John Searle

1
John Searle

• John Searle is Mills Professor of Philosophy of
  Mind and Language at the University of California
  at Berkeley
• His Chinese room argument is perhaps the most
  influential and widely cited argument against
  claims of artificial intelligence .

Chinese Room John Searle
2
John Searle

• This "infamous Chinese room argument" has been
  described by Stevan Harnad as having "already
  achieved the status of a minor classic" and as
  having "shook up the entire AI field" so
  considerably that "things still have not settled
  down since" .
• William Rapaport, for instance, deems it "a rival
  to the Turing Test as a touchstone of
  philosophical inquiries into the foundations of
  AI"
• On the other hand, Searle's argument has been
  decried, by Dennett as "sophistry" and, by
  Hofstadter, as a "religious diatribe against AI
  masquerading as a serious scientific argument"

Chinese Room John Searle
3
Weak versus Strong AI

• Weak AI modeling or simulating intelligence on a
  computer. Simulated hurricane does not blow down
  trees.
• Strong AI constructing actual intelligence on a
  computer. Must have intentionality

Chinese Room Weak Vs.Strong
4
Intentionality

• Intentionality is by definition that feature of
  certain mental states by which they are directed
  at or are about object and states of affairs in
  the world. Thus, beliefs, desires and intentions
  are intentionality states undirected forms of
  anxiety or depression are not

Chinese Room Weak Vs.Strong
5
Strong AI

• Thinking is a species of computation.
  (Computationalism/Functionalism)
• Universal Turing Machines can compute any
  computable function. (TuringChurch thesis)
• Digital computers are Universal Turing Machines.
• therefore,
• Digital computers can think. (Possible AI)

Chinese Room Weak Vs.Strong
6
Chinese RoomSimulating Simulation

• Suppose a computer program can simulate the
  understanding of Chinese by examining all the
  Chinese symbols it receives, consulting an
  internal look-up table and determining what
  symbols to send back as output.
• Suppose this program could pass the Turing Test
• Suppose Searle himself (who understands no
  Chinese) replaces the program
• Since Searle obviously would not understand
  Chinese, neither could the computer simulation.

Chinese Room Simulating Simulation
7
Chinese Room Main Issue

• The real issue is whether the right kind of
  computation / program is sufficient for having a
  mind. Does something think in virtue of having
  the right software?

Chinese Room Main Issue
8
Chinese Room Objections

• Systems Room knows Chinese, not man Searle
  suggests the man in the room internalize
  everything, then he is the whole system. But
  there is a real problem of scale.
• Robot Room lacks connection to world. The
  computer can only point to other data (even the
  mechanisms to control a robot are, internally,
  just device addresses) while human
  intentionality points outside the brain (e.g. to
  actual horses). But brains dont directly
  interact with the world either.
• Brain Simulator As long as it simulates only
  the formal structure of the sequence of neuron
  firings at the synapses, it wont have simulated
  what matters about the brain its ability to
  produce intentional states.

Chinese RoomObjections
9
Chinese Room Objections

• Combination Of 1, 2 and 3
• Other Minds How do you know that anyone has a
  mind?
• Many mansions Future developments

Chinese RoomObjections
10
Main Points

• But the main point of the present argument is
  that no purely formal model will ever be by
  itself sufficient for intentionality, because the
  formal properties are not by themselves
  constitutive of intentionality and they have by
  themselves no casual powers except the power,
  when instantiated, to produce the next state of
  formalism when the machine is running.

Chinese RoomObjections
11
Main Points

• Because formal symbol manipulations by
  themselves dont have any intentionality. They
  are meaningless-they arent even symbol
  manipulations, since the symbols dont symbolize
  anything. In the linguistic jargon, they have
  only a syntax but no semantics.

Chinese RoomObjections
12
Main Points

• Searle's primary argument remains the linking
  between the simulation of intelligence, with
  "real" intelligence. In the Chinese room he
  attempts to put this on a firm footing, but in
  the discussion he makes this claim more clear
  when he notes no one would expect to get wet
  jumping into a swimming pool full of ping-pong
  balls simulating water".

Chinese RoomObjections
13
Searle is NOT saying

• Machines cannot think
• Thinking organisms have to be biological
• Thinking is not symbol manipulation
• Searle is saying
• Hardware with the right casual powers AND
  software important
• Consciousness is necessary

Searle is NOT saying
14
Chinese RoomScientific American Article January
1990

• (A1) Programs are formal (syntactic).
• (A2) Minds have mental contents (semantics).
• (A3) Syntax by itself is neither constitutive of
  nor sufficient for semantics
• to the following, conclusion (p.27)
• (C1) Programs are neither constitutive of nor
  sufficient for minds.
• Searle then adds a fourth axiom (p.29)
• (A4) Brains cause minds.
• from which we are supposed to "immediately
  derive, trivially" the conclusion (p.29)
• (C2) Any other system capable of causing minds
  would have to have causal powers (at least)
  equivalent to those of brains.
• Finally, from the preceding, Searle claims to
  derive two further conclusions (p.29)
• (C3) Any artifact that produced mental
  phenomena, any artificial brain, would have to be
  able to duplicate the specific causal powers of
  brains, and it could not do that just by running
  a formal program.
• and
• (C4) The way that human brains actually produce
  mental phenomena cannot be solely by virtue of
  running a computer program.

Chinese Room Axioms
15
Some Observations

• Realistic implementation? What is the scale?
• Chinese room and computer cannot cause mental
  event but he never explains how the brain can
• Agrees consciousness is an emergent property, but
  claims physical medium must be right

Final Arguments
16
Some Observations

• The Chinese Room begs the question. The very
  existence of "understanding" is at stake the
  question is whether the Chinese Room
  "understands". Yet "understanding" is simply
  postulated of other entities-- human minds, and
  the human inside the Chinese Room. Searle cannot
  have it both ways.

Final Arguments
17
Some Observations

• The man in the Chinese Room represents the
  computer-- more precisely, the CPU, the tiny chip
  which actually executes the instructions that
  make up a program.
• It seems obvious the CPU really doesn't
  understand, and no one who believes in "strong
  AI" thinks it does.
• But note Searle's sleight of hand. The Chinese
  Room story tells us that the CPU doesn't
  "understand". But a page later he's trying to say
  that programs don't understand. ("Programs are
  neither constitutive of nor sufficient for
  minds.")

Final Arguments
18
BlockComputer Model of the Mindhttp//www.nyu.ed
u/gsas/dept/philo/faculty/block/papers/msb.html

• The mind is the program of the brain
• For cognitive scientists intelligent computers
  offer the chance to learn about the mind

BlockModel of the Mind
19
BlockDefinitions of Intelligence

• Two definitions
• the word..behavioral like the Turing test
• construction.. how it works (functionalism)

Block Definitions of Intelligence
20
BlockDefinitions of Intelligence

• Functionalism is a theory in philosophy developed
  as an answer to the mind body problem  because of
  objections to both identity theory and logical
  behaviorism. Its core idea is that the mental
  states can be accounted for without taking into
  account the underlying physical substrate (the
  neurons), instead attending to higher-level
  functions such as beliefs, desires, and emotions

Block Definitions of Intelligence
21
BlockHomunculi

• The mind consists of smaller and smaller and
  stupider and stupider homunculi until you finally
  reach to level of a mechanical homunculi

Block Homunculi
22
BlockIntelligence

• Intelligent capacities are understood via
  decomposition into a network of capacities
  executed by primitive processes (reductionism).

Block Intelligence
23
Block Multiplication
Multiplication is broken down into repetitive
addition
Block Intelligence
24
BlockPrimitive Processes

• For primitive processes the question of how they
  work is not a question for cognitive scientist.
• They are the only computational devices for which
  behaviorism is true.

Block Primitive Processes
25
Block Gates The primitive Processes for Adder
Block Primitive Processes
26
Block Gates The Primitive Processes for Adder

• Primitive processes considered equivalent if they
  have the same input and output
• Here gates can be realized in different ways
  different ways of representing bistable states
• Hardware irrelevant to computational description

Block Primitive Processes
27
Block Gates The Primitive Processes for Adder
Electrical AND gate open 0 closed 1
Block Primitive Processes
28
Block Gates The Primitive Processes for Adder
Cat and Mouse AND Gate hungry mouse 0 mouse
fed 1
Block Primitive Processes
29
Block Modeling the Mind

• Cognitive science attempts to model the primitive
  processes of the mind
• At this level hardware is irrelevant

Block Intelligence and Intentionality
30
Block Intentionality

• Intentionality is aboutness
• The thought that the moon is full and the
  perceptual state that the moon is full are both
  intentional states
• Pain is not an intentional state

Block Intentional Levels
31
Block Functionalism and Intentionality

• The method of functional analysis that explains
  intelligent processes by reducing them to
  unintelligent mechanical processes does not
  explain intentionality
• The parts of an intentional system can be just as
  intentional as the whole system the component
  processors of an intentional system can
  manipulate symbols that are about just the same
  things as the symbols manipulated by the whole
  system

Block Intentional Levels
32
Block Functionalism and Intentionality

• The multiplier was explained via a decomposition
  into devices that add, subtract and the like.
• The multiplier's states were intentional in that
  they were about numbers. The states of the adder,
  subtractor, etc., are also about numbers and are
  thus similarly intentional.

Block Intentional Levels
33
BlockIntentional Levels

• There is, however, an important relation between
  intentionality and functional decomposition
• Though the multiplier's and the adder's states
  are about numbers, the gate's representational
  states represent numerals
• In general the subject matter of representations
  shift as we cross the divide from complex
  processors to primitive processors.

Block Intentional Levels
34
BlockIntentional Levels

• Lowest intentional level are the primitive
  processes
• This level is the realization level where input
  and output are about things but they do not
  have parts that are about things

Block Intentional Levels
35
Block Symbols and Representations

• At the primitive process level, there is a shift
  of subject matter from abstractions like numbers
  or from things in the world to the symbols
  themselves .
• Difference between numbers and numerals are like
  the difference between the city of Boston and the
  word Boston.

Block Intentional Levels
36
BlockIntentional Levels

• The algorithm used by the multiplier notation is
  independent. It depends on the properties of the
  numbers represented, not the representations
  themselves.
• By contrast, the internal operation of the adder
  depends on binary notation, and its description
  speaks of numerals
• The adder works in binary, but not in other
  standard notations.

Block Intentional Levels
37
BlockIntentional Levels

• The designer has found a machine which has
  physical aspects that can be interpreted
  symbolically
• Under that symbolic interpretation, there are
  symbolic regularities that are isomorphic to
  rational relations among the semantic values of
  the symbols
• It is the isomorphism between these two functions
  that explains how it is that a device that
  manipulates symbols manages to add numbers.

Block Intentional Levels
38
BlockIntentional Levels

• The idea of the brain as a syntactic engine
  driving a semantic engine is just a
  generalization of this picture to a wider class
  of symbolic activities, namely the symbolic
  activities of human thought.
• The idea is that we have symbolic structures in
  our brains, and that nature (evolution and
  learning) has seen to it that there are
  correlations between causal interactions among
  these structures and rational relations among the
  meanings of the symbolic structures.

Block Intentional Levels
39
BlockComputational System

• As a computational system is decomposed reach a
  point where there is a shift in subject matter
  from the things in the world to the symbols
  themselves

Block Computational System
40
BlockSymbolic Representation

• Remember
• The map is not the territory

Block Symbolic Representation