Abstract

Cognitive scientists use computational models to represent the results of their experimental work and to guide further research. Neither of these claims is particularly controversial, but the philosophical and evidentiary statuses of these models are hotly debated. To clarify the issues, I return to Newell and Simon’s 1972 exposition on the computational approach; they herald its ability to describe mental operations despite that the neuroscience of the time could not. Using work on visual imagery (cf. imagination) as a guide, I examine the extent to which this holds true today. Does contemporary neuroscience contain mechanisms capable of describing experimental results in imagery? I argue that it does not, first by exploring foundational achievements in imagery research then by showing that their neural basis cannot be specified. Newell and Simon’s methodological position accordingly stands, even 50 years later. Computational — as opposed to physiological — descriptions must be retained to characterize and study mental phenomena, even as we learn high-level details of their implementation via brain data.

中文翻译：

---

符号和基底：认知科学计算的方法论

摘要

认知科学家使用计算模型来表示他们的实验工作的结果并指导进一步的研究。这些说法都没有特别有争议，但这些模型的哲学和证据地位引起了激烈的争论。为了澄清这些问题，我回到 Newell 和 Simon 1972 年关于计算方法的阐述；他们预示着它有能力描述心理运作，尽管当时的神经科学无法做到这一点。以视觉意象（参见想象力）方面的研究为指导，我研究了这一点在今天的适用程度。当代神经科学是否包含能够以图像方式描述实验结果的机制？我认为事实并非如此，首先探索图像研究的基础成就，然后表明它们的神经基础无法被指定。因此，即使在 50 年后，纽厄尔和西蒙的方法论立场仍然成立。即使我们通过大脑数据了解其实现的高级细节，也必须保留计算（而不是生理）描述来表征和研究心理现象。