Searle - Chapter 2 Flashcards
What is strong AI?
The belief that minds are a computational system; according to strong AI proponents, it’s estimated that minds can be uploaded onto hardware in about 40 years
What is weak AI in contrast to strong AI?
- Strong AI seeks to create artificial persons: machines that have all the mental powers we have, including phenomenal consciousness
- Weak AI seeks to build information-processing machines that appear to have the full mental repertoire of human persons
What is another way of saying strong AI?
Computational Theory of Mind (CTM)/ computational functionalism
What are Boolean operators or ‘logic gates’? What are the lessons of this?
- Where the function of ‘and’, ‘or’ and ‘not’ can be coded using a device called an ‘and gate’
- Shows (1) close connection between computers and human thought and (2) we can refer to physical devices which carry out/mirror the very function of components central to thinking, reasoning, and deductions
Which two questions does Searle seek to answer?
a) Can computers think?
b) Are minds computational systems? (Is CTM true?)
- N.B. if we say yes to a, doesn’t mean b is true as the mind is more than a thinking thing (e.g. consciousness, qualia…)
What is Crane’s definition of a computer? **
‘A device that processes representations (symbols) in a systematic way’
processing in a systematic way representation
What is the key idea of ‘processing in a systematic way’? What is significant about this?
Algorithms - we know what they are and we can, and have, create them
What is the significance of the Turing machine?
They reduce anything which we naturally recognise as an effective procedure to a series of simple steps performed by a very simple device (could we do with neurons??)
What did Hilary Putnam say about Turing machines?
As the first proponent of machine functionalism, he suggested that we should think of mental states as identical to the states of a Turing machine
What idea does Chomsky’s universal grammar support?
The general idea of systematic processes or rule-following are encoded in the brain
What is Chomsky’s universal grammar?
- Babies go through same stages in development no matter what language they’re learning
- Infants master language far faster than they should if they were completely blank slates
- Leads him to believe that everybody is born with an abstract set of rules that tell them what’s possible and what’s not
Distinction between Functionalism and CTM?
The involvement of symbols separates CTM from causal role Functionalism
- Causal role Functionalism is merely committed to the view that mental states are defined by their causal structures
- CTM functionalism says that this causal structure is computational i.e. a disciplined series of transitions among representations
What is special about representation through language?
Syntax and semantics creates meaning = ‘semantic compositionality’
What did Haugeland say completely in contrast to Searle?
‘if you take care of the syntax, the semantics will take care of itself’
What is Fodor’s Language of Thought Hypothesis?
LOTH: mental representations are literally words in a language of thought (‘mentalese’); thinking is a universal language that uses biologically fixed code analogous to computer machine code
How do advances in AI support CTM?
Eugene defeating the Turing Test/ ‘deep learning’ where computers learn new skills over time = human performances stimulated
With reference to Fodor’s LOTH, why might minds be computers?
- ‘Compositionality’ of semantics also applies to thinking itself; whatever it takes to think a thought, nothing more is needed to think a second; same as language
- Reasoning and deduction; brain is designed to grasp logic and follow rules; syntax enables deduction
- Any language depends on a set of rules (grammar)
- Chomsky’s Universal Grammar Theory = internal rule following drives learning
- Universal Turing Machine shows basic things can carry out complex things
Strengths of CTM?
- Avoids major criticisms of MBIT e.g. chauvinsim
- Advancements in AI
Issues for CTM (without reference to Searle)?
- Relies heavily on deductive reasoning as support for systematic mental processes BUT most reasoning is
abductive and inductive which don’t follow rules; CTM doesn’t account for how these can be reduced to ‘formal rules’ - Moreover, the mind involves more than thinking and so even if thought is computational, doesn’t follow that there’s nothing more to the mind than a device that manipulates systems systematically i.e. consciousness, qualia, intentionality e.g. feeling of pain isn’t always ‘about’ something so not a representation at all therefore can’t be manipulated into symobols
In chapter 2, what is Searle’s general response to the view that minds are computers?
- ‘A computer has syntax but no semantics’; Syntax is not sufficient for semantics
- A mind needs syntax and semantics
How does Searle use the Chinese room argument against strong AI?
- Somebody had a programme/rule book to put individual Chinese symbols together and create words in a way that they could give an outcome indistinguishable from that of a native speaker
- Searle argues it’s clear that there’s no understanding involved here; nobody would say this person understands Chinese
- Therefore, even if a computer manipulates symbols to give an outcome that passes the Turing Test, there’s no understanding involved and so supports Searle’s conclusion that syntax isn’t sufficient for semantics
CTM falls under the umbrella of functionalism - what are the main issues for functionalists?
- Fail to adequately deal with hard problem of consciousness/intentionality as illustrated by ‘absent qualia’ (Chinese nation)
- Possibility of Chalmers’ philosophical zombies which are functionally identical to us
- Jackson’s knowledge argument where Mary knows all functional facts about red
What is the system reply to the Chinese room? Is it strong?
- The system reply charges Searle with committing the fallacy of composition…
- Searle requires a focus on the man himself failing to understand Chinese but it’s the system as a whole that understands
EVALUATION
What is the systems reply to the Chinese room? Is it strong?
- The system reply charges Searle with committing the fallacy of composition…
- Searle requires a focus on the man himself failing to understand Chinese but it’s the system as a whole that understands
EVALUATION
What is the robot reply to the Chinese room?
- AI researchers say symbols must be grounded in experience of the physical world
- CR therefore fails because it’s disconnected from the external world and if a computer could interact with the outside world it could have understanding
Why does the robot reply fail to salvage CTM (Searle’s response)?
- Searle replies saying if the computer program within the robot is manipulating symbols, it still has no understanding as it’s merely a syntactical device
- Even if Searle accepted this objection, strong AI still fails because the criticism goes further than string AI’s assertion that all you need is symbol manipulation; the response itself implies it’s not enough to be a program to be a mind and so contradicts CTM
What is the brain simulator reply to the Chinese room?
If we could modify the program to simulate the actual sequence of neuron firings at the synapses of a native Chinese speaker; then understanding would take place
Why does the brain simulator reply fail to salvage CTM (Searle’s response)?
Searle, correctly, defines this as an argument from functionalism and so says that the man could operate water pipes with valves connecting them, where water connections correspond to a synapse in the Chinese person’s brain but ‘the man certainly doesn’t understand Chinese and neither do the water pipes’
Why is Searle’s response to the brain simulator reply somewhat weakened?
Searle may be accused of chauvinism by claiming only neurons can give rise to consciousness
Combination reply?
Suppose the systems, robot and brain simulator replies are all combined; Searle: ‘zero times three is naught’
How did Churchland/Pinkers object to CR thought experiment?
The danger of untutored intuitions: intuitions are unreliable when it comes to the nature of reality e.g. light and magnets are connected
Dennett in support of Churchland/Pinkers?
‘philosopher syndrome’: mistaking a failure in imagination for an insight into necessity
Compare the logic of BN with the logic of CR conclusion
BN: no neuron is conscious but macro brain states are conscious
CR: since nothing in the room understands Chinese, no understanding takes place
How is Searle inconsistent with his own reasoning in chapter 2?
- If no single neuron is conscious yet they can come together and create an aggregate of consciousness, it seems possible that a program with purely syntactical components can generate semantics with semantics being the macro/surface feature of the micro syntax
- Therefore, on his own logic, Searle must concede that syntax is sufficient for semantics