[Martin Taylor 940830 12:30]
Avery andrews 940830.1120
Hmm. So may problem is exactly what and where is the CEV defined by
a PIF: the output of the lower-order PIF's feeding into it? the
pattern of sensory stimulation impingeing on the organism? the pattern
of energy flowing into the organism, the external circumstances causing
that pattern? I see the PIF as responding to properties of all of these
things at once, and therefore a signalling multiple CEVs.
I'm not sure my earlier posting of today in response to Bill P. answers
this, so here's my take on it. Bill P. says that there is no environmental
concept corresponding to the CEV, so he will differ:
The PIF is a function of its sensory inputs. They could be connected
to anything, so far as the ECU is concerned that contains the PIF. The
PIF function outputs a value that changes moment by moment as the values
on its input lines (its arguments) change. It is a scalar function of
scalar arguments. If those arguments happen to be the outputs of sensor
systems, there are in the outer environment some physical variables whose
values determine the sensor outputs (always remembering that sensors
adapt and affect one another laterally, so that the relation between the
environmental variable and the sensor output is time-variant). The
output of sensor Sk is a signal 0sk, which is some function of environmental
(physically measurable) variables 0V1...0Vn. That sensor function is usually
weighted heavily on just one physical variable, 0Vk, and we say for
convenience that the sensor provides an intensity estimate of the single
physical variable. Of course, the sensor might well perform a time
derivative or a spatial derivative, but whatever it provides, it is a
function of physically measurable variables.
If the PIF, 1Pq, has as its inputs 0s1...0sn, then the CEV is DEFINED as
the same function 1Cq == 1Pq, but its arguments are the PHYSICAL variables
rather than the sensor outputs. (I use == here to mean "is equivalent to").
If the PIF computes the difference between sensor intensities 0s1 and 0s2
(1Pq(0s1, 0s2) == 0s1-0s2) then if the sensors measure the intensities
of physical variables 0V1 and 0V2, the corresponding CEV is 1Cq == 0V1-0V2.
It does not "exist" in the outer world, any more than does any other
function of 0V1 and 0V2, such as the 0V1'th root of 0V2. But it is a
function significant to the perceiving organism, and "exists" in the
organism's perception. It could be measured in the real world by anyone
who happened to know that the defining PIF was 0s1-0s2 and that the
sensors measured the intensities of the physical variables. And it could
be disturbed by anyone who wanted to apply The Test to see whether it
corresponded to a controlled perception. Nobody can disturb a perception,
except by disturbing the value of the CEV that "exists" in the real world
courtesy of that PIF.
Now go up one level. Each of the level 1 PIFs defines a variable in the
real world. The variable's value depends on the values of possibly many
physical quantities and possibly the histories of those values. The
variable could, in principle, be measured or computed, if the PIF defining
it were known. So we can treat those variables as if they were physical
variables. They are the inputs to the next level of PIFs, 2P1...2Pn. The
same argument repeats. The CEV for a PIF is defined only once, not many
times.
If a PIF P == s1-s2, where s1 and s2 are its only two inputs, then the
value of its output can remain constant, no matter how one of its inputs
changes value, provided that the other changes to compensate. So while it
is true that the value of a perceptual signal "responds" to the values
at all lower levels, it is determined only by the specific relation among
those values determined by the PIF. One PIF, and the value of its output
corresponds to one and only one "thing" in the outer world, the value
of the corresponding CEV function acting on its (derived) physical variable
arguments.
> but you should perhaps make it
>more clear that any ESV represents a fact of the real world that is not
>accessible to the organism unless there is a corresponding PIF.
This isn't quite right, since typically the CEV's approximate the ESV's,
and various kinds of impairment, failure and death result when the
approximation isn't good enough (as is often the case with lower
organisms, and sometimes, it seems to me, with higher ones, such as
whales beaching themselves, or humans setting up genocidal social
systems). The most important point is that the ESV is an aspect of
an *ecological* analysis, of practical to conservation biology, but
not really relevant to psychology.
There's no contradiction here. I said that the ESV is not accessible
to the organism unless there is a corresponding PIF. I didn't say that
"what you don't know can't hurt you." There are lots of things out there
that you cannot perceive that can hurt you fatally. High intensity
gamma radiation, for example. That's an ESV for which none of us have
a PIF. It kills. What you say about ESVs is right on target.
Martin