Connectionists: How the brain works

[Fred Rothganger]

This thread is intriguing. Here are my thoughts on some of the questions 
that have come up:

* Why create a model of how the brain works?
     For me, it is mainly curiosity. If we can describe how the human 
mind works, we will make a great leap towards understanding ourselves. 
This is much the same motivation as Psychology, but in a more generative 
sense.
     I believe that the human soul is a physical phenomenon, and that it 
can be modeled and computed. I believe in strong AI, that is, it is 
possible to build a functioning human mind in a substrate other than 
living neurons. These are merely beliefs. They are difficult, if not 
impossible, to prove.
     It is not sufficient to write out a model. We must build one and 
see what it does. In practical terms, the end-point of brain modeling 
will be a sentient machine.

* Equations
     My current expectation is that a set of coupled differential 
equations will be sufficient to express a complete model of the brain. 
This will likely not be a small set.
     The equations need not be at the scale of membrane dynamics and the 
like. We may eventually understand higher-level processes and express 
those directly. (For example, the famous ART algorithm may be expressed 
as a set of differential equations.)

* Simplicity
     A large set of equations is impossible for a single human mind to 
understand, certainly not in a single instant. Humans are notoriously 
bad at predicting a dynamical system with even 2 or 3 variables. 
(Example: A faucet is pouring into a tub. The tub drains at rate 
proportional to the water level. What happens to the water level in the 
tub?)
     You may examine and understand any small part of the equation set 
you wish. To understand the system as a whole, we need the aid of 
computers. Computers help us keep track of all the moving parts and 
analyze how they fit together. This includes simulation, but also other 
forms of analysis.

* Community sharing
     Jim, I admire what you are doing with GENESIS. In general, 
Neuroinformatics is crucial to our success, because it provides a 
scaffolding on which all our disparate work can be combined into a 
single model of the brain. To succeed, we must go beyond simple model 
sharing, to models that connect with each other and some larger 
structure. NIF is helping in this regard, by organizing the work around 
an ontology.
     My small effort in this area is called N2A 
(http://journal.frontiersin.org/Journal/10.3389/fncir.2014.00001/abstract). 
I wish it were better developed, so you could see the potential scope we 
envision.