Predicting Human Brain Activity Associated with Noun Meanings

Recent brain imaging studies have shown that different spatial patterns of neural activation are associated with thinking about different semantic categories of words and pictures (e.g., tools, buildings, animals). As a next step we seek a general theory capable of predicting the neural activity associated with arbitrary words not yet included in experiments. We present here the first such predictive theory, in the form of a computational model that is trained using a combination of data from a trillion-word text corpus, and observed fMRI data associated with viewing several dozen concrete nouns. Once trained, the model predicts fMRI activation for thousands of other concrete nouns in the text corpus, with highly significant accuracies over the 60 nouns for which we currently have fMRI data. One sentence summary: We present the first computational model capable of predicting observed fMRI activity produced when humans think about an arbitrary concrete noun, along with experimental results showing strong prediction accuracy over the 60 nouns for which we have fMRI data.

[1]  Christiane Fellbaum,et al.  Book Reviews: WordNet: An Electronic Lexical Database , 1999, CL.

[2]  J. Winn,et al.  Brain , 1878, The Lancet.

[3]  T. Bonhoeffer,et al.  Current opinion in neurobiology , 1997, Current Opinion in Neurobiology.

[4]  K. Goodman,et al.  Encyclopedia of Language and Linguistics , 2006 .

[5]  S. Grossberg,et al.  Psychological Review , 2003 .

[6]  C. Franzen,et al.  THE INTERNATIONAL JOURNAL OF LEXICOGRAPHY , 1989 .

[7]  M. Gazzaniga,et al.  The new cognitive neurosciences , 2000 .

[8]  Thomas G. Dietterich What is machine learning? , 2020, Archives of Disease in Childhood.