ALAN TURING                   

In his short life, Alan Turing (1912-1954) made foundational contributions to philosophy, mathematics, biology, artificial intelligence, and computer science. He, as much as anyone, invented and showed how to program the digital electronic computer. From September, 1939, his work on computation was war-driven and brutally practical. He developed high speed computing devices needed to decipher German Enigma Machine messages to and from U-boats, countering the most serious threat by far to Britain=s survival during World War Two.

Because of official secrecy, his war time exploits were unknown until the 1980s. By then some of his inventions no longer seemed connected to a real human being. Literature buffs read classics. Scientists just cite them. For example, Turing=s 1936 paper, AOn Computable Numbers,@ was soon seen as the most important theoretical paper ever written on computation. So mathematicians, engineers, and computer scientists came to write of  Aturing machines@ and Auniversal turing machines@ almost or completely forgetting that there was an Alan Turing. Almost the same thing happened with his breezy1950 paper, AComputing Machinery and Intelligence,@ which set the agenda for cognitive science. From it scientists and philosophers extracted the goal of writing Aintelligent@ computer programs good enough to pass the Aturing test@ for simulating human intelligence. While some computer scientists boasted that they would program a Apasser@ within a decade, they have not come close after five decades. Turing also suggested several other ways, aside from programming, that might be used to simulate human intelligence. When he did so, he anticipated approaches that have been ballyhooed in the 1980s and 1990s C training connectionist nets, sending a Achild machine@ to school, perhaps equipping it with eyes, ears, and hands, etc. The centrality of natural language in Turing=s test, and in human intelligence, is reinforced by the continuing and blatant failure of machine translation from one human language to another. Computers now easily exceed the most talented humans in arithmetic calculation and chess playing, but none come anywhere close to the performance of ordinary human translators. On the other hand, some of Turing=s work was so far ahead of his time that it earned credit only in retrospect. His last published paper (1952) anticipated the most important new approach in the last half century of developmental biology. But it wasn=t until the late 1970s that scientists began to refer to Aturing structures@!

In his 1936 paper, Turing answered the deepest computational question C  whether there is a finite mechanical procedure for deciding whether any given mathematical statement is true or false. Turing realized that the key was to get clear about what a Amechanical procedure@ (or computation) was. Mathematicians tended to assume that it meant Aexplicit, step-by-step, requiring-no-creativity, etc.@ and left it at that. Turing saw that if a procedure were mechanical, it could be automated.

Turing=s answer is to imagine a starkly minimal machine. He called it a Atheoretical computing engine,@ for he had no intention of building one. This is a machine to think with. It is fed by an indefinitely long tape divided into frames like those on a roll of film. The machine has a Aread head@ which can tell whether the frame under it has a A/,@ a A\,@ or is blank. The read head is also a Awrite head@ that can erase, write in a A/@ or A\,@ or do nothing. The machine also can move one frame forward or backward or stay in place. Then we have another move and so on. At the beginning of each move, the machine is in one of a small number of Ainternal states@ and this may switch to another after the move. The machine is built to enact instructions in its machine table of the form Aif / is read and the internal state is 1, then erase, move one frame forward and go into state 2.@ A turing machine for adding would get, say, the input sequence A//   //@ and then automatically change it into A////@ through a long series of steps and then stop. For us this is the computation A2+2=4.@ Since Turing was thinking about a theoretical device, he didn=t mind that a million would be represented by a like number of A/@s. Indeed, he insisted that an actual and literal universal turing machine would be much too slow for practical computation. 

Turing goes on to show that anything mathematicians call a mechanical problem or computation can be represented by some turing machine. Further, he shows that there is a universal turing machine. Depending on the input tape sequence, this turing machine can turn itself into any particular turing machine, do a computation, and then turn itself into another turing machine and do a computation, and so on. In 1936, Turing thought up the general purpose digital computer and gave its definitive abstract description. The tape can also store data and programs (Amemory,@ Askills,@ Aplans@), represent incoming data (Asensory input@), and issue output instructions (Amotor outputs@). So Turing has also given the framework in which to describe any sort of individual thinker, you and me included. As many scientists have said since the 1960s, we are, more or less, universal turing machines and so are our digital electronic computers; for the same reason, we now think of thinking as computing or data-processing. (To return to his original question, Turing also showed that no possible computer (digital electric, human, or E.T.) can decide the truth or falsity of every mathematical formula. Turing broke the species barrier: given his minimalist physical description of computability, anything could be a mind whatever its physical composition).

In his 1950 paper that created the field of artificial intelligence, Turing asks you to imagine the following Aimitation game.@ We have some judges who communicate by a terminal to A and B=s terminals. One of these terminals is operated by a woman, one by a man. Under the judges= questioning, the woman tries to convince them that she is the woman, while the man tries to convince them that he is the woman. For example, the judges might ask, AHow do you do your hair?@ Turing comments that the best strategy for the woman is to tell the truth. Interrogators look for inconsistencies. Telling the truth is the simplest way to avoid them. The man is going to have a lot to keep track of. If the man manages to win, you might say he can think like a woman.

Turing then proposes that we substitute a computer for the man. If the computer Apasses,@ that proves that the computer can think like a human being and that proves that the computer can think period. Turing remarked that the test draws a clear line between the physical and mental. He likened putting the contestants in separate rooms, so the judges can=t see them, to musical contests where the players perform behind a screen, so the judges won=t be biased by their physical appearance. This is what we now call the Aturing test.@ Turing didn=t think it was going to be easy to create a passer. He provided a vivid and testable goal for artificial intelligence research.



Born 23 June 1912, London. Sherborne School, 1926‑31; Wrangler, Mathematics Tripos, Kings College, Cambridge,1931; Ph.D., Princeton University, 1938. Fellow, King's College, 1935‑45; Princeton University, 1936‑38; British Foreign Office, Bletchley Park, 1939‑45; National Physical Laboratory, 1945‑48; University of Manchester, 1948‑54.  Smith's Prize, Cambridge University, 1936; Order of the British Empire, 1946; Fellow, Royal Society, 1951. In 1952, Alan Turing was convicted of Ahomosexual acts@ and legally forced to take huge hormone doses that rendered a man who nearly represented Britain in the 1948 Olympic marathon impotent, femininely-breasted, and obese.  Died of self-administered cyanide, 7 June 1954, Manchester England.            


Selected Works

AOn Computable Numbers, with an Application to the Entscheidungsproblem.@ Proceedings of the London Mathematical Society, 1936.

 AIntelligent Machinery.@ In Meltzer, B. and Michie, D., eds., Machine Intelligence. New York: American Elsevier Publishing Co., 1947/1970.

AComputing Machinery and Intelligence.@ Mind, 1950. Also available in many other places including Hofstadter, D. and Dennett, D.,  The Mind=s I, New York: Basic Books, 1984.

AThe Chemical Basis of Morphogenesis.@ Philosophical Transactions of the Royal Society, 1952.


Further Reading.

Hodges, Andrew (1984). Alan Turing: The Enigma. New York: Simon & Schuster.

Leiber, Justin (1991). An Invitation to Cognitive Science. Oxford: Blackwell.

Justin Leiber



Philosophy of Language

Rather like scholastic philosophy of the late middle ages, twentieth century philosophers have seen philosophy as linguistic analysis, as the attempt to discern the logical structure of reality through discerning the formal structures, the superficial or deep grammar, of  the language in which we report or think reality. Philosophers are particularly interested in certainty, in necessary truths as opposed to mere happen-chance events. Nineteenth century mathematicians, similarly, honed a sense of mathematics as syntactically grounded in formal language or in a consistent set of linguistic conventions. Since philosophers conceived of their enterprise as a search for helpful logical truths and not dependent on any experiential truths (philosophers don=t do experiments), 20th philosophy became logico-linguistic analysis, and so philosophy of  language became, in short and for much of the past century,  philosophy.

Linguistic philosophy was not just seeing the features of language that revealed the world=s categorical structure but also seeing beyond language=s perhaps misleading surface features. To give a standard example, the verb to be in English plays three vastly different logical roles. Is can mean identity as in A2 = 2,@ ADubya is President George W. Bush,@ or Aheat = the average motion of molecular particles.@ Is however can also mean predication as in AThe sky is blue,@ ARoses are red,@ or AThe earth is spherical.@ Finally, Is can mean existence as in AThere is a Santa Claus@ or AGod is@ (as opposed to AGod is not@). Bertrand Russell, while imprisoned as an antiwar demonstrator in WWI, wrote in Principles of Mathematical Philosophy that it was important to keep these three senses wholly separate in linguistic analysis (so important that Russell added that he would declaim so even if he were Adead from the waist down and not merely in prison@). In the mathematical logical notation that Gottlob Frege and Russell created in the late nineteenth century, these notions were indeed wholly separate and syntactically marked. With a notation that capitalized predicates and lower-cased names and variables for individuals, Aa=b@ means two names stand for the same individual, APa@ means predicating P of individual a, and A  x[Dx]@ means, if D predicates divine, that there exists a god. Russell stressed that this notation commendably dissolved the Aontological argument for the existence of God,@ which mostly simply runs, God is that being who possesses all possible perfections; existence is a perfection; therefore, God exists. This argument confounds the hypothetical predications AIf something is Divine, then it has various Perfections@ and the existential ASomething exists that is Divine.@  

Russell was particularly praised for his Aparadigmatic@ analysis of  Adefinite descriptions,@ phrases of the form The so and so. Consider  AThe present King of France is bald.@ If you think this statement has a subject/predicate logical form, and believe that statements are either true or false, then you seem to have to say that either AThe present King of France is bald@ is true or AThe present King of France is not bald@ is true. The problem of course is that there isn=t any present King of Franc (when Russell wrote and subsequently). It is no solution to say AThe present King of France@ means nothing because then the AThe present King of the United States@ and AThe round square@ and countless other phrases would presumably also mean, or stand for, nothing, but it is evident that all such phrases differ in meaning. If  a revolution occurred in the United States in 2084, there might then be a King of  the United States but that would not mean there would be a round square and a French King as well (or that nothing had changed to something). Russell insists that AThe present King of  France@ does not mean anything all by itself. Rather, AThe present King of France is bald@ means AThere exists an x that bears the predicate of being kingly of France, and if any y also bears that predicate,  y=x, and x is bald.@

Following Russell=s lead, the 1920s logical positivists of  the Vienna Circle insisted that logico-mathematical truths provide the formal structure within which the observational truths of experience array themselves. Theoretical terms, such as Avital spirit,@ Aelectron,@ or Avirus,@ are only acceptable if they can be cashed out completely in observational terms. Vienna Circle Harvard philosopher W. V. O. Quine claims that by casting science into Russell=s austere predicate logic, you can determine what science says is real. Indeed, such a translation into predicate logical will strip off the possibly misleading superficial features of actual languages. Philosophical linguistic analysis, conceived in this fashion, cares nothing for the phonology of language, for literal physical sound streams and their transformation into the sharper, leaner, and deeper structures that fluent speakers hear. Moreover, the exclusive concern is with the truth or falsity of sentences that describe the world, collectively feeding what scientific generalizations we can muster about the world. As Quine put it, startlingly, all of science is held up as one sentence to nature C but that is just to insist that we collectively assert and understand this scientific sentential consensus, not that anyone actually says this one sentence. Philosophy of language, as so understood, has little concern with how language is used in the acts and interactions of everyday life.

By the mid 20th Century, however, philosophers began to shift from a concern with the true/false relationship between sentences and the world to a more expansive concern with sentences as actions of speakers, who carry individual responsibility to us for what they say, indeed for what they do in saying so. In the 1950s, Oxford philosopher J. L. Austin drew attention to performatives, sentences that, when uttered by the appropriate person in the appropriate  circumstances, do something. If I say, AI promise to return your $20 tomorrow,@ I am not describing, truly or falsely, some peculiar mental state; rather, by my saying, I make a promise. If I say, ABy the power vested in me by the State of New Jersey, I appoint you Port Commissioner,@ I make an appointment. AI promise X@ and AI appoint Z@ are explicit performatives, where the main verb of the sentence explicitly indicates what action is performed. However, AI certainly will return your $20 tomorrow@ will, given appropriate circumstances, constitute a promise. Similarly, if an officer says to his subordinate, AYou will move your men to that bridge,@ he has given an  order. Indeed, Austin maintained that every use of language, or every speech act, has a performative aspect. For example, if I say AThere=s a bittern in your garden,@ I should know a little something about bitterns and have had some opportunity to identify the bird. If I have no idea of what bitterns look or sound like, and indeed, have not been in or near your garden, I have no right to say what I did. Given that I meet those minimal requirements, I may demur if you ask, AHow do you know it=s a bittern?,@ pleading perhaps AWell, I don=t know it=s a bittern but it is a large, white-feathered marsh bird@ C or I may take a further plunge and say, AOh, I know it=s a bittern: I got a clear view and, growing up in the Fens, I=d know that booming anywhere.@ Philosophers concern with semantic, performative, and pragmatic aspects of language has meant some fruitful interaction with linguistic science, particularly given the concern with syntax and logical form stressed by Noam Chomsky and other generative linguists since the 1960s.

But still other philosophers, have come to feel that philosophy is well quit of  an exclusive emphasis on philosophy as linguistic analysis. In the middle decades of the 20th Century, philosophers concerned with values emphasized metaethics or the Alogic of the language of morals@; more lately, philosophers have addressed specific normative issues. Similarly, many recent philosophers have constructed rational choice, and social contract, theories. Thinkers as diverse as logician Saul Kripke and linguist Noam Chomsky have argued that necessary truths are more central to science than the real but often trivial analytic truths of language.   


Further Reading

Austin, J. L., Philosophical Papers, Oxford: Clarendon Press, 1961.

Ayer, A. J., Language, Truth, and Logic, London: Gollancz, 1936.

Chomsky, Noam, Knowledge of Language, New York: Praeger, 1986.

Kripke, Saul. ANaming and Necessity,@ in Semantics of Natural Language, edited by Gilbert Harman and Donald Davidson, Dordrecht, Holland: D. Reidel, 1972.

Quine, W. V. O., Word and Object, Cambridge, Mass.: MIT Press, 1960.

Russell, Bertrand, Introduction to Mathematical Philosophy, London: Allen and Unwin, 1919.