Electrophysiological correlates of the continued influence effect of misinformation: an exploratory study

Abstract

Misinformation often affects inferential reasoning even after it has been retracted. This is known as the continued influence effect (CIE). Previous behavioural research into the effect's underlying mechanisms has focussed on the role of long-term memory processes at the time misinformation is retrieved during inferential reasoning. We present the first investigation into the CIE using event-related potentials (ERPs). Participants completed a continuedinfluence task whilst electroencephalographic data were recorded. Analysis was guided by previous ERP research investigating post-event misinformation effects. ERPs elicited for retracted misinformation were more negative at a frontal-midline region of interest (300-500 ms) and more positive at a left-parietal region (450-600 ms) compared to correctly-accepted true information, though no differences were observed between rejected and accepted misinformation. This suggests that post-retraction reliance on misinformation may be driven by particularly strong recollection of the misinformation, ostensibly following poor integration of the retraction into the initial, partially invalid mental model.

Vaccination against misinformation: The inoculation technique reduces the continued influence effect

“…Hence, in contrast to the research mentioned, in this paper it is the belief in misinformation that predicted the inference score, not the belief in the retraction. This is in line with the ERP research results obtained by Brydges et al [ 95 ], who concluded that reliance on misinformation may be driven by the strong recollection of the misinformation following poor integration of the retraction into the mental model. Nonetheless, the mediation analysis showed that belief in retraction indirectly influences the reliance on misinformation by lessening the belief in misinformation.…”

The psychological drivers of misinformation belief and its resistance to correction

“…Neuroimaging studies have suggested that activity during retrieval, when participants answer inference questions about an encoded event -but not when the correction is encoded -is associated with continued reliance on corrected misinformation 110,111 . This preliminary neuroimaging evidence generally supports the selective-retrieval account of the CIE, although it Box 1 | Why people share misinformation online misinformation transmission involves both a receiver (the person encountering the misinformation) and a sender (the person making or sharing the misinformation).…”

Does explaining the origins of misinformation improve the effectiveness of a given correction?

“…Participant's event-related reasoning was influenced by misinformation despite being told that the misinformation originated from a lie or a mistake. Both the model-updating (Brydges et al, 2020;Gordon et al, 2017;O'Rear & Radvansky, 2020;Rich & Zaragoza, 2016), and selective retrieval (Ecker, Lewandowsky, Swire et al, 2011;Gordon et al, 2017; accounts, suggest that more detailed corrections which provide an explanation for why the misinformation is incorrect are more effective . On this view, explaining that the misinformation was initially conveyed in error or to deceive should increase the correction's salience and encourage more elaborate processing, making it more likely to be integrated during encoding, or successfully retrieved later.…”

“…The selective retrieval account holds that the CIE occurs when the misinformation is successfully retrieved but the correction is not (Ecker et al, 2017;Ecker, Lewandowsky & Apai, 2011;Ecker, Lewandowsky, Swire et al, 2011;Ecker et al, 2010;Ecker, Swire et al, 2014;Lewandowsky et al, 2012). The model-updating account alternatively argues that the CIE is driven by a failure to integrate the updated information into a mental model of the described event constructed around the misinformation, unless an alternative explanation is available (Brydges et al, 2020;Gordon et al, 2017;Kendeou et al, 2014;Wilkes & Leatherbarrow, 1988). The selective retrieval account implies the misinformation can be retrieved without the correction whereas the model-updating account entails that although the specific information can be corrected the mental model is not.…”

“…Our chief aim was to examine whether corrections that explain the misinformation as intentionally or unintentionally misleading are more effective than corrections that do not explain how the misinformation originated. Providing an explanation for why misinformation is incorrect could facilitate correction processing (Brydges et al, 2020;Gordon et al, 2017;Lewandowsky et al, 2012), or enhance later retrieval of the correction (Ecker, Lewandowsky, Swire et al, 2011;Gordon et al, 2017;. Alternatively, an explanation of deception or an oversight may be insufficient for individuals to reconsider the story and correct for inaccuracies (e.g., Green & Donahue, 2011).…”