[From Bill Powers (2011.07.25.1940 MDT)]
Gavin Ritz 2011.07.26.12.01NZT –
BP earlier: If they represent only discrete values, whole numbers,
then the whole problem of Diophantine equations has to be dealt with. But
if the phenomena are basically continuous, there’s no problem. Neural
firing frequencies are continuous variables; they’re not quantized even
though the individual impulses are. So the Diophantine equations are
irrelevant for dealing with neural signals.
GR: Okay, I dont understand why you say neural firing is continuous. The
output variables in PCT look discrete to me, are they
not?
The frequency of neural firings, the main dimension in which a
neural signal can change, is continuous – that is, it doesn’t
necessarily jump suddenly from one frequency to a different frequency,
but can change slowly and smoothly, gradually increasing from one moment
to the next.
The output variables in PCT are also continuous for the most part, though
some are not. A muscle tension, for example, changes from one amount of
force to a different amount, passing through all the intermediate amounts
of force rather than jumping from one to another. An arm or leg changes
from one position to another, going through all intermediate positions –
the limb doesn’t disappear from one position and reappear instantly in a
different position, which is what a discrete variable would do.
GR: Im going to
delve into this a bit deeper to see if I understand it a bit better. The
variables from input to outputs and the functions of the Diophantine
Computer Equation look very similar to the output variables and input
variables of PCT equations to me.
Yes, they do, but they’re continuous variables. Discrete variables can
change instantly from one value to a different value without going
through all the intermediate values, which is what is meant by
“discrete.” When you look at the equations, however, there is
nothing to indicate what kind of variables make it up. That information
is given separately as a condition on the solutions that are acceptable.
The condition that makes the equation Diophantine is that all variables
must be expressible as whole numbers, without any fractional numbers
between them at all.
In Diophantine equations, we have things like
x^2 + y^2 = z^2, which reduces to
z = square root of (x^2 + y^2).
You might think that
if x is 2 and y is 3, z would be equal to the square root of 13,
but the square root of 13 does not come out to an exact integer (as the
square root of 16 would) so it’s not a Diophantine solution, which allows
only whole numbers. However, if the same variables are stated to be
continuous, then the value of z is the square root of 13.000… or
3.60555. Those numbers to the right of the decimal place are
available only if this equation is in the domain of continuous variables,
where you can have all values between 3 and 4.
Best,
Bill P.