Effects of associative inference on individuals’ susceptibility to misinformation.

Abstract

Associative inference is an adaptive process of memory that allows people to recombine associated information and make novel inferences. We report two online human-subject experiments investigating an associative inference version in which participants viewed overlapping real-news pairs (AB&BC) that could later be linked to support inferences of misinformation (AC). In each experiment, we examined participants' recognition and perceived accuracy of snippets of news articles presented as tweets across two phases. At Phase 1, only real-news tweets were presented, which were associated with political news of Phase 2 at three levels: real, fake, and fake with inference. In Experiment 2, participants' cognitive ability was also assessed. Participants recognized more but gave lower accuracy ratings for the fake news with inference than the fake news in both experiments. The effect of associative inference was more evident in the perceived accuracy ratings for participants of higher cognitive ability than those of lower cognitive ability. We conclude that associative inference can make people become susceptible to misinformation. We also discuss the results in terms of why associative inference made people susceptible to misinformation in the relatively automatic familiarity judgment (i.e., recognition) but not the relatively controlled and effortful semantic judgment (i.e., accuracy rating). Public Significance StatementThe present study shows that associative inference, an adaptive cognitive mechanism, can make participants susceptible to misinformation in the recognition judgment. However, participants, especially those of higher cognitive ability, gave lower accuracy ratings for fake news with associative inference than that without associative inference, indicating less susceptibility to the semantic judgment.

Associative Inference Can Increase People’s Susceptibility to Misinformation

“…Using the same two-phase procedure, Xiong et al (2023) conducted two online human-subjects experiments examining the effect of associative inference with a within-subjects design. Critically, each participant viewed three types of news articles (i.e., real, fake, and fake with associative inference) at Phase 2.…”

“…Compared to the CRT, substantially fewer MTurk workers have been previously exposed to the CRT-2 questions (Thomson and Oppenheimer 2016). Lee et al (2020) and Xiong et al (2023) evaluated participants' cognitive ability with the CRT-2 and the Wordsum. Both studies revealed the impact of participants' cognitive ability on the perceived accuracy ratings, but in the opposite direction.…”

“…They found that participants who saw political posts with hashtags were less motivated to know more about related issues compared to participants who viewed the same posts without hashtags. On the contrary, Xiong et al (2023) Figure 2: An overview of flow chart for all experiments. "Phase 1" and "Phase 2" boxes show different types of news at each phase.…”

“…Few studies have been conducted on people's susceptibility to associatively inferred fake news (Lee et al 2020;Xiong et al 2023). While Xiong et al found that associative inference could increase participants' susceptibility to misinformation in the familiarity judgment (i.e., recognition), the existing studies do not provide conclusive results on participants' misbelief evaluation.…”

Memory sins in applied settings: What kind of progress?

“…A study from Xiong et al (2022) suggests that the process of associative inference impacts the recognition of fake news items. Associative inference is an adaptive process that allows people to combine information from distinct episodes (A-B, B-C) to make novel connections that have not been directly experienced (A-C; e.g., Zeithamova & Preston, 2010).…”