iiinova — where your digital self keeps evolving

Thanks for reading! Comments very much welcome. As always, huge thanks to my awesome coauthors and advisors @DG_Rand & @RaBhui Link to paper: osf.io/preprints/ps... 14/14

osf.io

OSF

So what? Ppl’s beliefs may not be as fallible as we think. Ppl efficiently use info & can limit the harm of “biases” to avoid falling for misinfo, propaganda, & political persuasion (see thread). Another thread from the old site: x.com/ROrchinik/st... 13/

x.com

The illusory truth effect appears to be an adaptation to high-quality info sources. With a high-quality source, the standard illusory truth effect appears. With a low-quality source, people learn to interpret repetition in different ways. 12/

We also find evidence that the illusory truth effect is stronger for implausible headlines. Rather than being a bias that prevents the processing of other info, repetition appears to form a prior (towards truth) that is integrated with what people know about the item. 11/

Second, intuitive participants (measured by CRT) show a much stronger illusory truth effect in the high-quality condition. However, they show almost minimal effects of rep in low-quality. Deliberative participants show small illusory truth in both. 10/

Are intuitions adapting? We think yes. First, response times are much faster for repeated headlines, a hallmark of processing fluency. But, the effect of rep on RTs is identical by condition. Repetition/fluency is intuitively interpreted even in the low-quality condition. 9/

But how do we adapt? In prior work, intuitions adapt to sources allowing for quick approx. Bayesian inference. Here, we argue that repetition is processed intuitively, both in the standard illusory truth effect and its reinterpretation in low-quality. Thread from old site: x.com/DG_Rand/stat... 8/

x.com

While there is a diff by condition, the avg is + in low-quality. But, when we look at ind-level effects, most ppl in low-quality show NO illusory truth. We find frequent use of a new strategy in low-quality: decrease belief in repeated items. As ppl learn, they begin to adapt. 7/

Is the effect of repetition moderated by source credibility? YES! The effect of repetition is about ¼ the size in the low-quality condition. This moderation occurs for true and false items. 6/

140

To test this prediction, ppl see 3 True & 3 False headlines repeated in 3 exposure phases + 1 judgment phase. Ppl randomized to a high-quality condition see many novel headlines that are largely true. Those in low-quality see mostly false. Feed quality -> source credibility. 5/

The model unifies 4 findings in the lit: a) baseline illusory truth, b) each additional repetition has a smaller effect on beliefs, c) repetition effects are larger for implausible items, d) novel items are believed less when repetition is common. More explanation in fig 4/

140

Our model captures the informational value of repetition: When a source repeats a piece of info, it is less likely to have been sent by mistake. When the source is credible, repetition signals this info is more likely to be true. 3/

Despite what it feels like, ppl consume mostly true info from credible sources (fig from @jennyallen.bsky.social ). Sources are usually good but sometimes err – friends lie, credible news sources retract. In a formal model, we investigate what this implies for repeated info. 2/

Reed Orchinik @rorchinik.bsky.social · 17/11/2023

New WP! The illusory truth effect (repetition -> belief) is core to psych of beliefs, & thought to be a deep bias impacting misinfo, persuasion & advertising Why would cognition include such a flaw? We argue it is a rational adaptation to high-quality info environments 🧵1/

47430

Very excited about this paper! And make sure to check out my talk tomorrow if you’re at SJDM

071

Reed Orchinik @rorchinik.bsky.social · 31/10/2023

Check out our new working paper! While Elon Musk’s embrace of the right has had horrible effects, it puts him in a place to be a credible messenger to Reps about the importance of climate change. Showing Rs his pro-climate tweets causes meaningful changes in pro-climate beliefs & intended actions.

152

Perceptions of scientists (or other experts) play a key role in the public response to science communication. By mapping out people's belief systems, we can more clearly see how the same message can be interpreted in different ways, and design even better interventions.

These patterns of belief updating are consistent with a model of hierarchical Bayesian inference. When presented with consensus, people update their beliefs about climate change and scientists in a principled manner.

100

We test 2 interventions in conjunction with consensus information. One of these interventions, which focused on the long-established history of climate science, proved particularly effective at boosting climate belief above and beyond scientific consensus.

100

Given the relationship between scientist credibility and the effectiveness of the consensus message, does it follow that boosting perceptions of the skill of scientists has a downstream impact on persuasive effect? Yes!

100

The consensus message is effective for individuals who attribute consensus to the skill of scientists. This is even true for many Republicans who were a-priori skeptical of climate change. Consensus messaging can be persuasive among skeptics because of scientist credibility!

110

However, we find that people attribute consensus to the skill or bias of scientists in similar ways. All parties think that scientists are more skilled as consensus increases. But, perceived bias also increases. These attributions are predictive of the effect on climate belief.

100

Across parties, there are marked differences in the perceived skill and bias of scientists. Democrats think that pro-consensus scientists (those who say climate change is occurring) are largely skilled and unbiased. Republicans, however, think the opposite.

100

We vary the level of scientific consensus in a large, nationally-representative sample to explore how individuals attribute consensus to climate change and the attributes of scientists. First, we find that higher consensus increases belief in climate change across parties!

100

Beliefs about climate scientists can meaningfully change the impact of consensus messaging. At the same time, consensus, in turn, shapes beliefs about scientists. This is suggestive of the occurrence of hierarchical inference, as predicted by influential theory.

111

But, consensus means different things to different people. Is there such high agreement because climate scientists are skilled experts? Or, are they biased to reach a foregone conclusion? The weight placed on each possibility can determine if one believes the consensus position.

100

One of the most common climate communications is the consensus message, like this great example from Cook and Lewandowsky (2016), where people are told that 97% of climate scientists agree that human-caused climate change is happening.

110