This paper presents an incremental learning algorithm (ILA) that attempts to combine inductive
learning from examples with prior knowledge and reasoning. ILA is essentially a nearest hyperplane
learning model. It incorporates prior knowledge in the form of precepts that are hyperplanes whose
semantics are that of default rules. Finally, ILA implements a simple form of default reasoning,
combined with similarity-based reasoning, thus allowing it to perform many plausible inferences.
Results of simulation of ILA on many real-world datasets and commonsense protocols are presented
and discussed. They demonstrate promise.