Standard logical systems model logically omniscient agents that can recognise every consequence of their assumptions. LUCI investigates how to approximate ideal logics so they remain normatively meaningful while still modelling the limits of realistic agents.
The problem
A resource-bounded agent cannot be expected to perform all valid inferences of a logic L. A better model is a hierarchy of approximating systems that converges to L while respecting minimal rationality constraints.
The idea
LUCI studies hierarchies of approximations that treat the depth of virtual information and nested case analysis as a plausible measure of deductive difficulty.