Sebastian Scholz
Heinrich-Heine-University
Goodman’s New Riddle of Induction is that we cannot easily distinguish “good” (viz. projectible) predicates that are allowed for inferences from peculiar ones that are prohibited – such as the infamous Goodman-predicate ‘grue’. This is at least one common way of framing the problem. But the larger epistemological concern at stake is this: What is the justification of our inferential practice?
Peter Gärdenfors claims to have solved the riddle, namely with his cognitive naturalness criterion of convexity in conceptual spaces. His approach is inspired by Quine and is built on the key innovation of representing similarity as spatial distance. I will argue, however, convexity can explain, but not properly justify inferential reasoning. My argument takes the form of a dilemma: If we take Gärdenfors’ instrumentalism and cognitivism at face value, then epistemic relativism follows suit – a consequence he tries to avoid. But if we interpret in terms of scientific realism, then the convexity criterion is sidelined, as the source of the justification becomes independent of cognition. Even the Quine-inspired pragmatism Gärdenfors champions is not a way out of the dilemma, but collapses into the first horn. This is because, as I will argue, evolution does not function as a bridgehead between cognitive and ontological naturalness, as the account requires. There is a significant difference between evolutionary utility on the one hand and the fundamental building blocks of reality on the other. Consequently, the approach can at best offer an instrumentalist justification with which even David Hume would have agreed.
Chair: Cristina Somcutean
Time: September 6th, 15:20-15:50
Location: SR 1.007
SOPhiA 2026 