【摘要】 不同于学界对知觉问题的长久关注，把推理作为一种独特的认知能力来研究是新近才出现的哲学热点。为什么一个人要说“某人由Г推理得出Φ”,而非“想到”或“算出”Φ呢?如果说推理是“基于Г而相信Φ”,那么，这种“基于”究竟意味着何种关联?一方面，有别于心理联想，推理不只是一种被称作前提的态度在因果上引起被称作结论的另一态度。关键但隐蔽的一点是：推理者以为前提与结论之间具有支持关系，且不论此种支持力的强弱，也不论推理者所以为的支持关系是否符合实际。该“以为”要件赋予人类推理一种能动性内核，并令其外显出“奠基关系”“区分好坏”“负责任”等直观特征。另一方面，推理也并非依规则计算而已，它是一种有目的的循规活动：凡是不受“探知求真”目标指引的活动，哪怕是在某种意义上遵循规则的，也不能算作真正的推理。立足“以为”和“有目的的循规”要件，可以自然地解释推理概念何以内在地包含某种“好”(即“推理的摩尔现象”)但又不至于因此而认为凡推理都是合理的。对此限定性的推理概念，一种元哲学辩护是：理性不同于智力，是一种能动、负责任、可控制的“思维习惯”,而推理被认为是人类理性的一个重要衡量；维系推理与人类理性之间的这种纽带，不仅是“推理”一词日常用法的要核，也是推理在公众生活和科学研究中备受推崇的根本原因。更多还原

【Abstract】 In contrast to philosophy’s long-standing focus on the problems of perception,the study of reasoning as a distinct cognitive capacity has only recently gained attention. Why would one say “someone infersΦ from Γ”rather than“someone associates Φ with Γ”or“someone calculates Φ from Γ”? If reasoning is “believing Φ on the basis of Γ ”,what kind of relation does this “on the basis of ” entail?Philosophical engagement with these questions has not only advanced our understanding of human reasoning itself but has also stimulated systematic reflection on human cognition and rationality within novel theoretical frameworks. While acknowledging substantive disagreements and responding to potential objections,this paper defends two closely related core theses about the nature of human reasoning:( 1) Unlike psychological association,human reasoning is not merely a causal process in which an attitude called the “premise”leads to another attitude called the “conclusion”. A crucial yet often overlooked point is that the reasoner presupposes a supporting relationship between premises and conclusion,regardless of the strength of this support.( 2) Human reasoning is not just rule-following calculation,but a purposeful rule-following activity: any activity not guided by the goal of “seeking truth”cannot count as genuine reasoning,even if it follows rules in some sense.First,regarding the requirement of “presupposition ”: any attitude transition properly termed“reasoning”necessitates the agent’s presupposition that some justification relationship holds between premises and conclusion. However,such second-order beliefs( about what counts as good reasoning)are typically spontaneous in most reasoners rather than reflectively formed,exhibiting two characteristic imperfections:( I) These normative beliefs may be objectively erroneous,as the distinction between“someone presupposes P”differs from “P actually is true”constitutes a basic fact of human cognitive life.( II) The reasoner may fail to properly apply their second-order belief to the current reasoning context—they may profess adherence to certain inferential norms,yet their actual reasoning process systematically deviates from these professed standards.Second,as purposive rule-following activity,all human reasoning involves normative governance directed toward epistemic ends. Yet relative to the constitutive aim of truth-seeking,not all reasoning rules are equally valid—hence the natural distinction between good reasoning and bad reasoning. This framework generates a novel implication of the resulting normative theory: evaluating the quality of a reasoning act involves neither assessing the truth-value of its conclusion nor examining whether the premises provide necessarily or probabilistically justified support for the conclusion.Third,from a metaphilosophical perspective,rationality is distinct from mere intelligence: it is an accountable and governable “habit of thought ” that epitomizes agential autonomy. Reasoning,in contrast to memory or empirical common sense,manifests a certain degree of control over our cognitive lives. It constitutes one of our autonomous decision-making modes and is thus regarded as a crucial metric of human rationality. Precisely because reasoning is a controllable and accountable activity,it can be meaningfully appraised in the literal sense as “good”or “bad”—not merely evaluated against some standard as “satisfactory”or “unsatisfactory”. This intrinsic connection between reasoning and human rationality lies at the conceptual core of the ordinary usage of “reasoning”and constitutes the fundamental reason why reasoning is held in such high regard in both human ordinary life and scientific research. The present paper ’s advocacy for a strictly defined conception of reasoning is ultimately grounded in this theoretical commitment.更多还原