Toward an experimental account of argumentation: the case of the slippery slope and the ad hominem arguments