Pink elephants serenading the world's tallest woman might be an improbable situation, but our brain is able to comprehend this thought. Humans can generate an infinite set of ideas from a finite set of words, but how the brain accomplishes this feat remains unclear. Now, a new study by US scientists suggests the human mind flexibly combines the meanings of individual words to compose structured thoughts. Researchers in the Department of Psychology at Harvard University, devised two experiments to identify regions in the brain that encode for meaning where the structure of the sentence is critical; and then how the brain represents this meaning. In the first experiment, 18 participants undergoing a functional MRI scan read simple sentences that could be conveyed in either the active or passive voice. These sentences could also be arranged to have mirror meanings: ‘the dog bit the man’ or ‘the man bit the dog’.
Through this they were able to identify that the left mid-superior temporal cortex in the brain plays a key role in decoding of sentence meaning and predicting the required response based on this information. In the second experiment, four nouns -- man, girl, dog, cat -- and five verbs (chased, blocked, approached, bumped and scratched) were used in various combinations to create sentences. For example the 34 participants would read sentences while undergoing fMRI such as ‘the dog chased the man’ and ‘the girl was scratched by the cat’. An analysis of the scans from this experiment show one distinct area of the left mid-superior temporal cortex is responsible for processing "Who did it?" (the agent) while a neighbouring separate region encoded for "To whom it was done?" (the patient). The researchers say the findings support long-held theories that the brain acts like a computer processing data.