Science Quote by Marvin Minsky

Minsky frames intelligence as rummaging, not ratiocinating. In that deceptively plain description of case-based reasoning, the mind isn’t a pristine logic engine deducing from first principles; it’s a well-stocked attic of old troubles and half-remembered fixes. The verb choices matter: “stored,” “find,” “similar,” “analogous.” This is cognition as retrieval plus clever matching, a story that pushes against the heroic mythology of AI as pure mathematics and instead privileges the messy, human-looking stuff: precedent, pattern, and the sideways leap of analogy.

The intent is pragmatic, almost engineering-minded: build programs that work by leaning on what they’ve already seen. But the subtext is philosophical. If “understanding” can be approximated by mapping a new situation onto an old one, then intelligence starts to look less like an inner essence and more like an efficient reuse strategy. That’s a very Minsky move: demystify the mind by breaking it into mechanisms, then dare you to admit that the mechanisms might be enough.

Contextually, it sits inside the long tug-of-war in AI between rule-based systems (explicit logic, brittle certainty) and memory-based approaches (examples, heuristics, fallible but flexible). Case-based reasoning reads today like an early sketch of what modern machine learning operationalizes at scale: systems that perform by exploiting vast archives and measuring “similarity” in high-dimensional space. The line also quietly exposes the risk: if your database encodes the past, your future will rhyme with it. Analogy is powerful, and it’s also how errors, biases, and lazy thinking get upgraded into “solutions.”