神經科學建模歷程的多元知識價值

Pluralistic Epistemic Values in Neuroscientific Modeling

近二十年來，神經科學哲學家習於用科學說明的知識價值來評價神經科學模型，並分別發展出機制說明論與非機制說明論的觀點。這兩個觀點各自闡述了何謂科學說明，以及如何使用科學說明的知識標準來評價一個神經科學模型。本文將以神經科學中的正則模型為案例研究，並論證既有的（非 ）機制說明論架構過度專注於分析模型物件本身的表徵內容，而較不重視模型的建模歷程以及建模者的知識態度，因而忽略了一些在建模過程中，不屬於機制說明與非機制說明這兩種科學說明的知識價值。本文將重構正則模型的建模歷程與相關建模者的知識態度，並藉此主張，建模者在不同的建模脈絡，可以對同一類但不同個例的正則模型，採用不同的知識價值來評價模型的知識品質。此外，本文將進一步指出，以建模歷程與建模知識態度的分析架構來處理神經科學建模實作，不僅能夠同時容納機制說明與非機制說明的觀點，也能捕捉到兩方觀點所遺漏的建模知識價值。

Philosophers of neuroscience have been employing scientific explanation as an epistemic value to evaluate neuroscientific models for the past twenty years. Consequently, they have developed mechanistic and non-mechanistic accounts of neuroscientific explanation. These two types of accounts explicate how to use a specific kind of explanatory value to evaluate the epistemic value of neuroscientific models. This paper presents a case study involving the canonical models from mathematical and computational neuroscience. This case study will show that the above mechanistic and non-mechanistic framework overly focuses on analyzing neuroscientific models as objects with representational contents. As a consequence, it pays less attention to the process of modeling and the epistemic attitudes of modelers; moreover, it can miss some important epistemic values used by modelers. By reconstructing their modeling process, I will identify the relevant modelers’ epistemic attitudes and argue that these modelers use different kinds of epistemic values to evaluate the same type of canonical models. Furthermore, among them, one epistemic value is not captured by the relevant mechanistic and non-mechanistic accounts. I develop a processual framework that centers on the modeling process and modelers’ epistemic attitudes. This process framework is better because it it accommodates both the mechanistic and non-mechanistic accounts of neuroscientific explanation regarding the canonical model in addition to capturing what both accounts leave out.