June 25, 2012 12:45 PSP Book - 9in x 6in 09-Junichi-Takeno-c09
Proposed Concept Model of Consciousness 163
interaction between infants and the environment and expect that
consciousness and the mind will emerge from these learning
procedures.
While partly agreeing with the ideas of these evolutionary robot
researchers, the author believes that rather than attempting to cre-
ate consciousness with evolutional techniques, we should develop
artificial cell networks step by step while defining a new paradigm
regarding consciousness. As mentioned before, the author doubts
that consciousness can be created using evolutional techniques.
I believe that a new paradigm must be established to generate
consciousness in a machine. Descartes advocated a mind–body
dualism, but my standpoint is a materialistic mind–body monism.
I believe that something like human consciousness can be con-
structed in a machine. A robot with something like consciousness
would behave as though it had a will.
9.2.2 Expectations for a New Paradigm
The author believes that the imitation function and the function to
distinguish between the self and others are important factors of
human consciousness. Humans can distinguish the self from others
using a function that discriminates between the self and others. They
learn the behavior of others by just looking at them, thanks to the
imitation function. These two functions are the prime mover in the
evolution of humans into higher organisms.
The research studies by Meltzoff et al. have shown that infants
instantly imitate the behavior of their mothers. It was quite strange
for the researchers to learn that the infants performed the same
behavior in a short period of time (Meltzoff and Moore, 1977).
When a mother sticks out her tongue, the child instantly does the
same. The child seems to move to action immediately by just seeing
the mother. When I touch the keyboard of a computer, I mostly strike
the keys subconsciously but I never think that the fingers striking
the keys are those of someone other than myself. When a person
resembling me stands nearby, I never mistake him for myself.
Self-consciousness that tells that the image of oneself reflected
in a mirror is the self seems to be deeply related to the function of
distinguishing between the self and others.
June 25, 2012 12:45 PSP Book - 9in x 6in 09-Junichi-Takeno-c09
164 New Architecture of Robot Consciousness and the Robot Mind
The author is trying to discover the source of consciousness.
There are many researchers studying the problem of “What is
consciousness?” Nevertheless, no assertion has ever been acknowl-
edged to be decisive enough to solve this problem. This is because a
unified paradigm regarding the phenomena of human consciousness
had been missing.
The author has devised a new paradigm of consistency of
cognition and behavior as the source of human consciousness.
This chapter shows that the new paradigm can describe almost
all of the phenomena of human consciousness in a computationally
feasible manner. The author’s arguments may contain, in various
points, findings that have already been discussed by other re-
searchers, but what is asserted by the author as a whole is a unified
paradigm capable of explaining many phenomena related to human
consciousness and the mind.
The underlying idea of the system introduced herein is the con-
struction of neural networks essentially defined between inputs and
outputs based on this new paradigm. The system continually evolves
by capturing information about the surrounding environment.
The author believes that thought is generated from conscious
activities, i.e., a stream of consciousness. Consciousness here in-
cludes the subconscious. Explicit consciousness, generally referred
as simply consciousness, occurs when the subconscious satisfies
certain conditions, which will be discussed in detail later.
The mind is activated by the stream of consciousness and the
accompanying stream of kansei. Kansei is a Japanese word that
means both emotion and feelings. Consciousness is considered to
occur within the brain according to the internal state of the brain
and on receiving stimuli from the body and external environment.
The new paradigm is capable of explaining the root source of
consciousness and also describes imitation learning, distinction
between the self and others, self-consciousness, and qualia. It can
also describe imaging and creativeness. Furthermore, this new
paradigm makes it possible to draw a distinction between sense and
hallucination.
The methodology based on the author’s paradigm resembles
Minsky’s idea that in the human brain, higher-level areas have
evolved biologically from the old areas, i.e., the old areas are
June 25, 2012 12:45 PSP Book - 9in x 6in 09-Junichi-Takeno-c09
Proposed Concept Model of Consciousness 165
(a)
(b)
(c)
Figure 9.4. Development of the brain and self-recurrence.
supervised by the newly evolved areas. This idea may remind
the reader of a centrally controlled system where the new brain
dictates the activity of the old brain, but the author’s methodology
is different. The author’s methodology is the same to the extent that
the new and old brain areas co-exist, but the difference is that rather
than the new brain controlling the old brain, information circulates
between them. This idea may not sound smart, but using this idea it
is easy to explain the function of self-consciousness that is defined
as the state of being conscious of one’s own consciousness.
It is easier to understand the model of self-consciousness by
looking at Fig. 9.4a and then Fig. 9.4b. In Fig. 9.4a, the self is looking
at the self, i.e., the self-loop type. This is a typical self-recurrent
model. In Fig. 9.4b, the self is looking at others, i.e., the co-routine
type.
The reason why the author selects the co-routine type is the
difficulty of explaining the internal circulation of information using
Fig. 9.4a. With the co-routine type, it is easier to understand that
information circulates through both areas. If this description is
possible and successful, we may move to the smarter self-loop
type.
When one of the two routines is combined with the other (Fig.
9.4c), the result is the same as the model shown in Fig. 9.4a.
Accordingly, the author hereby declares that he is not using the
centrally controlled model.
9.2.3 Where Does Consciousness Come From?
What kind of state is it in which we are conscious of something?
June 25, 2012 12:45 PSP Book - 9in x 6in 09-Junichi-Takeno-c09
166 New Architecture of Robot Consciousness and the Robot Mind
I am now writing letters on a notebook with a pen. Or I am
thinking of my family, solving math problems, walking down a street
in Paris, etc. It is appropriate to express this awareness as being
conscious of something. It is also possible that I was walking while
thinking about my family, but I was not conscious of walking. When
this phenomenon is to be written into a computer program by
analogy, a number of software agents are working in the computer
and one of them is selected by means of some function.
Even if the computer displays a message on its screen that one
of the simultaneously running multi-tasking programs is currently
active, it would be difficult to say that the computer now possesses a
certain consciousness. To overcome this difficulty, some researchers
argue that consciousness requires embodiment (Pfeifer and Sceider,
2001) but this proposition would be promptly refuted by the opinion
that computers can also have embodiment such as by the keyboard
and the hard disk. A computer may be aware, through its functions,
thatitsharddiskisnowbeingaccessed.Evenso,itisdifficultto
accept that consciousness occurred in the computer. Why?
The author believes it is possible to define the relationship
between the operation of the computer’s hard disk and conscious-
ness, but this, by no means, indicates that I am ready to admit the
occurrence of consciousness in the computer. This is because the
existence of consciousness is a subjective phenomenon felt only
by an individual or by myself. I just feel that others also have
consciousness as I do, but its existence is difficult to prove.
The mirror neurons described earlier seem to make us believe
that other people also have consciousness. Since computers do not
have a function similar to mirror neurons, it is natural that we cannot
feel that computers have consciousness.
Some researchers thought about relationship between Turing
tests and the human consciousness (Mogi and Taya, 2003). This
means that they expect the action of the mirror neurons of humans.
From these observations, one might say that for humans to be able
to readily feel the existence of consciousness in a machine, the
machine must constantly stimulate the mirror neurons of humans.
This means that to create consciousness similar to that of humans
in machines, the functions of mirror neurons must be built into the
machine.
June 25, 2012 12:45 PSP Book - 9in x 6in 09-Junichi-Takeno-c09
Proposed Concept Model of Consciousness 167
A clear definition of consciousness capable of describing the
various functions of human consciousness must be established on
the basis of the functions of mirror neurons. Since mirror neurons
are related to the smooth perception of consciousness in others,
the shape of the machine may have to be a humanoid or human-
like robot. In the case of machines that are not human in shape, the
existence of consciousness may be judged by whether the machine
physically has the functions described in the above definition of
consciousness.
9.2.4 How Do We Define Consciousness?
Let us consider the existence of consciousness referring to an
example: The sound of a voice is input into the ears, and in
response to that stimuli, the action of speech from the mouth is
performed.
This example does not spoil generality. The area responsible
for consciousness is assumed to be located in the region with the
question mark in Fig. 9.5a. This is the area where neural networks
exist.
The above thinking is behavioristic because the description reads
that inputs entail outputs. If we apply an evolutionary algorithm to
this region, neural networks with the functions of consciousness can
be created in this region at some time.
Given the complexity of the phenomena of consciousness,
it is difficult in reality to use the evolutionary approach and
determine what functions are appropriate for creating neural
networks equivalent to consciousness. If consciousness consists of
a redundant system as I suggested earlier, it is quite doubtful that
the evolutionary technique would ever be useful. If the mechanism
of consciousness is a redundant system, an evolutionary technique
that relies on computers must find suitable functions by trying
out infinite possibilities. The author believes that the mechanism
of consciousness is a kind of redundant system. I know that the
evolutionary technique is excellent as already described, but it
is impossible to find the mechanism of consciousness using this
technique. The evolutionary technique may be effectively used once
the mechanism of consciousness is found.
..................Content has been hidden....................
You can't read the all page of ebook, please click here login for view all page.