The original paper can be found here. It’s very readable and worth going through in full.

These notes gloss over the proof of each of these points to a great degree. In case of disbelief, you should check the study for greater detail.

  • Previous studies have shown that reasoners can draw upon information in memory even as they acknowledge that it is false.
  • Failures to edit inferences in memory:
  • People were told of a fire in a warehouse and that cans of oil paint and gas cylinders were in the warehouse. They were later told that no paint or cylinders were present.
  • Those subjected to the message and the correction were much more likely to mention paint and gas as possible reasons for the fire than the control group who never saw the initial misinformation.
  • Those who saw the correction were as likely to mention the paint and gas as those who saw the false message, but not the correction.
  • There was a hypothesis that people updated the information, but not the inferences made using that information. This was tested by varying whether the message stating that the previous message was false was shown immediately or after some time had elapsed. It was found that this resulted in no improvement in the use of misinformation, and so it is not likely that this hypothesis holds water.
  • Note of course, that this is exactly the kind of thing that the article talks about, where it tells you an incorrect explanation that is immediately shown to be false. The full paper goes into more detail as to why it is false though.
  • The finding that people make new inferences using facts know to be false was replicated in a different study. </li>
  • Processing Contradictions:
  • The questionnaire made it clear that people understood the falseness of the information that they were making inferences from. </li>
  • Access to Misinformation:
  • People were exposed to the ideas of paint and gas incidentally, and it made no difference.
  • People were asked to come up with a bunch of alternative reasons for the fire after completing the account. This did not result in a meaningful difference.
  • It appears as though the effect is not just due to the misinformation or the lack of a cause.
  • It is possible at this point that it is due to the causal role of the misinformation in the narrative. </li>
  • Replacing Misinformation:
  • A theory at this point is that people fall back on misinformation because no other alternative presents itself.
  • They tried having the correction show evidence for arson and this resulted in a meaningful difference.
  • The existence of an alternate cause reduced the reliance on misinformation.
  • The plausibility of this alternative explanation was tested and found to have no impact on whether people used the misinformation. That is, people were as likely to accept an implausible alternative explanation as a plausible one. </li>
  • The Pragmatics of Correcting Misinformation
  • They then tried changing the form of the correction to be more emphatic and to contain an explanation for the misinformation.
  • The results were undisputed reference > correction by negation > correction explaining the reason for the mistake > no misinformation (with the comparison being the number of people using the misinformation.)
  • Explaining why the misinformation was stated is better than a simple negation, but not as good as an alternate explanation for reducing people’s reliance on misinformation. </li>
  • Successful correction of misinformation must address not only the informational content, but also the reader’s experience of contradictory information.