Pulvermueller et al. (1994) review the cell assembly theory and discuss theoretical explanations for their observed experimental findings during high-level cognitive processes such as word processing. To find empirical evidence for their hypothesis, ERP and MEG experiments on word and nonword comprehension were conducted. It is not clear, however, whether the observed suppression at 30 Hz following pseudoword presentation reflects real gamma-band dynamics rather than so-called event-related desynchronization in the beta band of the EEG with a similar time course.
2. The importance of the phenomenon of gamma-band activity for cognitive processing, observed mainly in animals, is still under debate (Singer 1993b). Even so, there are anatomical evidence for clustering cortical neurons (Pandya & Yeterian 1985); the function of these neurons does not need to be hard wired. It is more likely that a certain cortical neuron contributes to several META-UNITS, depending on the kind of cognitive process, and is flexibly switched by neuromodulation. This participation of a certain neuron within several functional groups in short time periods would explain some of the difficulties in correlating cognitive tasks with the response profile of neuron clusters. In the cell assembly theory on which Pulvermueller et al. base their argument, a figure of 106 neurons for one assembly is discussed. However, it is not yet clear which level of processing such a cell assembly would represent in the hierarchy of brain functions. There is no evidence for the tantalizing statement that "the hierarchy of linguistic structures (phoneme, morpheme, word, sentence) has its biological equivalent in a hierarchy of cell assemblies corresponding to these cognitive entities" [para. 4] especially when it is pointed out that these linguistic structures are artificial. It is not yet known in what way these constructions are related to real cognitive components during language comprehension. Phonemes are very useful entities in a structuralistic approach to language analysis, but that is a mentalist view (early Prague School). There is also evidence that the linguistic concept of a syllable, for example, could reflect the biological processing pattern. In addition, it should be kept in mind that concepts like "word" or "sentence" are strongly related to written language. So far, in linguistics, there is no consensus about the basic units of language comprehension.
3. Apart from a certain neuron functioning in different functional units, it seems reasonable that those units themselves should function within the next hierarchical unit. This means that a certain kind of processing can be accomplished by several neuronal units and not only by one functional META-UNIT. The enormous plasticity of cortical brain functions suggests that uniform response patterns from a certain group of cortical neurons in a certain high-level cognitive task such as the processing of one word are rather unlikely. It is difficult enough to get an identical response to repeated stimuli from third-order sensory neurons in lower vertebrates. Why should one expect identical response patterns from a small cortical cell assembly in humans while reading a certain word, as proposed in the target article: "after word presentation, the ignition of exactly one assembly takes place (while competing assemblies are inhibited)" [para. 30]? It is difficult to imagine that one word will elicit a specific assembly of neurons.
4. In fact, Pulvermueller et al. predicted stronger gamma-band responses to words compared to pseudowords. The following was shown:
(i) Around 400 ms post stimulus onset, pseudowords elicited more negative ERPs compared to words. This larger negativity has been taken as evidence that pseudoword presentation elicited more cortical activity.
(ii) Word presentation did not change 30 Hz power compared to the baseline. In contrast, pseudowords generated a reduction of 30 Hz cortical responses over the left hemisphere.
(iii) Higher frequency bands (35-45 Hz, 55-56 Hz) did not reveal a similar interaction in any of the time windows.
5. The authors assume that these results are consistent with the view that fast, coherent and periodic activation of large neuronal assemblies takes place after word presentation but not after pseudoword presentation. This view may or may not be correct; however, on the basis of the data presented, it remains pure speculation, for the following reasons: The activity in the high beta range of the human EEG/MEG was reported as suppressed, but there is no reason to infer from a suppression under a certain condition that an almost stationary signal under a different condition reflects an increase. Moreover, the suppression observed in the 25-35 Hz frequency range resembles the well known phenomenon of event-related desynchronization (ERD) in the alpha band of the EEG (Pfurtscheller et al., 1988). More recent studies have demonstrated a correlated decrease in power within the beta band during suppression of alpha-band activity although the spatial distribution of alpha- and beta-band suppression differs (Kaufman et al., 1990; Pfurtscheller & Klimesch, 1990; Dijk et al., 1992). The degree of suppression depends on attention, task complexity and effort resulting in higher cortical activation. We are therefore convinced that a similar interaction between word and pseudoword presentation in the beta frequency band should likewise be observable, if not in the alpha range. Recently published MEG/EEG studies show significant decreases at 20 Hz associated with motor responses (Makeig, 1993; Jokeit & Makeig, 1994) and long lasting suppression of 40 Hz activity to visually presented priming stimuli (Jokeit et al., 1994).
6. We must accordingly conclude that Pulvermueller et al. presented data reflecting event-related desynchronizations which can be observed up to 30-40 Hz. Relatively small effect sizes may demonstrate the upper frequency limit of the ERD phenomenon, or may stem from harmonics of lower frequencies which can show much stronger effects. Our own studies reveal that using short duration Fourier spectra, dynamics in the alpha band are also reflected at harmonic frequencies because of nonsinusoidal signals and finite time windows. We wonder why the authors did not present dynamics at alpha and beta EEG bands and why the rich literature on event-related desynchronization was overlooked. Finally, the speculation that the presentation of pseudowords may result in uncorrelated oscillatory activity in several assemblies has not been validated; the EEG spectrum would show a relatively broad band increase in power at higher frequencies, as has been shown by neural network simulations studying transitions from synchronous to asynchronous activity (H. Gluender, personal communication).
Bullock, T.H. & McClune, M.C. (1989) Lateral Coherence of the Electrocorticogram: A New Measure of Brain Synchrony. Electroencephalography and Clinical Neurophysiology, 73, pp. 479-498.
Dijk, J.G.V., Caekebeke, J.F.V., Jennekens-Schinkel, A. & Zwinderman, A.H. (1992) Background EEG reactivity in auditory event-related potentials. Electroencephalography and Clinical Neurophysiology, 83, pp. 44-51.
Gerstein, G.L., Bedenbaugh, P. & Aertsen, A.M.H.J. (1989) Neuronal Assemblies. IEEE Transactions on Biomedical Engineering, 36, pp. 4-14.
Gray, C.M., Koenig, P., Engel, A.K. & Singer, W. (1989) Oscillatory responses in cat visual cortex exhibit inter-columnar synchronization which reflects global stimulus properties. Nature 338, pp. 334-337.
Jokeit, H. & Makeig, S. (1994) Different event-related patterns of gamma band power in brain waves of fast and slow reacting subjects. Proceedings of the National Academy of Sciences of the United States of America, 91, pp. 6339-6343.
Jokeit, H., Goertz, G., Kuechler, E. & Makeig, S. (1994) Event related changes in the 40 Hz electroencephalogram in auditory and visual reaction time tasks. In: Oscillatory event related brain dynamics, C. Pantev, T. Elbert, & B. Lutkenhoener, eds., New York: Plenum (in press).
Kaufman, L., Schwartz, B., Salustri, C. & Williamson, S.J. (1990). Modulation of Spontaneous Brain Activity during Mental Imagery. Journal of Cognitive Neuroscience, 2, pp. 124-32.
Makeig, S. (1993) Auditory event-related dynamics of the EEG spectrum and effects of exposure to tones. Electroencephalography and Clinical Neurophysiology, 86, pp. 283-293.
Pandya, D.N. & Yeterian, E.H. (1985) Architecture and connections of cortical association areas. In: Cerebral cortex. Vol. 4. Association and auditory cortices. A. Peters & E.G. Jones, eds. London: Plenum Press, pp. 3-61.
Pfurtscheller, G., Steffan, J. & Maresch, H. (1988) ERD-mapping and functional topography - temporal and spatial aspects. In: Functional Brain Imaging. G. Pfurtscheller & F. H. Lopes da Silva, eds., Toronto: Huber, pp. 117-130.
Pfurtscheller, G. & Klimesch, W. (1990) Topographical display and interpretation of event-related desynchronization during a visual- verbal task. Brain Topography, 3(1), pp. 85-93.
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.
Sherrington, C. (1941) Man on His Nature. Cambridge: University Press.
Singer, W. (1993a) Neuronal representations, assemblies and temporal coherence. Progress in Brain Research, 95:461-474.
Singer, W. (1993b) Synchronization of cortical activity and its putative role in information processing and learning. Annual Review of Physiology, 55. pp. 349-374.