Just out in Collabra: although people generally don't usually fully revise their beliefs when misinformation is corrected, we show that it's different when the information is about people. And this is true whether the information is positive or negative. /1 - ThreadSky

andyperfors.bsky.social • 385 days ago

Just out in Collabra: although people generally don't usually fully revise their beliefs when misinformation is corrected, we show that it's different when the information is about people. And this is true whether the information is positive or negative. /1

Comments

andyperfors.bsky.social•385 days ago

This work was led by amazing PhD student Amy Mickelberg, Nic Fay, and Ullrich Ecker, they just let me tag along 😊

it builds on previous work exploring the continued influence effect, or CIE: the phenomenon where people robustly fail to update their misinformed beliefs when corrected. /2

andyperfors.bsky.social•385 days ago

This is well-replicated when the topic of beliefs is general news/events, but it was a bit ambiguous in the literature whether it was also true when it came to information about people, whether positive ("John saved a life") or negative ("John had an affair"). /3

andyperfors.bsky.social•385 days ago

Over four studies we presented people with information about "John", spread over multiple trials. On trial 13 we told them with some info (either positive or negative) and on trial 19 people in the retraction condition were told that it was not true. We tracked their impression of John over time. /4

andyperfors.bsky.social•385 days ago

Here's the result from Experiment 1, but they replicate throughout (we varied the stimuli, whether the retraction was causal, the congruency to the other behaviours, etc). As you can see, people are more affected by negative evidence, but they always "recover" to baseline if it's retracted.

andyperfors.bsky.social•385 days ago

I think this is especially interesting in light of how robust the CIE is in other domains -- suggests that social cognition is different, either because of our theories about the mechanism of correction or why the info was provided, or more experience doing that reasoning, or something else.

andyperfors.bsky.social•385 days ago

Anyway, here's the abstract. The article is open access so check it out if you're interested!

/totally screwed up the numbering, but this is the end of the thread

Comments

Posting Rules

Reply