribution of harmonics of alpha activity to activity in the beta and gamma ranges. The target article and the response to the commentary of Mueller & Jokeit (1994) do not present enough evidence to invalidate the hypothesis that the observed gamma-band effects are merely due to changes in alpha power.
2. Alpha activity is known to be closely linked to cognitive processing (Michel, Kaufman & Williamson, 1994; Pfurtscheller & Klimesch, 1991; Vijn, 1992). A reduction of spectral power in the lower alpha range (8-11 Hz) was found in many psychophysiological experiments after onset of a stimulus, during cognitive processing, and with increased attention. A stimulus related increase of power in the upper alpha band (11-13 Hz) was also reported (Gale, Davies & Smallbone, 1977; Roesler, 1975). Therefore, it is most likely that stimulus related changes of alpha activity were also present in the experiments described by Pulvermueller et al. (1994). Moreover, it has to be expected that alpha band effects were different for words and pseudowords, because processing demands, as reflected by the phasic slow negative wave, were higher for pseudowords than for legal words.
3. As Mueller and Jokeit (1994) already noted, Fourier analysis of nonsinusoidal periodic activity does not only produce one single peak corresponding to the oscillation period but also peaks at integer multiples of the base frequency. Power coefficients at these harmonics depend on the shape of the periodic process. A symmetric triangular waveform, for example, has harmonics at odd multiples of the base frequency while the power at even multiples is zero (Brigham, 1975).
4. A recent experiment of our group on memory storage and retrieval confirms that strong interrelations can exist between power in the alpha and in higher frequency ranges (Juergens, Roesler, Hennighausen & Heil, 1995). In the first part of this experiment, associations between line drawings and two types of mediators (grid positions and concrete nouns) had to be learned. In the retrieval experiment subjects should decide whether two drawings were associated via a common mediator or not (Heil, Roesler & Hennighausen, 1994; Roesler, Heil & Hennighausen, 1995). During both experimental sessions, the EEG was recorded between 0 and 100 Hz from seven electrodes. Fourier spectra were calculated in successive overlapping time windows of 1 s length. For one subject, a stimulus-locked increase of power at 12 Hz was found, while in the other two subjects alpha power decreased at the individual peak frequencies of 9 and 11 Hz, respectively. The spectra of all subjects revealed corresponding changes of power in the higher frequency ranges but only at harmonics of the individual dominant alpha frequencies. These observations could be substantiated by an analysis of covariance of the time courses at harmonic and non-harmonic frequencies in the alpha, beta and gamma ranges.
5. The target article does not discuss the possible influence of alpha activity on the reported findings in the gamma range. In their response to Mueller & Jokeit (1994), Pulvermueller & Lutzenberger (1994) reported that an analysis of variance did not reveal a significant Wordness-effect or interactions of Wordness with other factors in the 10-20 Hz range. However, this finding does not invalidate the hypothesis that the changes in the gamma-range are due to equivalent Wordness-effects in the alpha range, for two reasons: Firstly, the lower alpha band (8-9 Hz) was not examined at all; and secondly, the filter used to restrict the frequency range was cosinus shaped (Lutzenberger et al., 1994). This implies that filter coefficients at the edges of the passband (10, 11, 19 and 20 Hz) are small. Consequently, the respective frequencies could bear only very little influence on the power of the filtered signal. The exclusion of the lower alpha band may be the reason why possible differences in lower alpha activity between word/pseudoword conditions did not become manifest as significant statistical effects in the analysed frequency band (10-20 Hz). It should be noted that the influence of the first harmonics (16-22 Hz) of the lower alpha band might be non-significant because of the specific shape of the alpha oscillation. Power decreases at the first harmonics of lower alpha activity could also be masked by increases of higher alpha activity. The second harmonics of the lower alpha activity is located in the frequency range of 24-33 Hz overlapping the 25-35 Hz range where a significant Wordness effect was found. This might be the reason for the observed wordness effects in the gamma-band.
6. Further steps in data analysis are necessary for a decision whether effects in the gamma-band are genuine or just epiphenomena of alpha activity. In particular, it is necessary to evaluate how different sub- bands of the alpha range are related to the Wordness effect. In Juergens et al. (1995) it was found that stimulus related changes in the alpha-band are often restricted to very narrow frequency ranges with a bandwidth of one Hz only. Higher spectral resolution is a prerequisite for a more convincing analysis. Using narrow band filtering, frequency peaks at integer multiples of the dominant alpha frequency with high temporal covariance would reveal the existence of harmonics. Another problem is caused by the fact that alpha oscillations show substantial interindividual variation in frequency, depending on experimental manipulations and most likely also in the pattern of harmonics (Deakin & Exley, 1979; Gale, 1973). Therefore, alpha-gamma-interrelations have to be evaluated separately for each subject.
7. A further problem in Pulvermueller et al. (1994) concerns the fact that there were absolutely no differences in gamma-band activity from the pre- to the post-stimulus epoch when words had to be processed. Processing of words has a precise timing in relation to stimulus onset. Consequently the proposed assembly model would predict a difference in cell assembly activity between pre- and post-stimulus epoch. While in the prestimulus baseline "a continuous stream of thoughts" prevails, evoking search processes which are similar to the functional state of recognizing pseudowords, the presentation of a legal word should ignite a particular cell assembly only. This should become evident as a short increase of gamma-activity.
Brigham E.O. (1975). The fast fourier transform, Prentice Hall Englewood Cliffs, NJ.
Deakin, J.F.W. & Exley, K.A. (1979). Personality and male-female influences on the EEG alpha rhythm. Biological Psychology, 8, 285-290.
Gale, A. (1973). The psychophysiology of individual differences: Studies of extraversion and the EEG. In P. Kline (Ed.), New Approaches in Psychological Measurement (pp. 211-256). London, New York, Sydney, Toronto: John Wiley & Sons.
Gale, A., Davies, R. & Smallbone, A. (1977). EEG correlates of signal rate, time on task and individual differences in reaction time during a five-stage sustained attention task. Ergonomics, 20, 363-376.
Heil, M., Roesler, F. & Hennighausen, E. (1994). Dynamics of activation in long-term memory: The retrieval of verbal, pictorial, spatial, and color information. Journal of Experimental Psychology: Learning, Memory and Cognition, 20, 185-200.
Juergens, E., Roesler, F., Hennighausen, E. & Heil, M. (1995). Stimulus induced gamma oscillations: harmonics of alpha activity? Neuroreport, 6, 813-816.
Kutas, M. & Kluender, R. (1994). What is who violating? A reconsideration of linguistic violations in light of event-related brain potentials. In H.T. Heinze, G.R. Mangun & T. Muente (Eds.), Cognitive electrophysiology (pp. 183-210). Boston: Birkhaeuser Inc.
Lutzenberger, W., Pulvermueller, F. & Birbaumer, N. (1994). Words and pseudowords elicit distinct patterns of 30-Hz EEG responses in humans. Neuroscience Letters, 176, 115-118.
Michel, C.M., Kaufman, L. & Williamson, S.J. (1994). Duration of EEG and EMG alpha suppression increases with angle in a mental rotation task. Journal of Cognitive Neuroscience, 6, 139-150.
Mueller, H.M. & Jokeit, H. (1994). Word Processing and Gamma band Activity. PSYCOLOQUY 5(60) brain-rhythms.5.mueller.
Pfurtscheller, G. & Klimesch, W. (1991). Event-related desynchronization during motor behavior and visual information processing. In C.H.M. Brunia, G. Mulder & M.N. Verbaten (Eds.), EEG Suppl: 42. Event-related brain research (pp. 58-65). Amsterdam: Elsevier.
Pulvermueller, F. & Lutzenberger, W. (1994). Specific gamma band depression and linguistic units. PSYCOLOQUY 5(68) brain- rhythms.9.pulvermueller.
Pulvermueller, F., Preissl, H., Eulitz, C., Pantev, C., Lutzenberger, W., Elbert, T. & Birbaumer, N. (1994). Brain Rhythms, Cell Assemblies and Cognition: Evidence From the Processing of Words and Pseudowords. PSYCOLOQUY 5(48) brain-rhythms.1.pulvermueller.
Roesler, F. (1975). Die Abhaengigkeit des Elektroenzephalogramms von den Persoenlichkeitsdimensionen E und N sensu Eysenck und unterschiedlich aktivierenden Situationen. Zeitschrift fuer experimentelle und angewandte Psychologie, 22, 630-667.
Roesler, F., Heil, M. & Hennighausen, E. (1995). Distinct cortical activation patterns during long-term memory retrieval of verbal, spatial and color information. Journal of Cognitive Neuroscience, 7, in press.
Vijn, P.C.M. (1992). Coherent neuronal activity underlying the EEG. Unpublished Dissertation, Universitaet Amsterdam.