Download An architectural analysis of emotion and affect

Survey
yes no Was this document useful for you?
   Thank you for your participation!

* Your assessment is very important for improving the workof artificial intelligence, which forms the content of this project

Document related concepts

Ayin and Yesh wikipedia , lookup

Holocaust theology wikipedia , lookup

Wiccan views of divinity wikipedia , lookup

Jews as the chosen people wikipedia , lookup

Re-Imagining wikipedia , lookup

Divine providence in Judaism wikipedia , lookup

Misotheism wikipedia , lookup

Transcript
Naturalizing the Spiritual:
Lessons from Cognitive Science
Ron Chrisley
COGS/Informatics
University of Sussex
Yale Divinity School
November 13th, 2007
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
QuickTime™ and a
TIFF (Uncompressed) decompressor
are needed to see this picture.
Tease: Artificial intelligence &
spirituality
•
•
•
Creating minds: artificial intelligence (AI) as
Blasphemy?
Designing for autonomy: Can theodicy help
AI? Vice versa?
External naturalization of spirituality: Will
robots believe in God?
Will discuss these at end, if time permits
First: Topic as given in title
Terms: Cognitive science
• One can define cognitive science to be
any scientifically-based attempt to
naturalize the mind: show how it is
possible for the mental to be part of the
natural world
• Can we use same techniques to
investigate whether the spiritual is
also?
Naturalizing the mind
• Mental states and processes seem to be very
different from physical ones: e.g., thoughts
can be true or false, atoms can't
• How can it be that a physical thing can also
be a thing with a mental life?
• Or, conversely, why is it that my physical body
behaves in a way which matches up with my
intentions? (cf "The Miraculous Coincidence
Thought Experiment", Cussins 1987)
Terms: Naturalization
•
Making it intelligible how multiple views of
the world:
– Can be of the same world
– Can each give understanding of that same world
•
Need not give any one view special authority
– Though some approaches (asymmetric ones) do
•
cf "The Limitations of Pluralism" (Cussins
1992)
External vs. internal
naturalization
1. External naturalization of spirituality: E.g.,
why do people have spiritual beliefs? How
could this have evolved?
•
•
spiritual anthropology (van Pelt)
cognitive science of religion (Bloom)
2. Internal naturalization of spirituality: How
can spiritual views be reconciled with, e.g.,
scientifically physical views?
This evening, only discussing number 2
When (internal) naturalization
fails
Either:
• Elimination (of the exotic)
• Relativism (e.g., post-modernism,
quietism)
Terms: The spiritual
 Which notion of the spiritual?
 Any suggestions?
 Fortunately, a definition is not needed for my
purposes
 Proposed naturalization strategies should work for
most or all notions of the spiritual
 Am not assuming that the spiritual is
necessarily supernatural
 Begs the question
 Makes naturalizing the spiritual impossible by
definition
Methods of naturalization
1.
2.
3.
4.
5.
6.
Interactive dualism
Reduction
Supervenience
Interpretation
Intelligible construction
Conceptual change
What are the things to be
naturalized?
• Two or more ways of understanding the world
– Here we will only consider two at a time (P and Q
in the case of symmetric naturalization)
• Call these "discourses", although need not be
linguistically expressed (or even be
expressible)
• Some means of naturalization are
asymmetric:
– One discourse is deemed privileged or familiar (F)
– The other is unfamiliar or exotic (E)
– E is naturalized to F, not the other way around
Interactive dualism
• Symmetric: P and Q are "discourse peers"
• The states, processes, etc. of P and Q are
incommensurable
– e.g., non-extended vs. extended substances (Descartes)
• But there are non-explanatory (usually causal)
relations between P and Q
– e.g., perception and action
• Many famous problems, e.g.:
– If P or Q (e.g. physics) is causally closed, no room for the
other to have any effect on it
– Epiphenomenalism is no solution: Denying causal links in
one direction (e.g., mental to physical) prevents knowledge
(e.g. of the mental)
– Same problems apply if P or Q is spiritual discourse
Reduction
• Asymmetric: Reduction of E to F
• Requires being able to define each of the concepts in
E in terms of the concepts of F
– E.g. The temperature of a substance is the mean kinetic
energy of its molecules
• Many problems, e.g.:
– Such reductions can be hard to find
– If E (e.g. mind) is multiply realizable, reduction appears
impossible
– If reduction of E to F is achieved, eliminativism threatens: no
causal work left for E to do
– Same problems apply if E is spiritual discourse
Supervenience
• Asymmetric: Supervenience of E on F
• Weaker than reduction: Only requires that F
fixes E; a change in E implies a change in F
• So avoids the problems of reduction (e.g.,
Kim 1990)
• But too weak: can't explain, e.g., the causal
efficacy of the mental
• Same applies if E is spiritual discourse
Interpretation
• Asymmetric: interpretation of F in terms of E
• E.g. for E = the mental, the "Intentional Stance"
(Dennett 1984): A system S has mental states B and
D if the non-mentally construed behaviour of S can
be explained by:
– Ascribing to S the beliefs B and desires D S should have,
given S's interactions with the world
– Assuming that S will behave in a way that would achieve D if
B were true
• Not all systems will be explicable this way; those that
are are the ones with mental states
Interpretation: Problems
• Merely re-states what we want to explain
– We know that our bodies can be interpreted as intentional
systems, but why?
• Doesn't explain the source of mentality of the
interpreter
– Presupposes an ur-interpreter that grounds everything?
• Relative, and therefore non-deterministic
– Whether I am interpretable depends on who is doing the
interpreting; there is no fact of the matter
– But surely the contents of my mind (and that I have one at
all!) are independent of who is interpreting me
Interpretation: Problems
• These problems might also apply when
E = spiritual discourse
– But not clear how to modify Dennett's
proposal to yield the "Spiritual Stance"
– (Would be ironic if one could do so, given
Dennett's views on, e.g., theism!)
Intelligible construction
• Perhaps similar to interpretation, but with some key
differences
• In particular, symmetric: intelligible construction
between E and F
• Naturalization is achieved when one has a practical
capacity to act in terms of P in a manner appropriate
for achieving one's goals in Q, and vice versa (cf
Cussins 1990)
– E.g., an architect with respect to blueprints vs. construction
materials
Intelligible construction
• In the case of the mental, this practical
capacity is the result of :
– Possessing set of practical capacities to negotiate
a number of discourses
– These discourses together constitute an intelligible
construction from P to Q, and vice versa
– No longer a mystery why something with these
physical properties is also something with these
mental properties, and vice versa
– The intermediate discourses are, in the main,
computational
Naturalizing the mind
Mental description
intelligibility (skill mediated)
?
Computational description
intelligibility (skill mediated)
Physical description
Analogy: Chess computers
• Interpretable as having beliefs, desires,
goals, etc.
• But also understandable as a physical system
• But no mystery; Why not?
• Answer: Analysis in terms of representations,
and computations over them
• Computational properties are not reducible to
physical ones: multiple realizability
• So: Naturalistic explanation without reduction
Naturalizing the mind via
intelligible construction
• AI/Cognitive Science: What goes for the
chess computer goes for us
• We can naturalize our mental states
without reducing them
• Do so by finding a level of
description/explanation in between the
mental and physical: computational
Naturalizing the spiritual via
intelligible construction?
• Does this suggest a way for naturalizing the
spiritual?
• Suppose (as seems likely) that spiritual
events, properties, entities, etc. cannot be
reduced to physics
• Does this mean naturalists have to abandon
or eliminate the spiritual?
• No; not if some other means of naturalization
(e.g., intelligible construction) can be
achieved
Naturalizing the spiritual
• E.g., need not eliminate the spiritual if
one can find an intermediate level of
description that allows one to nonreductively naturalize the spiritual
• But what is this level?
– A new kind of discourse/conceptual
scheme?
– Or perhaps computation again?!
Naturalizing the spiritual
Spiritual description
intelligibility (skill mediated)
?
intelligibility (skill mediated)
Physical description
Conceptual change
• All preceding naturalization methods assumed that
the discourses to be related are fixed in advance
• But there can be change in the concepts one
employs without changing the topic
– I.e., same reference, different sense
• This may be a way to remove some conceptual
obstacles to naturalization (Chrisley 2007)
– E.g., the apparent possibility of zombies: creatures that
behave just like us but with no phenomenal consciousness
– A change in our concept of mind/consciousness might reveal
this apparent possibility to be an illusion
• So also for naturalizing the spiritual?
Changes to our conception of
the spiritual?
• E.g., must God be finite to be naturalized?
– No: there are infinite idealizations in many
discourses: e.g., the Turing machine (computation)
• But perhaps other changes are required, and
can be made, while still being concepts of the
spiritual
• A case where lack of precision ("slippage") in
the concept of spirituality may be an advantage
Could there be a science of
spirituality?
• Not asking about an external science (e.g.,
evolutionary account of why people have
spiritual views)
• But internal: a science that tests hypotheses
about the relation between spiritual events,
provides explanatory, even predictive theories
of them, etc.
• Perhaps required for naturalization?
• Just as a science of mind requires familiarity
with experience, a science of the spiritual
would require a familiarity with the spiritual
Coda: AI and God
The artificial/natural distinction
• What do we mean by "artificial"?
• Made by humans?
– No, because then making babies would be
a case of doing artificial intelligence
• Made by humans, by design?
– Better, but rules out possibility of other
species making artefacts
• Rather: Made by an agent, by design
Adam: The first AI?
• Consider: Judeo-Christian creation story
• Adam was made by an agent, by design
– In particular, he was "made, not begotten"
• So Adam was artificial: The first AI!
• But then everything is artificial?!
• Yes, but some artefacts are also
natural; natural = made by God
Is AI an oxymoron?
• Intelligence implies
autonomy
• Autonomy implies
responsibility
• So for something to be
intelligent, it must be
responsible for its
actions
• Artificiality implies
having been designed
• X having been designed
implies the
responsibility for X's
behaviour lies with the
designer
• So for something to be
an artefact, it must not
be responsible for its
actions
Responsibility in the case of
design
Designer
designs
Artefact
generates
is responsible for
Behaviour
Responsibility in the case of
no design
brings into being
Agent 1
Agent 2
is responsible for
Actions
Design and theodicy
• How can God be good, and yet be responsible for
evil?
• Traditional answer: God is not responsible for evil, we
are
• That implies that we were not designed by God (see
preceding diagrams)
• But traditional theology sees us as products of God's
design ("made, not begotten")
• If AI is a contradiction is terms, so also is the notion
of God as creator?
AI and theodicy
• To make systems more intelligent, AI researchers
elaborate their design
• But the more they do this, the more they limit the
autonomy of the systems they build
• How can AI researchers create an artefact, through
design, without thereby usurping its responsibility,
autonomy and intelligence?
• In other words:
– Can AI show how to resolve the problem of evil?
– Or: can proposed solutions to the problem of evil suggest a
way to do AI?
A middle way:
evolution/learning/adaptivity
nurtures/structures/guides
Evolved
Evolver
is responsible for
Actions
A twist in the debate
• If this is right, then not only is Darwinism
compatible with the view of God as creator…
• …but is in fact required for it!
– Any other means of divine creating would make
our actions too closely related to God's intentions
– This would render them his actions, not ours
– That would be not only a blow to our freedom, but
a blow against the omnibenevolence of God
Is attempting AI blasphemy?
Isn’t doing AI "playing God", and therefore blasphemy? Perhaps,
but consider:
• On the Judeo-Christian view, God created us in his/her image
• It follows immediately that part of that image is being a designer
of intelligent life
• So by striving to do such, we are only trying to fulfil the divine
potential God instilled in us
• Compare: Are we "playing God" when we strive to be like God in
other ways?
• On this view, we are only "playing God" when doing AI in the
same sense that we are "playing God" when we try to do good,
and to love
Thank you!
This lecture will be available soon, in video
("PodSlides"), audio and PowerPoint format,
at:
http://e-asterisk.blogspot.com
Comments welcome: [email protected]