諾頓（Norton）提倡歸納法的實質理論（the material theory of induction），以特定事實而非普遍法則來證成歸納推論。他進一步論證，這個方案可以解消著名的歸納法難題。1不過，部分學者指出，即便採取實質理論，歸納推論的證成依舊會陷入無窮後退，諾頓的方案因而並不令人滿意。對此，諾頓提出「非階層式經驗主義」來回應無窮後退的指控，並以「科學假設的鷹架觀」來支持這個證成結構。2本文首先指出，基於科學假設的鷹架觀無法保證一個獨一無二的科學體系，而且諾頓對鷹架觀的進一步辯護也有問題，因此非階層式經驗主義並沒有得到很好的支持。然後，筆者將從因果難題與歸納法難題的關連性入手，以「因果關係的鷹架觀」與「因果獨立性」來為實質理論提供另一種可能的合理性基礎。

Norton proposes a brand new approach called the material theory of induction (MTI) to justify inductive inferences. According to MTI, particular inductive inferences are justified by domain-specific facts rather than universal rules. Norton claims that the problem of induction (Hume’s Problem) can be escaped if we adopt MTI. However, some argue that MTI is untenable for the reason that it still encounters the infinite regress of justification. Norton uses the “non-hierarchical empiricism” to respond to this challenge, and further uses the “scaffolding analogy of scientific hypotheses” to support this response. However, Norton’s response is not satisfactory due to the “scaffolding analogy of scientific hypotheses” is not the unique explanation of the scientific history. In this paper, I first argue that Norton’s further justification for his scaffolding analogy is problematic. And then I am going to justify Norton’s theory in virtue of “the scaffolding analogy of causation” and “causal dependence”.