Seeing similarity and dissimilarity

[Martin Taylor 2014.01.11.14.51]

One of the reviewers of the book ridiculed the notion, by pointing

out that this implied that there should exist “sinister” people with
RH lesions who could not see jokes, since jokes depend on disrupting
an apparent similarity of the joke situation to the normal
situation. But, as was mentioned in the book, such unfortunately
literal people do indeed exist.
Perception of similarity is not one of the HPCT levels. Nor is
perception of dissimilarity. But they are types of perception. You
can put piles of mixed-up stuff into boxes of what you perceive to
be “the same kind of thing” without worrying about the fact that all
the items are actually different. That’s category perception. You
can create metaphors and similes all over the place, and their value
depends on the reader/listener perceiving what in the topic is like
and what is unlike the analogy. Similarity seems to me to be a
perception of how much alike two things are, whether they are
events, intensities, programs, or systems. Dissimilarity is a
perception of how unlike they are. Categories are distinct because
of the differences in “kind” among them. Metaphors are appropriate
because of the likeness between the topic and the metaphor.
Logically, the two measures ought to be inverses of each other, but
it seems that the brain does not treat them that way, and does not
even treat them in the same part of the brain. The canonical control loop in PCT has always compared one perceptual
signal value to a corresponding reference signal value. The output
of the comparator is an error signal value, which (again speaking
logically) should be a measure of dissimilarity. Might there be
separate control loops in which there is a function equivalent to a
comparator, but in which the signal corresponding to the error is a
measure of similarity? Or do the similarity and dissimilarity
perceptions refer only to relations between vectors of controlled
perceptions?
To me, this issue seems to relate to the control of avoidance and
the perception of “not”. If you say “The boy did not give the book
to the girl”, what actually happened? Did he give her a flower? Did
he sell her the book? was it not a girl to whom he gave the book? If
you instead say “The boy should not give the book to the girl” you
are controlling for avoidance of the situation in which he does give
her the book. But what dissimilarity of situation is the reference
state that you want to perceive? Is it necessary that the boy
doesn’t communicate with the girl at all? Is it sufficient that he
gives her a different book? What is the reference value that is
satisfied by the perception of what the boy does? And how different
from the avoided value must the perceived situation be in order to
satisfy that reference?
In the Crowd demo, the avoidance problem is finessed by controlling
for “proximity”, the inverse of physical distance, with a threshold
and a zero reference value for perceived proximity above the
threshold (if I remember correctly). This doesn’t seem to me to be a
general solution to the problem, though it may point the way to one.
I consider this issue to be one that future development of PCT
should consider. It should be possible to devise experiments to ask
people to control similarity and to control dissimilarity, but at
the moment I have none in mind. Maybe the control of avoidance is a
different problem with a simple solution, but at the moment the two
issues seem to me to be closely linked.
Martin

···

I’d like to get back to enquiring about a topic I raised in a
message under a less apposite subject line on New Year’s Eve.
Unfortunately, Rick [From Rick Marken (2013.12.31.2210)]
immediately steered the thread away from the question of the
different ways the brain deals with similarity and dissimilarity,
as follows:

    I propose that the question we address

should be something like this:

    What is the value of seeing similarity of non-PCT research

and/or theory to PCT research and/or theory?

  Now I want to get back to the issue I originally raised, of how

PCT should deal with perception of similarity and dissimilarity.
I’m not offering solutions, but suggesting a puzzle that might
have an obvious solution I haven’t considered.

  I quote the relevant part of my New Year's Eve message below.

[Martin Taylor 2013.12.31.12.40]

  Although they seem logically to be opposites, seeing similarities

and seeing dissimilarities are probably two different functions,
located in different places in the brain. At least, when my wife
and I were writing our 1983 Psychology of Reading (which one
reviewer called "the best book on reading since Huey, 1908) both
the then-known physiology and the psychology seemed to say so. To
read well, we thought that one had to be good at both seeing
similarity and seeing dissimilarity. Much of Chapter 11 is on that
topic under the heading “the Bilateral Cooperative Model of
Reading”.

  One example of problems that come with a right-hemisphere temporal

lesion is that the injured person more accurately sees
dissimilarity than does a normal reader, whereas someone with a
lesion on the corresponding part of the left hemisphere more
easily sees similarity and misses differences. Here’s a figure
from our book:

  The "control" (i.e. apparently normal) subjects can handle

misprints and malapropisms to get the sense of what the writer
intended better than can the reader with a RH temporal lesion. The
RH lesion reader would be likely to say “that’s not a proper word,
at least not one that makes sense in this context”, whereas the LH
lesion reader would be likely to say “I know that’s what they
mean” without properly reading to see whether “that” is actually
excluded by what was written.