摘 要 「類疇投射推論」這認知機制的運作可以帶給認知主體新的知識，亦 即關於某個物體尚未觀察到的表面特徵的知識。本文探討這個認知機制可靠 性的問題，主要檢討心理本質論以及最小假設兩個學說。心理本質論預設認 知主體擁有「類疇具有本質」這件心理表徵，以及「物體具有的本質在因果 上造成了物體具有的表面特徵」這件心理表徵。另一方面，最小假設否認第 一件預設，認為個體具有的第二件表徵就足以解釋「類疇投射推論」的可靠 性。這兩套進路都主張相似進路是不足的。 本論文首先論證，心理本質論與最小假設這兩套進路面臨重大的理論困 難，其次論證相似進路能夠解釋類疇投射推論的可靠性，最後則論證相似進 路比起前兩套進路在解釋上更為精簡。本論文據此結論，對於類疇投射推論 可靠性的解釋，相似進路是最為可取的。

Abstract Category-related projection brings new knowledge to cognitive subjects about some as yet unobserved surface feature of an object. This essay investigates the reliability of such a cognitive mechanism, focusing on both psychological essentialism and the minimal hypothesis. Psychological essentialism explains the reliability of this cognitive mechanism by positing the representation that kinds have essences and the representation that such essences causally underlie the surface feature, whereas the minimal hypothesis explains it by positing only the latter causal belief. Both approaches think that the resemblance approach that explains the reliability of projection in terms of similarity is never adequate. In this essay, I argue first that there are serious difficulties with both psychological essentialism and minimalism, second that similarity-based projection is a reliable mechanism, and third that the resemblance approach is explanatorily parsimonious. I thus conclude that the resemblance approach is to be favored to the other two approaches in explaining the reliability of category-related projection.