Spoken language is perhaps the most notable marker of our species. Our ability to call forth words with universal meaning is so natural that we take it for granted, and yet scientists still know very little about how it works, or from where the ability derives.
Because current neuroimaging technologies have relatively low resolution, it has been extremely difficult to study language in human subjects. Researchers have generally relied on observational models from linguistics to try to predict how the brain represents the meaning of words.
Recently, a team of researchers used a creative methodology to get past these technological hurdles by using a text corpus, a linguistics tool that shows how often certain words co-occur with other words. By combining information from the text corpus with previous fMRI data gathered while subjects thought of specific nouns, a model was able to predict patterns of brain computational activation with remarkable accuracy for words that had never before been imaged.
Predicting Human Brain Activity Associated with the Meanings of Nouns
Science May 30, 2008
