A computational model is introduced that predicts functional magnetic resonance imaging (fMRI) neural activation associated with the meanings of concrete nouns. The model is trained using a combination of data from a trillion-word text corpus and observed fMRI data from viewing concrete nouns. It successfully predicts fMRI activation for thousands of other concrete nouns with high accuracy, even for nouns for which no fMRI data are available. The model is based on the assumption that the semantic representation of concrete nouns is related to their distributional properties in a large text corpus. The model uses a two-step process: first, it encodes the meaning of a word as a vector of intermediate semantic features derived from the text corpus; second, it predicts fMRI activation as a weighted sum of neural activations contributed by each of these features. The model's predictions are validated by comparing them with observed fMRI data, showing significant accuracy. The results suggest a direct predictive relationship between word co-occurrence statistics in text and neural activation associated with word meanings. The model's success indicates that semantic features and their neural activation signatures can span a diverse semantic space. The study also highlights the importance of sensory-motor features in neural representations of concrete nouns and demonstrates that the model can accurately predict fMRI activation for words in new semantic categories. The findings contribute to understanding how the brain represents and processes conceptual knowledge.A computational model is introduced that predicts functional magnetic resonance imaging (fMRI) neural activation associated with the meanings of concrete nouns. The model is trained using a combination of data from a trillion-word text corpus and observed fMRI data from viewing concrete nouns. It successfully predicts fMRI activation for thousands of other concrete nouns with high accuracy, even for nouns for which no fMRI data are available. The model is based on the assumption that the semantic representation of concrete nouns is related to their distributional properties in a large text corpus. The model uses a two-step process: first, it encodes the meaning of a word as a vector of intermediate semantic features derived from the text corpus; second, it predicts fMRI activation as a weighted sum of neural activations contributed by each of these features. The model's predictions are validated by comparing them with observed fMRI data, showing significant accuracy. The results suggest a direct predictive relationship between word co-occurrence statistics in text and neural activation associated with word meanings. The model's success indicates that semantic features and their neural activation signatures can span a diverse semantic space. The study also highlights the importance of sensory-motor features in neural representations of concrete nouns and demonstrates that the model can accurately predict fMRI activation for words in new semantic categories. The findings contribute to understanding how the brain represents and processes conceptual knowledge.