# Catalogue of Artificial Intelligence Techniques

## Non-monotonic Reasoning

**Keywords:**
PLANNER, closed world assumption, non-monotonic logics

### Categories: Inference and Reasoning , Knowledge Representation

Author(s): **Luis Jenkins , Dave Plummer**

Non-monotonicity is a common feature of ordinary reasoning. For
instance, if we are told that Tweety is a bird we assume Tweety can
fly, but withdraw this when we are told that Tweety is a penguin.
Non-monotonic reasoning can be modelled using non-monotonic
logics. Such logics differ from normal
deductive logics in the effects of adding new axioms. In any type of
logic, new axioms may give rise to new theorems, so that the set of
theorems grows with the set of axioms. In non-monotonic logics,
however, the set of theorems may *lose* as well as gain
members when new axioms are added.
Default Reasoning is a way of overcoming the
problem of insufficient information; the system is told that unless it
has information to the contrary, certain defaults are assumed to be
true. Examples of systems utilising default reasoning are
PLANNER and the
Negation as Failure of
Prolog. In the case of Prolog,
any proposition which cannot be shown to be true from the statements
in database is assumed to be false: this is the so-called
closed world assumption.
See also Truth Maintenance System,
Default Logic,
and Circumscription.

### References:

- Ginsberg, M.L., ed.,
*Readings in Non-monotonic Reasoning*, Morgan Kaufmann, Los Altos, California, 1987. - Ginsberg, M.L., ed.,
*Readings in Non-monotonic Reasoning*, Morgan Kaufmann, Los Altos, California, 1987.

### Comments:

No comments.