Via Neuromarketing, here's a story in Wired about efforts to create a connectome, a highly-detailed map down to the synapse level, of the human brain.

I'm not sure exactly how useful this is going to be. I like the analogy at Neuromarketing:

The analogy isn’t entirely apt, but is it possible that one might be creating a detailed map of Cleveland to better understand traffic patterns in Dallas? Obviously, our brains all share basic structural similarities, but recent work in neurogenesis, brain plasticity, and even cultural differences suggest that as brain maps get more granular individual differences will become more evident.

From what I know of neuroscience, I'd predict greater similarities between individuals with more genetic similarities (like twins) in their subcortical structures and in their cerebral cortices early in development. I'd then expect such maps to diverge (predominantly in the cerebral cortex) as individuals get older.

Why? Because the subcortical structures are evolutionarily older, and contain more "hard-wired" behaviors, such as controlling your heartbeat and breathing. My recent post on biological motion detection is another good example. The neocortex is relatively uniform, though. I believe it's most a "blank slate," able to learn virtually any spatio-temporal pattern via experience. I believe this massive flexibility is the reason for its success, and its massive expansion in primates over time.

Because it is very generalized, you can learn many different concepts, and the order in which you learn them is largely unconstrained (except by prior knowledge). For example, it doesn't really matter whether you learn what a horse is before you learn what a cow is. It doesn't matter if you learn to ride a bike when you're 5 or when you're 9. You need a certain amount of prerequisite knowledge to learn new things, but once that knowledge is acquired, you can learn new concepts in arbitrary order. I could teach you how to play a game you've never played in five minutes.

And that feature of the neocortex means that there are probably going to be very large individual differences at the synaptic level, because everyone is going to have learned most of what they know through different experiences, in different orders, to different degrees, and all those differences will be reflected at the synapse level.

We know stereotypical patterns of connectivity in the neocortex already. It's a six-layered structure, about as thick as six business cards. And we know that cells in layer 4 are the primary input cells as information flows up the cortical hierarchy. We know that cells in layers 2 and 3 are the output cells to higher levels, with output from layer 2 going to the same side of the cortex, and output from layer 3 going to the opposite hemisphere. I think it's doubtful that such studies are going to buy us much more insight into the broad patterns of connectivity in structures like the neocortex.

However, as I said earlier, I think they will help shed insights into the anatomy and physiology of subcortical structures, and help determine the extent to which cortical regions really are hard-wired or soft-wired. More than likely though, they'll generate gobs of data (we already have a wealth of neuroscience data) which will sit around for years waiting to be integrated into strong, workable theories.