Belief can be viewed as unsupported fact, i.e. data that are believed to be true (or false), in contrast to knowledge, which is often considered to be known to be true (or false). Partial knowledge thus indicates missing or incomplete data, but partial belief implies a degree of doubt about some given data. Methods for representing degrees of belief have been developed in probabilistic reasoning. Other techniques include reason-management systems, which record dependencies between data and thus act as justifications that support beliefs (see truth-maintenance systems).

In distributed artificial intelligence, the local collections of belief of individual agents in a multiagent environment are called belief spaces.