4 ms·
We‘ve known for over thirty years, first based on lesion studies then neuroimaging. For instance, we saw that language didn’t map to one region, but depended on
by robg 5y ago
We‘ve known for over thirty years, first based on lesion studies then neuroimaging. For instance, we saw that language didn’t map to one region, but depended on how language was being used and then word meanings relying on sensory vs more abstract brain regions (e.g., cannon vs cannoli vs carve).
Think of features with distributed representations, patterns of connectivity re-using bits in endless ways. Another oversimplification is that neurons can be only representing 1’s and 0’s. The true computational power is every state in-between and strength of connections between processing units.