ラベル 数学 の投稿を表示しています。 すべての投稿を表示
ラベル 数学 の投稿を表示しています。 すべての投稿を表示

2016年11月22日火曜日

【メモ】 Vincenzo De Risi / Geometry and Monadology: Leibniz's Analysis Situs and Philosophy of Space

Preface

That space is actually constituted by points, though here abstractly meant as terms of situational relations, is perhaps the highest result of Leibniz’s geometrical investigation and, at the same time, it also marks the core of Leibniz’s theory of phenomenal expression. It shows in fact that a set of non-spatial relations (such as those occurring between monads) can be isomorphic to (“expressed by”) a set of situational relations that per se suffice to produce phenomenal extension and thus, ultimately, faithfully represent the supersensible through the sensible. (p. xii)

Chapter I. Historical Survey

"Having thoroughly inquired, I have found that two things are perfectly similar when they cannot be discerned other than by compresence, for example, two unequal circles of the same matter could not be discerned other than by seeing them together, for in this way we can well see that the one is bigger than the other. You may object: I shall measure the one today, the other tomorrow, and thus I will discern them even without seeing the two of them together. But I say that this still is a way of discerning them not by memory, but by compresence: because the measure of the first one is not stored in your memory, for magnitudes cannot be retained by memory, but in a material measure marked off on a ruler or some other thing. In fact, if all of the things in the world affecting us were diminished by one and the same proportion, it is evident that nobody could make out the change." (Letter to Jean Gallois, September 1677; GM I.180, A III.ii.227-28, A II.i.380) (pp. 58-59)

Chapter II. Geometry

He [Leibniz] says that two figures are congruent not only provided they are only distinguishable through the simultaneous perception of them (which would be the similarity definition), but also provided such a perception requires the presence of a third object (which was not a condition for similarity). For example, two spheres are similar, for you can distinguish them only by seeing them both simultaneously (otherwise, seen one at a time, they prove totally indiscernible). Furthermore, if they are also congruent, you can distinguish them only through their reciprocal situation (one being on the right, the other on the left), which calls in the use of a reference element external to the two spheres such as, for instance, a third object that fixes right and left, or proximity and distance, or an ideal object such as given set of coordinates (Specimen Geometriae luciferae, GM VII.275) (p. 143)

On the notion of homogony and in general on the importance of motion in the foundation of the continuum, Cassirer has forcefully insisted in his Leibniz' System: «Man begreift jedoch, dass es für Leibniz schwierig sein musste, die reine Auffassung der Kontinuität, die er zunächst am Problem der Veränderung gewonnen hatte, auch dann festzuhalten, wenn er für die Definition vom Sein der Ausdehnung ausging. Denn gerade dies erwies sich als das Originale von Leibniz' Gedanken, dass er die Unmöglichkeit zeigt, die Stetigkeit als Eigenschaft an einem gegebenen Sein erschöpfend zu bestimmen. Erst aus dem Gesetz des Werdens verstehen wir das Kontinuum. Die Unbestimmtheit, die wir auch in Kants Definition noch fanden, lässt sich gleichfalls darauf zurückführen, dass Kant die Kontinuität hier noch als Eigenschaft der Grösse, nicht als Prinzip ihrer Entstehung sucht.» [It is understood, however, that it would be difficult for Leibniz to maintain the pure notion of continuity, which he had first gained from the problem of change, even if he had assumed the definition of being as extension. For precisely this proved to be the originality of Leibniz's thinking: that he showed the impossibility of exhaustively determining continuity as a property of a given being. Only from the law of Becoming can we understand the continuum. The indeterminacy which we have also found in Kant's definition can also be attributed to the fact that Kant seeks continuity here as a property of magnitude rather than a principle of its origin.] (Werke, vol. 1, p. 167). Here we need however to remark that, no matter how much you look into kinematics, never will you extract from it nor from its metaphysical foundation even the least proof that it should be continuous. Nor let us be deceived by the fact that Leibniz has assumed it to be true since his young years for definition purposes («motus est mutatio situs continua»), as he should then demonstrate (and he does not) that a continuous change of situation is also possible, which is untrue if one assumes space to be discontinuous. The answer Cassirer seems to suggest to such a puzzle is that continuity of motion is however dynamically founded, i.e. founded on the continuity of force, which in turn may be founded on the continuity of monadic activity. Yet, it is precisely in this way that one goes out of kinematics and thus geometry, so that all objections about using motion in geometry present themselves again—in fact motion cannot be used in geometry because the continuity of changes of a monad is (perhaps) only a contingent one. Hence, it is not motion in itself, let alone the motion generated by the system of forces, that can guarantee continuity of space, but rather, as already suggested, some transcendental operation of production of space. (pp. 184-85n)

Chapter III. Phenomenology

"That is said to express a thing in which there are relations [habitudines] which correspond to the relations of the thing expressed. But there are various kinds of expression; for example, the model of a machine expresses the machine itself, the projective delineation on a plane expresses a solid, speech expresses thoughts and truths, characters express numbers, and an algebraic equation expresses a circle or some other figure. What is common to all these expressions is that we can pass from a consideration of the relations in the expression to a knowledge of the corresponding properties of the thing expressed. Hence it is clearly not necessary for that which expresses to be similar to the thing expressed, if only a certain analogy is maintained between the relations." (G VII: 263–64) (p. 298)

What most matters here about the relational character of the distinctive properties of monads is that an isomorphism which fully or partly expresses intermonadic relations as relations occurring between representational elements whatsoever is clearly a faithful isomorphism, in the sense that beyond such relations there is nothing else that may further characterize monads. In other words, saying that a perceptual isomorphism expresses the relations between monads is equivalent to saying that it expresses monads themselves, because monads consist in those relations or, at least, are perfectly individuated by them. Leibniz's substance, as it were, ends up as a function. (p. 323)

As however supersensible and phenomenal elements, at least in their distinctive features, are reducible to two sets of relations, an isomorphism for which all relations between monads are preserved would only express the logical identity between the noumenal world and the phenomenal one—hence, it would radically deny any phenomenalism. As we can see, some relations must exist that are not preserved by the situational isomorphism or at least (distributive) by the totality of homomorphisms. (p. 326)

[M]ost divergences between Critical Idealism and Leibnizean Phenomenalism can be reduced to the following. In Leibniz, the relation between phenomena and noumena is regulated by a structural morphism, whereas in Kant by a simply functional relation (not a determination, but only the ground of it, the Bestimmungsgrund). Thus, for example, no real intentionality can be recognized in Berkeley's idealism, because here, owing to the lack of any partial isomorphism, the noumenal world is simply overruled (i.e., identified with the phenomenal one): «But, say you, though the ideas themselves do not exist without the mind, yet there may be things like them, whereof they are copies or resemblances, which things exist without the mind in an unthinking substance. I answer, an idea can be like nothing but an idea; a colour or figure can be like nothing but another colour or figure.» (Treatise, §8). It is evident that here Berkeley deals with similarity in a most ingenious way and without having the notion of a structural identity that can be preserved beyond material distinctions of genus between phenomena and noumena. What he misses is the very concept of expression. Relevantly, even though no morphism is involved in Kantian philosophy, some functional relationship however subsists there linking a phenomenon to the thing-in-itself. (p. 327)

[T]he limited expressivity of the perceptual isomorphismwhich is required in order to establish a distance between phenomena and noumena—finds its further specifications in certain limits of aesthetic comprehension. Furthermore, as it is the only representational content distinguishing monads from one another, only the limitedness of aesthetic comprehension guarantees the possibility for a plurality of substances to be determined, which substances at this point can be distinguished through the different limits of their aesthetic comprehension. In effect, since monads also express such a difference phenomenally through each one's different situs, the most accurate specification of the limits of their finite aesthetic comprehension is to be found in the situation of the phenomenon that expresses the representing monad. Clearly, a monad superordinated to another one will have a broader aesthetic comprehension. We have arrived thus at Leibniz's well-known simile of a town multiplied perspectively. ... [E]ach sight sees the whole town (however variously deformed according to different points of view)—which means that it is not different isomorphisms we are dealing with here, but just one isomorphism variously specified into different homomorphisms. (p. 334)

[I]n order for the various (real) homomorphisms to be determined as genuinely expressive, i.e., referred to a unique intentional object (the noumenal world as a given set of relations), or also, say, as harmoniously perceptual homomorphisms, the ideal isomorphism needs to be determined as a regulative limit of the set of such homomorphisms. It plays actually no other role, in Leibniz's system, than that of an indispensable ideal unity that makes universal harmony possible. (pp. 339-40)

"In fact, even though a circle of a foot [pedalis], one of half a foot [semipedalis], etc., all are different between them, no definition of a foot can nevertheless be given, but there must be some fixed and permanent sample; this is why usually measures are made of a long-lasting material, and it has even been suggested that the Egypt Pyramids, which have lasted for many centuries and presumably will still last for a very long time, be employed for this purpose. Thus, as long as we assume that neither the globe of the Earth nor the motions of stars are significantly to change, our descendants will be able to recognize the same quantity of the Earth's degrees as we have. And if any forms keep their sizes throughout the world and the centuries, as is the case, according to many, with the cells of a honeycomb, these may also provide a constant measure. And as long as we assume that nothing is going significantly to change in the causes of gravity and the motions of the stars, our descendants will be able to learn about our measures through a pendulum. If on the contrary, as I have already said elsewhere, God changed everything and yet preserved the proportions, every measure would fail us, nor could we ever know exactly how much things have changed, because measure cannot by any means be defined, nor can it be kept by memory: and its real preservation is necessary. From this argument, I think, the difference between size and form, or between quantity and quality, appears most clearly." (Specimen Geometriae luciferae, GM VII.276) (pp. 356-57)

"If we conceive of two points existing together, and wonder why we say that they exist together, we will think that this is so either because they are perceived together or because it is just possible to perceive them together. When we perceive an object as existent, for this reason we perceive it in space, that is, we perceive that an indefinite number of other objects absolutely indiscernible from it can exist. Or, still in other words: we perceive that such an object can move, and thus come to be either in one place or in another one; but, since it cannot exist in a plurality of places at the same time, nor can it move in just one instant, this is why we perceive this place as continuous." (Characteristica geomterica §108) (p. 412)
→ もし運動がなければ空間もない(?)

"There is, moreover, a definite order in the transition of our perceptions when we pass from one to the other through intervening ones. This order, too, we can call a path. But since it can vary in infinite ways, we must necessarily conceive of one that is most simple, in which the order of proceeding through determinate intermediate states follows from the nature of the thing itself, that is, the intermediate stages are related in the simplest way to both extremes. If this were not the case, there would be no order and no reason for distinguishing among coexisting things, since one could pass from one given thing to another by any path whatever. It is this minimal path from one thing to another whose magnitude is called distance."(Initia rerum mathematicarum metaphysica, GM VII.25) (p. 423)
→ 二点間の移行のしやすさの度合いを「距離」と呼ぶ。

Chapter IV. Metaphysics

In many places we read that the possible must possess an internal tendency to existence, without which nothing would ever exist (De rerum Originatione radicali G VII.303) (p.440)

2016年6月17日金曜日

リンド数学パピルスのある奇妙な問題

私は仕事でパースの批判版著作集、Writings of Charles S. Peirce: A Chronological Editionの第9巻(近刊)の註釈を執筆しているのだが、科学史に関する1892–1893年のローウェル連続講義の中でパースが面白い話をしているので、それを紹介したい。

古代エジプトの数学文書に「リンド数学パピルス」(Rhind Mathematical Papyrus; 以下RMP)というのがある。紀元前1650年頃にアーメスという書記官が筆写したパピルスだが、冒頭の記述によると、第12王朝のアメンエムハト2世の時代に遡る、今は失われてしまったより古いテクストを書き写したものらしい。現代のエジプト学ではアメンエムハト2世の治世は紀元前19世紀中頃とされているから、RMPのオリジナルもその頃に成立したことになる。

RMPは87問の問題とそれらに対する解答によって構成されている。その中でとりわけ異彩を放つのが第79問である。ただ「第79問」といっても問題がはっきりと述べられているわけではなく、以下のような二つの表が与えられているだけである[註釈1]

公式に従った場合の和:2801に7をかける

1         2801
2         5602
4         11204
計     19607

足していった場合の和

家        7
猫        49
鼠        343
スペルト小麦  2401
ヘカト      16807
計        19607

二つ目の表では7の倍数が5つ列挙され、その和が示されている。現代の用語で言えば公比が7の幾何級数の和の、項数が5の場合を求めているのだ。こう考えれば一つ目の表の意味も分かる。幾何級数の和は、初項をa、公比をr、項数をnとすればa(rn - 1)/(r - 1)だから、a = 7、r = 7、n = 5のとき、和は7(75 - 1)/(7 - 1) = 7(16807 - 1)/6 = 7 * 2801となる。したがって一つ目の表は、幾何級数の和の公式を使って和を計算していることが分かる。紀元前19世紀に幾何級数の和の公式が知られていたというのは驚くべき事実だが、気になるのは二つ目の表の数字の横にある「家」、「猫」、「鼠」といった言葉である。RMPを初めて出版したドイツのエジプト学者August Eisenlohrは、パピルスの筆者が意図していたのは次のような問題なのではないかと推測している:「7つの家があり、それぞれの家に7匹の猫がおり、それぞれの猫は7匹の鼠を食べ、それぞれの鼠は7穂のスペルト小麦を食べ、それぞれのスペルト麦の穂は7ヘカトの麦を産出する[ヘカトは古代エジプトの体積の単位]。ではこれらのものの合計はいくらか」[註釈2]

幾何級数の和の問題がヨーロッパで再び登場するのは、13世紀初頭のフィボナッチ(ピサのレオナルド)のLiber Abaciまで待たねばならないが、興味深いことに、フィボナッチもRMPとほぼ同じ仕方で問題を提示しているのである[註釈3]


Septem uetule uadunt romam; quarum quelibet habet burdones 7; et in quolibet burdone sunt saculi 7; et in quolibet saculo panes 7; et quilibet panis habet cultellos 7; et quilibet cultellus habet uagines 7. Queritur summa omnium predictorum. [ローマへ向かう7人の老いた女性がいる。それぞれの女性は7匹の騾馬を連れており、それぞれの騾馬は7つの大袋を運んでおり、それぞれの大袋には7斤のパンが入っており、それぞれのパンには7本の包丁が付いており、それぞれの包丁は7本の鞘の入っている。さて、女性、騾馬、大袋、パン、包丁、鞘、これらを全部足したらいくつあるか?]
 
しかも公比7まで見事にRMPと一致している(もちろん、フィボナッチがRMPの問題を知っていたとは考えられない)。さらに、イギリスの有名な伝承童謡に「セイント・アイヴズに向かっていたところ」(As I was going to St. Ives)というのがある。この童謡は次のような「なぞなぞ」の形になっている:


As I was going to St. Ives,
I met a man with seven wives,
Each wife had seven sacks,
Each sack had seven cats,
Each cat had seven kits:
Kits, cats, sacks, and wives,
How many were there going to St. Ives?
[セイント・アイヴズに向かっていたところ、七人の妻を連れた男に出会った。それぞれの妻は七つの大袋を提げており、それぞれの大袋には七匹の猫が入っており、それぞれの猫には七匹の仔猫がいた。仔猫、猫、大袋、妻、セイント・アイヴズに向かっていたのは合わせていくつ?]

ここでもRMPやLiber Abaciと全く同じような問題が提示されている。さらに言えば、イギリスの別の童謡に「これはジャックが建てた家」(This is the House that Jack Built)というのがある。その冒頭の部分は次のようになっている:


This is the house that Jack built. This is the malt that lay in the house that Jack built. This is the rat that ate the malt that lay in the house that Jack built. This is the cat that killed the rat that ate the malt that lay in the house that Jack built. [これはジャックが建てた家。 これはジャックが建てた家にある麦芽。 これはジャックが建てた家にある麦芽を食べた鼠。 これはジャックが建てた家にある麦芽を食べた鼠を殺した猫。]

ここでは、RMPの第79問にも登場した「家」、「麦」、「鼠」、「猫」が出てくる。もちろんこれだけで確定的なことは何も言えないが、さきほどのLiber Abaciの問題にしても、「セイント・アイヴズに向かっていたところ」にしても、「これはジャックが建てた家」にしても、単なる偶然で片付けるにはあまりにも奇妙な一致ではなかろうか。


パースが提案する仮説は、アメンエムハト2世の時代の童謡ないし「なぞなぞ」が、様々に形を変えながら現代まで脈々と伝わっている、というものである。実際、「これはジャックが建てた家」は、ユダヤ人が過ぎ越し祭のセドルの最後の締め括りで歌うアラム語とヘブライ語の歌、「ハド・ガドヤー」に起源を持っているという説があるから、その原型はかなり古い可能性がある。  


 
[註釈1] 以下の日本語訳は、Arnold Buffum Chaceによる1927年の英訳、The Rhind Mathematical Papyrus: The British Museum 100057 and 100058, Vol. 1 (Oberlin, OH: Mathematical Association of America, 1927), p. 112に基づいている。

[註釈2] August Eisenlohr, Ein mathematisches handbuch der alten Aegypter (Papyrus Rhind des British Museum), Vol. 1 (Leipzig: J. C. Hinrichs, 1877), pp. 202–4.

[註釈3] Scritti di Leonardo Pisano matematico del secolo decimoterz, Vol. 1 (Roma: Tipografia delle Scienze Mathematiche e Fisiche, 1857), p. 311.

2014年2月15日土曜日

確率は客観的か?Peirceの主観確率批判を手掛かりに

確率は客観的に存在するのか、それとも人間の頭の中にだけあるのか。この問題は確率の哲学的解釈における長年の論争のテーマである。ここでは、"The Probability of Induction" (1878)においてなされた、C. S. Peirceによる主観確率の批判を手掛かりにこの問題を考えてみたい。Peirceの批判が成功しているかどうかに関しては、私には判断を下すことはできないが、もし成功しているとしても(つまりもし確率の客観的解釈が正しいとしても)決定論はなお保持できるのか、それとも、Peirceとともに偶然主義tychismを支持するしかないのか、という問題についても併せて考えてみたい。

【Peirceによる主観確率批判】

Peirceの立場は頻度主義frequentismである。頻度主義によれば、ある事象の確率とはその事象の相対度数の極限値である。これは、ある事象の確率とはその事象に対する人間の信念の度合いであるとする主観確率の立場(Peirceの用語ではconceptualist view)と対立する。Peirceによれば、確率の主観的解釈は「全く道理に合わない」。なぜだろうか。

まず、確率は客観的か主観的かという問題を考える前に、確率とは「何の」確率か、という(関連する)問題を取り上げたい。確率が帰せられるのは事象についての信念なのか、事象そのものなのか、それとも命題か。この問題に対するPeirceの答えは明快である。すなわち、彼によれば確率が帰せられるのは「もし~ならば…」という形の推論である。「もし賽子を振るならば、1の目が出る」という推論の確率は6分の1である、というようにである。この見方に従えば、条件節がなければ確率概念はそもそも意味をなさない。我々はしばしば事象そのものの確率について語るが、Peirceによれば、これは条件節が自明であるために省略されているだけである。換言すれば、すべての確率は条件付き確率である。これはAndrei Kolmogorovが『確率論の基礎概念』で行った確率の公理化と一致するものであり、極めて現代的な見方と言える。

さて、Peirceによる主観主義批判はBayesの定理の使い方を巡るものである。Bayesの定理は次のように書かれる:

P(Z | Y∧X) = P(Y | Z∧X) • P(Z | X) / P(Y | X)

あるいは右辺の分母を展開して

P(Z | Y∧X) = P(Y | Z∧X) • P(Z | X) / [P(Y | Z∧X) • P(Z | X) + P(Y | ¬Z∧X) • P(¬Z | X)]

Peirceの批判はBayesの定理そのものに対する批判ではない。この定理は、確率論の公理から演繹的に導かれるものである。Peirceの批判はあくまで、この定理の帰納推論への適用、すなわち確率の逆算法Method of Inverse Probabilitiesと呼ばれる方法に向けられている。彼によれば、確率の逆算法は帰納を演繹に帰着させようとするが、そんなことは不可能である。

まず、確率の逆算法を説明する前に、確率の通常の算法を見てみよう。黒玉50個と、白玉50個の入った壺を考える。この中から10個の玉をランダムに、かつ引いた玉を壺に戻しながら引いていくとする。このとき、例えば黒玉5個と白玉5個を引く確率を、演繹的に求めることができる。Uを壺の構成を与える命題(黒玉50個、白玉50個)、Xを引き方に関する背景情報(ランダムに、かつ戻しながら)、Sを引いたサンプルを与える命題(黒玉5個、白玉5個)とすると、求める確率はP(S | U∧X)と書ける。これはBernoulli試行であるから、

P(S | U∧X) = 10C5 • (1/2)^5 • (1/2)^5 = 63/256

と求めることができる。ここで重要なのは、壺の中身があらかじめ分かっていることである。確率の逆算法は、壺の中身が分からない場合に、壺の中身を推定するための方法である。すなわち、Bayesの定理を適用することによってP(U | S∧X)を求める方法である。これはサンプルから母集団への帰納推論S∧X ⇒ Uの確率であるから、Pierre-Simon Laplace、Augustus De Morgan、Adolphe Queteletらは、この方法によって帰納の問題を解決したと信じていた。実際、Bayesの定理によって、

P(U | S∧X) = P(S | U∧X) • P(U | X) / P(S | X)

と書ける。しかし、左辺を求めるには、右辺のP(U | X)があらかじめ分かっていないといけない。このP(U | X)を「事前確率」prior probabilityと呼ぶ。しかし、事前確率を事前に確定する方法はない——P(U | X)が分かっていればそもそも壺の中身を推定する意味がない——から、確率の逆算法の使用者は「不充足理由律」Principle of Insufficient Reason、あるいは後にJ. M. Keynesによって「無差別の原理」Principle of Indifferenceと呼ばれるようになる原理に訴える。これは、相互に排他的かつすべての可能性を尽くす事象(推論)がn個あり、それぞれの事象(推論)の確率についての情報が全くないとき、それぞれの事象(推論)に1/nの確率を割り振る、という原理である。壺の中身を述べる推論の場合、X ⇒ UとX ⇒ ¬Uの二通りの場合があるから、P(U | X) = P(¬U | X) = 1/2とするわけである。

Peirceが噛み付くのはここである。確率の逆算法の使用者が求めようとしているのは、「事実(この場合は壺の構成)、帰納推論の結果適合する確率」である。つまり彼らは、壺の中身についての可能世界を立ち上げ、それぞれの可能世界が我々の帰納推論の結果に適合している確率を求めようとしているのである。Peirceによれば、求めることができるのは——そして帰納推論の場合、求めるべきは——逆に「帰納推論の結果、事実適合する確率」である。すなわち、壺の中身を述べる命題Uを事実ないし「仮説」として一旦承認した上での、サンプルがこの仮説に適合する確率P(S | U∧X)である。このP(S | U∧X)が、上で述べた確率の通常の算法におけるP(S | U∧X)と異なるのは、ここではUは既知ではなく仮説として承認されているに過ぎないという点である。この方法が、統計学で「仮説検定」hypothesis testingと呼ばれるようになる方法である。帰納推論に確率を付与することはできないが、その代わりに、ある誤差の範囲内で帰納が成功している確率である「確からしさ」verisimilitudeを付与することができる、とPeirceは言う。他方で、「事実が、帰納推論の結果に適合する確率」を求めることが可能であるのは、我々の世界だけでなくすべての可能世界を見渡し、それぞれに確率を割り振ることができる場合だけである。Peirceは次のように言う:
自然のあの配列やこの配列の相対確率について語る権利が我々にあるのは、宇宙が黒苺のように豊富に存在し、それらのある量を袋に入れ、よく混ぜてサンプルを抽出し、ある配列を持つ宇宙と別の配列を持つ宇宙のそれぞれの割合を調べることができる場合[だけ]である。[The relative probability of this or that arrangement of Nature is something which we should have a right to talk about if universes were as plenty as blackberries, if we could put a quantity of them in a bag, shake them well up, draw out a sample, and examine them to see what proportion of them had one arrangement and what proportion another.] (Collected Papers 2.684)
しかし、宇宙はもちろん一つだけである。ゆえに「自然のあの配列やこの配列」について語るのはナンセンスである。そして仮に、宇宙が複数あったとしても、それらを含むより大きな宇宙が存在するはずであるから、事情はやはり同じである。要するに、事前確率に正しい値(客観的な値)を割り振るにはすべての可能世界を見渡す必要があるが、そんなことは不可能であるから、確率の逆算法の使用者は正しい値の代わりに主観的な確率を忍び込ませている、というわけである。

【批評】

以上、Peirceによる主観確率批判を見てきたが、彼の批判は論点先取りであるように見える。客観的な値の代わりに、「無差別の原理」に基づいて主観的な値を事前確率に割り振ることがいけないのは、Peirceがすでに頻度主義の立場に立っているからであろう。しかし、主観確率がナンセンスであることを示すために彼は次のような例も挙げている。少々長いが全文引用しよう:
確率の概念的[主観的]見方では、仮説を支持する方向にも反対する方向にも傾くべきではないような完全な無知[の状況]は、1/2という確率によって表される。では、土星人の髪の色について我々が完全に無知であったとしよう。そして、あらゆる可能な色が示されたカラー表があったとしよう。この表では、すべての色が互いに知覚不可能な度合いでグラデーションをなしており、異なる色のクラスによって占められる領域の相対的な面積は全く恣意的である。こうした領域の一つを線で囲い、土星人の髪の色がこの領域内の色である確率を、概念的見方の諸原理に従って考えてみよう。我々は何らかの信念状態にあるはずだから、答えは不定であってはならない。実際、概念主義に立つ論者たちは不定な確率を認めない。確実なことは何も言えないので、答えは0と1の間である。そして、データからいかなる数値も得ることができないので、値はデータの計算ではなく確率の目盛りそのものの性質によって決めるしかない。ゆえに答えは1/2でしかありえない。なぜなら、[我々の]判断は仮説を支持するわけにも反対するわけにもいかないからである。さて、この領域について成り立つことは他のどの領域についても成り立ち、この二つの領域を含む第三の領域についても同様である。そして、二つの小領域の確率はそれぞれ1/2であるから、この第三の領域の確率は少なくても1であるはずである。しかしこれはナンセンスである。[In the conceptualistic view of probability, complete ignorance, where the judgment ought not to swerve either toward or away from the hypothesis, is represented by the probability 1/2. But let us suppose that we are totally ignorant what colored hair the inhabitants of Saturn have. Let us, then, take a color-chart in which all possible colors are shown shading into one another by imperceptible degrees. In such a chart the relative areas occupied by different classes of colors are perfectly arbitrary. Let us inclose such an area with a closed line, and ask what is the chance on conceptualistic principles that the color of the hair of the inhabitants of Saturn falls within that area? The answer cannot be indeterminate because we must be in some state of belief; and, indeed, conceptualistic writers do not admit indeterminate probabilities. As there is no certainty in the matter, the answer lies between zero and unity. As no numerical value is afforded by the data, the number must be determined by the nature of the scale of probability itself, and not by calculation from the data. The answer can, therefore, only be one-half, since the judgment should neither favor nor oppose the hypothesis. What is true of this area is true of any other one; and it will equally be true of a third area which embraces the other two. But the probability for each of the smaller areas being one-half, that for the larger should be at least unity, which is absurd.] (Collected Papers 2.679)
確かに、この結論はナンセンスである。このような結論が得られたのは、実際は確率が等しいわけではない(どころか存在すらしていない)事象に、「無差別の原理」によって等しい確率を割り振ってしまったからである。他方で、Bayes推定を繰り返すことによって、誤った確率(主観確率)を正しい確率(客観確率)に近づけることができるのもまた事実である。しかし、これもやはり主観確率が収束するべき客観確率が存在するということだろうか?

【偶然主義】

確率は客観的か主観的か、という問題に関して私は最終的な判断を保留するが、主観確率には収束する傾向が存在するということで、私はPeirceとともに客観的解釈の方にやや傾いている。しかし、確率が客観的に存在するとはどういうことだろうか?決定論が破れるということだろうか?実際、Peirceは「純粋な偶然が存在する」という偶然主義の立場を採っているが、これを避けることはできないのだろうか?

確率の「客観性」は、それが述定する推論が成功する(十分長期的な)相対頻度から来る。すなわち、確率が客観的と言えるのは、推論が実現するかどうかが、我々がコントロールできない事実によって決まるからである。もちろん、どんな確率も収束するというわけではない。先のPeirceの引用における土星人の髪の色の確率は当然収束しないが、賽子の出目の確率は一つの極限値へ収束する。

とはいえ、賽子の出目は、よりミクロのレベルで見れば決定論的な法則に従っていることを我々は知っている(ここでは量子効果は考えない)。ゆえに、賽子の出目の確率が客観的であったとしても、自然の階層性を区別することによって、決定論と両立させることができるように思われる。この見方に従えば、結局、確率は観測者の側における情報不足に帰せられることになり、そういう意味では主観的であるが、一つの極限値へ収束するという意味で、確率はやはり客観的な側面をも持つと言える。では、主観確率と客観確率ではすべての可能性が尽くされていないということだろうか?これを考えるには、「自然の階層性」という概念の中身をもっとはっきりさせる必要がある。

参考文献
  • Burch, Robert. 2010. "If Universes Were as Plenty as Blackberries: Peirce on Induction and Verisimilitude." Transactions of the Charles S. Peirce Society 46 (3): pp.423-452.
  • Peirce, Charles Sanders. 1878. "The Probability of Induction." in Collected Papers of Charles Sanders Peirce vol. 2, ed. Charles Hartshorne and Paul Weiss. Cambridge, Mass: Harvard University Press.

2014年2月2日日曜日

【メモ】 Stewart Shapiro / Thinking About Mathematics: The Philosophy of Mathematics

【論理主義】 

一つ前の記事参照。

【直観主義】

・直観主義は排中律の拒否に尽きるわけではない。直観主義の根本にあるのは「数学は心的に構成される」という哲学であって、排中律の拒否はその一つの帰結に過ぎない。その他にも例えば、被定義項を含む集合への言及を含む非可述的定義の拒否がある。直観主義の理解では、言及される集合は構成に先立って存在するわけではないから、こうした非可述的定義は循環していることになる。これは排中律の拒否とは論理的に独立。また、可能無限の立場も排中律の拒否だけからは出てこない。確かに ¬∀xPx ⇒ ¬¬∃x¬Px ⇒ ∃x¬Px という推論は排中律の拒否によってブロックされるが、これが直ちに可能無限の立場を意味するわけではない。

・まだちゃんと考えたわけではないものの、直観主義論理と古典論理のどちらが「正しい」かという問いには答えがない気がする。現象論のレベルでは古典論理が有用だけれど量子のレベルでは分配律が破れるのと同様、記述対象に応じた使い分けの問題なのかもしれない。 例えば宇宙の状態は一意的に確定できないというRovelli流のRQMには直観主義論理が適しているように見える。

・Dummetの直観主義は謎。

【プラトニズム】

"But, despite their remoteness from sense experience, we do have something like a perception of the objects of set theory, as is seen from the fact that the axioms force themselves upon us as being true. I don't see any reason why we should have any less confidence in this kind of perception, i.e., in mathematical intuition, than in sense perception, which induces us to build up physical theories and to expect that future sense perceptions will agree with them." (Kurt Gödel, "What is Cantor's Continuum Problem")

[F]or Gödel, the independence of CH from ZFC shows that "these axioms do not contain a complete description of reality." (p.210)
→ ゆえにCHの真偽を確定するためには、ZFCを超出しなければならない。不完全性定理の証明において、算術体系の命題を決定するために算術体系を超出したのと同様に。

the "notion of analyticity ... just subsides into the humbler domain where its supporting intuitions hold sway: the domain of language learning and empirical semantics." (W. V. O. Quine, "Reply to Geoffrey Hellman")

To echo Mill, we know by experience that [mathematical] intuition is reliable. (p.224)

【虚構主義】

Arithmetic and real analysis can be simulated in space-time, using certain space-time points and regions as surrogates for numbers. One can also formulate a version of the continuum hypothesis in Field's synthetic physics. (p.232)
→ しかしこれは構造が等価と言ってるだけで、CHは構造そのものだとすれば、結局Fieldは数学をやっていることになるのでは、ってツッコミながら読んでたら、後でShapiroが同じこと言ってた。

Let P be a scientific statement that makes reference to say, real numbers, and let P' be a nominalistic reconstrual of P. The hermeneutic nominalist claims that P and P' have the same meaning, and so despite appearances, P does not really make any reference to real numbers. Burgess and Rosen point out that if P and P' do have the same meaning, then one is justified in the opposite conclusion: despite appearances, P' does make reference to real numbers, since it has the same meaning as P. After all, synonymy is a symmetric relation ... (p.248)