Brain, Mind, Body and Society: Autonomous System in Robotics

advertisement
Brain, Mind, Body and Society: Autonomous System in Robotics
Motomu SHIMODA
Kyoto Women's University
shimodamotomu@gmail.com
Introduction
Issues related to robots possessing a mind have been raised in the field of medical
and caring practices. Such robots are expected not only to process external information
using a built-in program and behave accordingly, but also to gain consciousness
activity responding to multiple conditions and interactive communication skills to cope
with unknown situations.
This prospect is based on the development of artificial intelligence in which
self-organizing and self-emergent functions have been available in recent years
especially in Japan. To date, controllable aspects in robotics have been restricted to
data making and programming of cognitive abilities, while consciousness activit ies
(volitional/emotional) and communication skills have been regarded as uncontrollable
aspects due to their contingency and uncertainty.
However, some researchers of robotics claim that every activity of the mind can be
recreated by engineering and is therefore controllable. Based on the development of
the cognitive abilities of children and the findings of neuroscience, researchers have
attempted to produce the latest artificial intelligence with autonomous learning
systems.
This article examines the possibility of robots with artificial intelligence focusing on
the autonomous system.
1. Robots possessing a mind as a previously programmed function of information
processing
People have become bored with entertainment or pet robots like "AIBO" by Sony
Corporation (1) and the therapeutic robot "Paro" by Daiwa House Industry Co., Ltd. (2)
because such robots react with limited patterns, and lack unexpectedness and
interactive communication even though they behave like pet animals. The robots’
reactions are totally programmed and almost predictable, so human beings cannot
regard such robots as real partners but as mere gadgets.
• So, what constitutes a mind? Cognitive scientists classify functions of the mind as
follows: [cf. Anzai 2011; Arita 2007; Shibata 2001; Nobuhara 2004]
- awareness: sensations, perceptions, consciousness, memories,
1
- thinking: understanding, judgment, reasoning,
- volition: willingness, acceptance/refusal,
- emotion: affections, feelings, passions,
- reflection: self-consciousness,
- imagination: fantasies, dreams,
- interaction: communication, friendships/conflicts.
Some researchers of robotics have attempted to create robots with artificial
intelligence by recreating mental functions using the following information
technologies: [cf. Maeno 2009; Taniguchi 2010]
(a) data creation of the cognitive function such as stimulus and reaction, data
processing, memory, feedback and learning,
(b) programming of neuro activities associated with consciousness such as sensation,
emotion, thinking, reasoning, imagination, volition and reflection,
(c) incorporating modules of the programs into the robots such as software for each
activity.
Such robots might have higher intelligence and be autonomously capable of
interactive communication with humans akin to them possessing a human-like mind.
By incorporating a learning system they could act in unpredictable and unexpected
situations. Even so, however, functions of information management are still evident -input, data processing and output --, and we regard these robots as instrumental
entities reflecting the intentions of their manufacturer.
In this context "mind" is exclusively regarded as the capability of handling
information, while data selection and input are performed by the creator of the robot.
The interactive or social contexts of such information remain alien to the information
processing of each robot.
2. Robots with an autonomous system
The basic principles of the latest artificial intelligence consist of the following
functions: [cf. Ishikawa 2006; Sena 2008; Taniguchi 2010; METI 2008]
(a) self-organizing function:
- incorporating the data of unexpected physical experiences, and modifying the
program in response to such experiences.
(b) self-emergent function:
- creating an entire autonomous system by each activity based on the previous program
and regulating each activity by the entire system.
(c) self-referential function:
2
- recognizing, interpreting and evaluating its own behaviors and coordinating them
according to the situation.
The autonomous system is based on cognition, behavior and communication
modeling using the module formation of each function. While the information
processing function mentioned above is mostly controlled by the creator and
programmer, the autonomous system is highly creative through its voluntary and
interactive activity that extends beyond the limits of the program.
Robots with an autonomous system would have the following abilities:
- to obtain information about the situation,
- to act without human guidance and assistance,
- to cope with unknown situations,
- to protect and maintain itself,
- to adapt to new situations, and
- to change the situation for its own purpose.
If a robot could possess such functions, interactive relations between humans and
robots would be established. Would such an autonomous robot really be a real human
partner instead of a mere gadget?
Although it is possible for such autonomous robots to cope with disasters such as a
fire, earthquake or nuclear power plant accident, to work in such severe situations
such as battlefields or deep sea, and to act as care workers or housekeepers, it is
necessary for them to be programmed and arranged by human beings before each
operation.
Autonomous activities demanding judgment according to circumstances can be
recreated by engineering based on accommodation to the specific purpose. They are
regarded as totally controllable in that they are modifiable if inconveniencies are
detected and they are equipped with a learning system that makes the updating of
programs available. Still, real partnerships and interactive relationships with robots
remain elusive because they are not agents with living bodies and specific perspectives,
but rather products programmed by the creators even if they are autonomous. For real
interactive communication it would be necessary not merely to interchange
information, but also to share understanding based on the contexts in which we
interpret the meaning of the situations and relations concerned.
3. Advanced stage of autonomous system of robotics
Some researchers on robotics have attempted to create interactive partners by
improving the artificial intelligence techniques of the robots. The achievements of
3
cognitive science recreating the functions of the human mind by computer technology
have reached a new stage. There is an automatic learning system, or self-updating
program, in which unexpected responses of a partner and uncertainty of a situation
can be dealt with.
These types of robotics are categorized as follows:
(a) Constructive approach is a methodology in which humans as a system are to be
understood by constructing the system and operating it by engineering and technology.
[cf. Taniguchi 2010; Kanda et al. 2002; Minato 2004]
(b) Cognitive developmental robotics aim to integrate neuroscience, cognitive science,
developmental psychology and robotics into interdisciplinary studies on the relation
between humans and robots. [cf. Asada et al. 2000; Asada 2010]
(c) Intelligence dynamics are an artificial intelligence using self-emergent functions
based on a human-made program, acquiring interaction between brain-body and the
environment in the manner of dynamics. [cf. Doi et al. 2006;]
Some researchers indicate that incorporating the physical functions of the body into
a robot is indispensable for acquiring such system or program. [cf. Dreyfus 1992; Asada
2010; Ishiguro 2009] Embodiment or human-shaped physicality plays a crucial role in
interactive relations. Each experience can be built into the program through the
data-processing of sensation. As human-shape physicality is extremely important in
terms of appearance, humans tend to treat such humanoid or android type robots as
humans in a human community.
Moreover, new findings of neuroscience have added to artificial intellig ence in terms
of autonomy and interaction. We can refer to the brain neuro system, in particular
neuro plasticity, as a model of an autonomous system. Neuro plasticity is the "capacity
of neurons and neural networks in the brain to chan ge their connections and behavior
in response to new information, sensory stimulation, development, damage, or
dysfunction." (3) This implies self-transfiguration of the neuro system.
If it is possible to incorporate neuro plasticity into a robot after recreating by
engineering, the self-organizing, self-emergent and self-referential functions of
artificial intelligence could be promoted to the advanced stage of the autonomous
system. This would make the agent (=robot) available to act self-consciously by
interpreting his/her own action in a social context and through interpersonal dialogue,
instead of behaving within the frame of the previously programmed system.
Such agent would no longer be a product of the creator of the program. Can we find
real partnerships with an agent of a human-shape physicality and neuro system?
4. Critical examination of autonomous system of the robot
4
Even if a robot could possess the program of such advanced autonomous activities as
physical functions, neuro plasticity, interpersonal dialogue and social interaction,
limitations would still exist as follows:
- physical functions are constructed by the data of sensation/motion,
- neuro plasticity is reduced to the feedback mechanism of information,
- interpersonal dialogue is merely an exchange of information,
- social interaction is equal to the role-playing demanded in the situation.
It is autonomous from the perspective of an outside observer, and as autonomous
activities are controlled by the program and classified into the various steps of
data-processing from a behavioral viewpoint, they therefore continue to be functions of
a product subject to the intention of the creators. For autonomy in the genuine sense, it
is necessary to obtain the perspective of the participant or performer within each
context:
- physical function is continually feedback and updated under inter-physical relations,
- neuro plasticity is a self-reflective and self-modifying system in the networking of
body and mind,
- interpersonal dialogue is performed by interpreting the meaning of what the other
person says,
- social interaction should be understood in terms of not the cause-effect but the
reason-consequence relationship.
Partnerships between autonomous agents from the perspective of the participant or
performer will be constructed by answering such questions as "why do you perform
such an action in such a way?" This does not constitute a causal relation of natural
phenomena. Each person is required to reply with a reason or ground s for their action
from his/her own viewpoint and perspective. [cf. Habermas 2003]
Although it is possible for autonomous robots to communicate with humans or other
robots, interchanged information could be provided for statistical transactions and
data-processing, not interpreted as experience full of implications and symbols
restricted by the physicality and social situations in the real world. As a product of the
creator, robots are observed and analyzed in their motion, thinking, interaction, etc.
If we shift the paradigm from the outside observer to the participant or performer,
we are no longer in a position where the mind is confined within the individual person
and consciousness is situated within the framework of subject-object, but rather we are
positioned from the standpoint of inter-subjectivity or the mind as an integration of
neuro functions, bodily sensations and social interactions.
Conclusion
5
Attempts have been made to create robots possessing a mind by incorporating an
autonomous system into a robot, which presupposes controllability of the cognitive,
reasoning and communicative abilities. However, controllability is inconsistent with
autonomy in the genuine sense, because it is based on the previously designed
programs by humans, including self-organizing, self-emergent and self-referential
functions. Autonomous robots recreated by engineering cannot be autonomous
partners of humans, because they are lacking in the perspective of the participant or
performer and the standpoint of inter-subjectivity.
Notes
(1) http://www.sony.jp/products/Consumer/aibo/
(2) http://www.parorobots.com/
(3) http://www.britannica.com/EBchecked/topic/410552/neuroplasticity
References
Anzai, Yuichiro(2011), Mind and Brain: an Introduction to Cognitive Science, Iwanami
Shoten [In Japanese]
Arita, Takaya(2007), Can We Program Mind?, Soft Bank Creative [In Japanese]
Asada, Minoru(2010), The Thought of Robot, NHK Books [In Japanese]
Asada, Minoru et al.(2000), Cognitive Developmental Robotics As a New Paradigm for
the Design of Humanoid Robots [In Japanese]
www.er.ams.eng.osaka-u.ac.jp/Paper/2000/Asada00c.pdf
Doi, Toshitada et al.(2006), Intelligence with Body, Springer Japan [In Japanese]
Dreyfus, Hubert(1992),What Computers Can't Do: A Critique of Artificial Reason, The
MIT Press; Revised
Habermas, Juergen(2003), The Future of Human Nature, Polity
Ishiguro, Hiroshi(2009), What is Robot?, Kodansha [In Japanese]
Ishikawa, Masato(2006), Mind and Cognitive Informatics: Making Robot, Knowing
Human, Keisoshobo [In Japanese]
Kanda, Takayuki et al.(2002),
humanoid robots,
A constructive approach for developing interactive
2002 IEEE/RSJ International Conference on Intelligent Robots
and Systems (IROS 2002), 1265-1270 [In Japanese]
http://www.ai.soc.i.kyoto-u.ac.jp/publications/02/kanda-iros2002.pdf#search='Constr
uctive Approach robot'
Maeno, Takashi(2009), Why "I" in the Brain Cannot be Found?, Gijutsu Hyoronsha [In
Japanese]
6
Minato, Takashi(2004),
A Constructive Approach for Studying Human -Robot
Communication, Proc. of 2nd International Symposium on Emergent Mechanisms of
Communication in the Brain, poster session, p.82.
www.geminoid.jp/~minato/papers/Minato04a.pdf'
Ministry of Economy, Trade and Industry(2008), Academic Roadmap concerning
Robotics. [In Japanese]
www.meti.go.jp/policy/economy/gijutsu_kakushin/kenkyu...pj/r_1.pdf
Nobuhara, Yukihiro(2004), Series Philosophy of Mind II: Robotics, Keiso Shobo [In
Japanese]
Sena, Hideaki(2008), Hideaki Sena's World of Robotics, Keiso Shobo [In Japanese]
Shibata, Masayoshi(2001), Mind of Robot, Kodansha [In Japanese]
Taniguchi, Tadahiro(2010), Can We Create a Communicative Robot?: Constructive
Approach towards Symbol Emergence System, NTT Press [In Japanese]
7
Download