Neurobiology of Language

The language science group is carrying out basic research to answer the following question: What brain mechanisms make it possible for humans to perform complex language tasks? The theoretical basis of the research is rooted in Hebb's concept of cell assemblies. Cell assemblies are strongly connected neuron sets made up of cortical (and subcortical) nerve cells that form as a consequence of simultaneous activation of neurons (Hebb's rule) Each assembly can be considered a closed functional system. It is important that a single assembly may include neurons of distant brain regions (transcortical assembly). It is postulated that the building blocks of cognition are transcortical assemblies. According to this view, cognitive entities, such as mental images, numbers, letters, or words, correspond to a specific transcortical assembly. Various predictions derived from this theoretical position are being tested in behavioral and physiological experiments with normals and neurological patients. The following hypothesis are of particular relevance: Words (such as "moon") correspond to transcortical assemblies, while so-called pseudowords ("noom") lack such brain representations. Assemblies corresponding to word forms are assumed to be (i) strongly lateralized to the left hemisphere and (ii) restricted to the perisylvian language cortices. Words of different grammatical or semantic classes have assemblies with distinct topographies. While assemblies representing grammatical function words ("the", "it", "and") are localized in the perisylvian cortices and lateralized to the left, content words ("fish", "run", "red") are likely to have assemblies with a much wider distribution. This latter assumption is based on the following con-sideration. In contrast to grammatical words, content words acquire meaning during language acquisition. A word, such as "fish", is frequently perceived and produced together with specific non-linguistic visual, gustatory, or somatosensory stimuli. The shape or smell of fish activates neurons in various cortices of both hemispheres. When such neurons are frequently co-activated with a perisylvian assembly, they develop into a higher-order assembly distributed over wide cortical areas of both hemispheres.

During the last year,we have been investigating electrocortical and biomagnetic differences between words and pseudowords, as well as between words of different grammatical and semantic classes (function/content, abstract/concrete, verbs/nouns). Three of these experiments are summarized below. Related behavioral experiments were conducted at the Psychology Department of UCLA in collaboration with Bettina Mohr, Janice Rayman and Eran Zaidel. In addition to empirical work, the members of the language science group have further developed the cell assembly theory in order to specify the neurobiological mechanisms underlying syntactic processes and language acquisition. This theoretical work is being conducted in collaboration with Prof. Valentin Braitenberg at the Max Planck Institute of Biological Cybernetics (now at our Insitute) and Prof. John Schumann at the Department of Applied Linguistics, UCLA.

Gamma-band responses of the brain and language processing: EEG

Neurobiologists have often claimed that fast oscillatory cortical responses in the gamma band (20 Hz and up) are of utmost significance for so-called "higher" or "cognitive" processing, such as gestalt perception, language processing, or reasoning. In order to test this hypothesis, it is necessary to compare higher integration processes to a baseline condition in which no (or reduced) integration takes place. In a series of experiments we compared evoked spectral EEG responses to meaningful words (such as "moon") and physically similar (matched) pseudowords (such as "noom"). Words and pseudowords were presented on a video screen. Subjects had to perform a "lexical decision task", that is, they had to communicate whether the item presented was a meaningful and regular word or not.

The results were the following: Words were associated with fast EEG responses (high power), while after pseudowords, there was a depression of gamma-band power around 30 Hz. This gamma-band depression was only observed over the left hemisphere.

This result can be explained on the basis of the Cell Assembly Theory. Word presentation leads to coherent and periodic activation of a large neuronal population, a cell assembly, and, therefore, to continuous gamma-band activity. In contrast, presentation of a pseudoword fails to activate an assembly and, therefore, gamma-band power is depressed. These results are consistent with the idea that gamma-band responses are an indicator of higher integrative processes taking place in the cortex.

Lutzenberger, W., Pulvermüller, F. & Birbaumer, N. (1994). Words and pseudowords elicit distinct patterns of 30-Hz EEG responses in humans. Neuroscience Letters 176, 115 -118.

Gamma-band responses of the brain and language processing: MEG

A similar study was conducted using the Magnetoencephalogram (MEG). Recording biomagnetic signals from both hemispheres is not yet possible in Tübingen. Therefore, we used a 2 x 37 channel biomagnetometer in San Diego, California (Biomagnetic Technologies Inc.). In collaboration with Prof. Elbert (University of Konstanz), Prof. Pantev and Dr. Eulitz at the Institute of Experimental Audiology in Münster, Germany, an experiment was conducted investigating high-frequency biomagnetic responses to spoken words and pseudowords. Again, the same results were found: Gamma-band depression after pseudoword presentation was observed in the 30 Hz range, while words led to continuous 30 Hz activity. Gamma-band depression after pseudowords was observed only over the left hemisphere in all five individuals tested. Figure 8 displays the results obtained from one experimental subject. Note the "valley" in the contour plot on the lower left. These data confirm the results obtained in EEG experiments. Again, the interpretation is the following: During the whole experiment, large neuronal assemblies - the brain-internal "word processors" - are active. However after pseudoword presentation, word assemblies are not ignited and, therefore, fast brain responses are not observed (have less power).

Pulvermüller, F., Eulitz, C., Pantev, C., Mohr, B., Feige, B., Lutzenberger, W., Elbert, T. & Birbaumer, N. (1995). Gamma-band brain responses reflect cognitive processing: an MEG study. Electroencephalography and Clinical Neurophysiology , EEG-Journal.

Cortical representation of content and function words

The classical theory of language representation in the brain goes back to aphasiologists of the 18th century. Broca, Wernicke and Lichtheim proposed that relatively small cortical regions process language. This was motivated by the observation that organic language disturbances, aphasias, occur primarily after lesions of these language areas. More recently, it has been found that lesions of brain areas outside the classical language areas also can cause aphasic disturbances. This indicates that large parts of the brain are involved in processing language. However, different linguistic processes may "use" different cortical areas. As detailed above, it has been proposed that cortical representations differ between word classes. Grammatical function words are assumed to be represented in the language regions of the dominant hemisphere, while cell assemblies corresponding to meaningful content words should be distributed over various cortical areas. This hypothesis was tested in an EEG experiment where matched content and function words were presented. While content words evoked almost identical responses over the left and right hemisphere, function words evoked asymmetric cortical potentials. After function words, potentials were more negative over the left hemisphere compared to the right (and compared to content words) between 150 and 300 msec after stimulus onset. This suggests that function word processing is lateralized to the language-dominant left hemisphere. In contrast, both hemispheres appear to be equally involved in processing content words.

Pulvermüller, F., Lutzenberger, W. & Birbaumer, N. (1995). Electrophysiological distinction of vocabulary types. Electroencephalography and Clinical Neurophysiology, 94, 357-370.

Home-Page
Hubert Preißl
Maintainer: hubert.preissl@uni-tuebingen.de(hubert.preissl@uni-tuebingen.de)