The properties of mechanistic constitution

当代神经生物学哲学的两个重要目标是：(a)发展一个描述充分的神经生物学解释的结构模型；(b)确定神经生物学理论构建的一般方法。这两个目标在某种意义上是联系在一起的，即神经生物学解释的逻辑结构的任何候选模型都应该与一套用于发现和扩展模型实例的规则相结合。试图满足这些需求的一个建议目前非常受欢迎，那就是神经生物学的“机械方法”（参见Machamer et al. 2000, Craver and Darden 2001, Craver 2002, 2007）。根据这一理论，神经生物学的解释基本上有因果关系和构成因素。这两个方面的建立都需要一定的方法论原则，其中因果推理比关于宪法结构的推理问题更少。关于后者，机械方法的支持者所支持的方法论建议是基于对宪法的“操纵主义”定义，根据该定义，一种机制构成另一种机制，当且仅当：(1)第一种机制在形态学上与第二种机制相关，(2)两者“相互可操纵”。最近有人认为，操纵主义者的定义面临某些概念问题（例如，Leuridan 2012）。作为另一种选择，Mark Couch和我独立提出了一种机制构成的规律性解释（Harbecke 2010, Couch 2011）。这种方法似乎在概念上是合理的，并且在某些范例神经生物学解释（目的(a)）的描述上是适当的。在这个项目中，我有三个目标。首先，(i)我想更好地理解机械构造的形式性质，（ii）将我的规律性解释与竞争性概念进行成本/收益比较。我已经研究了机制构成和监督之间的联系（cf. Harbecke 2013a），以及机制理论中的最佳系统问题（cf. Harbecke 2013b）。然而，我的叙述仍然缺乏对机械规则构成与科学形而上学和心灵形而上学中的其他经典概念（如实现、涌现、构成等）之间联系的全面分析。第三个目标（iii）涉及基于规则构成的神经生物学中构成推理的适当一般方法的发展（目标(b)）。目标将是为神经生物学解释的发展开发通用算法。由算法指导的规则将与布尔方法在因果关系的规则理论中讨论的因果推理有一定的对应关系。然而，转型将需要重要的调整。

Two important aims of the contemporary philosophy of neurobiology are (a) to develop a descriptively adequate structural model of neurobiological explanations and (b) to determine a general methodology for neurobiological theory building. The two aims are connected in the sense that any candidate model of the logical structure of neurobiological explanations should be combinable with a set of rules for the discovery and extension of the instances of the model. One proposal trying to meet these demands that currently enjoys much popularity is the ``mechanistic approach'' to neurobiology (cf. Machamer et al. 2000, Craver and Darden 2001, Craver 2002, 2007). According to this theory, neurobiological explanations essentially have a causal and a constitutional ingredient. The establishment of both aspects requires certain methodological principles, where causal reasoning is less problematic than the reasoning about constitutional structures. Concerning the latter, the methodological proposal favoured by the proponents of the mechanistic approach is based on a ``manipulationist'' definition of constitution, according to which a mechanism constitutes another one if, and only if, (1) the first is mereologically related to the second and (2) the two are ``mutually manipulable''. It has recently been argued that the manipulationist definition faces certain conceptual problems (cf., for instance, Leuridan 2012). As an alternative, Mark Couch and myself have independently proposed a regularity account of mechanistic constitution (Harbecke 2010, Couch 2011). This approach appears to be conceptually sound and descriptively adequate with respect to certain paradigm neurobiological explanations (aim (a)). In this project, my aim is threefold. First, (i) I want to understand better the formal properties of mechanistic constitution and (ii) to develop a cost/gains comparison of my regularity account with competing notions. I have already worked on the connection of mechanistic constitution and supervenience (cf. Harbecke 2013a) as well as on best-system problems within mechanistic theories (cf. Harbecke 2013b). However, my account still lacks a comprehensive analysis of the connection between mechanistic regularity constitution and other classical notions from the metaphysics of science and mind, such as realization, emergence, composition etc. The third aim (iii) concerns the development of an adequate general methodology of constitutional reasoning in neurobiology based on regularity constitution (aim (b)). The goal will be to develop general algorithms specified for the development of neurobiological explanations. The rules guided by the algorithms will bear a certain correspondence to Boolean approaches to causal inference discussed in connection to regularity theories of causation. However, the transformation will require important adjustments.