e-mail I sent to Cosmo about varying stimulation
2014-08-29azim58 - e-mail I sent to Cosmo about varying stimulation
Hey Cosmo, thanks for telling me about "Linked"! This book looks really
interesting, and I have added it to my audible.com wishlist. I think I
will most likely choose it as the next book I listen to.
One thing I am curious about is if the optimal way to challenge/stimulate
a system is through a power law type regimen. Perhaps adaptive systems
often come to resemble their environment. For example, the genie feral
child who grew up for 13 years strapped to a chair in LA had a very dull
environment and as a result her mind and body also seemed to distill into
a dull and/or deformed form. On the other hand, an adaptive system that
exists in a stimulating environment will develop to be capable of
responding to all of the stimulation. Perhaps too much stimulation is a
bad thing though. For example, professional weight lifters do not lift
the maximum possible weight that they can all of the time. They lift lots
of light weight, then a heavier weight fewer times, then a heavier weight
even fewer times, and then lift almost as much as they can very few
times, and then work their way back down again. Perhaps the optimal
number of reps and gradations in weight would follow some type of power
law or other pattern.
If one could discover the optimal way of stimulating a system perhaps
this knowledge could be used to improve all kinds of systems from
muscles, to minds, to digestion, to the immune system, to societies, to
any other complex adaptive system.
Just some thoughts.
===========================================================================
Cosmo's response
You constantly amaze me how quickly you pick up new ideas ;)
Follow the scale free stuff. Very fascinating field. His book discusses
the reason power law distributions arise is due to simple underlying
rules. Theres a field in math dealing with random graphs, and how they
are built. It turns out that it is very simple to produce scale free
graphs in the computer with very simple rules. Barabasi then discusses
that these simple rules are what give rise to complexity in the world.
The basic underlying premise is that the 'Rich get richer'. If your
thinking about the internet for example, its the popular websites out
there with lots of incoming links that have the highest rate in which
more websites link to them! Get it? If a site has more links pointing at
it, YOU have a higher chance of randomly finding it, and therefore have a
higher chance of linking to it from your own website. Therefore, the
websites that are the richest will get even richer. This also has
interesting implications that websites that are first to form have an
advantage.
If you wrote a simple program to hop around any initial graph, and then
grow that graph based on nodes that you randomly land on, then these
power-law distributions will arise. Iterate many times and you get a
graph with those scale free properties. Its fairly similar to the
emergent properties of fractals.
Lets also consider social networks. People in any social network that
have the greatest number of friends will likely meet new people at a much
quicker rate. I wonder if anyone has studied this as a timecourse.
The same princicples likely apply to biological networks. It could be
that signaling proteins and pathways that are the oldest, with the most
links, will develop more connections and links during evolution. It could
be that when life evolves, that it's just naturally easier to bolt new
signaling pathways onto existing framework. Its like a shortest path
thing. If you randomly find yourself with a new GPCR that senses some
novel analyte, its easier to link that signaling with an existing pathway
in the cell then evolve 10 more proteins to create a new dedicated
pathway. Hence biological networks clearly have proteins that are hubs
that link together several other pathways. I imagine these hubs are
probably the first to evolve in evolutionary history!
Ive always wondered if you could somehow combine phylogenetics and graph
theory to see if hub proteins in networks are also proteins with the
longest evolutionary history. There might be some neat way to use
molecular clocks, or just measure the degree of conservation in a given
gene and somehow correlate that with the genes "connectivity". I asked
Kumar about this once, but he isn't such a fan of Barabasi's work.
Anyways, neat book. It really shaped my view of the world when I read it
years ago.