贝叶斯认识论初探
|
摘要
贝叶斯认识论是以18世纪英国长老会牧师贝叶斯(Thomas Bayes,1702—1761)所提出的贝叶斯定理(即P(h|e)=P(e|h)P(h)/P(e))为基础,再加上简单规则(即一个人得到辩护地相信假说h,当且仅当h的概率非常高)而发展起来的一种认识论理论。广义而言,贝叶斯认识论则泛指关于概率归纳的所有主观主义理论。
传统的认识论认为能被我们接纳的知识必须是完全可靠的。反映在评判标准上,则表现为所评价的对象不是被证实就是被否证。但事实上,我们对世界的任何认识,包括知觉判断,其实都是一种假说,都存在着被否证的可能性,而被否证的假说也都存在着起死回生的可能性。关键是如何根据新的证据对已有的假说进行评价。另一方面,随着科学哲学的发展,随着概率论的发展,一种知识评价的盖然性的标准,即概率归纳逐步被哲学家们所接受。而贝叶斯认识论则为一个假说在新的证据(可能肯定也可能否定假说)的条件下对其进行评价提供了一种计算的方法和思路。这一方法作为一种决策论在各行各业得到了广泛的运用,显示了强大的生命力。但它本身也引起了许多争论。
本文试图对贝叶斯认识论的来龙去脉以及围绕它所展开的争论进行梳理,并阐明其理论和现实的意义。本文的前言部分阐明了本论文论题的研究意义和国内外研究的历史与现状以及研究方法和创新点。第一章着重介绍了贝叶斯认识论的有关理论背景。第二章着重介绍了各种概率概念,这是了解贝叶斯认识论的基础。第三章是本文的核心部分,具体介绍了贝叶斯认识论的各个方面,包括其核心内容,以及从核心内容(纯理论内容)到贝叶斯统计推断和决策理论(应用理论)的自然过渡。第四章集中讨论了对贝叶斯认识论的批评以及可能的回应。结语部分提出笔者对贝叶斯认识论总的评价和看法,并从理论和应用两方面说明其重大意义。
Thomas Bayes (1702-1761) is a British Presbyterian minister in 18th century. Bayesian epistemology is a theory of knowledge that developed from Bayesian theorem (i.e. P(h|e) = P(e|h)P(h)/P(e)) and the Simple Rule (i.e. a person is justified in believing hypothesish, if and only if the probability of his very high). In a broad sense, Bayesian epistemology makes a general reference about all of the subjectivist theories concerning probable induction.
According to traditional epistemology, our knowledge must be completely reliable. This is reflected in the standards of evaluation, according to which, what is evaluated is either confirmed or falsified. But in fact, all of our knowledge about the world, including perceptual judgments, consists of just hypotheses. Those hypotheses risk being falsified and even if they are defeated, they are still possibly to be restored. The most important thing is how to assess a hypothesis according to new evidence. On the other hand, in the wake of developments in the philosophy of science and in the probability theory, a kind of probable standard for knowledge evaluation, namely the probable induction, is gradually accepted by philosophers. Bayesian epistemology provides a computation method for assessing hypotheses when new evidence (may confirm or disconfirm a hypothesis) comes up. As a kind of decision theory, this method has obtained widespread application in various areas and has demonstrated the formidable vitality, though it has also caused a lot of controversy.
This article attempts to clarify the long and short story of Bayesian epistemology as well as the argument involved, and to expound its significance on theory and reality. The foreword part of this article has expounded researching significance, domestic and foreign researching situation, researching method and innovation spots. First chapter introduces the theory background relating Bayesian epistemology emphatically. Second chapter introduces each kind of probability concept emphatically. And this is the foundation for understanding Bayesian epistemology. Third chapter is the core part of this article. It introduces each aspect of Bayesian epistemology specifically, including its core content, as well as the natural transition from core content (pure theory content) to Bayesian statistical inference and decision theory (application theory). The fourth chapter discusses the criticism to Bayesian epistemology as well as the possible responses. The conclusion is author's general attitude to Bayesian epistemology, and it explains its great significance from the theory to application.
传统的认识论认为能被我们接纳的知识必须是完全可靠的。反映在评判标准上,则表现为所评价的对象不是被证实就是被否证。但事实上,我们对世界的任何认识,包括知觉判断,其实都是一种假说,都存在着被否证的可能性,而被否证的假说也都存在着起死回生的可能性。关键是如何根据新的证据对已有的假说进行评价。另一方面,随着科学哲学的发展,随着概率论的发展,一种知识评价的盖然性的标准,即概率归纳逐步被哲学家们所接受。而贝叶斯认识论则为一个假说在新的证据(可能肯定也可能否定假说)的条件下对其进行评价提供了一种计算的方法和思路。这一方法作为一种决策论在各行各业得到了广泛的运用,显示了强大的生命力。但它本身也引起了许多争论。
本文试图对贝叶斯认识论的来龙去脉以及围绕它所展开的争论进行梳理,并阐明其理论和现实的意义。本文的前言部分阐明了本论文论题的研究意义和国内外研究的历史与现状以及研究方法和创新点。第一章着重介绍了贝叶斯认识论的有关理论背景。第二章着重介绍了各种概率概念,这是了解贝叶斯认识论的基础。第三章是本文的核心部分,具体介绍了贝叶斯认识论的各个方面,包括其核心内容,以及从核心内容(纯理论内容)到贝叶斯统计推断和决策理论(应用理论)的自然过渡。第四章集中讨论了对贝叶斯认识论的批评以及可能的回应。结语部分提出笔者对贝叶斯认识论总的评价和看法,并从理论和应用两方面说明其重大意义。
Thomas Bayes (1702-1761) is a British Presbyterian minister in 18th century. Bayesian epistemology is a theory of knowledge that developed from Bayesian theorem (i.e. P(h|e) = P(e|h)P(h)/P(e)) and the Simple Rule (i.e. a person is justified in believing hypothesish, if and only if the probability of his very high). In a broad sense, Bayesian epistemology makes a general reference about all of the subjectivist theories concerning probable induction.
According to traditional epistemology, our knowledge must be completely reliable. This is reflected in the standards of evaluation, according to which, what is evaluated is either confirmed or falsified. But in fact, all of our knowledge about the world, including perceptual judgments, consists of just hypotheses. Those hypotheses risk being falsified and even if they are defeated, they are still possibly to be restored. The most important thing is how to assess a hypothesis according to new evidence. On the other hand, in the wake of developments in the philosophy of science and in the probability theory, a kind of probable standard for knowledge evaluation, namely the probable induction, is gradually accepted by philosophers. Bayesian epistemology provides a computation method for assessing hypotheses when new evidence (may confirm or disconfirm a hypothesis) comes up. As a kind of decision theory, this method has obtained widespread application in various areas and has demonstrated the formidable vitality, though it has also caused a lot of controversy.
This article attempts to clarify the long and short story of Bayesian epistemology as well as the argument involved, and to expound its significance on theory and reality. The foreword part of this article has expounded researching significance, domestic and foreign researching situation, researching method and innovation spots. First chapter introduces the theory background relating Bayesian epistemology emphatically. Second chapter introduces each kind of probability concept emphatically. And this is the foundation for understanding Bayesian epistemology. Third chapter is the core part of this article. It introduces each aspect of Bayesian epistemology specifically, including its core content, as well as the natural transition from core content (pure theory content) to Bayesian statistical inference and decision theory (application theory). The fourth chapter discusses the criticism to Bayesian epistemology as well as the possible responses. The conclusion is author's general attitude to Bayesian epistemology, and it explains its great significance from the theory to application.
引文
1 胡志强,作为认识论研究纲领的贝耶斯主义,哲学动态,2005年第3期
1 Fisher,R.A.,The Design of Experiment(4~(th)edition),Olive and Boyd,1947
1 江天骥,归纳逻辑导论,湖南人民出版社,1987年,第17页
2 有学者称逻辑主义为“逻辑贝叶斯主义”,笔者不太赞成。因为“贝叶斯主义”几乎是“主观贝叶斯主义”的代名词,而逻辑主义是和主观主义并列的两派。因此前面的叫法容易造成混淆。
1 彭漪涟、马钦荣,《逻辑学大辞典》,上海辞书出版社,2004年,第474页
1 尼古拉斯·布宁、余纪元,《两方哲学英汉对照辞典》,人民出版社,2001年,第871页
1 季爱民,《波普尔的概率观评析》,学术论坛,2007年第6期
2 波普尔,《走向进化论的知识论》,中国美术学院出版社,2001年,第209页
1 尼古拉斯·布宁、余纪元,《西方哲学英汉对照辞典》,人民出版社,2001年,第239页
2 约翰·波洛克和乔·克拉兹合著、陈真译,《当代知识论》,复旦大学出版社,2008年
1 Bowden,B.V.,A brief History of Computation(Faster than Thougt),Pitman Publishing,1953,p.4
1 Jeffreys,H.,Theory of Probability(3~(rd)edition),Clarendon Press,1961,p.3
1 Urbach,P.,The Scientific Standing of Evolutionary Theories of Society,The LSE Quarterly,1987b.vol.1.23-42.
2 Jeffreys,H.,Theory of Probability(3~(rd)edition),Clarendon Press,1961,p.4
1 陈小平,《归纳逻辑与归纳悖论》,武汉大学出版社,1994年,第49页
2 熊立文,《现代归纳逻辑的发展》,人民出版社,2004年,第222页
1 陈小平,《归纳逻辑与归纳悖论》,武汉大学出版社,1994年,第75页
1 Dunn,J.M.,and Hellman.G.,Dualling:A Critique of an Argument of Popper and Miller,British Journal for the Philosophy of Science,1986.Vol.37,P220-223
1 Kahneman,D.and Tversky,,A.subjective Probability:a Judgment of Representativeness,Cognitive Psychology,,1972.vol.3,P430-454
[1]江天骥,《归纳逻辑导论》,湖南人民出版社,1987年
[2]江天骥,《当代西方科学哲学》,中国社会科学出版社,1984年
[3]陈小平,《归纳逻辑与归纳悖论》,武汉大学出版社,1994年
[4][美]约翰·波洛克和乔·克拉兹合著、陈真译,《当代知识论》,复旦大学出版社,2008年
[5]张之沧,《科学哲学导论》,人民出版社,2004年
[6]邓生庆、任晓明,《归纳逻辑百年历程》,中央编译出版社,2005年
[7]熊立文,《现代归纳逻辑的发展》,人民出版社,2004年
[8]苏珊·哈克著、罗毅译,《逻辑哲学》,商务印书馆,2003年
[9]彭漪涟、马钦荣,《逻辑学大辞典》,上海辞书出版社,2004年
[10]尼古拉斯·布宁、余纪元,《西方哲学英汉对照辞典》,人民出版社,2001年
[11]波普尔,《走向进化论的知识论》,中国美术学院出版社,2001年
[12]波珀,《科学发现的逻辑》,科学出版社,1986年
[13][美]S.詹姆士·普雷斯著,《贝叶斯统计学:原理、模型及应用》,中国统计出版社,1992年
[14]梁彪,《推理与决策》,广东人民出版社,2002年
[15]波普尔,《波普尔思想自述》,上海译文出版社,1988年.
[16]王路,《逻辑基础》,人民出版社,2004
[17]卡尔纳普著、陈小山等译,《卡尔纳普思想自述》,上海译文出版社,1985年
[18]Peter Godfrey-Smith,Theory and Reality:An Introduction to the Philosophy of Science,University of Chicago Press,2003.
[19]John Earman,Bayes or Bust,The MIT Press,1992.
[20]Colin Howson and Peter Urbach,Scientific Reasoning:The Bayesian Approach,the university of Chicago Press,1993.
[21]Richard Jeffrey,Probability and the Art of Judgment,Cambridge University Press,1992.
[22]Mark Kaplan,Decision Theory as Philosophy,Cambridge University Press,1996.
[23]J.M.Joy,The Foundations of Causal Decision Theory,Cambridge University Press,1990
[24]Jeffreys,H.,Theory of Probability(3~(rd)edition),Clarendon Press,1961
[25]Fisher,R.A.,The Design of Experiment(4~(th)edition),Olive and Boyd,1947
[26]Bowden,B.V.,A brief History of Computation(in Faster than Thougt),Pitman Publishing,1953.
[1]陈晓平,《关于绿蓝悖论的争论》,自然辩证法研究,1991年第8期
[2]胡志强,《作为认识论研究纲领的贝耶斯主义》,哲学动态,2005年第3期
[3]陈晓平,《贝叶斯检验与库恩范式》,自然辩证法通讯,2005年第6期
[4]熊立文,《贝叶斯决策理论与归纳逻辑》,北京师范大学学报(社会科学版),2005年第2期
[5]季爱民,《波普尔的概率观评析》,学术论坛,2007年第6期
[6]Dunn,J.M.,and Hellman,G.,Dualling:A Critique o.fan Argument of Popper and Miller,British Journal for the Philosophy of Science,1986.Vol.37.
[7]Kahneman,D.,and Tversky,A.,subjective Probability:a Judgment of Representativeness,Cognitive Psychology,1972.Vol.
[8]Urbach,P.,The Scientific Standing of Evolutionary Theories of Society,The LSE Quarterly,1987b.vol.1.
1 Fisher,R.A.,The Design of Experiment(4~(th)edition),Olive and Boyd,1947
1 江天骥,归纳逻辑导论,湖南人民出版社,1987年,第17页
2 有学者称逻辑主义为“逻辑贝叶斯主义”,笔者不太赞成。因为“贝叶斯主义”几乎是“主观贝叶斯主义”的代名词,而逻辑主义是和主观主义并列的两派。因此前面的叫法容易造成混淆。
1 彭漪涟、马钦荣,《逻辑学大辞典》,上海辞书出版社,2004年,第474页
1 尼古拉斯·布宁、余纪元,《两方哲学英汉对照辞典》,人民出版社,2001年,第871页
1 季爱民,《波普尔的概率观评析》,学术论坛,2007年第6期
2 波普尔,《走向进化论的知识论》,中国美术学院出版社,2001年,第209页
1 尼古拉斯·布宁、余纪元,《西方哲学英汉对照辞典》,人民出版社,2001年,第239页
2 约翰·波洛克和乔·克拉兹合著、陈真译,《当代知识论》,复旦大学出版社,2008年
1 Bowden,B.V.,A brief History of Computation(Faster than Thougt),Pitman Publishing,1953,p.4
1 Jeffreys,H.,Theory of Probability(3~(rd)edition),Clarendon Press,1961,p.3
1 Urbach,P.,The Scientific Standing of Evolutionary Theories of Society,The LSE Quarterly,1987b.vol.1.23-42.
2 Jeffreys,H.,Theory of Probability(3~(rd)edition),Clarendon Press,1961,p.4
1 陈小平,《归纳逻辑与归纳悖论》,武汉大学出版社,1994年,第49页
2 熊立文,《现代归纳逻辑的发展》,人民出版社,2004年,第222页
1 陈小平,《归纳逻辑与归纳悖论》,武汉大学出版社,1994年,第75页
1 Dunn,J.M.,and Hellman.G.,Dualling:A Critique of an Argument of Popper and Miller,British Journal for the Philosophy of Science,1986.Vol.37,P220-223
1 Kahneman,D.and Tversky,,A.subjective Probability:a Judgment of Representativeness,Cognitive Psychology,,1972.vol.3,P430-454
[1]江天骥,《归纳逻辑导论》,湖南人民出版社,1987年
[2]江天骥,《当代西方科学哲学》,中国社会科学出版社,1984年
[3]陈小平,《归纳逻辑与归纳悖论》,武汉大学出版社,1994年
[4][美]约翰·波洛克和乔·克拉兹合著、陈真译,《当代知识论》,复旦大学出版社,2008年
[5]张之沧,《科学哲学导论》,人民出版社,2004年
[6]邓生庆、任晓明,《归纳逻辑百年历程》,中央编译出版社,2005年
[7]熊立文,《现代归纳逻辑的发展》,人民出版社,2004年
[8]苏珊·哈克著、罗毅译,《逻辑哲学》,商务印书馆,2003年
[9]彭漪涟、马钦荣,《逻辑学大辞典》,上海辞书出版社,2004年
[10]尼古拉斯·布宁、余纪元,《西方哲学英汉对照辞典》,人民出版社,2001年
[11]波普尔,《走向进化论的知识论》,中国美术学院出版社,2001年
[12]波珀,《科学发现的逻辑》,科学出版社,1986年
[13][美]S.詹姆士·普雷斯著,《贝叶斯统计学:原理、模型及应用》,中国统计出版社,1992年
[14]梁彪,《推理与决策》,广东人民出版社,2002年
[15]波普尔,《波普尔思想自述》,上海译文出版社,1988年.
[16]王路,《逻辑基础》,人民出版社,2004
[17]卡尔纳普著、陈小山等译,《卡尔纳普思想自述》,上海译文出版社,1985年
[18]Peter Godfrey-Smith,Theory and Reality:An Introduction to the Philosophy of Science,University of Chicago Press,2003.
[19]John Earman,Bayes or Bust,The MIT Press,1992.
[20]Colin Howson and Peter Urbach,Scientific Reasoning:The Bayesian Approach,the university of Chicago Press,1993.
[21]Richard Jeffrey,Probability and the Art of Judgment,Cambridge University Press,1992.
[22]Mark Kaplan,Decision Theory as Philosophy,Cambridge University Press,1996.
[23]J.M.Joy,The Foundations of Causal Decision Theory,Cambridge University Press,1990
[24]Jeffreys,H.,Theory of Probability(3~(rd)edition),Clarendon Press,1961
[25]Fisher,R.A.,The Design of Experiment(4~(th)edition),Olive and Boyd,1947
[26]Bowden,B.V.,A brief History of Computation(in Faster than Thougt),Pitman Publishing,1953.
[1]陈晓平,《关于绿蓝悖论的争论》,自然辩证法研究,1991年第8期
[2]胡志强,《作为认识论研究纲领的贝耶斯主义》,哲学动态,2005年第3期
[3]陈晓平,《贝叶斯检验与库恩范式》,自然辩证法通讯,2005年第6期
[4]熊立文,《贝叶斯决策理论与归纳逻辑》,北京师范大学学报(社会科学版),2005年第2期
[5]季爱民,《波普尔的概率观评析》,学术论坛,2007年第6期
[6]Dunn,J.M.,and Hellman,G.,Dualling:A Critique o.fan Argument of Popper and Miller,British Journal for the Philosophy of Science,1986.Vol.37.
[7]Kahneman,D.,and Tversky,A.,subjective Probability:a Judgment of Representativeness,Cognitive Psychology,1972.Vol.
[8]Urbach,P.,The Scientific Standing of Evolutionary Theories of Society,The LSE Quarterly,1987b.vol.1.