Chapter 13








Associative and Dissociative Networks in Cognition












The modern conception of a monadic mind in terms
of cogito was introduced by Descartes. Being
aware of the double nature of the sensory organs
of the brain, he had to suggest a seat for the
unity of the soul:




"...the other parts of the brain are all of them
double, just as we have two eyes, two hands, two
ears, and finally all the organs of our outside
senses are double; and inasmuch as we have but
one solitary and simple thought of one particular
thing at one and the same moment, it must
necessarily be the case that there must somewhere
be a place where the two images which comes to us
by the two eyes...can unite before arriving at
the soul, in order that they may not represent to
it two objects instead of one....there are no
other place in the body where they be thus united
unless they are so in the gland."(Descartes,
1649)3




His assignment of this unifying role to the
pineal gland was soon discarded. But the more
general mind-brain problem that Descartes raises
remains a controversial issue. Most of the time
one feels


oneself to be of a single mind, not divided,
irrespective of the facts of asymmetry and
partial complementarity between the two cerebral
hemispheres revealed by brain studies. Even when
Sperry and others have uncovered the mediating
function of corpus callosum one continues to ask
about how coherence of visual processing comes
about. We may still ask with Descartes: Why is it
that we (normally) see one image, not two?


He raised this question in "Passions of the Soul"
in 1649. Centuries later the almost opposite
question was raised with reference to figures
inviting visual ambiguity such as Necker's cube,
Rubin's vase, or the rabbit-duck figure: How is
it that we are able to shift between two rival
images of the "same" form? How does such
discriminational coherence and alternating come
about? What are the processes involved in
coherent perception and understanding? Is
conscious attention a unified or divided activity?


Such questions are pursued in cognitive
psychology in terms of schemata (Neisser, 1967,
1979), in psychobiology and neuroscience in terms
of emergent properties and downward causation
(Sperry, 1965, 1969, 1988) and cortical
self-organization (Singer, 1990), and in general
psychology in terms of association and
dissociation (Hilgard, 1986. A concern with
self-organizing systems and associative networks,
first proposed by James and Hebb may also be
recognized in neurocomputational modelling
(Fukushima 1975; Malsburg, 1973; Rummerhart and
Zipser, 1985). In the following I shall indicate
how such questions are dealt with in terms of
different frames of reference, refer to
neurocomputational explorations of visual
ambiguity, and relate to the thesis of the mind
as a self-organizing dyad involving the
participation of a virtual other. Towards the end
I shall indicate how the thesis may account for
the "hidden observer" disclosed and discussed by
Hilgard in terms of dissociation theory, and how
it relates to a controversial issue on the nature
of consciousness brought about by the studies of
"split-brain" paitents.




Neurocomputational approaches




A pioneer in the field of neurocomputation,
Warren Mcsuper4




Visual ambiguity




In his later work Malsburg has proceeded to
attack the problem of how neurons distributed
according to orientation preference in spatially
separate columns can give rise to vision of a
coherent figure, separated from a background, and
the even more difficult problem of separating
different superposed figures from a scene that
invites visual ambiguity. He proposes that a
coherent image comes about through temporal
coherence of the firing of neurons, exhibiting
synchrony in spite of their locations in distant
columns. When there are two superposed figures,
then cell assemblies coding for the different
figures should be expected to be activated in
alteration (Marlsburg et al., 1978, 1986; Singer,
1990).5 Later, a research team headed by Singer
found that neurons in the visual cortex of cat,
when responding to stimuli conforming to the same
figure, exhibit oscillatory responses and appear
to fire in synchronization across even distantly
located functional columns.6






Neurocomputational nets and classifiers facing
visual ambiguity




Visual ambiguity in the form of the famous cube
presented by L. A. Necker (Fig. 13.1 (i)) in 1832
has been subject to much discussion and many
attempts in terms of connectionist modelling and
neurocomputation in terms of parallel distributed
processing. Rummelhart and Zipser (163-164)
review the type of multi-layered architecture of
the competitive learning mechanisms usually
developed and used in parallel distributed
processing approaches:




- The units in a given layer are broken into
non-overlapping clusters. Each unit within a
cluster inhibits every other unit within the
cluster. The clusters are winner-take-all, such
that the unit receiving the largest input
achieves its maximum value, while all the other
units in the cluster are pushed to their minimum
value (Grossberg)




- Units may be active or inactive




- A unit learns if and only if it wins the
competition with other units in its cluster. A
unit learns by shifting its weights from it
inactive to its active input lines etc.




In his computational approach to perception Marr
suggests the Necker cube invites a neural network
with two distinct stable states. This is returned
to by Feldman (1985:87;92) in the context of
connectionist models and by Rummelhart,
Smolensky, McClelland and Hinton (1986) in their
parallel distributed processing (PDP) approach.
They7 report from runs of a simulation of Necker
cube processing by a connectionist network model
in terms of the two perspectives with the cube
front facing left or the cube front facing right.
They show that the system will (almost always)
end up in a situation in which all the units in
one subnetwork are fully activated and none of
the units in the other subnet are activated. That
is, the system settles in the stable state (or
fixed point) of interpreting the Necker cube as
either facing left or facing right. This occurs
when the input values are low relative to the
strength of the constraints among units, and
appears to be functional as long as somebody
knows that there is only one right
interpretation, and which one it is.


Under high input condition, the system
occasionally yield the interpretation that the
cube has two front faces. While this has the
merit of retaining both the left and the right
hand perspective in operation, it makes for an
"impossible" interpretation through fusing the
two.



































Necker's diagram invites two incompatible
cube-in-depth perspectives on the basis of the
same two dimensional source: one with the cube
front facing left when seen from above; the other
with the front facing right when seen from below.
Rubin's vase permits figure-ground reversal
between the image of a vase and two silhouetted
faces. And, finally, the well known figure, used
by Wittgenstein and others, permits reversal
between seeing a picture of a duck and of a
rabbit. While such perceptual reversal is
difficult to accomplish with regard to the
cumbersome Necker cube forms, it is relatively
easy to perform upon Rubin's vase, and easy on
the duck-rabbit figure. Our perceiving Rubin's
vase presupposes the ability to segregate the
visual forms into a coherent figure against a
coherent ground.


This has inspired some neurocomputational and
computer simulation explorations (Bråten and
Espelid, 1989; Ekornæsvåg, 1991).8


The basic idea is simply this. Define a
perspective P simply as a related set of
viewpoints, p1,p2,...,pn, which evoke companion
viewpoints, q1,q2,...,qn, as members of a
complementary set Q. Let these complementary
preferences in viewing the world be imposed by
the systems designer or "trainer". Train
competing networks, or a population of
classifiers, to operate according to different
viewpoints, for example, a perspective P that
makes one see the world only in terms of faces,
and a complementary or rival perspective, Q, that
makes another see the world only as composed of
vases or goblets. Permit these P- and Q-nets, or
populations of P- and Q-classifiers, to operate
concurrently, each from their own perspective,
when facing visual ambiguity. Then, when faced
with Rubin's vase, P will recognize the
silhouetted faces, while Q will see the vase.
Allow these particular P- and Q-viewpoints to
enter into conversation. This will allow the two
views to complement each other, or one of them to
win, given other clues that comes from other
sources, perturbing or supporting the viewpoints
in question.


In one of these projects a modified versions of
the Neocognitron model developed by Fukushima
(1986) has been modified and trained,
respectively, as P- and Q-networks. Fukushima's
model is a cybernetic self-organizational
multi-layered network model, developed for visual
pattern recognition, with backward paths and
capable of selective feature attention. It is
capable of recognizing different handwritten
versions of characters and digits, and can also
handle certain types of ambiguity. For example,
when trained to recognize the individual patterns
in this set (0, 1, 2, 3, 4) it will attend and
recognize selectively three distinct patterns in
this order (4, 2, 1) in the below stimulus
(Fig.13.2 (i):











































This is relevant for the kind of visual ambiguity
exhibited by technical document patterns for
which the immediate neighboring context.9






















T /-\ E          C /-\ T








Figure z. Selfridge's example of instant
disambiguation by virtue of available context
provides no clues for correct interpretation by
conventional automated means. For example, the
above label (Fig. 13.2 (i)) may be read as 125,
as IZS, as 12 S, as IZ 5, as I 25, etc.



















When a single Neocognitron network version,
trained to "recognize" both digits and
characters, was exposed to the above series, it
failed, as expected, in our trials to come out
with significant results. When, however, one
network was trained as a P-net to "recognize"
digital forms, and another was trained as a Q-net
to "recognize" letter forms, each came out with
clear forms according to their respective
"perspectives" when exposed to the partly
deformed patterns of Fig.13.3 (ii), and even to
highly distorted forms.10 That is, the P-net
generated forms conforming to the digits (1 2 5),
and the other forms conforming to the letters (I
Z S).


Thus, while a single implemented network11 But
there is also another possibility. The modified
neocognitron which we use as competing P- and
Q-nets is one of the most advanced of the about
50 different types of neural network models
currently being used around the world in research
and applications. Compared to other fairly
successful models, such as the Hopfield and the
Kohonen networks, and the Bolztman machine, used
in the retrieval of images from fragments and
pattern recognition, the Neocognitron is a
sophisticated network able to identify complex
patterns. But its weakness is its requiring large
numbers of processing units (PEs) and
connections. (Cf. Philip Trevleaven, 1989).
appear to defy "training to recognize" patterns
of similar form that conform to elements of both
the above series,12 the trained P- and Q-nets
appeared capable of "recognizing" even highly
distorted forms as conforming to viewpoints in
terms of their respective perspectives.13






























Figure 27 Examples of distorted patterns that are
recognized by the P- net and by the Q-net as,
respectively, conforming to the form Z and 2.


Their structures are indicated in Fig. 13.3.


So-called back-propagation, or regulatory
feedback based on some externally set criteria,
works within the limited perspective that each
net has been "trained" for, but not accross the
networks. Conversation in view of global or more
distant contextual information from the past may
be required for resolving their difference,14 for
example, in the case of exposure to the series in
Fig.13.2 (ii), about how labels in this document
tend to adhere to the rule (character, character,
digit).15


- P-networks, trained and committed to processing
viewpoints belonging to a P-perspective;




- Q-networks, trained and committed to processing
viewpoints belonging to a Q-perspective;




- N-networks, trained to recognize various
patterns for which conversation with a rival
perspective is not called for, and hence, without
commitment to the P- and Q-perspectives evoked by
other networks in the system; and


- P- and Q-network communication points where P
vs. Q conflicts are resolved (for the time being)
on the basis of conversation between rival
networks through access to contextual and
rules-based clues provided by an updated global
base (or as presently, provided by the trainer).




We term this the DIAD structure (for Dialogic
Interpretation of Ambiguous Data). Should later a
different solution be required, there may be
reversal to the discarded alternatives by virtue
of the parallel net.







-> R(P,Q) CONVERSATION ->


in view of past or global information



















Figure 13.3 Illustrations of the structure of two
implemented versions of the Neocognitron model, a
P-net and a Q-net, the first trained to see the
world in digit forms, the other in letter forms.
When exposed to the same source, each comes out
with a form conforming to their respective
"viewpoint".






This R(P,Q) philosophy has been explored in a
somewhat different version in computer
simulations of robots operating in a Necker cube
environment. In robot vision various paradigms
have been tried out with limited degrees of
success (Cf.Aliomonos and Rosenfeld, 1991).
Recently it has been suggested that simple and
primitive special purpose systems may be
preferable.16 A candidate paradigm for robot
vision has been offered by the Classifier system,
developed by Holland and coworkers (19??).17 It
shares the parallel and distributed processing
characteristics with connectionist models, while
being rules-based in the tradition of
production-oriented knowledge-based systems.
Classifier principles provide the basic operating
characteristics of simulational robots
implemented in ADA by Ekornåsvåg (1991). She
studies errors in vision and locomotion of
computer simulational robots in relation to a
Necker-cube.18 The cube has to be touched in
order for the simulated robot to resolve which of
the two perspectives, P (facing the left) or Q
(facing the right), that applies. In terms of
error frequencies the simulated robot version
operating most closely to the classifier
principles appears to do better than one in which
for any one single classifier in a P-population
there is a complementary classifier as a member
of a Q-population. But both simulated robots are
capable of evoking complementary P- and
Q-perspectives. They appear to cope with the
ambiguous environment, changing their dominant
viewpoints and adjusting their locomotion
accordingly.






Discussion




While the above Neocognitron and robot
simulations concern a referent domain of
relevance to the present thesis, they serve no
corroborational purposes. They merely serve to
demonstrate that it is possible to implement
simple network and classifier structures that are
capable of evoking complementary viewpoints. A
variety of structures allowing for this could
have been implemented.


Of relevance to the present thesis is the
phenomenon of human perceptual reversal in
itself. The fact that a human perceiver is able
to, when "setting one's mind to it", to alternate
between two such complementary viewpoints
suggests that she is able to house complementary
viewpoints in parallel or near-parallel. The kind
of perceptual alternating invited by figures (ii)
and (iii) in Fig. 13.1 appears to presuppose a
structure capable of embodying complementary
perspectives in concurrent or near-concurrent
operation and which may alternate as a
consequence of the mind setting. Such alternating
may be considered an extreme case of
complementary viewpoints in dialogue. Thus,
structural prerequisites for perceptual reversals
may relate to a structural prerequisite for the
mind's inner dialogue with itself, that is, with
the complementary perspective of the virtual
other.


While perception is mostly considered in terms of
mediating representations, Gibson and Heider,
each from different points of departure, invite
considerations in terms of ecological or direct
perception of the world and others in it. In
human perception, Heider, like Tomkins and
Whitehead, stress affects or feelings as basic in
perception. William James, a forerunner of
network modeling, points to affective interest as
basic to the operation of associative networks.
But then, this also entails dissociation, that
is, between networks based on complementary or
rival affective interests.


In this vein the concept of perspective may be
defined in terms of feeling, as a family of
viewpoints involving graded feelings that include
and generate a world and its complement as felt
on the basis of affective interests. In chapter 6
a participant perspective, P, was defined as a
related set of unsatisfied viewpoints, (p1, p2,
...,pn) which by virtue of gradation of feelings
distinguish that which is to be included by the
perspective, and which invite viewpoints
belonging to a companion perspective, Q, so that
the graded feelings of self-satisfaction in P and
Q feed upon the complementary relation to each
other.19 This aspect of feelings as affective
interests involved in the conversation between
complementary perspectives, escapes the above
crude "neurocomputational" and simulation
explorations.20


But the structures capable of realizing such
perspectives in concurrent operations in a human
perceiver - one view dominant and holding the
foreground, the other receding and maintaining
the background - may be relevant to the everyday
phenomenon of divided attention.






Divided attention and consciousness




In specifying dialogic closure in Part One I have
avoided the term "consciousness" which easily
evokes reference to processes of mediation and
reflection, in contrast to phenomena of felt
immediacy involved in dialogic closure. In the
affect attunement that may accompany
protoconversation between infant and caregiver
perception of the actual other, or what
Trevarthen terms alteroception, appears to be
immediate and direct, not indirect, mediated by
representations. The self-organizing dyad
realizes itself through the interlaced networks
involving the bodies and neurophysiological
structures of the two participants in direct or
immediate reciprocal operations. Mediation in
terms of the re-presented past other comes in at
the moment the affect attunement is broken. But
this does not entail that feelings are absent
from the way in which the child gazes at the
mother in wonder. She is included in the child's
perspective in another manner than during the
affect attunement, but still in a way that
involves feeling.


Take for example the toddler engaged in playing
with his toy, with his back to the mother who
accompanies his playing with noises in attunement
with his excitement. The toddler is devoting his
attention to the plaything, included in his
perspective. But this does not entail that the
mother escapes his attention. She remains present
in the background, included in his companion
perspective. The moment she comes out of tune or
is interrupted by something, the toddler stops
his playing and turn to look at her, pulling her
into what is now his foreground perspective.


Reflective awareness, for example of the feeling
of affect attunement between mother and child,
emerge only at the moment when affect attunement
breaks down. Then the child stops his activity
and gaze questioningly at the mother (Cf. Daniel
Stern, 1985: 150). There is acute awareness of
something lost in the past, bringing about an
inquiry about why things have changed. During the
affect attunement the mother was present to the
child, included in his companion space even when
his main attention was directed elsewhere. While
still in the same physical room, she is no longer
with the child in this immediate sense when the
attunement has been broken, even though the child
now devotes all his attention to her.






The illusory unity of consciousness?




Independent of each other, Ulrik Neisser (1976)
and Ernest Hilgard (1977) have considered anew
everyday assumptions and experimental evidences
on divided or dual attention.


Experiments in the 1950s on selective listening
to concurrently presented messages showed the
subjects to almost entirely ignore the secondary
message, but not quite. On closer inspection that
more was picked up from the unattended voice than
previously assumed. Already at the turn of the
century experiments on attending two things at
once were carried out. Neisser concludes from his
review that it seems clear the amount of
information picked up from one source while
attending to another is not limited by any fixed
mechanism, but appears to depend on skill.
Practiced subjects can successfully carry out
divided attention which the novice cannot.21
Neisser models perceptual processes in terms of
perceptual cycles involving anticipatory schemata
which are plans for explorative activity as well
as readiness for particular kinds of optical
structures. Some are innate, operating outside of
attention, while ready to detect their proper
signals no matter what other perceptual cycles of
attention may be in progress. Thus, permitting
the concurrent operation of perceptual cycles,
some more autonomatic than others, Neisser can
account for phenomena of dual listening and
doings. As for consciousness, he makes this
declaration:




"It is often said that a person can only be
conscious of one thing at a
time.........Introspection does not necessarily
show that one is aware of only a single thing at
a time. I think that people report the singleness
of consciousness largely because the
philosophical assumptions of our culture require
it.."(Neisser, 1976, p.104).22




Hilgard declares that the unity of consciousness
is illusory. We may pay acute attention to
something in the present while being aware of
several things at the same time. Even divided
attention is familiar in everyday life. For
example, in the process of a conversation, while
paying attention to the what the other is saying,
the listener may at the same time be preparing
his reply. While replying he may be watching the
other for confirmation of his being understood,
and perhaps changing the direction that his reply
is taking him as a consequence of the expression
in the other's face. Sometimes, for example,
while driving a car and talking to the passenger
the car driver may devote most of his attention
to the conversational topic until some traffic
incident calls upon his attention. When there is
no awareness at all of the parallel process of
thought or movements the latter is said to be
dissociated from the conscious awareness of the
person.




Explaining "the hidden observer"




In the late 50s Hilgard put into operation a
research laboratory on hypnosis experiments,
re-examining the assumptions in dissociation
theory. The breakthrough occurred when a double
train of thought in a subject unexpectedly was
revealed in a classroom demonstration on hypnotic
inducement of deafness.


Hilgard (1977, 1986) reports from a series of
experiments in which the subject's senses of
hearing and of feeling physical pain, partly or
wholly is being set aside through hypnosis. The
first disclosure of what he came to term a
"hidden observer" occurred in this way: Using an
experienced subject who was blind, Hilgard
induced deafness, telling him that he would be
able to hear when a hand was put on his shoulder.
The class was shown how unresponsive the subject
under hypnosis was to noise or speech. Then
someone in the class asked whether the subject
really was as unresponsive as he seemed.


Hilgard then addressed the hypnotically deaf
subject in a quiet voice, mentioning that there
perhaps is some part of the subject that might
hear his voice. "If there is," Hilgard continued,
"I should like the index finger of your right
hand to rise as a sign that this is the case". To
the surprise of the class, and the instructor,
the finger rose. When the subject was asked what
he remembered, he mentioned that everything had
become still, he had got bored and started
thinking about statistics. Then he felt his
forefinger rise. He wanted to know why. Before
telling him, Hilgard asked for a report from
"that part of you that listened to me before and
made your finger rise".


A detailed report was given by the hidden other,
which Hilgard metaphorically terms the "hidden
observer", and which later has been found to
recur in other hypnosis experiments on
suggestions on pain, as well as hearing.23 These
phenomena, and the phenomenon of divided
attention, are taken as an indication of
consciousness as being divided, described in
terms of dissociation and a double train of
thought in the mind of the individual.






Disscussion and explanation in terms of the
present thesis




Although such "hidden-observer" experiments have
been replicated in several laboratories and
clinics one may ask whether the reported
occurrences reflect some underlying phenomenon or
are merely the outcome of experimental
constructions. An obvious ground for critique is
to attribute to the subject a wish to comply with
the experimenter's request and to simulate being
hypnotized. This has been tested out by Spanos
and Hewitt (1980). They asked subjects of proven
low hypnotizability to simulate hypnosis, and
compared their behavior to higly responsive
subjects. Naturally, the simulators could report
on the process, as if they harboured a hidden
observer. Their reports, however, have been
judged by Hilaard and associates to differ
significantly from the reports provided by the
"hidden observers" of actually hypnotized.24


An important but unresolved problem in the
neodissociation theory that have emerged from
these experiments, concerns the question why some
highly responsive subjects do not appear to have
access to a hidden observer:




"A finding not yet satisfactorily explained is
that among highly responsive hypnotic subjects,
that is, those who can reduce both pain and
hearing substantially when appropriate
suggestions are given, only half have access to
covert pain or hearing by the hidden observer
technique."(Hilgard, 1986: 195)25




The above finding, as well as the original case,
can be explained in terms of the virtual other.
For those subjects who come up with a
"hidden-observer" report, the hypnotist may be
seen to control the subject's self, but not the
subject's virtual other. For the subjects who are
without any concurrent "hidden observer" during
the hypnosis, the hypnotist may be seen to have
replaced the virtual other, filling the subject's
companion space and excerting his control from
there, much in the same way in which the adult
may fill the infant's companion space and replace
the virtual other, but without the dialogic
closure.


Let B denote the subject's self and *A the
subject's virtual other, functioning as the
hidden observer in some of the subjects. Let A
denote the hypnotist as the subject's actual
other in the experiment, and <B> the hypnotic
control of B by A. In line with the concepts and
notation introduced in chapter 3, the different
hypnotic transforms may be succinctly indicated
as follows:






(3) For a subject with a hidden observer: B,*A :=
<B>,*A




(4) For a subject without a hidden observer: B,*A
:= <B>,A






The state indicators to the left of the sign :=
indicate the original dyadic constitution of both
kinds of subjects. To the right of the sign (for
becoming) the different outcomes are indicated.
In the case of (3) B, but not *A, is monitored by
A. In the case of (4) B is monitored by A, who is
also replacing *A. That is, in both cases the
subject's self B is monitored by the hypnotist A,
but in the latter case (4) the hypnotist excerts
his control from the subjects companion space,
replacing the subject's virtual other *A. Since
this actual controlling other has replaced the
subject's virtual other in case (4), there is no
participant virtual other to "observe" or
"report" anything.


For the other subjects (3), whose selves are also
passivated or "put into brackets" by the
experimenter, the virtual other remains
unmonitored by the hypnosis, and hence continues
to operate. But the experimenter creates a
barrier to the mind's internal dialogue, takes
control over the self perspective, while the
virtual other remains as a hidden observer of the
events, without means to enter into conversation
with the passivated self-perspective. In
contrast, for the highly responsive subjects
(case 4) in whom no hidden observer continues to
operate, the experimenter has filled the
companion space, replacing the virtual other as a
controlling actual other.


It is consistent with such an explanation that
highly responsive subjects unable to come up with
a hidden-observer report are reported to be more
compliant to suggestions to regress to feel like
children again:


"Subjects without a hidden observer are more
"complient" to age-regression suggestions - that
is, they report feeling like children again -
whereas those with a hidden observer inevitably
report a persistent duality of awareness. During
age regression they see themselves simultaneously
as adult observers and as children."(Atkinson,
Atkinson and Hilgard, 1983: 185)26




This duality appears to be spontaneous, not
suggested by the hypnotist. Again, this fits with
the above account. For those who report such a
duality, the hypnotist while calling upon the
child self, has not replaced the virtual other
who remains in an observer's position. They
conform to (3). For those that do not report any
duality, the hypnotist has replaced the subject's
virtual other, conforming to (4). The latter
suggests a sequence whereby the subject submits
to the complementary perspective imposed by the
dominant actual other, filling the companion
space of the subject's virtual other. Apparently,
the persons in whom there is no "hidden observer"
to report anything of the hypnotic event have
admitted the controlling agency to fill the
companion space of the subject's virtual other.
This may be compared to the way in which the
infant's companion space can be filled by the
actual adult other in a direct mode of
relationship.






Does a dual brain entail a double mind?




A few years after Hilgard's laboratory for
experimental explorations of assumptions from the
last century about divided consciousness had been
put in operation, another decisive event occured
that was to reactivate the issue of a single or
double mind. This was Sperry's publication of his
findings in 1961.


In her study of the history of ideas on the
double brain, Anne Harrington (1987) finds it
difficult to uncover any article or essay on the
dual brain after 1920 and before this publishing
event.27


Already in 1844 Wigan published his work about
"The Duality of the Mind". A few years before
Hewitt Watson(1836) had asked: "What Is the Use
of the Double Brain?". He argued that the two
halves could engage in cooperative activity and
permit ideas and feelings of others to be
apprehended. He even suggested that communication
with others may just be "self-communing between
the hemispheres of our brain".28 In French
neurophysiology and psychiatry at the turn of the
century, Binet wrote a book on double
consciousness and Pierre Janet viewed alternating
consciousness in terms of dissociation. But
later, when Janet and Piaget participated in a
symposium on individuality in Paris 1931, Janet
saw the challenge to explain how a multiplicity
of diverse elements could constitute a
psychological unity, while Piaget's concern was
the formation of reason.29 By then single-minded
consciousness, except in pathological cases,
appeared no longer to be questioned.


The split-brain studies reactivated the question
of whether a dual brain entailed a divided mind.
We tend to attribute to consciousness a
phenomenological unity that denies a simple
identity or one-to-one correspondence with the
partial complementarity of the brain hemispheres.
Is there a structural basis for a single-minded
conscious self, or can the brain, when divided,
permit the realization of two co-conscious selves?


Those were the kinds of questions that Roger
Sperry (1985) has raised on the basis of his
experimental findings from split brain studies.
They were reported in Science30 three decades
ago. Ten years earlier he had begun with Myers to
study cats after section of optic fiber systems
and corpus callosum.31 It was followed up by
experimental studies on monkeys, and finally on
human patients who, for medical reasons, had
undergone surgical section of forebrain
commissures (commissurotomy). The first patient
was a 48-year old war veteran who in 1961
underwent the operation of disconnecting the two
halves of the cerebral cortex to prevent seizure
activity.32 Split brain patients' response to
visual and tactile stimuli presented in each
half-field of vision and touch were
experimentally studied. For example, presented
with an object in the left visual field the
patient reports that he cannot see the object
(projected to the right hemisphere) while he can
find it with his left hand, but not using his
right hand.


The two disconnected hemispheres were found to
function concurrently, but independently, in
parallel when different stimuli were presented
simultaneously to the two hands or to the two
visual half-fields. That is, things felt with one
hand or seen in the corresponding half visual
field constituted a world of its own independent
of the world of things felt, seen, and processed
by the other hemisphere. The two worlds could be
grossly incompatible or mutually contradictory
without either hemisphere noticing that anything
was wrong. Each surgically disconnected
hemisphere was revealed to function, to a
significant extent, independently of the other.
As Sperry (1985) reports, each appeared to "have
a mind of its own, but each cut off from, and
oblivious to, conscious events in the partner
hemisphere". In tests on self-consciousness and
levels of social awareness, it turned out that
even the right hemisphere, despite language
deficits, harbors a well-developed, seemingly
normal conscious self with a level of social
awareness in close accord with the presurgical
character of the patient, and also with the other
hemisphere.33


Joseph Bogen, one of the medical doctors who
performed the first split brain operation, draws
this conclusion from the subsequent studies of
the patients: First, it takes only one hemisphere
to have a mind. Second, following surgery, the
two hemispheres can sustain the activity of two
separate spheres of consciousness. The
controversial issue is this:




"Does the surgical splitting of the brain create
a duality of consciousness or duality of
mind?...Or is it an experimental maneuver that
makes it possible to demonstrate a duality that
was previously present..?"(Bogen, 1985:38)34




Bogen is of the opinion that such a duality is
present before the cerebral commissurotomy.
Puccetti (1973) argues in the same vein: we do
not normally notice our being of two different
minds or two different "persons" because when the
commissures are intact the two brain hemispheres
operate in close synchrony.35


Sperry himself has favored a different view.
While each surgically disconnected hemisphere
appears to have a mind of its own, he considers
the conscious mind to be normally single and
unified. It is mediated by brain activity that
spans and involves both hemispheres, which
function together as a very closely integrated
whole. Consciousness is seen to be a holistic or
emergent, functional property of high-order brain
activity, but with causal potency. With regard to
its subjective qualities it involves a system of
organizational properties that are select and
special to operations at top level of the brain
hierarchy and supercedes in brain causation the
powers of their neuronal and other
infrastructures36:




"Consciousness appears in the causal chain of
brain event at higher cognitive levels in the
form of irreducible emergent properties of brain
activity. Mental events interact on a holistic
'functionalistic' basis at their own cognitive
level and also exert a concomitant supervenient
form of downward control over lower level brain
events." (Sperry and Trevarthen, 1990).37




Sperry (1968, 1985) has elsewhere described the
structure of the conscious system to be Y-shaped,
that is, divided in its more structured level but
undivided with regard to organizational
properties.38






In terms of the present thesis




How does the above relate to the thesis of the
virtual other partaking in the mind's dialogue
with itself? The thesis attributes an inherent
duality to the mind involving a self and a
virtual other. In this sense it relates to
Bogen's point. But the self and the virtual other
are seen to mutually complement each other as
inter-dependent constituents of a self-organizing
whole. In this sense the thesis relates to
Sperry's and Trevarthen's point about an
irreducible holistic organization. The present
thesis presupposes that complementary mental
activities can be concurrently realized in the
same brain, even when not subjected to
commissurotomy. But it does not imply that there
normally in the unperturbed brain be realized two
separate minds in independent operations as Bogen
and Puccetti would have it. It entails that the
mind as a self-organizing dyad recreates itself
as an operational unity involving interdependent
and mutually constituting complementary
perspectives as participants in the
self-organizing dyad of the mind. As a
self-organizing dyad the mind recreates itself as
a unity through dialogic closure or communion.


Thus, the present thesis need not be inconsistent
with Sperry's emphasis on the unity of the
organizational whole, provided this be considered
a dialogical unity that presupposes structural
prerequisites for the concurrent co-operation of
complementary perspectives. Such a structural
prerequisite may be satisfied by the way in which
"the two hemispheres in the normal intact brain
tend regularly to function closely together as a
unit" (Sperry, 1985:19).


It may, however, also be satisfied by the way in
which two brains can be involved in the
realization of dialogic closure, for example, by
mother and infant in protodialogue and affect
attunement, or by two intimate friends in
dialogue or communion. The near-perfect
synchronization, co-ordination of gestures and
sound, and bodily "postural echo" sometimes
exhibited by such dyads demonstrate how two
participants, while complementing each other, can
function closely together as a unity.


In a film shot at 48 frames per second, for
example, one may see how sudden, small mutually
matching movements start on the same frame of
film, as if the participants were of one body,
not two.39 Also infant-adult interplay may
sometimes exhibit a near-perfect synchrony,
albeit the more typical pattern is the mutual
completion of gestures, with clear indications of
bodily attunement.


According to the thesis brought forth in this
book this comes about without any operational
jump when each, as an actual other, replaces the
virtual other. The virtual other is not
postulated as a structural entity, entailing a
specific brain location, but as an organizational
and operational complementary participant
perspective that partakes in the self-organizing
mind which recreates itself as a unity through
dialogic closure. But the thesis does imply that
the single brain be capable of realizing the mind
as a self-organizing dyad that recreates itself
as a unity through dialoging with a virtual
other, and permits the co-operative and
reciprocal involvement of another brain without
an operational jump when the actual other
replaces the virtual other in dialogic closure.
In such dialogic closure there is an immediate
feeling of unity, irrespective of whether it
arises in communion with an actual other or from
the mind's dialoging with itself, that is, with
the virtual other.


I shall return to the above example of what
Morris terms "postural echo" in the concluding
chapter to follow, concerned with the issue of
time.




































131191 NOTES TO CHAPTER 13




1G.N.A. Vesey (ed.): Body and Mind. Readings in
Philosophy. George Allen and Unwin, London, 1964,
p.47. Cited in : A. Harrington: Medicine, Mind,
and the Double Brain, Princeton University press,
Princeton 1987, p.6.


2W. S. McCulloch: "Recollections of the many
sources of cybernetics", ASC FORUM, VI, No.2,
1974.


3G.N.A. Vesey (ed.): Body and Mind. Readings in
Philosophy. George Allen and Unwin, London, 1964,
p.47. Cited in : A. Harrington: Medicine, Mind,
and the Double Brain, Princeton University press,
Princeton 1987, p.6.


4Torsten N. Wiesel: Development of the visual
cortex. In: K. B Hell, O.D. Lærum and H. Ursin
(eds.): The Nansen symposium on new concepts in
neuroscience, Sigma, Bergen 1987, pp.59-60.


5C. von der Marlsburg and W. Schneider, Biol.
Cybern. 28, 223, 1978; C. von der Marlsburg and
E. Bienenstock, Europhys. Lett., 3, 1243, 1986,
referred in: W. Singer: Search for coherence: A
basic principle of cortical self-organization,
Concepts in Neuroscience, vol. 1, no.2, 1-26,
1190.


6C.M. Gray, P. Konig, A.K. Engel, W.Singer:
Oscillatory responses in cat visual cortex
exhibit inter-columnar synchronization which
reflects global stimulus properties. Nature 338,
334, 1989. See also Science, 249, Aug.24 1990,
pp.856-858.


7D.E.Rumelhart, P.Smolesnky, J.McClelland &
G.Hinton in: J.L.McClelland and D.E.Rumelhart
(1986),vol.II,pp.11-17.


8Stein Bråten and Rune Espelid: ............. We
wish to thank Kunihiko Fukushima for providing
updated papers on his Neocognitron model, which
is the basis for the adapted P- and Q-networks
used in the simulations; and Fredrik Manne for
assisting in the computer simulations (on IBM
3090/200 VF at BSC) on the basis of C-versions
implemented by him and Petter Möller. Affiliated
institutions are acknowledged for project
support, especially BSC IBM, and Dept. of
Cognitive Psychology, Bergen, during the period
(1987-89) in which the first-listed author was
guest professor in cognitive science at the
University of Bergen and consulting scientist at
BSC. We also wish to thank ............


9Normally, one use the context to disambiguate,
for example when the same distinct pattern, such
as this /-æ, in one instant may be seen as "H"
and in the next instant may be seen as "A" (Fig.
z).


10With the respective strength values 2.41; 2.58;
2.73 for the p-net, and 2.51; 3.59; 0.12 for the
q-net.


11The competing networks have been implemented in
C. Since a crucial feature is the conversation
between competing networks, object-oriented
languages have been considered for implementation
of the DIAD structure: Ada, Simula, and C+.


12In the first implementational stages, two
different adapted versions of the Neocognitron
model have been implemented in C and run on IBM
3090/200VF and on an IBM 6150(RT/PC). Trained to
recognize nine clearly distinguishable patterns
conforming to the elements in this series,
(0,1,2,..9), they are capable of recognizing
partly deformed input patterns in the Gestalt of
the respective numbers, and even to extract
superposed forms. However, when confronted with
ambiguous stimulus that may permit recognition
both as characters and numbers, we were unable to
train a particular network to recognize
characters and numbers in such a manner, even
when a variety of inhibition values were tried
out, that it could recognize a character pattern
and a number pattern in the same distorted
stimulus.


13In Figure x, three such increasingly distorted
forms turn out to be recognized by each of the
nets with these respective strength
values.................


14Such a conversational structure has not yet
been implemented. Ronald Chrisley (of Oxford and
Xerox PARC) has mentioned that at Oxford they
explore competition between perspectives in terms
of 'cognitive maps'. He has put the question to
us about whether these competing P- and Q-nets
have to be physically different from each other:
Isn't it sufficient for them to be conceptually
different only? Initially we thought so. But we
are no longer sure, when being trained to
recognize the world in terms of different
perspectives, they do become structurally
different from each other. In any case, they put
heavy demands on a complex program environment
that permits an exceedingly high number of
processing units (PEs) and interconnections, in
addition to the demand for an environment that
permits their inter-communication, when offering
their rival interpretations.


15The complete structure comprises ideally:


16Y. Aloimonos and A. Rosenfeld: Computer Vision.
Science, 253, Sept 13 1991 1249-1254. R.A.
Brooks: A new approach to robotics, Science,
ibid., 1227-1232.


17J.H.Holland, K.J.Holyak, R.E.Nisbett &
P.R.Thagard (1986).


18Anne-Grethe Ekornåsvåg: Simulation in a
Necker Cube Environment. Student thesis in
informatics. University of Bergen, Dept. of
Informatics, Bergen 1991. The model has been
implemented in Ada.


19This definition is introduced in S. Bråten,
...1988..with reference to Whitehead and others...


20In cognitive science and computer simulation
approaches, there are several early attempts to
include aspects of 'hot cognition' and feelings
in simulation models, for example, by Robert
Abelson (196?; 197?), Stein Bråten (1968),
Kenneth Mark Colby (198?); Roger Schank (19..);
and quite recently, Phil Johson-Laird (198?).....


21U. Neisser, op.cit., p.92.


22U. Neisser, ibid, p. 104.


23Ernest Hilgard: Divided Consciousness. John
Wiley, New York 1986, p.186.


24R.L.Atkinson, R.C.Atkinson and E.R.Hilgard:
Introduction to Psychology, Harcourt Brace, San
Diego 1983, p.185. Cf. also D.O Hebb:
'Understanding Psychological Man: A
state-of-the-science report' Psychology Today 16,
pp.52-53, and N.P.Spanos and E.C. Hewitt (1980):
The hidden observer in hypnotic alagesia:
Discovery or experimental creation? Journal of
Personality and Social Psychology, 39:1201-14
(1980).


25E. Hilgard, op.cit., p.195.


26Atkinson, Atkinson and Hilgard, op.cit., p.
185, referring to J.R. Laurence (1980): Duality
and Dissociation in Hypnosis. Unpublised MA
thesis, concordia University, Montreal 1985.


27Anne Harrington: Medicine, Mind, and the Double
Brain., Princeton University Press, Princeton,
New Jersey 1987, p.248.


28Harrington, op.cit., quotes Watson (1936:610)
stating: mental communication with others, as it
is commonly expressed, may just be self-communing
between the hemispheres of our own brains,
accompanied by signs and sounds addressed to the
senses."


29P. Janet: L'individualite' en psychologie; J.
Piaget: l'individu et la formation de la raison.
In: L'individualite', Libraire Felix Alcan, Paris
1933, pp.39-50;67-116.


30R.W. Sperry: Cerebral organization and
behavior. Science, 1961, 133: 1749-1757.


31R.E. Myers and R.W. Sperry: Intercular transfer
of visual form discrimination habit in cats after
section of the optic chiasma and corpus callosum.
Anat. Rec., 1953, 115: 351-352.


32M. S. Gazzaniga: The split brain in man.
Scientific American, vol. 217, no.2 1967 pp.24-29


33R. W. Sperry: Consciousness, identity, and
divided brain. In: D.F. Benson and E. Zaidel
(eds.): The Dual Brain, The Guilford Press, New
York 1985, pp.11-26.


34J. E. Bogen: The Dual Brain: Some historical
and methodological aspects. In: D. F. Benson and
E. Zaidel (eds.): The Dual Brain, The Guilford
Press, New York 1985, pp.27-43.


35R. Puccetti: Brain bisection and personal
identity. Br. J. Philosophy of Science, 24 1973
339-355 (Referred to by R. Sperry (1985) op.cit.
p.21.


36R. W. Sperry, op.cit., p.21-24.


37Roger W. Sperry and Colwyn Trevarthen:
Turnabout on Consciousness: New paradigm for
causation. Conference abstract August 1990 for
the ...............Italy....


38R. W. Sperry: Mental unity following surgical
disconnection of the cerebral hemispheres. In:
The Harvard Lecture Series, Vol. 62, Academic
Press, New York, 1968: 293-323. Referred to by
Sperry (1985), op.cit., p.22.


39Desmond Morris:..........W.S. Condon and W.D.
Ogston: Sound Film Analysis of Normal and
Pathological Behavior Patterns, Journal of Mental
and nervous Disorders, 143, p.338-347.












TO SEE THE INDEX OF TEXTS IN THIS SITE, CLICK HERE