Slashdot Mirror
Adult Brains More Flexible Than Previously Thought
stemceller passed us a link to the official site for Johns Hopkins, which is reporting on some research into cognition. Generally, doctors have understood our best learning to be done at a young age, when the brain has a 'robust flexibility'. As we get older, our brain cells become 'hard-wired' along certain paths and don't move much - if at all. Or, at least, that was the understanding. Research headed by the hospital's Dr. Linden has taken advantage of 'two-photon microscopy', a new technique, to get a new picture inside a mouse's head. "They examined neurons that extend fibers (called axons) to send signals to a brain region called the cerebellum, which helps coordinate movements and sensory information. Like a growing tree, these axons have a primary trunk that runs upward and several smaller branches that sprout out to the sides. But while the main trunk was firmly connected to other target neurons in the cerebellum, stationary as adult axons are generally thought to be, 'the side branches swayed like kite tails in the wind,' says Linden. Over the course of a few hours, individual side branches would elongate, retract and morph in a highly dynamic fashion. These side branches also failed to make conventional connections, or synapses, with adjacent neurons. Furthermore, when a drug was given that produced strong electrical currents in the axons, the motion of the side branches stalled.'"
There's a branch of neural net studies that focuses on a technique called entropic topography. Essentially, it involves random evolution of just the fringes of a digital neural net. That is, much as this John-Hopkins study has found, a rigid core is kept. It is only the neural subnets branching off that undergo synthesis and morphing.
While there are various deterministic algorithms that are used to evolve neural nets, it's only recently that we've begun seeing randomness used. This has an added benefit of bringing in unexpected mutations, which really don't happen with the deterministic algorithms.
Some advances from the study of Lei topographies have also lead to breakthroughs recently, where some of the more complex, yet deterministic, algorithms have had entropic terms introduced in order to bring in an element of randomness. These neural nets are probably the closest to the human brain, as they introduce the random mutation that is so prevalent within the human species, while also following the constraints of this new-found core neural path.
I hate this view that some how results of tests on animals don't apply to humans at all. It's simply not true, almost every major medical advance has been tested or researched on animals like mice first. the simple fact is mammals bodies all work in very similar ways.
Having worked in a lab (disclaimer: not as a scientist) I learned that there are loads and loads of promising treatments for cancer and such that work great in mice, and never translate beyond. Even a casual glance at immunology from a layman's perspective reveals your statement to be utter bullshit; there are many, many diseases and afflictions that are species specific, sometimes highly so.
Anyway...it is entirely plausible that this ability to re-purpose brain cells is a plus for mice in survival/adaptation, where they have very little brain capacity at their disposal. We have loads at our disposal, and tend to build a lot of generally useful knowledge..ie, we build tools, literally or figuratively, and apply those 'real' tools or knowledge/skill 'tools'. Mice do not do either. We're more "general purpose", so maybe we don't *need* the ability to re-learn, since our learned skills are so broadly applicable in a survival sense.
Please help metamoderate.