This touches on my area of research. The 'pyramid' idea of brain function is essentially a straw man to jazz up the Science Daily article. It's been known for a long time (a century) that recurrent loops occur throughout the brain. The mystery is what these loops do, given the apparent hierarchical buildups of representation that seem to occur in sensory systems, particularly the visual system. (Also in the hippocampus, e.g. place cells, and other areas in all likelihood.)
I haven't read the paper thoroughly (http://www.pnas.org/content/early/2010/07/28/1009112107), but looks like the authors are using a strategy of systematic and tightly circumscribed injections of anterograde & retrograde tracers to map out, with unusually high precision, the inter-areal projections for a particular circuit they're interested in. This mapping of inter-areal connectivity is a little different from what I'm interested in, which is to understand how the many-to-many anatomical wiring diagram between individual neurons in a neuronal network relates to the transformation of information carried out by that network. In other words, they are mapping connections between areas, but I think it's likely that an additional level of resolution -- the mapping of connections at the level of individual neurons -- will likely be necessary (and probably insufficient) to crack brain circuits.
I think that this view has been around longer than 2003. Steven Pinker describes how integrated as well as parallel thought is in his book The Blank Slate, which came out in 2002 (see: http://en.wikipedia.org/wiki/The_Blank_Slate). Assuming that he took 2-4 years writing it. He must have been working on it in the 1990s.
This is an important thing as it isn't an assumption anymore. This idea of a complex and rich network of networks is now starting to be seen in computer simulations of actual neural networks. The best bit is that people can do deeper experimentation based upon those two facets like the blue brain project.
Although, a lot of questions remain unanswered. A good starting part, I think, would be to see how information itself is transferred and stored at the 'nodes' of the network. For example, recently Dr. Thomas Sudhof (see: http://neuroscience.stanford.edu/research/laboratories/Sudho...) from Stanford won a prize because he decoded the protein that causes synaptic exchange to happen (see: http://med.stanford.edu/ism/2010/june/sudhof-0603.html). That's just one piece, but it's a crucial piece.
What I love about the current state of neuro-science is that it's like a scattered jigsaw puzzle waiting to be put together. Almost every month / week (?) some small, but crucial piece is found and it gets added to the box. It's simply beautiful.
Another really interesting thing is that everything that's usually here on HN from map reduce to manipulating huge amounts of data is laying the foundation for creating tools that tackle this problem. It's like everybody is inadvertently working to complete the jigsaw without realizing it, which is really, really mind blowing.
When do you think we'll have a very basic understanding of how the brain works? E.g, How are memories stored, how does pattern recognition happen, how does generalization work?
Um, I have no clue as I am pretty uneducated. I don't know anything beyond what I read in light books. So, I doubt if I am qualified at all to say anything unsubstantiated in any field.
On the other hand, I also doubt if anyone actually does have a clue about the when.
I really can't. I don't even have a day of college education, and I wasn't exactly a stellar student in HS either. So, it's quite likely that you know 10^10^10^10 times more than me.
Also, I am really not as smart as I sound. In fact, I am the most boring and dumbest# person you could possibly ever meet.
I doubt its much like either, although a distributed network might be our current best but still poor guess. At any given time period people are always comparing the brain to their most salient piece of current technology. While at each step more of the brain's essence is arguably captured, the metaphors still leave alot to be desired - glossing over gigantic swathes of functionality.
The brain has been compared to spirits, hydraulics, clockwork systems, computers and now the internet. I think it takes hubris to not expect people in the future to think our current theories of the brain to be as quaint as how we now view ideas of the brain as a mechanical system.
Well, it's actually sort of hard for "distributed network" to be wrong. That's not because it's a brilliant insight, but because it's so vague it's almost inconceivable it does not include the truth in the immense number of organizations that can be called "distributed networks".
> Neuroscientists are split between a traditional view that the brain is organized as a hierarchy, with most regions feeding into the "higher" centers of conscious thought, and a more recent model of the brain as a flat network similar to the Internet.
"Higher" centers of consciousness seems to be a predictable bias given our view of humans in relation to the outside world.
What are these ambiguously "higher" centers of conciousness that are most advanced in humans? I think it is succinctly described as the following: analogical reasoning. The capacity for not just associating relations between things, but the capacity for relating relationships.
What makes humans unique is our massive capacity for analogical reasoning. Humans have by far the highest capacity for analogical reasoning in the animal kingdom, with bottle-nose dolphins coming in a more distant second and great apes third.
My thoughts instantly go to Deleuze & Guattari's rhizome Idea in A Thousand Plateaus. Why wouldn't our brain run small connected pieces of software that together emulate how the world works and especially how other people are expected to act/behave. This way the network of networks can run simulations. Also, when new bits of information arrive - see Dawkins' communication-is-influence idea - the software gets updated.
I haven't read the paper thoroughly (http://www.pnas.org/content/early/2010/07/28/1009112107), but looks like the authors are using a strategy of systematic and tightly circumscribed injections of anterograde & retrograde tracers to map out, with unusually high precision, the inter-areal projections for a particular circuit they're interested in. This mapping of inter-areal connectivity is a little different from what I'm interested in, which is to understand how the many-to-many anatomical wiring diagram between individual neurons in a neuronal network relates to the transformation of information carried out by that network. In other words, they are mapping connections between areas, but I think it's likely that an additional level of resolution -- the mapping of connections at the level of individual neurons -- will likely be necessary (and probably insufficient) to crack brain circuits.