I’ve heard it explained that the wrinkles increase surface area and so…smarter. Now shut up and eat you broccoli.
So maybe the more appropriate questions (with the mandatory caveat of: because evolution, dumbass–there is no intent, only results that increased fitness):
More surface area = what? What’s surface area got to do with braining?
More surface area increases distance between points. Seems like a solid sphere would be more economical and efficient, but evidently the “effort” of evolving wrinkles yielded better results. Explain benefits of wrinkles vs. smooth brains.
For whatever evolutionary reason, the most active areas of the brain seem to be near the surface. (Kind of like potatoes have the most vitamins near the surface, I guess.) The interior of the brain is packed full of the long-distance wiring and trunk lines – the nerves that connect various parts of the brain to one another. (These interior connecting nerves, by the way, are the ones that get damaged in Multiple Sclerosis patients. Everything I think I know about the brain, I learned while reading library books about MS.)
So, as brains got smarter, they got more convoluted, increasing the surface area where the “braining” happens, the better for us to understand such convoluted topics as Relativity, Quantum Anything, Economics, and International Relations.
I think the natural course of evolution missed a chance here. If brains had evolved into four dimensions, there would be all the space we could ever need. Four-dimensional geometry (even a simple 4-D cube or “tesseract”) is infinitely more spacious than 3-D geometry, in some sense, just as 3-D geometry is infinitely more spacious than 2-dimensional plane geometry.
Each of these basic processing units is six layers deep, and is surrounded by six neighboring columns.
Maybe there could be other way of organizing neurons to process information, but this is the basic unit that evolved and increasing capacity was accomplished by increasing their numbers by increasing the surface area for them populate. Positioning them to be able to transmit information efficiently (with white matter tracts as short as possible) was also important but secondary.
Take that “maybe” back. There are other ways. Several species of birds (corvids for example) are remarkably intelligent and with relatively small brains. They are organized not in that same columnar fashion but as clusters of nuclei. Octopuses also are quite intelligent with a completely other solution - they have a central brain but also peripheral ones for each arm and for the eyes that do much of the processing locally.
But we had to work with what we evolutionarily had to work with.
It’s fun to watch a trained octopus do its tricks. I saw a video once – they put a divider in the tank with several holes in it. Then they trained the octopus to go through the divider to the other side when a light was shined.
All the legs tried to go through different holes, whatever hole each leg came across first. Eventually, when the head got through one of the holes, all the other legs followed.
This reminds me of one of the problems in high performance computing. Scalability of compute is limited by communication between the computing elements. The precise amount of communication you need versus the amount of computation varies with the problem at hand, but in general, eventually scalability becomes limited by the ability of the units to communicate. Communication comes as the bandwidth (amount of data per unit time) and latency (time to transmit the data down the link.) This has obvious parallels with the brain’s organisation. You get bandwidth with more interconnections (which chews space) and your latency goes up as the distances increase (which happens when there isn’t the room to get interconnected unit close together.
Eventually you are limited by the physical dimensions you have to fit the mix of compute and communication into. Older supercomputers had the luxury of a relatively small number of compute elements, and some were able to have very high degrees of connectivity (CM-2 with a 10 dimensional hypercube, CM-5 with a “fat-tree” that provided constant bandwidth and only slowly increasing latency.) But eventually things outgrow such luxuries, and you are left with interconnects that fit in 3 dimensions, and these tend to be limited to three dimensional topologies. The entire game becomes one of tweaking the topology in such a way as to balance the mix of compute and communication in a physical space to effect the best performance. Bird brains seem to have reached a particular optimum, but I suspect they cannot scale it. Ours is less efficient (perhaps) but allows greater scalability.
But remember what the 4th dimension means. What good is it if the result of your computation about the jumping tiger happens in the brain ply that’s 3 minutes after (or before!) you see the tiger. Either way is not conducive to survival.