Justin Leiber took the trouble to put pen to paper and express his discontent concerning my article "What Kind of Turing Test did Turing have in Mind?" and for this, I am grateful to him. Contrary to what the late Professor Feynman used to say ("Don’t listen to what I say, listen to what I mean!"), Leiber courageously says what he means, notwithstanding the academic etiquette which recommends to keep one’s anger for oneself. I’ll try to follow Leiber’s example, first because I think he is wrong in his criticisms even if I am perhaps guilty of not stating my own line of argument as clearly as I should have done, and secondly because everybody has to gain from a clarification of both our points of view. Since we have the opportunity, thanks to Tekhnema’s editors, to express ourselves, I’ll try to make the most of this dialogue and compare what I said with what Leiber claims I said. In studying Leiber’s line of thought, I’ll do my best to draw a sharp line between his rational arguments (they are three) and his emotional ones (they are plenty).
*
Let us start with emotions.
Justin Leiber repeatedly argues in his nine-page long harsh response that my line of argument is "absurd" (five times), "perversely" argued (four times), "preposterous," "silly," "uncharitable" and a "manic parody of psychoanalysis," which presents Turing as a "sexist," "scientistic machine-lover," impaired with a "perverse psycho-pathology." Last but not least, Justin Leiber hints that I should pay more respect for Turing as a person because of his great work in cryptanalysis during World War II which kept Britain (and finally the rest of the world) free from Nazi dominion. Leaving aside the sarcastic tone Leiber amply uses, I shall start with a brief comment on the last aspect of these very passionate criticisms, which deals with Turing as a person.
Justin Leiber then suggests that I tampered with Turing’s texts in order to fool the gullible reader. I shall of course comment on what I consider a most unfair criticism but considering this is not the main point of our disagreement, I shall first quote the whole paragraph written by Turing which Leiber and I scrutinize, and then leave my defense to a footnote.
1. First, I am most respectful of Turing as a person and as a scientist and there is not a single line in my article that could suggest I tried to bring into disrepute either his person, his theory of mind or his magnificent work during World War II for which he was decorated with an O.B.E.1 Nowhere in my article did I write, for example, that Turing was a "pervert" or had "pathological" tendencies, as Leiber lets the reader suppose I did. Leiber is hence the only one to be held responsible for these depreciatory terms and this can easily be checked by any unbiased reader. Nowhere did I try either to "foist" the impression on the reader that my point of view should allow anyone to label Turing as a "sexist scientistic machine-lover pervert."
The only reason I can figure out why Justin Leiber supposes that this was what I had in mind when writing my article is that he thinks my point of view is incompatible with the respect we should owe to science. But this is wrong. It means only that Leiber and I do not share the same idea about the nature of science, not that I despise Turing, Turing’s works or science in general. I should perhaps have emphasized more in my article the fact that the respect towards science is indeed a prerequisite anyone is entitled to demand from me before listening to what I have to say. I should therefore insist on the fact that I have been fascinated for years by Turing’s tremendous creativity in so many different conceptual fields (mathematics, logic, computer science, artificial intelligence, theoretical biology and philosophy) and that this is precisely what my point of view is about: what is creation or, more precisely, how can we account for conceptual creation? I had no intention of denigrating Turing or science, or both, and I must say I was quite stunned, considering what I wrote, that Leiber could have believed that that was my intention.
2. As I just said, I shall now quote the paragraph written by Turing which is at the core of our discussion (pp. 433-434 in the original Mind edition). It deals with what Turing calls the "imitation game":
The new form of the problem can be described in terms of a game which we call the "imitation game." It is played with three people, a man (A), a woman (B), and an interrogator (C) who may be of either sex. The interrogator stays in a room apart from the other two. The object of the game for the interrogator is to determine which of the other two is the man and which is the woman. He knows them by labels X and Y, and at the end of the game he says either "X is A and Y is B" or "X is B and Y is A." The interrogator is allowed to put questions to A and B thus:
C: Will X please tell me the length of his or her hair?
Now suppose X is actually A, then A must answer. It is A’s object of the game to try and cause C to make the wrong identification. His answer might therefore be
"My hair is shingled, and the longest strands are about nine inches long."
?In order that tones of voice may not help the interrogator the answers should be written, or better still, typewritten. The ideal arrangement is to have a teleprinter communicating between the two rooms. Alternatively the question and answers can be repeated by an intermediary. The object of the game for the third player (B) is to help the interrogator. The best strategy for her is probably to give truthful answers. She can add such things as "I am the woman, don’t listen to him!" to her answers, but it will avail nothing as the man can make similar remarks.
According to Leiber, I managed to quote only parts of this paragraph in order to draw the most extravagant conclusions from it. My remarks concerning this accusation can be found in the next footnote.2 Since the reader is now fully acquainted with the text that Turing actually wrote, we can leave these appetizers aside and come to the main course, the rational part of Leiber’s arguments.
*
As I said in the beginning, there are three arguments which Leiber opposes to my "perverse misreading." The first one deals with "an elementary example of the genetic fallacy"; the second one has to do with the interpretation of the woman role in the imitation game and the third one with the difference between the first and the second game. Let us start with the "genetic fallacy."
1. The genetic fallacy
Let me quote Leiber’s words first: "If the ‘Turing test’ has a clear, useful and established sense in the cognitive science, it is simply irrelevant to cognitive science whether Turing mis-formulated it originally or expresses it bizarrely through some unconscious, self-revealing, pathological cause." If I understand Leiber’s label, the "genetic fallacy" arises when one believes that the actual content of a concept depends on the subjective conditions of its discovery. I shall make two remarks about this.
1.1. Cognitive science is not just another objective theory, it is a theory of mind and this implies a very specific attitude towards the objects of the theory. I think this is where Leiber and I do not agree. Two points should be stressed here.
1.1.1. First, as a science of the mind, cognitive science should be able to describe (and hopefully explain) what is going on in the mind, especially the way individuals manage to create new concepts which must be worked out by real human beings.
Let me take an example in Turing’s biography. Since his sabbatical year of 1947-1948, Turing became more and more involved in theoretical biology. He finally wrote an article, published in 1952,3 which was considered by him as having the same value for biology as his 1936 article for computability. In this article, he builds a mathematical model of a specific biological reaction for which he planned the future use of computer simulation. But why was he so preoccupied by morphogenesis? After all, he did also pioneering computer simulation in a field which was much more familiar to him as a mathematician, the computation of the Riemann zeta function.4 Why morphogenesis then? Let us look at the example Turing chose to study, i.e. the properties of Hydra. What is striking in this example is the fact that a part of Hydra,5 once cut from its main body, regenerates itself. This is not a very common feature in biology. So there may be a more subjective motivation than objective knowledge in Turing’s attempt to modelize this phenomenon specifically. And we know, thanks to Andrew Hodges’ book, that Turing had heard of this very astonishing property in his childhood, while reading Brewster’s book Natural Wonders Every Child Should Know which had a very profound influence on his future career as a scientist. Since I am arguing that the question of gender difference receives a very special attention in the imitation game and is important to get rid of in order to make intelligent computing machinery possible, I would also argue that there might have been a phantasmatic motivation in studying the very peculiar case of Hydra which exemplified the biological possibility of parthenogenesis, this kind of auto-genesis which does not need gender difference to generate new individuals. Is this interpretation "perverse"? Or do I try to foist the impression that Turing was a "pervert"? Not at all. I only stressed the coherence of Turing’s personal motivation both in what became called artificial intelligence and in theoretical biology. And what I am arguing is that, in noticing the general drive which is implemented in Turing's scientific behavior, I do gain some insight concerning what a mind is like, and hence concerning the science of the mind in general.
Of course, in artificial intelligence as well as in theoretical biology, we can now describe, use and improve the concepts of a "Turing machine" and of a "Turing structure" without knowing all these biographical details and more or less unconscious motivations. But these are not just details from the point of view of a theory of mind: they exemplify from a general point of view the way individuals use their own personal history and even their phantasms to create new concepts. Is this irrelevant to cognitive science? If it is, I don’t know what cognitive science is talking about but obviously not of the mind. A theory of mind must find a way of taking into account how individuals participate in the creation of universal concepts. Such a theory has to deal with how specific biological organisms express their own thinking through (natural) language because language is the means through which universality can become conscious to individuals and be shared in human community. That is why my analysis is focused on every detail of Turing’s text and that is also why I do not agree with Leiber when he says that I fall into the trap of "genetic fallacy": this would indeed be the case if I was trying to derive the objectivity of concepts from an individual source but I am not talking about concepts once they have acquired their objectivity, I am talking about concepts as they are produced by individuals through language in order to become objective. To me, this is what a science of the mind is about and it makes a real difference.
1.1.2. But this has at least one very important consequence: a theory of mind must develop conceptual tools which can justify its own genesis, i.e. its own process of becoming objective. This reflexive fold is necessary to a science of the mind, it would otherwise fail to describe its object of study, the mind.
As a matter of fact, the genetic description of the conquest of objectivity is exactly what Turing does in his 1950 article. We must not forget that Turing’s aim is actually to ruin, by means of the imitation game, the reader’s prejudices against intelligent machinery. The text itself is hence a process through which the reader will hopefully be able to imitate Turing’s position. Therefore, the aim of Turing’s paper is to show the way the reader should follow in order to conceive intelligent machinery as possible. And the best way Turing found to make his point of view accessible to the reader is to describe his own process of creating new concepts. This is what the imitation game is about. And this is also why the question of creation is linked to Turing’s own interpretation of sexuality, pace Leiber and many other commentators who would be happy to forget that the aim of the imitation game, as it is explicitly set up by Turing, is to question the nature of gender difference.
In the classical interpretation of the imitation game, the actual process of the game is completely left aside: the reader is expected to perform silently the thought experiment of the imitation game and to ignore the process dealing with gender difference through which Turing’s point of view will finally be conceived as possible. The imitation game becomes therefore an "objective" (Turing) test and almost an empirical proof that intelligent machinery is possible. I do not want to suggest that artificial intelligence is not possible but I think the interpretation of the imitation game as a preliminary to artificial intelligence leaves aside what is at stake in Turing’s paper: the status of a science of the mind and the specific problems that arise from such a status.
1.2. The second point I wanted to make on this issue is rather secondary compared to the first one. Even if my point of view were a typical example of "genetic fallacy" (which I doubt), there would be no reason to identify the imitation game and the Turing test. But I have never found this distinction explicitly acknowledged in the cognitive literature and as far as I know, Leiber’s response to my paper is the first one which made the distinction between the imitation game considered as the "genesis" of the Turing test and the "Turing test" itself6 reasonably clear.
Following Leiber’s line of thought, it would therefore be advisable to stress the fact that cognitive science made some progress since the 1950s as the clumsy presentation of the argument made by Turing can now be replaced by a clear-cut new setting to be called, in honor of the great pioneer, the "Turing test." As a matter of fact, it is true that this kind of improvement happens everyday in science: for example, the long and hard-reading proof of the absence of a mechanical solution for the halting problem as it was originally written by Turing in his famous 1936 article7 is now replaced even in ordinary textbooks by one or two pages of a much neater, simpler proof, though the general problem is still called "Turing’s Halting Problem."8 But, as I said earlier, what is common rule in other parts of science may not be common rule in the science of the mind and that is why I think Leiber is wrong in accusing me of "genetic fallacy."
Let us now focus our attention on the imitation game itself and more precisely, on the role played by the woman since this is also a point where Leiber and I disagree.
2. The "woman" strategy
I must admit that my line of argument was not particularly clear on this issue, though I still think I am right. According to Leiber, I made a mistake in discussing the part played by the woman in the imitation game. I assumed that "the woman is supposed to have the same reciprocal imitative task (imitating the man) as the man," contrary to what Turing suggests, namely that the task of the woman "is to help the interrogator" which is, Leiber adds, "clearly the best possible strategy." Let us have a look at the nature of the woman strategy.
2.1. If we follow Turing’s description, her strategy must include two features: first, it must be coherent; secondly, it must be truthful.
2.1.1. Why should it be coherent? Is it possible to imagine the contrary? Well, we could of course imagine a completely chaotic behavior on the woman’s part which would prove her gender to be undetectable. Such a strategy reminds me of Stanislaw Lem’s novel His Master’s Voice where a crypted message received from outer space cannot be deciphered because of its purely chaotic appearance (something like a segment of the decimal expansion of pi which nobody could recognize as belonging to pi). But I agree that this would be a dangerous strategy for the woman because it is hard to imagine a purely chaotic series of answers given to the interrogator’s questions. Let us then assume that the woman strategy must follow a certain line of discourse and that she would better stick to it. So I agree with Leiber on this point but I contest the fact that telling the truth is her "best possible strategy." It may well be the best possible strategy of one of the two players but not of the woman specifically. Let me explain this point.
2.1.2. Why should she speak the truth? We could first imagine that this strategy derives from the rules of the game itself: as a matter of fact, there must be someone to tell truthful answers, otherwise the other player could not imitate these answers and imitation is what it is all about in this game. But why should this truthful part be played by the woman and not by the man? This is what I would like Leiber to tell me because we could easily imagine a game in which the woman would try to imitate the man. So the real – and to me, unanswered – question runs as follows: what is the relationship between the logical level which rests on the difference between truth and falsehood and the biological level which rests on the difference between feminine and masculine? The only thing Turing suggests is: "The best strategy for her is probably to tell truthful answers." But clearly, this is not enough and nowhere else in the 1950 article did Turing explain why it is the woman who is supposed to tell the truth. And Leiber does not give any answer to that either. On the contrary, I do. So, if we stick to Turing’s explanations, we can figure out as I wrote in my article a kind of "hierarchy in the players’ responses."
According to Leiber, it is "silly" to suppose that the first and the second game are "dis-analogous." First, it is only fair to notice that this silly idea was not originally mine but was first put across by Andrew Hodges in one of his papers.9 Secondly, I think there is a confusion on this issue that can rather be clarified.
Leiber is right in recalling that Turing’s aim in the imitation game is to show that since the most important physical difference between humans is not apparent in game 1, the even bigger difference between human beings and computers is not apparent in game 2 either. But how does he manage to reach this conclusion? It is here that the process of the game should be studied carefully and this is where Leiber and I disagree.
3.1. If we reach the intermediate conclusion concerning game 1 that the most important physical difference between humans is not apparent, it means that the presupposition the reader is supposed to have that there is a connection between the physical and the mental has already been ruined. And this conclusion can only be reached if by game 1 one can establish that there is no connection between the physical and the mental aptitudes of the human players. But this is precisely what game 1 cannot establish because there is no way in game 1 by which one could connect the physical and the mental in order to show that this connection is illusory: the fact that the man can imitate the woman cannot establish by itself that there is no connection between the physical and the mental, it is simply neutral about this issue. So where does the prejudice that the presupposition that there is no connection between the physical and the mental has been destroyed come from? It comes from the final conclusion of game 2. But this conclusion itself is not yet reached in game 1. So there is a difference between the two games, contrary to what Leiber thinks.
3.2. That said, how could we reach the conclusion of game 2? Only if the imitation is successful, that is to say, if we let the interrogator play indefinitely, otherwise we could never be sure that the interrogator is not going to ask the very question that would bring the decision. Hence, if we agree in saying that the conclusion of game 2 can be actually attained, we have to suppose that the difference between the mental and the physical has been established before the beginning of the game.
This is exactly what I was aiming at when I was discussing what Leiber calls the "genetic fallacy" I was supposed to have made: in fact, the prejudice that there is a clear cut difference between the physical and the mental is necessary for an objective science of the mind to take place. But what my interpretation of Turing’s imitation game shows, contrary to this too simplifying interpretation, is that there is another science of the mind possible, if only we pay attention to the details of what is said, no matter how embarrassing they are.
*
I think I can now recall the three points I made in the conclusion of my article:
•First, the "Turing Test" as it stands in the cognitive community is not feasible. This does not entail that cognitive science is not feasible; it only means, from a more Socratic than a dogmatic point of view, that we have to question the belief that a science of the mind should be of the same type of objectivity as artificial intelligence: as a science of mental processes, the science of the mind should not be too quickly akin to the point of view of full blown objectivity.
•Secondly, the way gender difference is taken into account plays a role in the imitation game that we cannot ignore and certainly also in the process of conceptual creation in general. Therefore a psychoanalytical interpretation is not to be ignored on an a priori basis and it does not imply reducing the mind to its pathological aspects.
•Thirdly, a reflection on the relationship between the biological and the linguistic aspects of the human mind is a necessary prerequisite to the constitution of a science of the mind precisely because the distinction between the biological and the logical is not given, contrary to what seems to be presupposed by Justin Leiber, but is the result of a construction which can be studied for its own sake.
Let me end this response by mentioning that I dedicated the article that enraged Justin Leiber so much to the memory of Edward Gumbel (1913-1995) who, as a German-born British citizen, knew what persecution meant but nonetheless managed to keep through his life this light-hearted sense of humor – a very Turingian virtue indeed – which seems to me one of the most civilized features produced by British culture all of us should try to put into practice.
Reference matter
Hodges, A.
- (1983). Alan Turing; The Enigma of Intelligence. London, Burnett Books Ltd, Counterpoint. Other editions: Alan Turing; The Enigma.
- (1988). "Alan Turing and the Turing Machine", The Universal Turing Machine; a Half-Century Survey. R. Herken ed., Oxford University Press, Oxford: 3-15.
- http://www.wadham.ox.ac.uk/~ahodges/main.html
Leiber, J. (1991). An Invitation to Cognitive Science. Oxford, Basil Blackwell.
Lassègue, J. (1996). "What Kind of Turing Test did Turing have in Mind?", Tekhnema, 3, Spring 1996: 37-58; also available at: http://www.gold.ac.uk/tekhnema/
Minsky, M. (1967). Computation: Finite and Infinite Machines. Englewood Cliffs, N. J., Prentice-Hall.
Turing, A. M.
- (1936). "On Computable Numbers with an Application to the Entscheidungsproblem." Proceedings of the London Mathematical Society 42: 230-265 reprint in [The Undecidable, M. Davis ed., Raven Press, Hewlett, New-York, 1965: 115-154].
- (1950). "Computing Machinery and Intelligence." Mind LIX (236; Oct. 1950): 433-460 reprint in [Collected Works of A. M. Turing vol. 3: Mechanical Intelligence, D. C. Ince ed., Elsevier Science Publishers, Amsterdam, 1992: 133-160].
- (1952). "The Chemical Basis of Morphogenesis." Phil. Trans. Roy. Soc. B 237: 37-72 reprint in [Collected Works of A. M. Turing vol. 4: Morphogenesis, P. T. Saunders ed., Elsevier Science Publishers, Amsterdam, 1992: 1-36].
- (1953). "Some Calculations of the Zeta-function." Proc. London. Math. Soc (3) 3: 99-117 reprint in [Collected Works of A. M. Turing vol. 1: Pure Mathematics, J. L. Britton ed., Elsevier Science Publishers, Amsterdam, 1992: 79-97].
Notes
1 Hodges 1983, p. 338. Concerning Leiber's note (# 3) to my error of reference, the immensely valuable book written by Andrew Hodges is indeed called Alan Turing: The Enigma of Intelligence in my Unwin Counterpoint paperback edition bought in 1987 and is indexed under this title in the British Library Cataloguing in Publication Data. This edition is now out of print. I am therefore guilty of quoting the edition I am used to and I apologize for my laziness: I should give the more common title of Hodges' book since, in other editions, it is called Alan Turing: The Enigma. All of this can be checked on Andrew Hodges' web site at http://www.wadham.ox.ac.uk/~ahodges/main.html
2 I should advise Leiber to be very cautious with this accusation for at least two reasons.
First, he must know what tampering is about since he misquoted four times the very paragraph he blames me to have misquoted. Describing the imitation game in his 1950 article, Turing writes: "It is played with three people" and not "between three people," as Leiber writes; Turing also writes: "The object of the game for the interrogator is to determine which of the other two is the man and which is the woman" and not "which of the two"; he finally writes: "We now ask the question, 'What will happen when a machine takes the part of A in this game?'" and not "in the game." I won't make a fuss about this because I don't think it alters the general meaning of the text. But I must say it partakes of the same carelessness we find in the cognitive "unanimous" community which tends insidiously to transmute, as if by magic, the imitation game in a more presentable "Turing test." Before accusing me of God knows what kind of intellectual manipulation, Leiber himself should pay more attention to the text because I consider it as the minimal form of respect we owe to its author. To paraphrase Feynman's motto: "Do not read what you suppose was written, read what is written!" In the same way, Leiber should for instance have a look at p. 77 of Andrew Hodges' biography of Turing and try to make sense of the following sentence: "Alan told his friend about how much he resented having been circumcised, and also of his earliest memories of playing with the gardener's boy (presumably at the Ward's house), which he thought had perhaps decided his sexual pattern." But of course this would imply taking into consideration the "absurdist charm" of a psychoanalytical interpretation Leiber prefers to ignore. The strategy consisting of ignoring anything which would not fit in your own point of view ("I will not follow Lassegue into this latter exercise for it soon attains what is almost self-parody") is in itself purely irrational since no discussion can ever take place. I can only regret it and claim that since my point of view consists in studying the ways we can account for conceptual creativity, I think it is necessary to take into consideration the way people justify their own attitude towards creation, which implies to take into account their own attitude towards sexuality since this is precisely why creation is a driving force for the speaking animals that we all are.
Secondly, it is true that I did not quote the whole paragraph describing the imitation game but Leiber does exactly the same. The following sentence written by Turing was deleted in Leiber's response to my article: "She can add such things as 'I am the woman, don't listen to him!' to her answers, but it will avail nothing as the man can make similar remarks." Furthermore, contrary to Leiber, I quote the missing lines later. So Leiber can accuse me of quoting some sentences out of context but not of "crucially" leaving out some of the material that Turing wanted to be read and whose excision fosters my "perverse misreading." As a matter of fact, I did not quote the entire paragraph the first time precisely because I knew I would quote the missing sentences later in the article and I did not want the reader to get bored with the repetition. I don't think anyone can honestly draw definitive conclusions against my point of view from this purely pedagogical setting.
3 Cf. Turing, 1952.
4 Cf. Turing, 1953.
5 Cf. Turing, 1952, § 11, p. 68: "Hydra is something like a sea-anemone but lives in fresh water and has from about five to ten tentacles. A part of a Hydra cut off from the rest will rearrange itself so as to form a complete new organism."
6 As I said in my article, Leiber himself in his book disposes of this important distinction in only two sentences. Cf. Leiber, 1991, p. 110. Let me stress the fact that there was nothing personal against Leiber in my quoting the way he presents the imitation game. It was just an example of the classical way it is described in the cognitive community and this also can easily be checked by any unbiased reader.
7 Cf. Turing, 1936, § 8.
8 For example, Minsky, 1967, p. 148-149.
9 Cf. Hodges, 1988, p. 10. This point is also stressed in his web site at the entry: "The Turing Sex Test."