[ExI] The digital nature of brains (was: digital simulations)

Eric Messick eric at m056832107.syzygy.com
Fri Jan 22 18:18:04 UTC 2010


Gordon writes:
>PS. Like most everything in the natural world including apples, the
> human brain appears to be a non-digital object.

This is only true in the very narrow sense in which a 74LS00
integrated circuit is a non-digital object.  The digital nature of
neurons is one of the things which makes them most useful as
information processors.

The key characteristic of digital circuits is that they quantize an
analog signal into discrete values.  That 74LS00 accepts a wide range
of analog voltages on it's input pins, but the output is insensitive
to small changes in the input voltage.  It divides the input signal
space into low voltages which are considered zeros, and high voltages
which are considered ones.

The action potential of a neuron looks more like the output of a
digital logic gate than like the output of an analog circuit.  It's an
all-or-nothing response.  When the integrative portion of the neuron
crosses a threshold voltage the output voltage of the neuron spikes in
a pulse which is relatively uniform and insensitive to small changes
in the input voltages.

Successive gates in a digital circuit design restore the voltage level
of signals to their discrete values, preventing losses from
accumulating during multiple stages of processing.  Successive neurons
in a brain restore signal levels in much the same way, and enable
multi-stage processing via the same mechanism.

This is an important way in which the brain acts very much like a
digital computer.  It is not incidental to the functioning of the
brain, it is fundamental.  It's also one reason why analysis of the
brain at the neural level makes sense, and the details of what happens
inside each neuron can be abstracted.

Another reason why it makes sense to draw an abstraction boundary
around neurons is that there is already a natural boundary around
neurons: the cell wall.  Most of the interesting behaviors of neurons
happen because the cell wall is only selectively permeable to various
molecules and ions.  That barrier reduces the complexity of the
interactions which can occur across it, so it is a perfect place to
put an abstraction layer.

As far as digital simulations and copies are concerned, consider the
following substitution:

>For example you might create a digital simulation of an apple on your
> digital computer. Your digital simulation of an apple will appear
> very much like a real apple, but you will find it difficult to
> eat. The reason you cannot eat that apple should be pretty obvious:
> it's not really an apple. It's merely a digital simulation of a
> non-digital object.

For example you might create a digital simulation of a thermostat on
your digital computer. Your digital simulation of a thermostat will
appear very much like a real thermostat, but you will find it
difficult to use it to regulate temperature. The reason you cannot
regulate temperature with that thermostat should be pretty obvious:
it's not really a thermostat. It's merely a digital simulation of a
non-digital thermostat.

Most thermostats installed today are digital simulations of analog
thermostats.  They manage to get the job done anyway.

And yes, a digital simulation of a person should enjoy eating a
digital simulation of an apple.  If not, it's not a very good
simulation.

-eric



More information about the extropy-chat mailing list