We are poor lie detectors, with accuracy only marginally better than guessing. Raters in social situations such as these must battle with their own uncertainty if they are to make a judgment, yet lie detection research has given little attention to the effects of uncertainty. We present evidence suggesting prior knowledge and expectations have an early influence on the judgment process, but only when forced into judgment. If able to abstain, raters do not rely on prior knowledge but rather indicate their uncertainty by withholding judgment until all the available information has been presented. After the speaker has presented their statement, and when no additional new information is available, we provide evidence from a number of experiments that lie detectors integrate their pre-existing biases and experience about deception in general with more specific information about the statement at hand.