How we work.
In one study, for instance, liberal and conservative subjects were asked whether government health care should be extended to a hypothetical group of AIDS victims who were responsible for their own fates (they'd contracted the disease while knowing the risks, and having unprotected sex anyway). Liberals who were not under load not distracted wanted to help such people, despite the fact that they were personally responsible for their plight. But liberals under load were much more like conservatives, appearing to reason that this group of AIDS victims had gotten what they deserved. (Cognitive load did not appear to change the view of conservatives in the study.)
Drunkenness. Alcohol intoxication is not unlike cognitive load, in that it cuts down the capacity for in-depth, nuanced thinking, and privileges economical, quick responses. Sure enough, in a recent study of 85 bar patrons, blood alcohol content was related to increased political conservatism for liberals and conservatives alike.
The drinkers still knew whether they were liberal or conservative, of course. But when asked how much they agreed with a variety of statements of political principles like, "Production and trade should be free of government interference: higher blood alcohol content was associated with giving more conservative answers.