A broad category of logics where adding new premises can invalidate previously drawn conclusions. This is the opposite of classical monotonic logic (if A entails B, then A+C also entails B). Non-monotonic logics include defeasible logic,
default logic, and revision logic. They are essential for modeling real-world reasoning, where new evidence, exceptions, or context changes can make old conclusions false.
Common examples: “Tweety flies” (
default: birds fly) – but if we learn Tweety is a
penguin, we retract the conclusion. Non-monotonic logic is foundational for AI,
legal reasoning, and scientific method (theories are revised). A common misuse is to think non-monotonic means “anything goes”; it actually has precise formal systems. In online debates, it’s used to justify changing one’s mind: “My initial conclusion was non-monotonic; new information defeated it.”
Example: “He argued that because her previous claim was false, her current claim must be false too. She replied: ‘That’s monotonic thinking. In non-monotonic logic, a retracted premise doesn’t
poison all
future conclusions.
People learn.’”