UBC Theses and Dissertations

UBC Theses Logo

UBC Theses and Dissertations

Machines cannot think 1966

You don't seem to have a PDF reader installed, try download the pdf

Item Metadata

Download

Media
UBC_1966_A8 G4.pdf [ 3.44MB ]
UBC_1966_A8 G4.pdf
Metadata
JSON: 1.0105248.json
JSON-LD: 1.0105248+ld.json
RDF/XML (Pretty): 1.0105248.xml
RDF/JSON: 1.0105248+rdf.json
Turtle: 1.0105248+rdf-turtle.txt
N-Triples: 1.0105248+rdf-ntriples.txt
Citation
1.0105248.ris

Full Text

MACHINES CANNOT ' THINK by Robert George G e l l B . S c , U n i v e r s i t y of B r i t i s h Columbia, 1 962 A THESIS SUBMITTED IN PARTIAL FULFILMENT OF THE REQUIREMENTS FOR THE DEGREE OF M.A. <' i n the Department of PHILOSOPHY We accept t h i s t h e s i s as conforming to the req u i r e d standard THE UNIVERSITY OF BRITISH COLUMBIA A p r i l , 1966 In presenting t h i s t h e s i s i n p a r t i a l f u l f i l m e n t of the requirements f o r an advanced degree at the U n i v e r s i t y of B r i t i s h Columbia, I agree that the L i b r a r y s h a l l make i t f r e e l y a v a i l a b l e f o r reference and study. I f u r t h e r agree that permission f o r extensive copying of t h i s t h e s i s f o r s c h o l a r l y purposes may be granted by the Head of my Department or by h i s r e p r e s e n t a t i v e s . I t i s understood that copying or p u b l i c a t i o n of t h i s t h e s i s f o r f i n a n c i a l gain s h a l l not be allowed without my w r i t t e n permission. Department of Philosophy The U n i v e r s i t y of B r i t i s h Columbia Vancouver 8, Canada Date, 29 A p r i l 1966 i ABSTRACT This paper i s a c r i t i c a l essay on the question "Can machines t h i n k ? " , w i t h p a r t i c u l a r a t t e n t i o n paid to the a r t i c l e s appearing i n an anthology Minds and Machines, A. R. Anderson e d i t o r . The general c o n c l u s i o n of t h i s paper i s that those arguments which have been advanced to show that machines can t h i n k , are i n c o n c l u s i v e . I begin by examining ra t h e r c l o s e l y a paper by H i l a r y Putnam c a l l e d "Minds and Machines" i n which he argues that the t r a d i t i o n a l mind-body problem can a r i s e w i t h a complex cybernetic machine.- My argument against Putnam's i s that e i t h e r there are no problems w i t h computers which are analogous to the ones r a i s e d by mental s t a t e s , or where there are problems wi t h machines, these problems do not have at bottom the same d i f f i c u l t i e s that human experiences r a i s e s . I then continue by' showing that a cybernetic machine i s an i n s t a n t i a t i o n of a formal system. This leads to a d i s c u s s i o n of.the r e l a t i o n s h i p between f o r m a l i t y and p r e d i c t a b i l i t y i n which I t r y to show that some types of machine are i n p r i n c i p l e p r e d i c t a b l e . In the next s e c t i o n I attempt to prove that any d i s c u s s i o n of outward signs of i m i t a t i v e behavior presupposes that some l i n g u i s t i c theory, such as a type r e d u c t i o n , has been s u b s t a n t i a t e d . The f o r c e of t h i s argument i s that such a theory has not i n f a c t been su b s t a n t i a t e d . I o f f e r some general theory about the complexity of concept-property r e l a t i o n s . i i F i n a l l y I give a demonstration that no t e s t or set of t e s t s can be found that w i l l be l o g i c a l l y s u f f i c i e n t f o r the a s c r i p t i o n of the concept '"capable of thought." I f t h i s i s s u c c e s s f u l , then I have shown that no t e s t can be found, which when a machine i s b u i l t to pass i t , i s l o g i c a l l y adequate f o r saying that that .machine can t h i n k . This argument, i s o f f e r e d as . f u r t h e r c r i t i c i s m of the I m i t a t i o n Game which A. M. Turing pro- posed as an adequate t e s t f o r t h i n k i n g s u b j e c t s . Besides the s p e c i f i c c o n c l u s i o n t h a t i n s u f f i c i e n t evidence,has been o f f e r e d to say that machines can t h i n k , t h i s paper o f f e r s a more general c o n c l u s i o n that most standard problems have at bottom a l i n g u i s t i c d i f f i c u l t y . However, t h i s general c o n c l u s i o n i s a broad s p e c u l a t i v e one to which the work i n t h i s paper, i s only a small e x e m p l i f i c a t i o n and as such r e f l e c t s mainly the f u r t h e r ambitions of the author. i v ACKN OWLEDGEMSNT I wish to thank my two t y p i s t s , Sue Reeves and A p r i l Toupin, and also Steye Porche who proof-read the manuscript and o f f e r e d some va l u a b l e suggestions. U n f o r t u n a t e l y I cannot acknowledge the help or encouragement of any of the members of the department i n the pr e p a r a t i o n of t h i s work. I should l i k e to dedicate t h i s t h e s i s to L o i s , whose e encouragement, though seldom acknowldged, was everpre.sent. V TABLE OF CONTENTS Section Page I. I n t r o d u c t i o n 1 I I . The Analogy Between Men and Machines h I I I . F o r m a l i t y and P r e d i c t a b i l i t y 19 IV. What i s Behavior? 27 V. A Test f o r Thinking 39 VI. Conclusion ...50 Footnotes -5*+ B i b l i o g r a p h y • " . . . 56 1 S E C T I O N I I N T R O D U C T I O N T h i s i s a p a p e r o n t h e q u e s t i o n " C a n m a c h i n e s t h i n k ? " a n d i t s g e n e r a l c o n c l u s i o n i s n e g a t i v e . I t i s d i f f i c u l t t o g i v e a n e x a c t c h a r a c t e r i z a t i o n o f t h e p r o b l e m s t h a t p h i l o s o p h e r s a r e i n t e r e s t e d i n w h e n t h e y d i s c u s s t h i s q u e s t i o n . H o w e v e r i t w o u l d b e f a i r l y s a f e t o s a y t h a t t h e p r o b l e m s a r e t h o s e p o s e d b y t h e r e c e n t a d v a n c e s i n d i g i t a l a n d a n a l o g u e c o m p u t e r s . T h e s e m a c h i n e s h a v e b e e n b u i l t t o p e r f o r m a g r e a t v a r i e t y o f h u m a n t a s k s a n d t h e q u e s t i o n n a t u r a l l y a r i s e s a s t o w h e t h e r o r n o t w e m u s t s a y o f s o m e ' s u p e r ' c o m p u t e r t h a t i t t h i n k s . I n t h i s r e s p e c t , o f c o u r s e , i t i s o f i n t e r e s t t o c o n s i d e r t h e d e f i n i t i o n o f a m e c h a n i c a l c o m p u t e r t o s e e i f t h e r e a r e a n y l i m i t a t i o n s s e r i o u s e n o u g h t o j u s t i f y u s i n w i t h h o l d i n g t h e d e s i g n a t i o n , ' c a p a b l e o f t h o u g h t 1 . B e f o r e w e c a n d e c i d e w h e t h e r o r n o t a m a c h i n e t h i n k s , a g r e a t n u m b e r o f s e c o n d a r y p r o b l e m s m u s t b e t a c k l e d a n d t h e s e p r o b l e m s a r e o f w i d e g e n e r a l p h i l o s o p h i c i n t e r e s t . F u r t h e r m o r e t h e p h i l o s o p h i c i m p o r t a n c e o f r e c e n t d e v e l o p m e n t s i n m a t h e m a t i c s a n d p h y s i c s m u s t a l s o b e a s s e s s e d . S o p o t e n t i a l l y t h e p r o b l e m " C a n m a c h i n e s t h i n k ? " c o u l d l e a d u s i n t o v e r y g e n e r a l p h i l o s o p h i c s p e c u l a t i o n . H o w e v e r , a n a r t i c l e b y A . M . T u r i n g 3 " i n 1950 s p a r k e d a w h o l e s e r i e s o f p a p e r s i n t h e p h i l o s o p h i c j o u r n a l s , s o m e o f w h i c h w e r e c o l l e c t e d b y A . R . p A n d e r s o n , i n a n a n t h o l o g y c a l l e d M i n d s a n d M a c h i n e s . T h i s p a p e r i s a c r i t i c i s m o f t h e m a i n a r g u m e n t s p r e s e n t e d b y t h o s e w h o f e e l t h a t m a c h i n e ' s c a n t h i n k w i t h p a r t i c u l a r a t t e n t i o n g i v e n t o t h o s e a r t i c l e s i n M i n d s a n d M a c h i n e s . T h e r e a r e s e v e r a l c o n c l u s i o n s a r r i v e d a t i n t h i s p a p e r . 2 The argument i n the second s e c t i o n attempts to show that there . i s no serious analogy between men and machines. That i s to say, no serious analogy i n the sense that those problems which are r a i s e d because of the uniqueness of human experience, are not r a i s e d w i t h very complicated computers. .In "the t h i r d s e c t i o n I show that a Turing machine i s f o r m a l , and as such i s , i n the important sense, p r e d i c t a b l e . The f o u r t h s e c t i o n i s an at t a c k upon the p o s s i b i l i t y of b u i l d i n g a computer to i m i t a t e human behaviour. The argument i s , that u n t i l c e r t a i n things are shown about our behavioural concepts, then the problem of i m i t a t i o n cannot a r i s e . Of course the f o r c e of t h i s argument i s that these things have not been shown. F i n a l l y i n the f i f t h s e c t i o n I t r y to show that no t e s t can ever be constructed which w i l l l o g i c a l l y be adequate f o r the a p p l i c a t i o n of the concept, 'capable of t h i n k i n g ' . This argument i s meant to undercut the long debate which has gone on c r i t i c i z i n g Turing's " I m i t a t i o n Game", which was proposed as a t e s t f o r t h i n k i n g . Most of the arguments of t h i s paper are an e x e m p l i f i - c a t i o n of a general p h i l o s o p h i c approach. This approach i s one i n which a t t e n t i o n i s focused on the concepts that we use. By doing t h i s a t t e n t i o n i s drawn to the complexity of these concepts, p a r t i c u l a r l y i n t h e i r l o g i c a l ' s t r u c t u r e . I t i s argued that too l i t t l e a t t e n t i o n i s given to the complexity of language, p a r t i c u l a r l y w i t h respect to our. behavioural concepts. At times i t i s argued-that u n t i l aome problems about the nature of concepts .are answered, then no d e c i s i o n about the p o s s i b i l i t y • 3 of c o n s t r u c t i n g robots can be made. So i n a sense, I t h i n k that the general nature of the t h e s i s of t h i s paper can be sa i d to be l i n g u i s t i c . There i s al s o a l a r g e r t h e s i s behind t h i s paper, but upon which none of the arguments depend. This i s the idea that most questions of p h i l o s o p h i c importance can be put i n the form of a problem about the l o g i c of concepts. I f t h i s i s so, and a systematic way can be found f o r d i s c o v e r i n g the l o g i c of concepts, then the main problems of philosophy can be solved w i t h i n a science of language. This paper does not attempt to e s t a b l i s h t h i s t h e s i s but r a t h e r i s meant to be i n some small way an e x e m p l i f i c a t i o n of i t . Thus the arguments of t h i s paper t r y to show that the problems connected w i t h t h i n k i n g machines can a l l be.given a l i n g u i s t i c i n t e r p r e t a t i o n ; although no attempt i s made to give a method f o r d i s c o v e r i n g the l o g i c of concepts. I should mention, f i n a l l y , t hat this'paper does not r e i t e r a t e i n any d e t a i l , the arguments which have already been made i n the many papers on t h i s subject. In f a c t i t i s assumed that the reader i s f a m i l i a r w i t h most of the arguments and i n p a r t i c u l a r that the reader i s very f a m i l i a r w i t h some s p e c i f i c a r t i c l e s . In some places t h i s paper i s an extension of some very thorough work by other philosophers. But i n general the c r i t i c i s m s of t h i s paper are very broad and are intended to undercut many of the standard ideas connected w i t h the problem "Can machines t h i n k ? " SECTION I I THE ANALOGY BETWEEN MEN AND MACHINES H i l a r y Putnam i n h i s paper "Minds and Machines" t r i e s to draw an analogy between the va r i o u s s t a t e s of a complex cybe r n e t i c machine ( c a l l e d a Turing machine) an d the correspon- ding s t a t e s of a human being. He maintains that a machine- "a?.s has l o g i c a l and s t r u c t u r a l s t a t e s , j u s t as a human has mental and p h y s i c a l s t a t e s , and also that those arguments which support the i d e n t i t y or n o n i d e n t i t y of mental and p h y s i c a l s t a t e s also show that the same t h i n g about l o g i c a l and p h y s i c a l s t a t e s . As we a l l know, a machine i s capable of a complete mechanistic (causal) e x p l a n a t i o n and has no hidden or otherwise mysterious p a r t s . Thus'if Putnam can s u s t a i n h i s analogy between men and machines, he t h i n k s that t h i s w i l l go some way (he does not th i n k i t would be conclusive) i n s u b s t a n t i a t i n g a mechanistic t h e s i s . I t i s my contention i n t h i s s e c t i o n that Putnam f a i l s to f i n d the analogy that he i s lo o k i n g f o r . Putnam's t h e s i s r e s t s on two main claims. He t r i e s to show that the p r o p o s i t i o n " I am i n st a t e A i f and only i f f l i p f l o p 3 6 i s on" i s , from the machine's point of view, s y n t h e t i c , or what i s taken to be the same t h i n g , at l e a s t e m p i r i c a l l y v e r i f i a b l e . ' This w i l l make i t analogous to the p r o p o s i t i o n " I am i n pain i f and only i f C - f i b r e s are sti m u l a t e d " and w i l l depend upon there being d i f f e r e n t methods of v e r i f i c a t i o n of sta t e A and f l i p f l o p 3 6 being on. His other c l a i m i s that the l o g i c a l - s t r u c t u r a l d i s t i n c t i o n i s analogous to the mind-body one i n that there can be a l o g i c a l d e s c r i p t i o n of the machine's computations j u s t as there i s a mental d e s c r i p t i o n of human 5 a c t i v i t y . I hope to show that even from the poi n t of view of the machine, the above p r o p o s i t i o n i s not s y n t h e t i c , and a l s o that the l o g i c a l - s t r u c t u r a l d i s t i n c t i o n i s not analogous to the mind-body d i s t i n c t i o n . Putnam considers a Turing machine 'T' which can be i n a number of s t a t e s , one of which i s named A. As he says, '"a Turing machine i s a device w i t h a f i n i t e number of i n t e r n a l c o n f i g u r a t i o n s , each of which i n v o l v e s the machine's being i n one of a f i n i t e number of s t a t e s , . . . " I presume that any p a r t i c u l a r s t a t e of T i s defined as a unique combination of c e r t a i n c i r c u i t s being a c t i v a t e d , c e r t a i n c i r c u i t .breakers being open, and c e r t a i n vacuum tubes operating and f u r t h e r that other c i r c u i t s are dead, other c i r c u i t breakers are closed and other vacuum tubes are not operating. I t may be the case however that the c o n d i t i o n of some components of the machine are i r r e v e l a n t i n the determination of some s t a t e , (say) s t a t e A. For the , d i s c u s s i o n i n t h i s s e c t i o n , l e t usi define s t a t e A as that s t a t e of a Turing machine i n which f l i p floip 36 i s on and a l l the other c i r c u i t s are i r r e v e l a n t . This l a s t c l a u s e , "and a l l the other c i r c u i t s are i r r e l e v a n t " can be expanded i n t o a f i n i t e l i s t . Instead of s p e c i f y i n g whether the other components of the machine should be closed or non-operational, we can say i f some c i r c u i t which i s i r r e l e v a n t that i t can be e i t h e r open or c l o s e d , e i t h e r o p e r a t i o n a l or non-operational. I n t h i s way we can expand the d e f i n i t i o n of s t a t e A i n t o a f i n i t e l i s t , such as, f l i p f l o p 36 i s . on, f l i p f l a p 1 i s e i t h e r on or o f f , f l i p flo\p 2 i s e i t h e r on or o f f , e t c . As I s a i d , t h i s d e s c r i p t i o n i s f i n i t e because there are a. f i n i t e number of components i n any 6 machine. We can now g e n e r a l i z e our d e s c r i p t i o n of what a state i s hy saying that any st a t e of a Turing machine i s equivalent to a l i s t of the va r i o u s components of the machine s t a t i n g e i t h e r that they are on, o f f , or e i t h e r on or o f f ; a c t i v e , i n a c - t i v e or e i t h e r ; a c t i v e or i n a c t i v e ; e t c . Thus any p a r t i c u l a r s t a t e could he p i c t o r i a l l y represented by a p l a n of the machine showing the c o n d i t i o n s of the various c i r c u i t s , c i r c u i t breakers, tubes, magnetic, f i e l d s , r e l a y s , e t c . ^ We could b u i l d i n t o T a sub-machine (sub-T) which could check the c o n d i t i o n of the various components of T and which would p r i n t - o u t (say) onto the input tape of T, the r e s u l t s that i t obtained'. I f we wish to check f o r the var i o u s s t a t e s t h a t T i s i n , i t w i l l s i m p l i f y our job considerably i f we determine what are the s u f f i c i e n t f e a t u r e s of each p a r t i c u l a r s t a t e which d i f f e r e n t i a t e i t from a l l the other s t a t e s of T. Then we could speak of the s u f f i c i e n t c o n d i t i o n s f o r any p a r t i c - u l a r s t a t e . There w i l l be many of the c h n f i g u r a t i o n s of st a t e B which are d i f f e r e n t from C or D but not from E or F. But that configurations;- of the va r i o u s components of T which i s s u f f i c i e n t to d i f f e r e n t i a t e some st a t e from a l l the others w i l l be c a l l e d the s u f f i c i e n t c o n d i t i o n s of that s t a t e . Now that we have t h i s machine, we can ask i t to v e r i f y the statement m I am i n state A when and only when f l i p flo.p 3 6 i s on."" To give a p l a u s i b l e s i t u a t i o n f o r t h i s to a r i s e , imagine that we have j u s t b u i l t T and t h e o r e t i c a l l y the p o s i t i o n of f l i p f l o p 3 6 should be the s u f f i c i e n t c o n d i t i o n f o r sta t e A. We ask the. machine to check (or as Putnam consid e r s , the machine i t s e l f considers checking) the above statement. The method . 7 would be t h e o r e t i c a l l y simple. The machine enters s t a t e A and sub-T repor t s the c o n d i t i o n of. f l i p f l o p .36. The machine then enters every other s t a t e (which i s a f i n i t e number) and compares the r e p o r t s of sub-T on f l i p f l o p 36 to the f i r s t . r e p o r t . I f the subsequent repo r t s are a l l d i f f e r e n t than the f i r s t one, the p r o p o s i t i o n i s t r u e . There i s however a vast p r a c t i c a l problem of g e t t i n g the machine to go through every other s t a t e , and making sure that none are missed. However, t h i s a s i d e , the statement seems open to an e m p i r i c a l s o l u t i o n , making i t s y n t h e t i c . • 7 Putnam wants to say' that i f some b r i g h t person r a i s e d the question of the i d e n t i t y of s t a t e A and f l i p . f l o p 36 being on, the same o b j e c t i o n s could be r a i s e d against i d e n t i t y i n the machine case as are r a i s e d i n the case of the i d e n t i t y of being i n p a i n and C - f i b r e s being s t i m u l a t e d . In the mind-boiy.case i t i s argued that since there are d i f f e r e n t ways of knowing about the s t a t e s to be i d e n t i f i e d , the two states could not be i d e n t i c a l . These same co n s i d e r a t i o n s hold i n the machine case. The way that T determines the s t a t e of f l i p f l o p 36, i s from the reports of sub-T, and the way that i t determines what.state i t i s i n , i s from the o r i g i n a l input order to enter the s t a t e . So there are two d i f f e r e n t ways of knowing about the two s t a t e s . Thus st a t e A i s not i d e n t i c a l w i t h . f l i p f l o p 36 being on". At t h i s ' p oint Putnam leaves the reader w i t h the choice of saying e i t h e r there i s - a 'mind-body' (or l o g i c a l - p h y s i c a l state) problem w i t h machines or e l s e the • human mind-body problem i s merely l i n g u i s t i c . Before we take Putnam's choice, l e t us go back and see whether or not the co n s i d e r a t i o n s are a c t u a l l y p a r a l l e l . I gave an example e a r l i e r .in which the statement '''I am i n st a t e A i f and only i f f l i p f l o p 36 i s closed" was s y n t h e t i c . But the example I gave to i l l u s t r a t e t h a t , was the case of checking the oper a t i o n of some machine which had j u s t been constructed. . I t i s a normal assumption i n the d i s c u s s i o n of machines that we are only considering ' t h e o r e t i c a l 1 machines; i . e . , those that never have mechanical f a i l u r e s . I assume that 8 9 Putnam i s t a l k i n g about the same machines that Turing, Church, and Davis"1"^ were, and these were t h e o r e t i c a l machines. I f T i s a t h e o r e t i c a l machine then, the case I gave to i l l u s t r a t e the s y n t h e t i c nature of the statement could not a r i s e . By dea l i n g w i t h t h e o r e t i c a l machines we e l i m i n a t e the p o s s i b i l i t y . o f m a l f u n c t i o n i n the machine, so the problem of seeing whether or not the machine f u n c t i o n s as designed cannot a r i s e . But perhaps there i s a f u r t h e r sense i n which the statement i s s y n t h e t i c . I s n ' t i t an e m p i r i c a l question as to whether or not the p o s i t i o n of f l i p f l o p 36 i s the s u f f i c i e n t c o n d i t i o n of state A, i . e . i s the p o s i t i o n of f l i p flo.p 36 the fea t u r e of the i n t e r n a l c o n d i t i o n of the machine which makes the sta t e A d i f f e r e n t from a l l other s t a t e s ? But t h i s question i s not the o r i g i n a l question but ra t h e r the one as to whether " I am i n st a t e A i f f l i p flo.p 36 i s cl o s e d . " This i s of course r a t h e r obvious because the necessary and s u f f i c i e n t c o n d i t i o n s are a complete d e s c r i p t i o n of the c i r c u i t s , c i r c u i t breakers, and tubes being on,off or e i t h e r , i n the proper c o n f i g u r a t i o n 9 f o r s t a t e A. That these c o n f i g u r a t i o n s are the proper ones i s not an e m p i r i c a l or s y n t h e t i c question but r a t h e r a question of naming or d e f i n i n g j u s t which c o n f i g u r a t i o n would be s t a t e A. Since t h statement " I am i n s t a t e A i f and only i f f l i p f l o p 36 i s c l o s e d " i s ane about the necessary and s u f f i c i e n t c o n d i t i o n s of s t a t e A, i t i s a matter only of the way the machine was set up; i . e . , a matter of the i n i t i a l s t i p u l a t i o n . That t h e d d e f i n i t i o n - o f s t a t e A i s a matter of i n i t i a l s t i p u l a t i o n though, does not prevent the question about the d e f i n i t i o n of s t a t e A being asked. The machine may consider, or some programmer u n f a m i l i a r w i t h T may consider, the t r u t h o f the proposal " I am i n s t a t e A i f and only i f - f l i p f l o p 36 i s on". This w i l l be a d i f f i c u l t , but not i n s o l u b l e problem, but t h i s alone w i l l not show the p r o p o s i t i o n to be synethic. The i n i t i a l assumptions - of any system, or the o r g i n a l c o n s t r u c t i o n a l correspondences of any machine, may be d i f f i c u l t to determine but t h i s does not prevent them from being s t i p u l a t i o n s (or axioms or d e f i n i t i o n s ) . Thus the f a c t that there i s q u i t e a problem, which one may f a i l to s o l v e , i n a s c e r t a i n i n g the i n i t i a l s t i p u l a t i o n s of the various s t a t e s of the machine., does not show that these c o r r e l a t i o n s (namings, s t i p u l a t i o n s ) are not a n a l y t i c . The main argument however i s , that the ways of determining s t a t e A and the p o s i t i o n of f l i p f l o p 36 are d i f f e r e n t , and thus i t . seems an e n t i r e l y contingent matter whether or not the two things are i d e n t i c a l . "For i n s t a n c e , " Putnam says, "the machine might b e ' i n s t a t e A and i t s sense organs might report that f l i p f l o p 36 was not on." In which case the machine would have to decide 10 whether to say the p r o p o s i t i o n was f a l s e or to a t t r i b u t e the discrepancy to o b s e r v a t i o n a l e r r o r . This problem which Putnam poses f o r the machine could never a r i s e w i t h a Turing machine because we are assuming that the machine f u n c t i o n s c o r r e c t l y • f o r as long as we want i t t o . So there i s no p o s s i b i l i t y of an ob s e r v a t i o n a l e r r o r i n a Turing machine, and i f there was an ''observation' of f l i p f l o p 36 being o f f when the machine was i n sta t e A, then the only c o n c l u s i o n i s that the given statement i s f a l s e . But i f the exact problem which Putnam r a i s e s can 1not a r i s e , s t i l l we have the f a c t that there are two independent ways of v e r i f y i n g each, part of the proposition., r.->\<; -.. ~ ; c S ' ^ : : T h e way the machine determines the p o s i t i o n of f l i p f l a p 36 i s from the.input r e p o r t of sub-T. But how does T determine which s t a t e i t i s i n ? The machine determines t h i s from the i n i t i a l input order which was given i t (or even which i t gave i t s e l f . ) .At no time does the machine d i r e c t l y observe that i t i s i n st a t e A as Putnam claims. The machine i n f e r s from the evidence of the input order to the a c t u a l i n t e r n a l c o n f i g u r a t i o n . Also the machine i n f e r s from the evidence of ' the input r e s u l t s of sub-T to the a c t u a l i n t e r n a l c o n f i g u r a t i o n . Thu's in^determining whether i t i s i n state A or whether c i r c u i t breaker 36 i s on, the machine makes an" i n f e r e n c e from evidence which i s presented to i t . Although the evidence i s d i f f e r e n t , the method of v e r i f i c a t i o n i s the- same i n both cases. Furthermore, sincerwe are d e a l i n g w i t h t h e o r e t i c a l machines, we assume that no mechanical f a i l u r e s occur and that there have been no mistakes i n programming. So, f o r a Turing machine i t i s not p o s s i b l e that T be given an order and f a i l to 11 execute i t or that sub-T report i n c o r r e c t l y . Thus both the input order and the rep o r t of sub-T become d e f i n i t i o n a l c r i t e r i o n f o r s t a t e A. Therefore, i f the p r o p o s i t i o n " I am i n stat e A" means that the machine has been given the order to enter s t a t e A, e i t h e r by i t s e l f or some programmer, and since f l i p flo.p 36 being on i s a necessary c o n d i t i o n f o r stat e A, then the- p r o p o s i t i o n " I am i n stat e A i f and only i f f l i p flo.p 36 i s on" i s a n a l y t i c f o r a Turing machine. On both accounts then the case of the machine i s d i f f e r e n t from the human case. The p r o p o s i t i o n '"I am i n stat e A i f and only i f f l i p flo.p 36 i s on'" i s a n a l y t i c whereas the analogous p r o p o s i t i o n " I am i n pai n i f and only i f my C - f l b r e s are s t i m u l a t e d " i s s y n t h e t i c . The ways i n which the machine v e r i f i e s both the s t a t e i t i s i n and the c o n d i t i o n of f l i p flo.p 3.6 are the same-. Whereas., i n the human case there i s an i n - p r i n c i p l e d i f f e r e n c e between the ways of v e r i f y i n g that one i s i n p a i n and that one's C - f i b r e s are st i m u l a t e d . So Putnam has not b u i l t an analogous case w i t h Turing machines. Putnam then turns to showing that the question of whether a machine 'knows' what s t a t e i t i s i n , i s a degenerate 12 question. I f he can show that i t i s degenerate .in a way that the s i m i l a r question about human knowledge of mental sta t e s i s , t h i s w i l l add more evidence to the analogy between l o g i c a l s t a t e s of a machine and mental s t a t e s of a human. So he compares the two questions "Does the machine ' a s c e r t a i n ' that i t i s i n s t a t e A?" and "Does Jones 'know' that he i s i n p a i n " i n order to show that questions about the meithod of a t t a i n i n g knowledge of i n t e r n a l machine s t a t e s . He hopes to show that 12 .they are both degenerate, but I s h a l l argue that the questions about machine methods are e i t h e r not degenerate or i f they are, not f o r the same reasons that questions of method are f o r mental s t a t e s . There i s one obvious sense i n which i t can e a s i l y be s a i d that the machine computed s t a t e A, and that i s the case where the machine goes through a s e r i e s of c a l c u l a t i o n s which terminates i n s t a t e A. But I take i t that Putnam i s i n t e r e s t e d i n the case of whether or not a machine can be s a i d to compute that i t i s i n s t a t e A from stat e A alone. Before considering the question though, we must add one more fe a t u r e to our' machine T, by supposing that' whenever the machine i s i n one p a r t i c u l a r s t a t e (say s t a t e A), i t p r i n t s out the words " I am i n s t a t e A". This can be done i n two ways: e i t h e r every time we give the machine an i n s t r u c t i o n to enter s t a t e A, we next give i t the i n s t r u c t i o n to p r i n t out "I am i n s t a t e A", or e l s e we can have the machine so constructed that every time i t enters s t a t e A i t a l s o p r i n t s out " I am i n s t a t e A". The question may now a r i s e "Does the machine ' a s c e r t a i n ' that i t i s i n s t a t e A?" According to Putnam, ' a s c e r t a i n ' i s synonymous w i t h 'compute' or 'work out'; so the question can be rephrased as "Does the machine ' a s c e r t a i n 1 (or compute or work out.) that i t is. i n a s t a t e 13 A?" I f we have a machine i n which a f u r t h e r i n s t r u c t i o n i s given i t to p r i n t out " I am i n ' s t a t e A", then the answer to the above question i s yes, and. the answer to the f u r t h e r query about how i t a s c e r t a i n s or w o r k s • i t out i s given by showing the programming r e q u i r e d . In t h i s p a r t i c u l a r case i t i s a matter of the i n s e r t i o n of a sub-routine (granted i t i s a short one of 1 3 one i n s t r u c t i o n ) a f t e r the i n s t r u c t i o n to enter s t a t e A. So i f we -have t h i s type of machine, the question i s not degenerate. But i f we have a machine that has b u i l t i n t o i t a programme such that every time i t enters s t a t e A i t p r i n t s out "I am i n st a t e A", then the p r i n t i n g out becomes part of the d e s c r i p t i o n , and thus a d e f i n i t i o n a l c o n d i t i o n of .the machine being i n state": A. (Mechanical e r r o r s are t h e o r e t i c a l l y e l i m i n - ated.) I f t h i s i s the case then i t l o s e s i t s analogy w i t h the human s i t u a t i o n of someone 'evincing' " I am i n p a i n " , f o r the v e r b a l statement i s not^'part of. the d e s c r i p t i o n of p a i n and not a d e f i n i t i o n a l c o n d i t i o n of being i n pain. The question about the machine a s c e r t a i n i n g or computing that i t i s i n state A becomes degenerate because the f a c t that the machine p r i n t e d out " I am i n st a t e A" i s a d e f i n i t i o n a l c r i t e r i o n of the machine's being i n s t a t e A. Putnam says that the d i f f i c u l t y of degeneracy has, i n both cases the same cause: "namely, the d i f f i c u l t y i s occasioned by the f a c t that the v e r b a l r e p o r t ( I am i n state A and I am i n pain) issued d i r e c t l y from the state i t r e p o r t s . . . But the p r i n t out " I am i n state A" i s not a r e p o r t , but a part of what i s s t i p u l a t e d as being i n st a t e A; r e p o r t s can be mistaken, but not d e f i n i t i o n a l c r i t e r i o n . The question about the machine computing " I am i n st a t e A" from stat e A i s a d e s c r i p t i o n because part of what i s set up i n t h i s machine asstate A i s a d e s c r i p t i o n of the p r i n t - o u t mechanism p r i n t i n g " I am i n st a t e A", and not as Putnam t h i n k s because '"I am i n st a t e A" issues d i r e c t l y from the machine's being i n st a t e A. However the statement '"I am i n p a i n " , i f i t i s degenerate, i s not; so f o r these reasons. In the human case, a person saying t h a t they are i n p a i n i s not. a necessary c o n d i t i o n e i t h e r f o r them knowing themselves'that they are i n p a i n nor f o r someone el s e knowing that they are i n pain. The r e l a t i o n between the statement " I am i n p a i n " and the p a i n i s q u i t e contingent, and i t i s t h i s f a c t which gives r i s e , i n the human s i t u a t i o n , to. the question of knowing about the p a i n i n order to 'evince 1' " I am i n pain'"1. This analogous s i t u a t i o n does not a r i s e i n a Turing machine. So the question of how a machine computes or works out what s t a t e i t i s i n , i s not u s u a l l y degenerate, but when the question i s , i t i s not degenerate f o r the reasons that questions of knowing pa i n ( i f those questions are a c t u a l l y degenerate) .are. To continue h i s analogy between machines and humans, Putnam shows that there are two types of machine s t a t e s , l o g i c a l s t a t e s and s t r u c t u r a l s t a t e s , and that these are analogous t o t t h e mental and p h y s i c a l s t a t e s of human beings. As I mentioned e a r l i e r , any t h e o r e t i c a l Turing machine i s capable of being i n a f i n i t e number of, s t a t e s , A, B, C, and i f the va r i o u s programmes of t h i s machine are already i n memory, then the machine w i l l change from one s t a t e to another according to i t s programming. But as Putnam says "a given 'Turing machine' i s an a b s t r a c t machine which may be p h y s i c a l l y r e a l i z e d i n an almost i n f i n i t e number of d i f f e r e n t ways," y and, f o r ' any p a r t i c u l a r manufactured machine the p h y s i c a l c o n d i t i o n of i t mayvyary from one c o n d i t i o n to another. Thus any a c t u a l machine may be i n a number of p h y s i c a l or s t r u c t u r a l states and yet may ' s ' t i l l be i n the same l o g i c a l s t a t e . So f o r any p a r t i c u l a r machine i t can be thought of or described as a f i n i t e number of l o g i c a l s t a t e s or as a number of s t r u c t u r a l s t a t e s , and the f u n c t i o n i n g of the. machine 15 can be-expressed e i t h e r e n t i r e l y i n terms of l o g i c a l s t a t e s , or again, e n t i r e l y i n s t r u c t u r a l s t a t e s . This i s , according to Putnam,' analogous to the human s i t u a t i o n i n which the f u n c t i o n i n g of the human can be explained i n terms of mental occurrences ( e . g F r e u d i a n explanation) or i n terms of p h y s i o l o g i c a l changes (e.g., complete behavioural d e s c r i p t i o n ) . In order to assess t h i s analogy, l e t us b a c k t r a c t to the d i s t i n c t i o n between l o g i c a i a a n d s t r u c t u r a l s t a t e s and consider b r i e f l y again j u s t what are l o g i c a l s t a t e s . When we set up a Turing machine, we said that i t could enter a f i n i t e number of s t a t e s , A, B, C, ... e t c ' These sta t e s r e f e r r e d to something more or l e s s e x p l i c i t ; namely the i n t e r n a l c o n f i g u r a t i o n of some h y p o t h e t i c a l machine. These sta t e s of the machine, A, B, C, ... must be e x p l i c i t , at l e a s t to the extent that we can see that we can b u i l d some machine that w i l l enter, these s t a t e s . Thus i f the p a r t i c u l a r s t a t e we are t a l k i n g about i s one i n which the machine places the input data i n t o memory space ^683, we must be able to show that a machine can be b u i l t which w i l l f u l f i l t h i s f u n c t i o n a n d consequently be able to enter t h i s s t a t e . This could be done by l a y i n g out on the d r a f t i n g board the p o s s i b l e c o n f i g u r a t i o n s of c i r c u i t s , r e l a y s , and vacuum tubes such that any machine which was b u i l t from these plans would be able to enter t h i s p a r t i c u l a r s t a t e . This requirement that the states of a Turing machine r e f e r at l e a s t to one p o s s i b l e c o n f i g u r a t i o n of a machine, i s a b s o l u t e l y e s s e n t i a l . Otherwise we would beg the e n t i r e question. I f we simply s a i d that the machine could f u l f i l such-and-such f u n c t i o n and we d i d not s p e c i f y how t h i s could be accomplished mechanically, then we would simply be 16 saying that machines can do whatever humans can and I presume that i t i s j u s t t h i s question of whether machines can do every- t h i n g humans can do that-we are t r y i n g to answer. So unless we beg the question, we must be able to s p e c i f y at l e a s t one mechanical c o n f i g u r a t i o n of a p o s s i b l e Turing machine f o r every s t a t e that we a t t r i b u t e to machine T. When we say that the i n t e r n a l c o n f i g u r a t i o n of s t a t e A must be s p e c i f i e d , we do not mean that i t must be e x p l i c i t l y l a i d out i n every minute d e t a i l . For example, i f i n s p e c i f y i n g s t a t e A we say that there must be a c i r c u i t j o i n i n g the scanner to the memory input compartment, we do not s p e c i f y the l e n g t h of the c i r c u i t , nor the chemical composition of the w i r e , nor even f o r that matter that i t must be a wire which c a r r i e s the impulse from one to'the other. In f a c t there i s no l i m i t to the v a r i o u s ways that such a c i r c u i t could be set up. (The c i r c u i t i s s p e c i f i e d by the f u n c t i o n , (or purpose, or g o a l ) , and thus there are an u n l i m i t e d number of a c t u a l mechanical ways of f u l f i l l i n g the p a r t i c u l a r purpose. We could a l s o have a messenger boy c a r r y the message, but t h i s would not be a mech a n i c a l s o l u t i o n . But we must show that there i s at l e a s t one mechanical s o l u t i o n . ) On the other hand, f o r any a c t u a l machine there w i l l be a complete p h y s i c a l d e s c r i p t i o n of the v a r i o u s c i r c u i t s , r e l a y s , tubes, e t c . , s p e c i f y i n g the a c t u a l p h y s i c a l make-up of the machine. But these s p e c i f i c a t i o n s must inc l u d e at l e a s t those s p e c i f i c a t i o n s which were l a i d down f o r the t h e o r e t i c a l s t a t e . That i s , those c o n d i t i o n s which we s p e c i f i e d f o r the T machine to be i n s t a t e A must be included i n (or d e d u c t i b l e from) the p h y s i c a l s p e c i f i c a t i o n s of t h i s - a c t u a l machine-, although these 1 7 p h y s i c a l s p e c i f i c a t i o n s w i l l a l s o , d e s c r i b e many p r o p e r t i e s which were not included i n the t h e o r e t i c a l c o n s i d e r a t i o n s of st a t e A. Our i n i t i a l s p e c i f i c a t i o n of the p r o p e r t i e s of st a t e A was ab s t r a c t i n the sense t h a t i t l e f t open to the engineer b u i l d i n g the machine many other p r o p e r t i e s to be s p e c i f i e d before the machine could be b u i l t . But the computer's p h y s i c a l or s t r u c t u r a l d e s c r i p t i o n of st a t e A w i l l d i f f e r from the t h e o r e t i c a l or l o g i c a l d e s c r i p t i o n of st a t e A only i n that i t describes more p r o p e r t i e s f o r the machine. Thus i f we t h i n k of the s t r u c t u r a l d e s c r i p t i o n as designating a set of p r o p e r t i e s and co n d i t i o n s of T, the l o g i c a l d e s c r i p t i o n w i l l be a sub-set of these. Now- i t i s u s u a l l y thought that the d i f f e r e n c e between mental s t a t e s and p h y s i c a l s t a t e s i s one of a more serious natura than j u s t that mental s t a t e s have the same but fewer p r o p e r t i e s than p h y s i c a l s t a t e s . I t i s g e n e r a l l y thought that the t e s t f o r determining p h y s i c a l p r o p e r t i e s are not a p p l i c a b l e to the p r o p e r t i e s of mental s t a t e s . Most of the-.philosophical specul- a t i o n of the l a s t few years has been an attempt to f i n d some i d e n t i t y p r i n c i p l e between the p r o p e r t i e s of our mental st a t e s and those, p r o p e r t i e s which are o b j e c t i v e l y a t t r i b u t e d to other people. .Putnam doesn't even need an i d e n t i t y p r i n c i p l e because there i s only one type of property. He has f a i l e d to f i n d two types of things between which we need to f i n d some bridge or connection. From a complete p h y s i c a l d e s c r i p t i o n of a machine we can deduce the t h e o r e t i c a l d e s c r i p t i o n , but u n t i l some i d e n t i t y p r i n c i p l e i s afforded by Putnam or someone e l s e , we cannot deduce the mental d e s c r i p t i o n of a person from h i s p h y s i c a l c o n d i t i o n . This i d e n t i t y p r i n c i p l e which would bridge 1 8 the g u l f between mental and p h y s i c a l s t a t e s may yet he found by philosophers,, nevertheless,, what i s c e r t a i n l y true i s that some p r i n c i p l e i s needed. In the case of a Turing machine there i s no p r i n c i p l e needed because Putnam has f a i l e d to show that there i s a type d i f f e r e n c e between the p r o p e r t i e s of l o g i c a l and p h y s i c a l s t a t e s . Therefore the d i f f e r e n c e between a l o g i c a l and p h y s i c a l d e s c r i p t i o n of a'machine i s not analogous to the d i f f e r e n c e between a mental and p h y s i c a l d e s c r i p t i o n of some person's p a i n (say). Thus I conclude that the l o g i c a l - s t r u c t u r a l d i s t i n c t i o n w i t h machines i s not analogous to the mental-physical d i s t i n c t i o n i n the human s i t u a t i o n . The c o n c l u s i o n of t h i s s e c t i o n i s not that there are no problems to be answered or d i s t i n c t i o n s to be made w i t h complex Turing machines. The co n c l u s i o n i s rat h e r that the problems .raised or the questions asked by a Turing machine a b o u t . i t s e l f are not.problems f o r the same reasons that s i m i l a r questions about humans are. The machine may ask i t s e l f questions of the same form as humans may,' but the d i f f i c u l t y i s not the same d i f f i c u l t y that a human has. S i m i l a r l y , many d i s t i n c t i o n s can be drawn i n d e a l i n g w i t h complex machines, but these a l s o , I conclude, are not the same d i s t i n c t i o n s which philosophers have noted i n the human case. Thus the problems which a complex Turing machine might face are not the same as those that humans t r y to answer, and i n t h i s sense the analogy between men and machines f a i l s . . " - ' 1 9 SECTION I I I FORMALITY AND PREDICTABILITY' •In t h i s s e c t i o n , I wish to show that a Turing machine i s a concrete i n s t a n t i a t i o n of a formal' system, and as such, i s p r e d i c t a b l e . My demonstration that Turing machines are formal i s not unique but I f e e l that i t i s important that i t should be shown ra t h e r e x p l i c i t l y . Many p h i l o s o p h e r s have argued that i f a Turing machine i s formal then Godel 1s Incompleteness Theorem can help us 1 6 to some i n t e r e s t i n g conclusions about machines. 'Some, siich as Lucas have argued that the Theorem r e f u t e s mechanism; others, such as 17 18' • ' Putnam.' and Turing have argued t h a t the Theorem has no bearing on the i n t e r e s t i n g p h i l o s o p h i c questions. I s h a l l argue, on the other hand, only that Turing machines are formal and that i n the important sense that philosophers have concerned themselves with,- these machines are ' predictable'. • Before entering the problem of showing any l i m i t a t i o n s of a Turing machine, we must demonstrate rather' c l e a r l y that any Turing machine can be represented a's a formal system. My demonstration of t h i s i s e s s e n t i a l l y the one used by M a r t i n Davis i n the f i r s t 1 9 chapter of h i s book, Computability and U n s o l v a b i l i t y . As I explained i n the f i r s t s e c t i o n , a machine can be i n . any one of a number of c o n f i g u r a t i o n s , q̂ ', qg, q-^,. . .' up to some-finite l i m i t . A tape, d i v i d e d i n t o d i s c r e e t u n i t s , i s fed i n t o the machine and i n each u n i t there appears a l e t t e r of a language comprising a number of symbols, S Q , S ^ , Sg,... up to some f i n i t e number. Furthermore the tape i s f i n i t e , but can be as long as i s needed. One of the e s s e n t i a l f u n c t i o n s of a Turing machine i s that i t i s a b l e , upon the r e c e i p t of a symbol, to change from one s t a t e (say) q^ to another s t a t e q^. Not only can a Turing machine change states but i t can also 20 change the symbol on the scanned u n i t or i t can move the tape along so that the next u n i t i s scanned. This p o s s i b i l i t y of changing can be represented by a quadruple, such as, q ^ S p ^ . The machine that t h i s i s a quadruple of,' w i l l , i f i t i s i n s t a t e q 1 ,and i s scanning symbol S^, change to s t a t e q 2 , and erase S 1 and put the Tsymbol S 2 i n the scanned tape u n i t . More g e n e r a l l y , a quadruple stands f o r a machine b u i l t to c a r r y out any i n s t r u c t i o n of the f o l l o w i n g form: when i n stat e q and the symbol S„ i s on the tape u n i t being scanned then change to s t a t e q (x^, y or y ^ x) and e i t h e r change the symbol on the scanned tape u n i t to S or e l s e scan the u n i t to the r i g h t or l e f t . I f a machine i s capable of f o l l o w i n g out an i n s t r u c t i o n of that form,- then i t can be represented by a quadruple. I t i s important to n o t i c e that a f t e r the machine has c a r r i e d out t h i s i n s t r u c t i o n , i t i s i n the o r i g - i n a l p o s i t i o n again i n that i t i s i n some st a t e w i t h a scanned u n i t i n f r o n t of i t . Thus the machine i s ready'to.carry out another i n s t r u c t i o n of the same form. However, i f . t h e r e i s no such i n s t r u c t i o n b u i l t i n t o the machine, then when i t reaches that s t a t e and symbol, the machine w i l l stop. Thus any machine which goes through a process or s e r i e s of changes from one p o s i t i o n to another can be represented by a s e r i e s of quadruples. Since the number of st a t e s and symbols i s f i n i t e , the number of quadruples w i l l a l s o •be f i n i t e . Therefore a l l the p o s s i b l e movements of the machine can be described by a s e r i e s of quadruples, so that t h i s s e r i e s a c t u a l l y defines the machine"s p o s s i b i l i t i e s . Any p a r t i c u l a r Turing machine can be represented, then, by a s e r i e s of quadruples. But as I s a i d , when the machine has f i n i s h e d one change i t i s i n a p o s i t i o n to c a r r y out another. This continuous change of the machine i s represented by a s e r i e s 21 of deductions. I f we take the -tape to he given f o r any p a r t i c u l a r machine, then by knowing which u n i t the machine w i l l scan f i r s t and the s t a t e that the machine i s i n when i t begins, we can deduce, using the l i s t of quadruples of that machine', the v a r i o u s steps that the machine w i l l go through to a r r i v e at the answer. So considering the q*s and S's as p r i m i t i v e words, and the o r i g i n a l tape as i n i t i a l axioms, and. the quadruples as r u l e s of i n f e r e n c e , we have constructed an axiomatic system which w i t h the a d d i t i o n of a few more s t i p u l a t i o n s can be made qui t e formal. And t h i s system rep r e s e n t s , i n symbolic terms the various'changes that a Turing machine would go through i n any a c t u a l problem. I s h a l l i n what f o l l o w s , s t a t e t h i s f a c t r a t h e r b r i e f l y by saying 21 t h a t a machine i s a concrete i n s t a n t i a t i o n of a formal system. F i n a l l y , any theorems which apply to formal systems, as formal systems, w i l l a l s o apply to Turing machines. I f we consider a computer.as a d i s c r e t e s t a t e machine whose motion f o l l o w s some formal system, then i t seems that whatever the machine does i s p r e d i c t a b l e . I f we know the i n i t i a l s t a t e of the computer and we know i t s complete l i s t of quadruples then we can p r e d i c t what the machine w i l l do once we see i t s tape. However, does i t f o l l o w from the f a c t that a machine i s formal that i t i s p r e d i c t a b l e , and f u r t h e r , i f the machine i s not p r e d i c t a b l e does t h i s show that i t i s not formal? Now there are s e v e r a l reasons to suggest that a computer i s not p r e d i c t a b l e . One reason may be'that we doiib-t have enough know- ledge of "the machine. For example, we carP'ctt p r e d i c t ( i n general) when a complicated piece of machinery w i l l break down because we don't know enough about the manufacture or s t r u c t u r a l 22 composition of the v a r i o u s p a r t s . But we a t t r i b u t e the i n a b i l i t y to p r e d i c t simply to our l a c k of knowledge which we f e e l that we could get, givah enough time and l a b o r a t o r y space. That i s , we hold that f o r these reasons machines are not ' i n p r i n c i p l e ' u n p r e d i c t a b l e . However, there are other reasons f o r the unpred- i c t a b i l i t y of computers which stem from our i n a b i l i t y to get knowledge. But t h i s i n a b i l i t y i s not a p r a c t i c a l matter but a t h e o r e t i c a l one. I'take i t - t h a t the' i m p l i c a t i o n of Heizenberg'-s U n c e r t a i n t y p r i n c i p l e i s that measurements below f i x e d amounts are not p o s s i b l e , f o r the more a c c u r a t e l y we measure the p o s i t i o n of a p a r t i c l e the more inaccurate w i l l be our measurement of i t s momentum. So much so that i f we ever d i d measure the p o s i t i o n of a p a r t i c l e completely a c c u r a t e l y then we would n e c e s s a r i l y have made an i n f i n i t e e r r o r i n : i t s momentum. Thus, considering measurements of the utmost accuracy, we must, i n p r i n c i p l e , have a f i n i t e magnitude of e r r o r , and we are unable to p r e d i c t g r e a t e r anything i n t o an accuracy^than the accuracy of the accumulated e r r o r s . However as I s a i d , we are d e a l i n g w i t h measurements of great accuracy and of course we w i l l be measuring sub-atomic s t r u c t u r e s . For i f we want to make a measurement of something to the greatest accuracy we w i l l have to consider the object as a c o l l e c t i o n of sub-atomic p a r t i c l e s . But i f we consider the object or machine as a macroscopic u n i t , then usihg-macroscopic measuring d e v i c e s , we can, w i t h i n experimental e r r o r , measure, t e s t , and..predict the movements of the mechanism. So i f we spent a great de a l of time t e s t i n g the v a r i o u s parts of some machines, the above reasons would not be s u f f i c i e n t to show that any machine i s i n p r i n c i p l e u n p r e d i c t a b l e i n macroscopic u n i t s . I t i s g e n e r a l l y contended, however, that computers 2 3 i z which c o n t a i n randoming devices'are i n p r i n c i p l e u n p r e d i c t a b l e . I want .to examine two types of randomizers, (a) a counter of the number of radium atoms to have d i s i n t e g r a t e d i n the half-minute previous and (b).the decimal expansion of TT . I take the counter as an example o'f a device which we can never, r e g a r d l e s s of how much-knowledge we had,- p r e d i c t , i . e . the number which the counter has o n - i t at any moment- i s i n p r i n c i p l e u n p r e d i c t a b l e . The reason f o r our i n a b i l i t y to p r e d i c t may be due to- the v a r i a t i o n s which a f f e c t the d i s i n t e g r a t i o n of radium atoms being of such a small magnitude that the U n c e r t a i n t y P r i n c i p l e l i m i t s our i n v e s t i g a t i o n . (This would only show that we cannot i n v e s t i g a t e the laws governing d i s i n t e g r a t i o n although there may be' some.) But granting that there are i n the world counters which are u n p r e d i c t a b l e i n the strong sense that no increase i n knowledge w i l l ever a v a i l i n p r e d i c t i n g them, what can we say about computers which contain these devices? Presumably, a computer w i t h a random device w i l l work as f o l l o w s , the machine i s given the i n s t r u c t i o n to look at the tape u n i t t o t t h e r i g h t and there i s no symbol on that u n i t . The symbol i s not w r i t t e n on the u n i t u n t i l the tape i s i n the scanner and then the symbol which i s w r i t t e n on the u n i t i s determined by the random device. In t h i s way no one could p r e d i c t how the machine would operate a f t e r t h i s i n s t r u c t i o n because we could not, i n p r i n c i p l e , know what symbol would be on the tape u n t i l the machine a c t u a l l y d i d scan the u n i t . However t h i s example i s j u s t another case of adding more i n f o r m a t i o n to the machine during, i t s c a l c u l a t i o n s . We can c e r t a i n l y b u i l d machines that w i l l do some c a l c u l a t i o n s and then come to a h a l t u n t i l more i n f o r m a t i o n i s given to i t . This would be the case 2k where the machine works out the i n i t i a l tape i n p u t , and when i t stops we a l t e r the tape, which i s j u s t the same as g i v i n g i t a new tape. Then the machine w i l l work again t h i s problem. We can make t h i s more s o p h i s t i c a t e d by having the machine i t s e l f add more i n f o r m a t i o n to the tape at c e r t a i n stages of i t s c a l c u l a t i o n s . And the case of having a randomizing device i n the machine i s an example of adding more i n f o r m a t i o n , but the i n f o r m a t i o n can not be p r e d i c t e d . When we o r i g i n a l l y thought of the problem of p r e d i c t i n g a computer, we were t h i n k i n g of a machine which was given some c a l c u l a t i o n s to do. In terms of the machines formal system, the case of p r e d i c t a b i l i t y arose where we had a f i n i t e l i s t of quadruples and a given s e r i e s of tape expressions. Then i t was asked whether or not the machine-' s • movements could be p r e d i c t e d . This i s a l l q u i t e analogous to the human s i t u a t i o n where we give someone a. problem and then t r y to f i g u r e out what t h e i r behaviour w i l l be. But the o r i g i n a l problem was not one of t r y i n g to p r e d i c t how a,•machine would r e a c t when given more in f o r m a t i o n l a t e r i n the problem, i n f o r m a t i o n which we could not get ourselves. No one would t h i n k that you had shown a machine to be unpred- i c t a b l e i f you proved that we cannot f i g u r e out i n advance how the machine would r e a c t when unknown in f o r m a t i o n was fed i n t o i t . When we ask whether or not machines are p r e d i c t a b l e we are asking whether or not, given a machine and the i n f o r m a t i o n fed i n t o it,'we can p r e d i c t the subsequent movements of the machine.' The randomizing device feeds i n f o r m a t i o n i n t o the machine from w i t h i n the machine. But I do not t h i n k that t h i s changes the case at a l l . The tape that the machine scans i s changed 25 and that creates a new axiomatic beginning f o r the machine. The f a c t that the source of the i n f o r m a t i o n i s some device w i t h i n the p h y s i c a l bounds of the machine does not make the case, d i f f e r e n t than the one. where more i n f o r m a t i o n i s fed i n from o u t s i d e . I t may be thought, however, that I am p r e j u d i c i n g the case by making the randomizing device p e r i p h e r a l to the actual machine, and that a c t u a l l y the device can be b u i l t i n t o the ' e s s e n t i a l ' workings of the machine. I myself cannot see how t h i s randomizing e f f e c t could be expressed i n terms of quadruples and tape expressions except i n a way.similar to the one suggested above. I f we b u i l d the device i n t o the e s s e n t i a l workings of the machine, then we would not have a computer but r a t h e r j u s t a super-randomizer. The purpose of a randomizer i s to supply random numbers when the machine r e q u i r e s that type of i n f o r m a t i o n , v i z . randjom numbers. Therefore, a computer w i t h a randomizer i s s t i l l q u i t e p r e d i c t a b l e as f a r as i t s movements are concerned during a problem. I t i s not p r e d i c a b l e , however, i f during the problem more unknown i n f o r m a t i o n i s fed i n t o the machine, but then no one ever thought, that a machine was p r e d i c t a b l e under those c o n d i t i o n s . I f .7the type of randomizer i s one that s e l e c t s numbers s u c c e s s i v e l y from the decimal expansion of TT ? then the computer i s completely p r e d i c t a b l e . I f we b u i l d the machine so that each time i t r e c e i v e s an i n s t r u c t i o n to ' s e a r c h 1 , i t s e l e c t s the next number s u c c e s s i v e l y i n the expansion, then the numbers which the computer s e l e c t s w i l l be random. However i f we know how many past searches the machine has done, and we know where i n the expansion the computer s t a r t e d , then we can c a l c u l a t e the next number and we w i l l know which a l t e r n a t i v e the machine w i l l 26 f o l l o w . Thus there are machines w i t h randomizers which are together completely p r e d i c t a b l e . We can conclude, t h e r e f o r e , from the d i s c u s s i o n of the two types of randomizers, that computers w i t h these devices i n them are s t i l l p r e d i c t a b l e i n the strong sense. Furthermore the f o r m a l i t y of the machine i s not upset, because we can e a s i l y a l l o w f o r a change i n the input tape, which we sai d was comparable to the axioms of a formal system. A l t e r i n g the axioms of a system does not destroy the f o r m a l i t y of the system, i t j u s t makes a new system that has d i f f e r e n t theorems. \ 27 SECTION IV WHAT IS BEHAVIOUR? 22 In h i s a r t i c l e "The Mechanical Concept of .Mind", Michael S c r i v e n presents the f o l l o w i n g argument: the outward signs ( i n c l u d i n g speech) are not i n f a l l i b l e i n d i c a t i o n s of consciousness. I t i s therefore q u i t e c e r t a i n that they arepV not, ... the same th i n g as consciousness.- : ^ This argument i s meant to show that consciousness cannot be reduced to outward signs or observable behaviour. S c r i v e n seems to have i n mind a d i s t i n c t i o n between the behavioural and the non-behavioural aspects of man. When he t a l k s about two d i s t i n c t t h i n g s , outward signs and consciousness, S c r i v e n seems to be d i s t i n g u i s h i n g between outward observable behaviour and something e l s e which i s inner and unobservable. In order to assess t h i s argument which I have quoted or any others l i k e i t , we.must make c l e a r e r t h i s d i s t i n c t i o n between outward signs and consciousness. In p a r t i c u l a r - , i t might be asked j u s t what are the outward signs? . What are the behavioural aspects of man? More g e n e r a l l y , t h i s i s j u s t the qu-stion "What i s behaviour?" When philosophers t a l k about the p o s s i b i l i t y of there being mechanical robots around, i t seems that they are also using the idea of a robot to mark the d i s t i n c t i o n between the behavioural aspects of human experience and the non-behavioural aspects. 1 The robot i s considered to be able to behave e x a c t l y l i k e a person, even, w i t h some w r i t e r s , to the p o i n t of being b a h a v i o u r a l l y i n d i s t i n g u i s h a b l e from other people;.so that whatever e l s e a man has besides behaviour, t h a t ' s what makes him d i f f e r e n t from a robot. No one ever considers a c t u a l l y b u i l d i n g a robot and philosophers are not i n t e r e s t e d i n some supposed f u t u r e problem of d i s t i n g u i s h i n g a c t u a l people from t h e i r mechanical robot s l a v e s ! When we conceive of mechanical robots, we are j u s t using a conceptual device to mark the d i s t i n c t i o n between those things which have.just behaviour and those which have something else besides. Again, however, before we can consider using t h i s conceptual device of mechanical robots, i t i s important to determine j u s t e x a c t l y what i s to be considered as behaviour. I t i s g e n e r a l l y thought that i f we could b u i l d a robot to i m i t a t e any human behaviour, that we would not be able to d i f f e r e n t i a t e the robot from other people as- f a r as i t s behaviour was concerned. However, even i f we grant that a machine could be b u i l t to i m i t a t e any human behaviour, t h i s would not .mean that i t was i n d i s t i n g u i s h a b l e from a human. The f a c t that we can b u i l d a robot to i m i t a t e any piece of human behaviour does . not prove that we can b u i l d a robot to behave the same as a human. We do not u s u a l l y equate 'acting l i k e ' someone el s e and ' i m i t a t i n g ' them. Take the case where X i s s a i d to be i m i t a t i n g Y. I f we could show that X was unaware of what Y was doing, then we could not say that X was i m i t a t i n g Y. Furthermore if we are correct, i n saying that X i s i m i t a t i n g Y, then we could c o r r e c t l y a t t r i b u t e some i n t e n t i o n to X; namely, the i n t e n t i o n to i m i t a t e Y. Whereas when we say that so-and-so i s a c t i n g l i k e another person we are implying only coincidence. Confusing 'acting l i k e ' and ' i m i t a t i n g ' i s tantamount to reducing c o i n c i d - e n t a l behaviour to conventional behaviour, l i k e confusing s i m i l a r and t y p i c a l . There i s c e r t a i n l y a d i f f e r e n c e between on the one hand, two people having s i m i l a r enough c h a r a c t e r i s t i c s to be •29 i n d i s t i n g u i s h a b l e and, on the other hand, people having c e r t a i n c h a r a c t e r i s t i c s the same but not having some others. I m i t a t i n g i s a case of having some c h a r a c t e r i s t i c s the same as whoever i s being i m i t a t e d but not having some f u r t h e r c h a r a c t e r i s t i c s . A c t i n g l i k e or being a l i k e i s a matter of doing s i m i l a r s orts of t h i n g s , things which are comparable enough to be c a l l e d the same. So i f we a l l o w that i m i t a t i o n of any piece of behaviour i s p o s s i b l e we can not move immediately to the co n c l u s i o n that robots and humans are i n d i s t i n g u i s h a b l e . Thus i f we al l o w that robots can be b u i l t that i m i t a t e human behaviour, i t by no ' means f o l l o w s that they are i n d i s t i n g u i s h a b l e , even b e h a v i o u r a l l y , from humans. This c l a i m , that i f we a l l o w that i m i t a t i o n i s p o s s i b l e does not prove that men and robots are i n d i s t i n g u i s h a b l e , i s qu i t e compatible w i t h the evident f a c t that during a performance an actor may be i n d i s t i n g u i s h a b l e from (say) someone .who i s r e a l l y mad. For to say that an actor i s i n d i s t i n g u i s h a b l e during a performance i s to admit ( t a c i t l y ) that there i s a d e f i n i t e l i m i t to the s i m i l a r i t i e s between ac t o r s and madmen. But to admit that there are l i m i t s i s to acknowledge that a c t o r s and madmen are r e a d i l y d i s t i n g u i s h a b l e i n a l a r g e r context. However, there may be cases of i m i t a t i o n which are done so w e l l that one may doubt whether there are any c h a r a c t e r i s t i c s which the i m i t a t o r has f a i l e d to d u p l i c a t e ; a s o r t of p e r f e c t i m i t a t i o n . But I f i n d t h i s case g e n e r a l l y i n c o n c e i v a b l e , since i m i t a t i n g presupposes a ( p a r t i c u l a r ) second-order i n t e n t i o n a l i t y on the part of the actor which the person i m i t a t e d doesiiapfc have. Unless one held that i n t e n t i o n a l i t y was e n t i r e l y non-behavioural., i . e . , 3 0 had no behavioural m a n i f e s t a t i o n s , then I cannot conceive of a case of p e r f e c t i m i t a t i o n . However i t i s c e r t a i n l y the case that i f we allo w that robots can be b u i l t to i m i t a t e any piece of human behaviour, we can not conclude from t h i s that they would be i n d i s t i n g u i s h a b l e , even as f a r as t h e i r behaviour i t s e l f i s concerned, from humans. However, l e t me t r y to make c l e a r e r the d i s t i n c t i o n that I drew above between the problems of s i m i l a r i t y and e x e m p l i f i c a - t i o n . In problems of s i m i l a r i t y we are t r y i n g , f o r example, to 1 determine whether some p a r t i c u l a r piece of behaviour can be c a l l e d a smile. We are .troubled because we do^not have any c l e a r t e s t f o r determining what c o n s t i t u t e s s m i l i n g . Or, we may be i n doubt about how s u c c e s s f u l one must be i n some proposed t e s t i n order to be said to have smiled. This i s the problem of t r y i n g to f i n d adequate t e s t s f o r the a p p l i c a t i o n of some c h a r a c t e r i s t i c s to given s i t u a t i o n s . By an adequate t e s t , I mean one that i s s u c c e s s f u l or p o s i t i v e when we say that the s i t u a t i o n has the c h a r a c t e r i s t i c , and unsu c e s s f u l or negative when the s i t u a t i o n . doesiGoi have the c h a r a c t e r i s t i c a t t r i b u t e d to i t . This means that the statement of the success of an adequate t e s t i s l o g i c a l l y necessary and s u f f i c i e n t f o r the statement.of the d e s c r i p t i o n of the c h a r a c t e r i s t i c to the given s i t u a t i o n . Thus when we r a i s e questions about- adequacy, :what i s i n doubt i s the r e l a t i o n s h i p between the c h a r a c t e r i s t i c s a t t r i b u t e d to some s i t u a t i o n and the t e s t s done on t h e - s i t u a t i o n . However, i n the other problem of f i n d i n g t y p i c a l examples, we may be i n doubt as to "whether two subjects have the same c h a r a c t e r i s t i c s because the t e s t we have w i l l not apply to one of them. Or, i f -we can see that they both have some character- i s t i c i n common, we may t r y to f i n d some other c h a r a c t e r i s t i c s which one has and the other haszipt* This problem may a r i s e , t h i n k i n g now of robo t s , i n which someone says they have produced an example of-'something c l a i m i n g that t h e i r product h a g a a l l the c h a r a c t e r i s t i c s of the other t h i n g s . This i s the problem of determining the c h a r a c t e r i s t i c s of any given s i t u a t i o n which one may s e l e c t to examine. Thus there are two d i s t i n c t problems: that of determining the. adequacy of t e s t s and that ofddetermining the v a r i o u s c h a r a c t e r i s t i c s of given s u b j e c t s . I do not t h i n k that these, two problems are u n r e l a t e d ; i n f a c t I s h a l l argue that one presupposes that the other has been answered. I t can r e a d i l y be seen, I t h i n k , that i n order to answer the question of whether or not some proposed subject i s to be admitted to another c l a s s of objects as a t y p i c a l example, we must have some way of determining the c h a r a c t e r i s t i c s of the members of the group and al s o of the proposed subject. I f . t h e proposed example'has a l l the c h a r a c t e r i s t i c s of the members of the group (that are r e l e v a n t to them being a group), then the example becomes a member. But t h i s problem could not be t a c k l e d u n t i l we have some adequate way of deciding when two subjects have the same c h a r a c t e r i s t i c s . And t h i s question of adequacy i s none other than the f i r s t problem we noted, that of determining s u c c e s s f u l t e s t s f o r c h a r a c t e r i s t i c s . Furthermore, unless we thought that the problem of determining success was at l e a s t capable of s o l u t i o n , then the second problem could not p r o p e r l y a r i s e . I f we could not i n p r i n c i p l e f i n d a t e s t f o r some c h a r a c t e r i s t i c , then we could.never t e s t some proposed example f o r that c h a r a c t e r i s t i c . The proposal to t e s t some example f o r 3 2 a property assumes that there i s an adequate t e s t f o r that property. Therefore to ask the second question presupposes that the f i r s t one of adequacy can be solved. Furthermore the second question of t e s t i n g examples could not even a r i s e unless i t was at l e a s t i n p r i n c i p l e p o s s i b l e to f i n d a t e s t . For i f we know a p r i o r i that no. t e s t could i n p r i n c i p l e be found, then the questions about t e s t i n g subjects f o r c h a r a c t e r i s t i c s could not a r i s e . Therefore before we can answer any questions about b u i l d i n g examples w i t h some c h a r a c t e r i s t i c s , -the p r i o r question of the p o s s i b i l i t y of f i n d i n g adequate t e s t s must be answered. When we t a l k about robots and t h e i r d i f f e r e n c e s from people, we are wondering whether.there are some c h a r a c t e r i s t i c s which people have that robots do not. This i s c l e a r l y the second problem; the one of determining the existence of c h a r a c t e r i s t i c s i n v a r i o u s s u b j e c t s . S i m i l a r l y any d i s c u s s i o n of the d i f f e r e n c e of the subjects which i l l u s t r a t e outward s i g n s , and others which may have more c h a r a c t e r i s t i c s , i s again a question of t e s t i n g some subjects to see i f they have the c h a r a c t e r i s t i c s which other given examples have. Thus to t a l k about robots and people, or outward signs of behaviour and consciousness, presup- poses that the f i r s t question i s capable of an a f f i r m a t i v e answer. That i s , i t i s assumed that we can f i n d adequate t e s t s of behaviour. In f a c t I don't t h i n k i t would.be going too f a r to say that the use of the conceptual device, robot, presupposes that behaviour, or examples of behaviour, can be adequately t e s t e d f o r . . Thus to assess the opening argument which was used by S c r i v e n , we must examine the p o s s i b i l i t y of e s t a b l i s h i n g adequate t e s t s f o r behaviour. . ' 33 So f a r I have stated the problem of adequacy In terms of c h a r a c t e r i s t i c s and t e s t s , and now I would l i k e to r e s t a t e i t i n a more general form i n order to show the fundamental character of t h i s problem. When we say that some s i t u a t i o n has a character- i s t i c we;are, speaking more g e n e r a l l y , using i n a meaningful way, some concept to t a l k about the s i t u a t i o n . The a t t r i b u t i o n of the c h a r a c t e r i s t i c 'smile' can be thought of as the meaningful use of the concept 'smile'. .Although I by no means intend to equate use and meaning, I do take use to be con c l u s i v e evidence that the. concept has. a meaning. On the other hand, however, when we t a l k of t e s t s we are, more a c c u r a t e l y , t a l k i n g about the r e s u l t s of t e s t s which i n d i c a t e the various p r o p e r t i e s of a s i t u a t i o n . The statement of the r e s u l t of some s u c c e s s f u l t e s t i s a statement saying that a given s i t u a t i o n has been tes t e d and found' to have a c e r t a i n property. So the r e s u l t s of a t e s t can be considered as the statement that a given s i t u a t i o n has a property. The question of adequacy can now be considered more g e n e r a l l y as a problem about the r e l a t i o n s h i p between the meaningful useoof a concept i n some s i t u a t i o n and the r e s u l t s of v a r i o u s t e s t s on that s i t u a t i o n . I s h a l l abbreviate the s t a t e - ment of t h i s problem i n what f o l l o w s to j u s t the problem of the r e l a t i o n s h i p between concepts and p r o p e r t i e s , but i t must be r e - membered that I am t a l k i n g about the meaningful use of a concept i n some p a r t i c u l a r s i t u a t i o n and the t e s t s which can be done on 2k that s i t u a t i o n . I have used Taylor's terminology i n t a l k i n g i n i t i a l l y about the adequacy question i n terms of t e s t s , but t h i s second f o r m u l a t i o n of the problem i n terms of concepts i n the one 2 6 that Hare uses i n h i s chapter on "Meaning and C r i t e r i o n ' . '3H I want to look at the p o s s i b l e r e l a t i o n s h i p s between the p r o p e r t i e s of given s i t u a t i o n s and the. concepts used to t a l k about these s i t u a t i o n s . (Note: t a l k about does not mean, e x c l u s i v e l y , to describe!) There are t h e o r e t i c a l l y q u i t e a number of r e l a t i o n - ships and I tend to group them under two main headings (a) l o g i c a l and (b) n o n - l o g i c a l . The l o g i c a l r e l a t i o n s h i p s are very numerous: ( i ) a property (p) i s necessary and s u f f i c i e n t f o r tiie- concept (c) ( i i ) p i s nece'ssary f o r c, . ( i i i ) p i s s u f f i c i e n t f o r c, (iv)drame group of p r o p e r t i e s (p n) are s u f f i c i e n t and necessary f o r e , ( v ) p n • i s necessary f o r c, ( v i ) • p i s s u f f i c i e n t f o r c, ( v i i ) some of a group of p r o p e r t i e s (p ). are necessary and s u f f i c i e n t f o r c, ( v i i i ) p n - k i s necessary f o r c, ( i x ) p n ~ k ; i s s u f f i c i e n t f o r c. The general form of those r e l a t i o n s h i p s which are necessary and s u f f i c i e n t i s p *-*c where' k<h and' n^l,k^.o. S i m i l a r generalforms can be found-for the necessary and f o r the s u f f i c i e n t r e l a t i o n - s h ips. I t i s therefore evident that there are, i n p r i n c i p l e , no l i m i t s to the number of l o g i c a l r e l a t i o n s h i p s between p r o p e r t i e s and concepts. And f i n a l l y those, p r o p e r t i e s which s a t i s f y or belong to one. of these r e l a t i o n s h i p s , I s h a l l c a l l a c r i t e r i o n f o r that concept. . Some p r o p e r t i e s however are only normally adquate f o r the a s c r i p t i o n of some concept. That i s to say that when a s i t u a t i o n contains a property, or s e r i e s of p r o p e r t i e s , the concept i s normally applicable.Ther^are e x c e p t i o n a l cases, of course, but ge n e r a l l y we are j u s t i f i e d i n using the concept when these p r o p e r t i e s e x i s t i n some s i t u a t i o n . The r e l a t i o n s h i p s between the p r o p e r t i e s and the concept i s not a l o g i c a l one becuase we are only normally j u s t i f i e d i n using the concept when the given s i t u a t i o n e x h i b i t s •3-5 these p r o p e r t i e s . This case may a r i s e when the p r o p e r t i e s ase good i n d u c t i v e evidence f o r the use.of the concept. Some p r o p e r t i e s may be (say) only s u f f i c i e n t i n normal circumstances, f o r the a p p l i c a t i o n of the concept. This means that the r e l a t i o n s h i p s between the concept and the property i s such that we are not normally j u s t i f i e d i n using a concept because of the r e s u l t s of a t e s t . However because i t i s a . s u f f i c i e n t r e l a t i o n s h i p , we can to g e n e r a l l y conclude from the r e s u l t s of a t e s t A t h e a p p l i c a b i l i t y of the concept. Furthermore, other p r o p e r t i e s may be (say) nec- essary, i n normal s i t u a t i o n s , f o r the a p p l i c a t i o n of the concept; i n which case we would be j u s t i f i e d i n concluding from the a p p l i c a b i l i t y of the concept to the r e s u l t s of some t e s t . So we can have p r o p e r t i e s which are, w i t h i n some normal range of s i t u a t i o n s e i t h e r necessary, or s u f f i c i e n t or p o s s i b l y both, for. the a p p l i c a t i o n of some concept. However the r e l a t i o n - ships are not l o g i c a l ' i n the formal sense because we can not s p e c i f y the range of normal s i t u a t i o n s , nor s p e c i f y the ranges that w i l l be normal i n the f u t u r e . But i n normal s i t u a t i o n s the p r o p e r t i e s could be necessary or s u f f i c i e n t or both. These p r o p e r t i e s which are r e l a t e d to concepts, I c a l l ( f o l l o w i n g a modified v e r s i o n of S c r i v e n ) 2 7 i n d i c a t o r s . There are thus as many r e l a t i o n s h i p s between i n d i c a t o r s and concepts as there are w i t h c r i t e r i o n , but the normal r e l a t i o n s h i p s are not l o g i c a l . Therefore i t seems that there are an u n l i m i t e d number of r e l a t i o n s between concepts and p r o p e r t i e s , and even although there are. two main d i v i s i o n s i n the types of r e l a t i o n s , even w i t h i n these types there is an u n l i m i t e d number of p o s s i b l e r e l a t i o n s . The question now a r i s e s q u i t e n a t u r a l l y as to what types of concepts our b e h a v i o r a l ones are? By b e h a v i o r a l concepts, I 3 6 mean those concepts which we use when t a l k i n g about how people behave; such as, smil e , smirk, g r i n , and grimace, to mention only a few from the.various f a c i a l expressions that people adopt. I t seems to me that many of the concepts are of a normal type; that i s , that there are normally j u s t i f i a b l e i n d i c a t i o n s when people are s m i l i n g , but no c r i t e r i o n f o r smiles. Granting that at l e a s t at present some of our b e h a v i o r a l concepts are of a normal type, i t may be thought that they could a l l be changed to a l o g i c a l type. That i s changed i n type; but the meanings remain the. same. In t h i s regard i t i s i n t e r e s t i n g to consider, as an example of type r e d u c t i o n a paper "Can Humans 'Feel'?" by Mr. S. Coval i n which •he argues that our b e h a v i o r a l concepts may become l o g i c a l types as we l e a r n more about the human organism. He argues (roughly) t h a t we w i l l develop behavior concepts, l i k e " t i r e d " which w i l l be i d e n t i f i e d by the cause of the c o n d i t i o n of the human. • Thus i f we could determine the exact t e s t s f o r the causes of a piece of behavior we would have the c r i t e r i o n f o r the use of that concept. Now t h i s suggests two a l t e r n a t i v e s , (a) that our present normal b e h a v i o r a l concepts could a l l be made l o g i c a l types by f i n d i n g the t e s t s which are c r i t e r i o n . But here no proof i s o f f e r e d t o show that t h i s i s p o s s i b l e i n p r i n c i p l e , and I see no reason to t h i n k that all-normal' type concepts could possibly be made l o g i c a l types. Or (b) that i f we do develop a set of l o g i c a l type behavior concepts,, we w i l l have two sets which are i r r e d u c i b l e , and I do not know what so r t of standard we should use to compare- them, as they are d i f f e r e n t types. Of course these remarks of mine about Mr. Coval's ideas are by no means meant as a r e f u t a t i o n , but on the other hand I do not see why, when we are con s i d e r i n g the r e l a t i o n s 3 7 between t e s t s and concepts, we should t a c k l e the question w i t h only one r e l a t i o n i n mind, that of l o g i c a l l y adequate. However more i m p o r t a n t l y , i t i s evident from an examination of Coval's paper, j u s t where a theory i s needed i n order to succeed i n a. re d u c t i o n . The r e d u c t i o n i s t must o f f e r e i t h e r some p r i n c i p l e of comparison between concepts;^ which are d i f f e r e n t i n type or e l s e prove a p r i o r i t hat a l l concepts we p r e s e n t l y use could be made l o g i c a l i n type without change i n meaning. In the absence of e i t h e r of these p r o o f s , we can not conclude that, a l l of our b e h a v i o r a l concepts which we now employ are r e d u c i b l e to a l o g i c a l type. Therefore we can assume i n the' absence of a r e d u c t i v e theory, that our present b e h a v i o r a l concepts do not have c r i t e r i o n . .Where does a l l t h i s leave Mr. S c r i v e n w i t h h i s mechanical r o b o t s i m i t a t i n g human behavior? Since a robot i s a mechanical device i t can be t a l k e d about e n t i r e l y i n terms of a l o g i c a l type. Nowhere i s any proof o f f e r e d e i t h e r by Sc r i v e n or anyone el s e who t a l k s about robots that OUT b e h a v i o r a l cnocepts are a l l of a l o g i c a l type. U n t i l they prove that the concepts we use to t a l k about how humans behave can be reduced to l o g i c a l type terms, then, I argue, the question of mechanical i m i t a t i o n cannot even a r i s e . Every concept that a p p l i e s to a a machine i s of a l o g i c a l type; probably even of the narrower c l a s s of l o g i c a l types c a l l e d nec- essary and s u f f i c i e n t . Thus i f some performance or movement (or act i o n ) i s to be accomplished by a mechanical device, then the e performance must be d e s c r i b a b l e i n l o g i c a l concepts. At present we recognize, t a l k about, and describe human behavior using normal type concepts. But the problem of mechanical i m i t a t i o n can only a r i s e when human behavior i s described i n l o g i c a l type terms. U n t i l '3:8 i t i s shown that a l l human behavior i s d e s c r i b a b l e i n these type of terms, then the problem of i m i t a t i o n does not and cannot a r i s e . Furthermore the opening argument about the i n f a l l i b i l i t y of out- ward signs of consciousness does not show that consciousness i s something other than behavior, i t only shows that our concepts about consciousness are not of a l o g i c a l type, but r a t h e r are of a normal type I Now i t becomes evident that the robot-man d i s t i n c t i o n i s not meant to mark something outer vs. something i n n e r , or separate outward v i s i b l e signs from inward p r i v a t e f e e l i n g s : but ra t h e r i s meant to mark the d i s t i n c t i o n between a d e s c r i p t i o n of human a c t i v i t i e s i n l o g i c a l type and n o n - l o g i c a l type terms. Or perhaps the robot-man distinction- can be thought of as d i s t i n g u i s h i n g those b e h a v i o r a l concepts which are l o g i c a l from those which are not. Here, of course the n o n - l o g i c a l type of concepts are those that we use to t a l k about consciousness. The question "What i s Behavior?" has become the question fWhat types of concepts do we use f o r b behavior?" and.now perhaps the f l y can get out of the f l y b o t t l e . '39 SECTION V A TEST FOU THINKING 29 • At the co n c l u s i o n of h i s paper "The I m i t a t i o n Game", K e i t h Gunderson tempers some of h i s previous c r i t i c i s m s w i t h the remarks: Neverthless...the general question would remain unanswered: what range of examples would s a t i s f y the i m p l i c i t c r i t e r i o n we use i n our or d i n a r y c h a r a c t e r i z a t i o n of subjects as "those capable of thought"? A c o r o l l a r y : I f we are to keep the question "Can machines t h i n k ? " i n t e r e s t i n g , we cannot withhold a p o s i t i v e answer simply on the grounds that i t (a machine) does not d u p l i c a t e human a c t i v i t y i n every respect. The question "Can a machine t h i n k i f i t can do everything a human being can do?" i s not an i n t e r e s t i n g question....30 However I do not t h i n k that these remarks j u s t i f y Mr. Gunderson i n q u a l i f y i n g h i s e a r l i e r c r i t i c i s m s . I s h a l l argue that the concept "capable of thought" has no l o g i c a l l y s u f f i c i e n t c r i t e r i o n . I f t h i s i s so then he need not worry about our i m p l i c i t ( l o g i c a l ) c r i t e r i o n f o r the concept. . Mr.'Gunderson does not f i n d the question about machines t h i n k i n g , i n t e r e s t i n g , i f we grant that machines can do everything humans do. But I should t h i n k that even i f a machine could do every t h i n g , we would s t i l l have s c e p t i c a l grounds f o r withholding our mental concepts. Machines are d i f f e r e n t from.humans and d i f f - erent in' a way that other humans do not d i f f e r from each other. Since a machine i s ;.by d e f i n i t i o n d i f f e r e n t than a human, even i f ' a machine could do everything a human does, the question of relevance of the d i f f e r e n c e s w i l l always a r i s e and I see no reason to r u l e i t out a p r i o r i as u n i n t e r e s t i n g . When we b u i l d a machine to do •„? everything t h a t humans can, we use d i f f e r e n t m a t e r i a l s to b u i l d w i t h . Even when we b u i l d a mechanical " b r a i n " we use d i f f e r e n t >+0 •materials than those the b r a i n i s made of. And because a machine i s d i f f e r e n t from a human i n ways that other humans are not, the s c e p t i c can always doubt the v a l i d i t y of the a p p l i c a t i o n of mental concepts to machines. Whether or not the s c e p t i c i s j u s t i f i e d i s another i n t e r e s t i n g question, but one that can always a r i s e w i t h machines despite the f a c t that they do everything. Gunderson's c o r o l l a r y that we cannot withhold a p o s i t i v e answer simply on the grounds' that machines do not d u p l i c a t e human a c t i v i t y i n every respect, seems to me to f a i l to n o t i c e t h i s ever present s c e p t i c a l ground. I f we could f i n d one a c t i v i t y which no machine could- do and' t h i s was a mental a c t i v i t y , then together w i t h the i m p l i c i t s c e p t i c i s m , there would .be good grounds f o r withho l d i n g a p o s i t i v e answer. .This i s the reason that some philosophers have been so impressed w i t h Godel 1s theorem. Godel showed that given any p a r t i c u l a r Turing machine, he could always f i n d a theorem which a human could prove was true but the machine could not. Thus there was at l e a s t one mental a c t i v i t y , ' i . e . , proving the Godelian statement of that machine, which-the machine could not do. When you couple t h i s f a c t w i t h the general d i f f e r e n c e s between machines and humans (or even b r a i n s ) , then there are good reasons f o r withholding mental concepts ( e s p e c i a l l y t h i n k i n g ) from machines. Gunderson f e l t however that there was a general unanswered question;, namely, what range of examples would s a t i s f y the i m p l i c i t c r i t e r i a we use i n our or d i n a r y c h a r a c t e r i z a t i o n of subjects as- "those capable of thought?" I t i s evident that, we use tone concept "capable of thought" w i t h some subjects i n some s i t u a t i o n s and not i n others.. Most people understand the concept and we can use i t , g e n e r a l l y , unambiguously. That i s to say, the concept has a meaning which most people comprehend. Now granting that a concept has meaning , and f u r t h e r that the meaning can be taught to others I should say, f o l l o w i n g W i t t g e n s t e i n that there must be paradigm instances of the use of the word. There must be some s i t u a t i o n s i n which the concept i s lised c o r r e c t l y and we know, g e n e r a l l y , which s i t u a t i o n s they are. The concept has been taught to us and i s taught by i t s use i n paradigm s i t u a t i o n s . However, granting . a l l t h i s , i t does not f o l l o w that there are c r i t e r i a , e i t h e r i m p l i c i t or e x p l i c i t , f o r the use of t h i s concept. More proof must be o f f e r e d than t h e - f a c t that the concept i s learned i n order ten prove that meaningful concepts have l o g i c a l l y r e l a t e d c r i t e r i a . Yet the attempt to f i n d a t e s t assumes j u s t t h i s p o i n t , namely, that there i s some t e s t which s a t i s f i e s the c r i t e r i a of the concept "capable of thought." There i s however, no proof o f f e r e d to show that the concept has c r i t e r i a . Some people who work wi t h computers contend that they can program a computer to do any task which any person could do. They may be q u i t e j u s t i f i e d i n th&s claim. They then argue that i f . we show them what the subjects do when we say that they are capable of thought, then they w i l l b u i l d a computer to do that job also.- Howgv-er t h i s l i n e of reasoning presuppdsse© that there are a d e f i n i t e number of s p e c i f i c tasks which, when completed, the l a b e l "capable of thought" cannot i n l o g i c a l consistency, be w i t h h e l d . But.we cannot a l l o w people to argue that because we are t e s t i n g a machine, the concept must be of a s p e c i f i c type. Rather i t can only be held that i f we are ever going to be able to f i n d a t e s t f o r the a p p l i c a t i o n of the concept, then the concept must have c r i t e r i a . However i f the concept does not have c r i t e r i a then we cannot f i n d a l o g i c a l l y s u f f i c i e n t t e s t f o r i t s a p p l i c a t i o n . S c r i v e n t h i n k s that i f we refuse to apply our mental vocabulary, each time they b u i l d a computer to do. more human achievements, then we w i l l be making a mistake. He says: •'The l o g i c a l trap i s t h i s : no one performatory - achievement w i l l be enough'to .persuade us to apply the human achievement vocabulary, but i f we refuse to use t h i s vocabulary i n each case se p a r a t e l y , on t h i s ground, we w i l l , perhaps wrongly, have committed ourselves to avoiding i t even when a l l the achievements are simultaneously a t t a i n e d . • . '31 • S c r i v e n seems to t h i n k that there'are a d e f i n i t e number, (namely, a l l of them) of achievements which one does to q u a l i f y f o r the human-achievement vocabulary. I f the number i s not d e f i n i t e (and t h i s does not mean the number i n f i n i t e ) then there i s no l o g i c a l • 3 ? t r a p . But where i s the proof that a l l of our human-achievement concepts are of a type that have a d e f i n i t e number of c r i t e r i a ? S c r i v e n does not o f f e r one, and ,1 intend to show that none can be given. I s h a l l aggue that the concept "capable of thought" i s an e v a l u a t i v e ooncept which does not have any l o g i c a l l y s u f f i c i e n t set of c h a r a c t e r i s t i c s so that no t e s t f o r character- i s t i c s of p e o p l e . w i l l . e v e r be found that i s l o g i c a l l y s u f f i c i e n t . I n order to prove t h i s , however, I must f i r s t begin by reviewing some of the conclusions that have been reached i n the; a n a l y s i s of e v a l u a t i v e language. •5-5 In the f i f t h chapter of The Language of Morals-^ Hare reformulates Moore's c r i t i c i s m of n a t u r a l i s m i n e t h i c s . In doing so Hare shows that any attempt to reduce our e v a l u a t i v e terms to the statement of a d e f i n i t e set of d e s c r i p t i v e c h a r a c t e r i s t i c s must be i n p r i n c i p l e mistaken. He s t a t e s : l e t us g e n e r a l i z e . I f P i s a good p i c t u r e 1 i s h e ld to mean the same as 'P i s a p i c t u r e • and i s C (where C.is a group of character- i s t i c s ) , then i t w i l l become impossible to to commend p i c t u r e s f o r beingC: i t w i l l be p o s s i b l e only to say that they are C. It- i s important to r e a l i z e that t h i s d i f f i c u l t y . , ' ' has nothing to do w i t h the. p a r t i c u l a r example I have chosen. ' I t i s .not because we have - chosen the wrong d e f i n i n g c h a r a c t e r i s t i c s ; • i t i s because whatever--defining character- i s t i c s we choose,- t h i s 'objection a r i s e s , that we can no longer commend an object f o r posse possessing those c h a r a c t e r i s t i c s . 3 ^ (my parenthesis added) As I s a i d , I accept e n t i r e l y Hare's proof that i f we are to evaluate or commend var i o u s subjects :for doing or being something,.then we must have e v a l u a t i v e concept's which are not j u s t equivalent to an a s s e r t i o n of a d e f i n i t e set of c h a r a c t e r i s t i c s or p r o p e r t i e s . I t i s a f a c t . t h a t we do value and commend, and as long as we , continue to , we must have value concepts. Thus i n the absence' of any prcoif a p r i o r i that at some time humans w i l l stop f o r e v e r to evaluate, i t can be assumed that we must have e v a l u a t i v e concepts. Thus we must have concepts which are not equivalent, to the a s s e r t i o n of a set of c h a r a c t e r i s t i c s . The question now a r i s e s as to whether or not when we say " X c a n t h i n k " , we are making an e v a l u a t i v e judgement. I n s e c t i o n . V I I I of h i s paper Gunderson says: A f i n a l p o i n t : the stance i s o f t e n taken that t h i n k i n g i s the crowning c a p a c i t y or achieve- --'-'''•-"'' mpnt.^of:..the:;human race, and that i f one denies 'that"machines can t h i n k , one i n e f f e e t . a s s i g n s them to some lower l e v e l of achievement than that a t t a i n e d by human being. But one might w e l l contend that machines can't .think f o r they do much b e t t e r than that.3 5 (my i t a l i c s ) I f we o f t e n say that t h i n k i n g i s the crowning c a p a c i t y , or the f a c u l t y which makes us b e t t e r , than to say of someone that they I t h i n k i s not only to say that they have some ca p a c i t y but that they are commendable'(or more valuable) because, they have i t . I f we c a l l some ca p a c i t y the crowning""one we..are i n - e f f e c t saying that .whoever has t h i s c a p a c i t y i s commendable because of i t . , .V..\\v And to o f f e r a reason f o r commendation i s simply to commend someone f o r the reason o f f e r e d . However I t cannot be denied that "X has the crowning c a p a c i t y , v i z . a b i l i t y to t h i n k " and "X can t h i n k " are d i f f e r e n t u t t e r a n c e s . ' I t i s a g e n e r a l l y accepted f a c t that people can t h i n k , and i f someone st a t e s a f a c t which everyone knows, then i t i s ' g e n e r a l l y assumed that he has some other purpose i n mind. For example when I t e l l my w i f e , t h a t she already knows, that- the house i s d i r t y , I am not j u s t s t a t i n g a f a c t but r a t h e r I am (say) condemning'this c o n d i t i o n of the house and thus recommending that she c l e a n i t . So i f someone st a t e s that people (or some person) can t h i n k and we a l l g e n e r a l l y assume t h i s , then we take i t that they have some other purpose' i n mind i n u t t e r i n g the sentence. Now when we remember that we o f t e n consider the a b i l i t y to t h i n k as a reason f o r commending people, i t i s not d i f f i c u l t to see that on some occasions at l e a s t , the purpose i n saying that someone can t h i n k i s to commend them. For i f i t i s assumed that L o i s can t h i n k as i t g e n e r a l l y i s and we o f t e n recommend people because they can t h i n k , then to say that L o i s can t h i n k i s to commend her because she can t h i n k . And I t h i n k ' t h a t the sentence "X can t h i n k " has j u s t t h i s use of commendation on some occasions I want to emphasize that a l l I wish to e s t a b l i s h i s that on some occasions . the sentence has t h i s use, while not' denying that on other occasions the sentence has other uses. But part of the meaning of the concept, i f we judge i t s meaning by i t s use, i s e v a l u a t i v e and as such w i l l have the c h a r a c t e r i s t i c s which Hare noted about e v a l u a t i v e s.fe&tements. I f we accept the v a l i d i t y of Hare's a n a l y s i s of our o r d i n a r y use of e v a l u a t i v e concepts, then we must conclude that the concept "capable.e of fcnought" i s not equivalent to the statement of a set of c h a r a c t e r i s t i c s about humans. The question now a r i s e s as to whether or not there i s a set of c h a r a c t e r i s t i c s which are l o g i c a l l y s u f f i c i e n t f o r the a s c r i p t i o n of the concept "capable of thought?" Since Hare has shown that there i s no set which i s e q u i v a l e n t , then perhaps • there is ..some set of p r o p e r t i e s wh$ch are s u f f i c i e n t f o r ascription!"!.. In t h i s case we would then set up a s e r i e s of t e s t s f o r the p r o p e r t i e s and we would have a l o g i c a l l y s u f f i c i e n t group of t e s t s which, when a machine passed them, would f o r c e us ( l o g i c a l l y ) to • say that the machine was aapable of thought. Gunderson seems to t h i n k that there ls_ a set when he asks f o r the range of examples which would s a t i s f y the i m p l i c i t c r i t e r i a ' : ; of the concept. But there i s no n e c e s s i t y that meaningful concepts have l o g i c a l l y s u f f i c i e n t criteria.}.. I argued i n s e c t i o n IY that there -,>was! an i n d e f i n i t e number of r e l a t i o n s between concepts and p r o p e r t i e s , some of which were l o g i c a l and others not. Granted that these r e l a t i o n s are conventional ones, 'this does not show that they must be l o g i c a l . The convention could be t h a t some set of c h a r a c t e r i s t i c s i s normally s u f f i c i e n t f o r the a p p l i c a t i o n of the concept,- but that we a l l o w e x c e p t i o n a l circumstances to j u s t i f y the withholding of the concept.As these circumstances can be n e i t h e r s p e c i f i e d nor f o r s e e n , i t i s evident, as I argued i n s e c t i o n IV, that the r e l a t i o n s h i p would not be a s t r i c t or l o g i c a l one. What i s the r e l a t i o n s h i p , then, between an e v a l u a t i v e concept and the c h a r a c t e r i s t i c s of s i t u a t i o n s ? Hare argues that i f we evaluate something, then we must be prepared to evaluate something r e l e v a n t l y s i m i l a r , the" same way, or el s e o f f e r a j u s t i f i c a t i o n f o r not doing so. And he says that the "must" i s a l o g i c a l one i n the sense that i f one refused to similarly..' evaluate without o f f e r i n g a j u s t i f i c a t i o n , then one would have committed a c o n t r a d i c t i o n . Thus to f a i l to o f f e r reasons i s to v i o l a t e the convention, and t h i s , Hare argues, i s tea "involve ©.heseltf din a l o g i c a l c o n t r a d i c t i o n , but t h i s i s f a r from showing that the conventional r e l a t i o n i s a l o g i c a l one. I t shows only that i f one v i o l a t e s or refuses to p a r t i c i p a t e i n t h i s language convention ( a f t e r e n t e r i n g i t by using an e v a l u a t i v e concept) then one commits a l o g i c a l f a l l a c y , but the convention i t s e l f could j u s t as e a s i l y be a normal one as a l o g i c a l one. I f one uses a concept which, as pa r t of i t s convention, r e q u i r e s a j u s t i f i c a t i o n , i n some cases and one subsequently refuses to acknowledge the demand f o r a . j u s t - i f i c a t i o n , then one c o n t r a d i c t s o n e s e l f , even i f the convention i s only one of a normal r e l a t i o n between the concept and the c h a r a c t e r i s t i c s of the s i t u a t i o n . But. Hare's a n a l y s i s of the a c t u a l conventional r e l a t i o n between e v a l u a t i v e concepts and the var i o u s p r o p e r t i e s of s i t u a t i o n s , was that i f an e v a l u a t i v e concept i s used to (say) commend a s i t u a t i o n then one must al s o commend another s i t u a t i o n or el s e j u s t i f y why one i s withholding the commendation. That the s i t u a t i o n s are both given and n u m e r i c a l l y d i s t i n c t i s proof enough that there are d i f f e r e n c e s between them, but the convention demands- a j u s t i f i c a t i o n f o r the relevance of the d i f f e r e n c e s i n •withholding e v a l u a t i o n . Furthermore the same c h a r a c t e r i s t i c may "be r e l e v a n t i n one s i t u a t i o n f o r an e v a l u a t i o n and not i n another s i t u a t i o n f o r the same e v a l u a t i o n . But a convention i n which some d e f i n i t e set of c h a r a c t e r i s t i c s are (say) s u f f i c i e n t f o r the a s c r i p t i o n of some concept except i n e x c e p t i o n a l circumstance, i.e.,. those circumstances i n which j u s t i f i c a t i o n can he found, i s the type of. convention I c a l l e d normal i n s e c t i o n l V . The convention f o r e v a l u a t i v e terms i s that the terms must he r e a p p l i e d or j u s t i — f i c a t i o n o f f e r e d f o r not rea p p l y i n g them, which means that normally they w i l l be used i n the same s i t u a t i o n s but we al l o w e x c e p t i o n a l l y j u s t i f i e d s i t u a t i o n s to be. exemp't. Therefore I conclude on the ba s i s ' o f Hare's a n a l y s i s and t f e . e d i s t i n c t i o n s I drew i n s e c t i o n IV that e v a l u a t i v e terms are of a normal type. Furthermore since the concept "capable of thought" i s an ev a l u a t i v e one, i t has t h i s n o n - l o g i c a l r e l a t i o n to the c h a r a c t e r i s t i c s of s i t u a t i o n s ; so that no set of t e s t s f o r the c h a r a c t e r i s t i c s of some proposed subject could be l o g i c a l l y s u f f i c i e n t f o r the a s c r i p t i o n of the concept. Thus no t e s t or set of t e s t s , could , i n p r i n c i p l e , be found which would be l o g i c a l l y s u f f i c i e n t to al l o w us to say '"Machines can t h i n k . " I t may be argued that i f we e l i m i n a t e the e v a l u a t i v e content from, the concept of t h i n k i n g then we s h a l l be able to. f i n d a t e s t . In the case where we f i n d a computer which s u c c e s s f u l l y passes t h i s t e s t , we w i l l then be able to say that i t t h i n k s , remembering that t h i s use of t h i n k i s non-evaluative. There are two r e p l i e s to t h i s type of c r i t i c i s m . I f i t i s thought that t h e • a p p l i c a t i o n of t h i s new concept to mechanical devices i s a step forward i n the problem of applying mental concepts to machines, then i t i s a mistake. By c u t t i n g out the troublesome p a r t of the concept, one does.not thereby make gains but rat h e r ore only saves up the tro u b l e u n t i l l a t e r . I n t h i s repect then, to change the concept i s only to by-pass the tro u b l e u n t i l l a t e r while t h i n k i n g t h a t one i s making gains. The second r e p l y i s -that i n cons i d e r i n g problems connected w i t h the concept of t h i n k i n g , the only way to l o c a t e the problem i s by considering our present concept and i t s or d i n a r y usage. When the problem "Can machines t h i n k ? " was o r i g - i n a l l y proposed, i t was assumed that people were wondering whether or not they could say of machines, what they say of l o t s of other t h i n g s ; namely that they can t h i n k . I f the concept of t h i n k i n g was not the one o r d i n a r i l y used and meaning what we o r d i n a r i l y mean, then what other p o s s i b l e meaning could i t have had? How should we have been able to f i n d any meaning f o r the question, i f the words were not used as we use them i n E n g l i s h ? I f i n • the s o l u t i o n to the problem, we change the meaning of the question, how can i t be argued that the o r i g i n a l question has been answered. Those people who change the concept have not answered the question "Can machines t h i n k ? " but ra t h e r some other problem that they have invented., Gunderson seems to have thought that h i s i n i t i a l ' c r i t i c i s m s of the I m i t a t i o n Game could be countered i f the i m p l i c i t c r i t e r i o n of the concept "capaMe of thought" could be found. He had'argued i n c r i . t i c i s m ? that the I m i t a t i o n Game was only one example, and a multitude of examples were needed to apply the concept. But he thought.that a set of t e s t s could be found which, when s a t i s f a c t o r i l y completed, would be l o g i c a l l y adequate f o r a s c r i p t i o n of the concept. However I have argued against t h i s , that there i s rio set of t e s t s which are l o g i c a l l y s u f f i c i e n t . Gunderson 1s e r r o r seems to have been that he mistook the type of concept that "capable of thought" i s . He thought i t was a l o g i c a l type concept, whereas I have argued that i t i s anormal or n o n - l o g i c a l type. By type I mean type of r e l a t i o n s h i p between the concept and the p r o p e r t i e s of s i t u a t i o n s . By mistaking the type of @0J3cept, some philosophers have assumed that i t had a l o g i c a l l y s u f f i c i e n t t e s t and set about f i n d i n g the t e s t (or t e s t s ) . However when we under- stand what type of concept "aapable of thought'" i s , I have argued, then"we can see that the search f o r a t e s t i s i n ' p r i n c i p l e f u t i l e . 50 • SECTION VI CONCLUSION In c o n c l u s i o n , I should l i k e to r e s t a t e some of the conclusions t e n t a t i v e l y a r r i v e d at i n the preceeding s e c t i o n s of the paper. I have argued i n s e c t i o n I I I that there i s good ' evidence that mechanical robots are p r e d i c t a b l e i n the important sense. Furthermore, i n s e c t i o n IV, I argued that even using the idea of a robot as j u s t a conceptual device, presupposed that c e r t a i n l i n g u i s t i c problems had been solved which indeed have not been solved. In the proceeding ^ % i ' 6 n , I t r i e d to show that we could never i n p r i n c i p l e f i n d a t e s t which was l o g i c a l l y s u f f i c i e n t f o r the utterance'"Machine X can t h i n k . " Together I t h i n k that .these conclusions add up to a r a t h e r serious c r i t i q u e of the general arguments advanced to show that machines can t h i n k . However I t h i n k that there are more f a r - r e a c h i n g i m p l i c a t i o n s to be drawn from the work i n t h i s paper. In order to p o i n t these i m p l i c a t i o n s , l e t us review the sources of e r r o r s that I suggested other philosophers had made. In arguing against the p o s s i b i l i t y of i m i t a t i o n , I showed that • philosophers had made an e r r o r by f a i l i n g to no,tice a l i n g u i s t i c question which the whole d i s c u s s i o n of i m i t a t i o n presupposed. I then.went on to i l l u s t r a t e the complexity of r e l a t i o n s that could e x i s t between a concept and the s i t u a t i o n s i n which they were used. F i n a l l y , I suggested that those philosophers who were concerned w i t h f i n d i n g a t e s t f o r t h i n k i n g had mistaken the type of concept that "capable of thought" i s . I have, i n f a c t , been c o n t i n u a l l y t r y i n g to show that the source of e r r o r s have a l l been of a l i n g u i s t i c nature. Thus one of the more general conclusions of t h i s paper i s that f a r more a t t e n t i o n must be given to language and the v a r i o u s l i n g u i s t i c problems that can a r i s e . "What i s needed i s a systematic method f o r t a c k l i n g these problems of language once they have been shown to be behind many of the more t r a d i t i o n a l problems. But even though we s t i l l l a c k a methodology, there i s a great need to focus more a t t e n t i o n upon our language,, i t s conventions, and concept types. There i s however another way of l o o k i n g at the r e s u l t s of t h i s paper. Much of my work has been i n an e f f o r t to change the form of the standard problems a s s o c i a t e d w i t h the question "Can machines t h i n k ? " . For example, I t r i e d to show that the problem of t r y i n g to f i n d a t e s t f o r t h i n k i n g i s j u s t the problem of d e t e r i n - i.ng types of concepts. In another s e c t i o n , I showed that the problem of c o n s t r u c t i n g a robot to. Imitate humans was at bottom, the problem of type reduction,, i . e . , the problem of changing a concept of one type to another without change of meaning. In making these changes, I have t r i e d to show that the problems which have bothered philosophers i n t h i s area are e s s e n t i a l l y l i n g u i s t i c i n nature; that i s , a l l the problems can.be r e s t a t e d as l i n g u i s t i c ones. When I say that I have r e s t a t e d a t r a d i t i o n a l or standard problem, I do not mean that I have given a synonymous rephrasing of the problem. I mean e i t h e r that the .standard problem can be shown to have a r i s e n because of a l a c k of c a r e f u l l i n g u i s t i c • a n a l y s i s , or that the t r a d i t i o n a l problem presupposes that some l i n g u i s t i c t h e o r i e s be s u b s t a n t i a t e d . Or even that i t can .be shown that the standard problems have as t h e i r main d i f f i c u l t y a confusion i n types'of concepts. I f a t r a d i t i o n a l problem i s r e l a t e d to a • .52 l i n g u i s t i c one i n one of these ways, then we can change i t i n t o a problem i n l i n g u i s t i c s . As I s a i d , I have t r i e d to do j u s t t h i s - r e s t a t i n g of the problems i n the area of minds and machines. The c o n c l u s i o n that I wish to suggest i s that i f the problems i n t h i s - area can be r e s t a t e d , then that i s some evidence that other problems may a l s o be r e s t a t a b l e i n t h i s way. I must, however, grant that t h i s paper i s not very s u b s t a n t i a l evidence to suggest the p o s s i b l e scope of t h i s r e s t a t i n g programme. I t i s my b e l i e f that most t r a d i t i o n a l p h i l o s o h p i c problems can be r e s t a t e d as l i n g u i s t i c ones. Thus another more general conclusion, of t h i s paper i s to suggest.the p o s s i b i l i t y of a general restatement of t r a d i t i o n a l p h i l o s o p h i c problems. Besides the more general c o n c l u s i o n s , there are the s p e c i f i c ones i n c r i t i c i s m of the arguments,for saying that machines can t h i n k . I have argued that there i s good evidence to doubt the p o s s i b i l i t y ofai.imitating robot and even i f t h i s evidence were mi s s i n g , the whole argument using i m i t a t i n g robots presupposes a l i n g u i s t i c d i f f i c u l t y which has not been answered. The analogy between men and r o b o t s , I argued, was empty, and I t r i e d to show that no l o g i c a l l y adequate t e s t of. t h i n k i n g can be found; so that these examples of machines p l a y i n g games were not c o n c l u s i v e but r a t h e r only persuasive evidence. As Gunderson s a i d : In the end, the steam d r i l l o u t l a s t e d John Henry as a digger of r a i l w a y tunnels, but that d i d not prove the machine had muscles; i t proved that muscles were not needed f o r digging r a i l w a y tunnels. There have been many i n t e r e s t i n g p o i n t s -made i n the arguments f o r .thinking machines, and these p o i n t s have had the e f f e c t of making most philosophers expand t h e i r concept of what a . • 53" machine i s . However, aside from t h i s merit of the arguments f o r the a f f i r m a t i v e , I have argued that we are s t i l l j u s t i f i e d i n saying that machines cannot t h i n k . • 5!+ FOOTNOTES 1 A.M.Turing,"Computing Machinery and I n t e l l i g e n c e " , Mind, v o l . L I X , No.236 -(1950) . 2 A.R.Anderson, ed, Minds and Machines, Englewood C l i f f s , New Jersey, P r e n t i c e - H a l l , Inc, 196 Lf. 3 Ibid'. , pp.h-5- k I b i d . , p p v 7 2 - 9 7 - 5 i b i d . , p.75 6 Another way of p u t t i n g the preceeding remarks i s to say that the 'only i f c o n d i t i o n i n 11T i s i n s t a t e A i f and only i f f l i p f l o p 36 i s on", can be cashed i n t o a f i n i t e l i s t ; such as, f l i p f l o p 1 i s e i t h e r on or o f f , f l i p f l o p 2 i s e i t h e r on or o f f , e t c . 7 A.R.Anderson, op. c i t . , pp.73-7 1+- 8 A.M.Turing, l o c . c i t . 9 A.Church, I n t r o d u c t i o n to Mathematical L o g i c , P r i n c e t o n , P r i n c e t o n U n i v e r s i t y P ress, 1956. 10 M.Davis, Computability and U n s o l v a b i l i t y , New York, McGraw-Hill Book Company, Inc, 1958. 11 A.R.Anderson, op. c i t . , p.7*+- 12 I b i d . , • p. 81 . 13 I b i d . , P- 77- 1̂ f I b i d . , P- 81 . 15 I b i d . , P- 8 2 . 16 I b i d . , P- ^ 3 . 17 I b i d . , P- 77- 18 I b i d . , P- 15. 19 M.Davis, op. c i t . , Chapter 1. 20 A.R.Anderson, op. c i t . , pp. L K3-59. 21 I b i d . , pp 22 I b i d . pp.31-^2. 23 I b i d . , p.3^ 2h C.Taylor, The E x p l a n a t i o n of Behavior„ London, Routledge 555 and Kegan P a u l , \96h, pp.82-87. 25 R.M.-Hare, The Language of Morals, Oxford, at the Clarendon Press, 1961, pp.9^-110- 26 I b i d . , pp.9^-'95- 27 M.Scriven, "The Logic of C r i t e r i o n " , The J o u r n a l of Philosophy, v o l . L V I , No.22, p.857. 28 S.C.Coval, "Can Humans 'Feel'?", unpublished. 29 A.R.Anderson, op. c i t . 30 I b i d . , p.70. 31 S.Hook, ed., Dimensions of Mind,- New York,. New York U n i v e r s i t y P r e s s , 1960, p.124-. 32 I suspect that S c r i v e n a l s o sees t h i s . Compare h i s a r t i c l e i n The J o u r n a l of Philosophy, op. c i t . , p.868. I use h i s argument only as an example of the i m p l i e d l o g i c a l trap argument wit h t e s t i n g . 33 R.M.Hare, op. c i t . , pp.79-93- 3h I b i d . , p.85. 35 A.R.Anderson, o p . c i t . , p.71• 36 Loc. c i t . 56 BIBLIOGRAPHY Books Anderson,A.R., ed. Minds and Machines. Englewood C l i f f s , New Jersey, P r e n t i c e - H a l l , Inc. , 1 96l+. (Contemporary P e r s p e c t i v e s i n Philosophy S e r i e s , .eds. J o e l Feinberg and W.C. Salmon, vol . 1 ) . Chappell,V.C., ed. The Philosophy of Mind. Englewood C l i f f s , New Jersey, P r e n t i c e - H a l l , Inc., 1962. Davis, M. Computability and U n s o l v a b i l i t y . New York, McGraw- H i l l Book Company, Inc.,,1958. Hare,R.M. The Language of Morals. Oxford, at' the Clarendon'" . P r e s s , 1961. Taylor,C. The E x p l a n a t i o n of Behavior. London, Routledge and Kegan P a u l , 196 .̂ ( I n t e r n a t i o n a l L i b r a r y of Philosophy and S c i e n t i f i c Method, ed.A.J.Ayer). . Witt g e n s t e i n , L . P h i l o s o p h i c a l I n v e s t i g a t i o n s . Trans.G.E.M. Anscombe. Oxford, B a s i l B l a c k w e l l , 1963* A r t i c l e s I • TT c o A l b r i t t o n , R . "On Wi t t g e n s t e i n s USe of the Term '"Cri t e r i o n * ". The J o u r n a l of Philosophy, v o l . L V I , No.22, pp.8>+5-856. Scriven,M. "The.L6gic of C r i t e r i o n " . The Jo u r n a l of Philosophy, v o l . L V I , No.22, pp.857-865. Wisdom,J.O. '"A New Model f o r the• Mind-Body R e l a t i o n s h i p " . The B r i t i s h J o u r n a l f o r the Philosophy of Science, vol . 2 No.8 (1951-52) pp.295-301. Wisdom,J.O. "The. Hypothesis of Cybernetics". The B r i t i s h J o u r n a l fo'r the Philosophy of Science, vol . 2 No. 5 (1951-52), pp. 1-2.M. 1

Cite

Citation Scheme:

    

Usage Statistics

Country Views Downloads
United States 3 0
Germany 1 35
Canada 1 0
China 1 26
City Views Downloads
Unknown 3 35
Beijing 1 0
Toronto 1 0
Ashburn 1 0

{[{ mDataHeader[type] }]} {[{ month[type] }]} {[{ tData[type] }]}

Share

Share to:

Comment

Related Items