In expert systems or knowledge bases having an inference engine, backward inference is inference that occurs at the time a question is posed to the system. For example, if a system had a rule stating that every raven is black, and it knew that Joe is a raven, but it waited until it was asked "What color is Joe?" before concluding that he is black in color, that would be backward inference. A terse way of explaining backward inference is that it is inference that happens at query time.

Contrast backward inference with forward inference, which is inference that happens at assert time.