The nature of illusion

[Martin Taylor 2018.]

  This message was inspired by

but is not a response to that message. Here's the bit from which I

am taking off:
The speed-curvature power law is an observation, a perception
produced in some analyst by an analytical process. It’s no more than
that. As Bill Powers never tired of reminding us, perceptions are the only
truths we have of which we can be sure. They are what they are,
nothing more and nothing less. But to say that a perception is an
“illusion” implies that the perception is untrue. What could that
possibly mean, if our perceptions are the only truths we can ever
As I see it, to say that a perception is untrue is meaningless if
only that perception is considered in isolation. But if one
perception of the world is inconsistent with some other way of
perceiving the same attribute of the world, the they cannot both be
true. Consider the Mueller-Lyer illusion (an interesting
juxtaposition of names in the current US political climate:-). Most people perceive the distance between the arrowheads at the left
to be less than that between the end dots in the middle, and still
less than between the “feather-tips” on the right. That’s one set of
three relationship perceptions (left<middle, middle<right,
left<right). There’s another way of perceiving distance relationships, by
comparing each with something else, such as a ruler. Only when “the
same” perception can be produced by two different routes with
different results can it be said that one is an “illusion” – not
properly representative of what is controlled by the actions that
influence it. You can set up an experiment with the Mueller-Lyer
figures and get someone to control the relationship so that the
“arrowheads” and “feathers” patterns look the same, but when that
person compares them against a ruler, or perhaps by haptic
exploration, the left will be perceived to be a longer distance. Another kind of illusion is the mirage. Someone in the desert may
perceive a distant lake, or on a hot day a pool on the road ahead,
and that perception is a truth. But when a different sense is used,
you can’t drink from the lake or splash on the road. Here what is
compared are two different attributes that “ought to be” associated
and are not. Every lake in our past experience has contained water,
one might say “by definition”, and has a shimmery reflective surface
when calm, but one can perceive a lake without putting a finger into
the water. Only when the perceived lake turns out not to contain
water does the “untruth” of the perceived lake turn the mirage into
an illusion. (Yes, the vanishing of the lake as one moves is also an
indication that it is an illusion, but that is of the same kind as
the one-attribute Mueller-Lyer illusion).
What of the power-law of curved movement? That is a perception, an
observation made by mathematical operations on measured data. There
have been no distinct ways of determining the “truth” of that
perception, so there is no basis for determining whether it might be
an illusion. But there is a related illusion, which is that when a
movement around an ellipse is perceived to be at a constant speed,
the measurements indicate that it slows around the sharp ends as
compared to the flatter sides. That’s an illusion of the same class
as the Mueller-Lyer illusion – one attribute perceived by two
routes with different results.
If you accept Bill’s assertion that your perception is the only
truth you can ever have, you cannot say that an illusion is a
perception whose value differs from the real-world truth. You can
say that if there is a real-world truth and you have a variety of
ways to perceive it, then if these ways produce different perceptual
values, there is somewhere an illusion. But it takes another
perceptual function that produces a perception of the relationship
among these different ways of “perceiving the same thing” to
perceive “illusion”, a deviation from a reference value of zero for
the difference among the “different ways” results, if one is
controlling for “truthful perception”.
So what are we to make of the concept of “behavioural illusion”?
This “illusion” is a pejorative comment on a non-PCT theory that
purports to account for a particular observation. In this case, the
two routes to a perception of “the same thing” are applications of
abstract theories, one of which happens to be PCT. Call the
observation “X”. According to PCT “X” has one cause, whereas
according to the other theory, “X” has a different cause, though
both agree that the observation is indeed “X”. So the conflicted
perceptions that prove an illusion exists are of “The Mechanism”
that produces observation “X”, exactly as is the case when the
observation is of a shimmery area that reflects the sky, and one
mechanism for this is that there is a lake, while the other is that
the surface is hot and light rays are bent to appear as though there
is a reflective surface.
According to PCT, the relation between the effect of a disturbance
on a CEV is directly opposed by the output of the controller acting
through the part of the environmental feedback path between output
and the CEV. If control were perfect, this opposition would be
exact, and if there was a consistent relationship between the output
and the disturbance, it would be due to consistency in the
environmental feedback path. The other theory claims that if there
is any consistency in the relationship, it is due to consistency
internal to the entity that PCT calls the controller. With only this
information, the proponents of either theory would be entitled to
call the other claim and “illusion”, neither having evidence to the
contrary, though both would perceive that an illusion existed
because both could not be true at the same time.
Unfortunately, with only this information, the observations could
not distinguish the theories. Any distinction must come from adding
more relevant perceptions into the mix. For example, completely
different processes lead one to perceive that if the relationship
between output and disturbance is sufficiently consistent, one can
use the techniques of mathematics to say that the output effect is
the negative of the disturbance, so there is a functional
relationship (output)=f(-disturbance). Very little more math is
required to be able to perceive the implication that the existence
of a loop through perception to output implies that the functional
relationship that expresses the consistency through this part of the
loop is (output) = g(disturbance) and that g(.) is the inverse
function f(.), so that (output)=f(-disturbance).
So where does the “second perception of the same thing” come from
that might allow a decision as to whether the main consistency is
f(.) or g(.)? It comes from experiment. Change f(.) to f*(.) and see
whether g(.) continues to be f(.) (Powers’s Behavioral
Law) or changes to f*(.). We know what happens – g
changes. Hence the “other theory” is the illusion. Nevertheless, it
remains true that the internal processes of the entity must be able
to perform f(.) and possibly f*(.). The
experiment does allow us to perceive something of the internal
working of the entity.
The “second perception” could also come from analysis (and usually
does). Since the entity is unable to affect f(.) but might be able
to affect g(.), the reason g(.) is f(.) must be caused
by g(.) adapting to f(.) rather than the other way round. As a
mechanism, PCT allows for this way of maintaining consistency,
whereas the other theory, on the face of it, does not, since it
would imply that f(.) would adapt to become g(.) and no
mechanism for that is proposed – though it is often possible for an
entity (person, animal) to deliberately alter the feedback path in
order to ease control, something done by quite other controllers
inside the entity.


On 2018/08/11 3:44 AM, Adam Matic
( via csgnet Mailing List) wrote:

[From Adam Matic]

        You say that trajectory is the controlled variable, but

that the speed-curvature power law is a side effect, …

        The meaning of "speed-curvature power law" is that there

is a regularity in the trajectory of an object…The
“behavioral illusion” from the 1978 paper is named that to
contrast “the behavioral law” as SR correlation…