Traditional probability values are on a scale of 0-1. Shortliffe argues this does not support ``ruling out'' reasoning of the kind done in medicine.
EMYCIN CF's are on a scale of -1 to +1. A CF combines both a ``positive probability'' and a ``negative probability''.
| MB | 0 - 1 | Measure of Belief |
| MD | 0 - 1 | Measure of Disbelief |
| CF = MB - MD | -1 to 1 | Certainty Factor |
| -1 | Definitely False | |
| 0 | No information, or cancellation | |
| +1 | Definitely True | |
When a data parameter is True/False, ``False'' is represented as ``True'' with a CF of -1.