The neuron as an analog NOT digital system

[Joseph Strout]

On 8 Aug 1995, Marc Anton wrote:

>      Though we tend to think of the neuron as essentially
> a digital processor (with its on and off firings) it's really
> closer to an analog processor with infinitely complex chemical
> and electrical functions above and beyond anything capable with
> digital reductionism.

I don't think anyone involved in neuroscience treats real neurons as 
purely digital processors, as you suggest.  There are at least two 
reasons why you may have gotten this impression:

1. Action potentials (neuron "spikes") are indeed all-or-none over most 
connections.  They are transmitted as a regenerative wave of potential.  
If the signal is too weak, it will quickly die and not travel far at 
all.  If it is just strong enough, it will be regenerated at each point 
and will propagate to the synapse.  If the signal was very strong, the 
result is exactly the same.  (There can be a *little* variance in action 
potential amplitude and shape, but not much.)  However, the real 
processing in neurons takes place in the dendrites and soma, where inputs 
from various synapses at various times are combined.  This is a complex, 
nonlinear, and analog function of space and time, and this is widely 
recognized.

2. Some researchers work with abstract "artificial neural networks" whose 
structure and operation are inspired by real neurons.  Some of these may 
be digital (indeed, they are almost always implemented on a digital 
computer).  But these are only cartoon sketches of real neurons, leaving 
out nearly all the complexity and detail found in a biological system, 
while retaining the bare minimum of properties capable of supporting the 
complex behavior of interest.  And again, I think any researcher in the 
field would readily acknowledge this.

,------------------------------------------------------------------.
|    Joseph J. Strout           Department of Neuroscience, UCSD   |
|    jstrout at ucsd.edu           http://sdcc3.ucsd.edu/~jstrout/    |
`------------------------------------------------------------------'