Researchers at UC Berkeley have announced early results of a very ambitious brain-mapping project. The results, so far, are remarkable.
The term “mapping” refers to any number of techniques used to determine what parts of the brain are responsible for which functions. In the past, most of what we knew about brain function came from lesional studies. Look at someone with brain damage, figure out what they are missing behaviorally, and then do an autopsy after they die and see what part of the brain was damaged. When CT and then MRI imaging came out, there was no longer any need to wait for autopsy results; we could get detailed images right away. But these were hardly living images; we could see how the brain looked, but we couldn’t see how the living brain functioned.
Today, we can. Functional MRI (“fMRI”) leverages the fact that the brain self-regulates its own blood supply to keep up with metabolic activity. Thinking is metabolism; the parts of the brain that are thinking at any given moment get more blood supply than the rest of the brain. Stick someone in an MRI scanner, tune it to detect regional blood flow, and you can take a snapshot of what part of their brain a person is using to any given time.
Galland and Huth, the principal authors, have published proof-of-concept for a technique they are using to map language at an unprecedented level of detail. See this overview at The Guardian, watch the video, and when you have a couple of hours to play, check the interactive web tool.
Subjects were put into an MRI scanner and were told stories. A snapshot was taken after each word, and results were averaged across several subjects. When a word registers in the subject’s consciousness, their brain lights up like a Christmas tree. What Galland and Huth found was that the pattern of activation was remarkably similar across subjects, down to an astonishing level of detail.
Note, we aren’t looking at where words are stored in the brain. The brain runs off of concepts, and concepts don’t have a physical storage site. Functionally speaking, a concept is a pattern of brain activity that arises in response to stimulation. Each concept represents any number of attributes that are linked together. Before this study came out, we figured concept retrieval would reveal several small areas of brain activation happening all at once, and figured the attributes would generally correlate with brain function, location-wise. So visual attributes would probably light up the visual part of the brain, and sounds would light up the sound part of the brain, etc.
What we didn’t anticipate is the degree of granularity, and the degree of consistency from one subject to another, considering the level of detail.
It’s difficult to draw specific conclusions from this study. It’s a “fishing expedition.”
Play around with the interactive tool, and you’ll see what I mean. After a while you find yourself asking, “What the heck were they reading?” The predominant themes are of crime and punishment. So we now have a good map of the brain’s guilt centers. Note how the word “Jesus” maps with curse words. Hopefully they will try out some different thematic content in the future. We will likely to continue to see the most robust word clouds in parts of the brain that light up during reading. Posterior parietal structures that will light up in response to novel stimuli, frontal regions that light up during concentration, and temporal auditory association cortex. Having the subjects read, instead of listen, will likely light up visual association cortex. Activation of the cingulate gyrus down in the inter-hemispheric fissure will likely vary with the emotional content of the material presented.
But even as submitted, there’s definitely something here. We see a clear picture of the back-to-front organization of the brain, where posterior regions tend to favor content we figure to engage with — themes of contemplation, knowledge and spirituality. Anterior regions tend to favor themes of guilt and control of behavior. Engage, disengage.
Key finding: concepts are grouped by meaning, and meaning tends to correlate with functionally relevant brain structures.