This experiment, to be published in Science, used priming (cues like hard-to-read fonts, showing participants the sculpture The Thinker) and problem-solving tasks to induce "analytical thinking" in the participants, and found that it seemed to reduce their degree of religious belief. Participants not given such tasks showed no such reduction.

Their methods of quantifying "religious belief" aren't given in detail (a questionnaire, probably), so it may be interesting to see the actual article when it comes out.