Memory and the Moses illusion: Failures to detect contradictions with stored knowledge yield negative memorial consequences

An initial accuracy focus prevents illusory truth

News stories, advertising campaigns, and political propaganda often repeat misleading claims, increasing their persuasive power. Repeated statements feel easier to process, and thus truer, than new ones. Surprisingly, this illusory truth effect occurs even when claims contradict young adults' stored knowledge (e.g., repeating The fastest land animal is the leopard makes it more believable). In four experiments, we tackled this problem by prompting people to behave like "fact checkers." Focusing on accuracy at exposure (giving initial truth ratings) wiped out the illusion later, but only when participants held relevant knowledge. This selective benefit persisted over a delay. Our findings inform theories of how people evaluate truth and suggest practical strategies for coping in a "post-truth world."

Neural basis of goal-driven changes in knowledge activation

Depending on a person's goals, different aspects of stored knowledge are accessed. Decades of behavioral work document the flexible use of knowledge, but little neuroimaging work speaks to these questions. We used representational similarity analysis to investigate whether the relationship between brain activity and semantic structure of statements varied in two tasks hypothesized to differ in the degree to which knowledge is accessed: judging truth (semantic task) and judging oldness (episodic task). During truth judgments, but not old/new recognition judgments, a left-lateralized network previously associated with semantic memory exhibited correlations with semantic structure. At a neural level, people activate knowledge representations in different ways when focused on different goals. The present results demonstrate the potential of multivariate approaches in characterizing knowledge storage and retrieval, as well as the ways that it shapes our understanding and long-term memory.

Beyond hypercorrection: remembering corrective feedback for low-confidence errors

Correcting errors based on corrective feedback is essential to successful learning. Previous studies have found that corrections to high-confidence errors are better remembered than low-confidence errors (the hypercorrection effect). The aim of this study was to investigate whether corrections to low-confidence errors can also be successfully retained in some cases. Participants completed an initial multiple-choice test consisting of control, trick and easy general-knowledge questions, rated their confidence after answering each question, and then received immediate corrective feedback. After a short delay, they were given a cued-recall test consisting of the same questions. In two experiments, we found high-confidence errors to control questions were better corrected on the second test compared to low-confidence errors - the typical hypercorrection effect. However, low-confidence errors to trick questions were just as likely to be corrected as high-confidence errors. Most surprisingly, we found that memory for the feedback and original responses, not confidence or surprise, were significant predictors of error correction. We conclude that for some types of material, there is an effortful process of elaboration and problem solving prior to making low-confidence errors that facilitates memory of corrective feedback.

Competing cues: Older adults rely on knowledge in the face of fluency

Consumers regularly encounter repeated false claims in political and marketing campaigns, but very little empirical work addresses their impact among older adults. Repeated statements feel easier to process, and thus more truthful, than new ones (i.e., illusory truth). When judging truth, older adults' accumulated general knowledge may offset this perception of fluency. In two experiments, participants read statements that contradicted information stored in memory; a post-experimental knowledge check confirmed what individual participants knew. Unlike young adults, older adults exhibited illusory truth only when they lacked knowledge about claims. This interaction between knowledge and fluency extends dual-process theories of aging. (PsycINFO Database Record

Expertise effects in the Moses illusion: detecting contradictions with stored knowledge

People frequently miss contradictions with stored knowledge; for example, readers often fail to notice any problem with a reference to the Atlantic as the largest ocean. Critically, such effects occur even though participants later demonstrate knowing the Pacific is the largest ocean (the Moses Illusion) [Erickson, T. D., & Mattson, M. E. (1981). From words to meaning: A semantic illusion. Journal of Verbal Learning & Verbal Behavior, 20, 540-551]. We investigated whether such oversights disappear when erroneous references contradict information in one's expert domain, material which likely has been encountered many times and is particularly well-known. Biology and history graduate students monitored for errors while answering biology and history questions containing erroneous presuppositions ("In what US state were the forty-niners searching for oil?"). Expertise helped: participants were less susceptible to the illusion and less likely to later reproduce errors in their expert domain. However, expertise did not eliminate the illusion, even when errors were bolded and underlined, meaning that it was unlikely that people simply skipped over errors. The results support claims that people often use heuristics to judge truth, as opposed to directly retrieving information from memory, likely because such heuristics are adaptive and often lead to the correct answer. Even experts sometimes use such shortcuts, suggesting that overlearned and accessible knowledge does not guarantee retrieval of that information.

Does Anyone Know the Answer to that Question? Individual Differences in Judging Answerability

Occasionally people may attempt to judge whether a question can be answered today, or if not, if it can be answered in the future. For example, a person may consider whether enough is known about the dangers of living close to a nuclear plant, or to a major electricity cable, for them to be willing to do so, and state-authorities may consider whether questions about the dangers of new technologies have been answered, or in a reasonable future can be, for them to be willing to invest money in research aiming develop such technologies. A total of 476 participants, for each of 22 knowledge questions, either judged whether it was answerable today (current answerability), or judged when it could be answered (future answerability). The knowledge questions varied with respect to the expected consensus concerning their answerability: consensus questions (high expected consensus), non-consensus questions (lower expected consensus), and illusion questions (formulated to appear answerable, but with crucial information absent). The questions’ judged answerability level on the two scales was highly correlated. For both scales, consensus questions were rated more answerable than the non-consensus questions, with illusion questions falling in-between. The result for the illusion questions indicates that a feeling of answerability can be created even when it is unlikely that somebody can come up with an answer. The results also showed that individual difference variables influenced the answerability judgments. Higher levels of belief in certainty of knowledge, mankind’s knowledge, and mankind’s efficacy were related to judging the non-consensus questions as more answerable. Participants rating the illusion questions as answerable rated the other answerability questions as more, or equally, answerable compared to the other participants and showed tendencies to prefer a combination of more epistemic default processing and less intellectual processing.

Recent study, but not retrieval, of knowledge protects against learning errors

Surprisingly, people incorporate errors into their knowledge bases even when they have the correct knowledge stored in memory (e.g., Fazio, Barber, Rajaram, Ornstein, & Marsh, 2013). We examined whether heightening the accessibility of correct knowledge would protect people from later reproducing misleading information that they encountered in fictional stories. In Experiment 1, participants studied a series of target general knowledge questions and their correct answers either a few minutes (high accessibility of knowledge) or 1 week (low accessibility of knowledge) before exposure to misleading story references. In Experiments 2a and 2b, participants instead retrieved the answers to the target general knowledge questions either a few minutes or 1 week before the rest of the experiment. Reading the relevant knowledge directly before the story-reading phase protected against reproduction of the misleading story answers on a later general knowledge test, but retrieving that same correct information did not. Retrieving stored knowledge from memory might actually enhance the encoding of relevant misinformation.

Memory for flip-flopping: detection and recollection of political contradictions

During political campaigns, candidates often change their positions on controversial issues. Does changing positions create confusion and impair memory for a politician's current position? In 3 experiments, two political candidates held positions on controversial issues in two debates. Across the debates, their positions were repeated, changed, or held only in the second debate (control). Relative to the control condition, recall of the most recent position on issues was enhanced when change was detected and recollected, whereas recall was impaired when change was not recollected. Furthermore, examining the errors revealed that subjects were more likely to intrude a Debate 1 response than to recall a blend of the two positions, and that recollecting change decreased Debate 1 intrusions. We argue that detecting change produces a recursive representation that embeds the original position in memory along with the more recent position. Recollecting change then enhances memory for the politician's positions and their order of occurrence by accessing the recursive trace.

A process-dissociation analysis of semantic illusions

We examine semantic illusions from a dual-process perspective according to which the processes that go into failing or succeeding to detect such illusions can be decomposed into controlled processes (checking the facts in the sentence against the information in memory) and automatic processes (the impression of truth that comes from the semantic associations between the elements in the sentence). These processes, we argue, make largely independent contributions to truth judgments about semantic-illusory sentences. The Process Dissociation Procedure was used to obtain estimates of these two kinds of processes. In Study 1, participants judged whether sentences were true or false while under high or low cognitive load. Cognitive load increased the rate of semantic illusions by specifically affecting controlled processing but not automatic processing. In Study 2, a previous paired-associate learning task also increased the rate of semantic illusions, but it did so by specifically affecting automatic processing, not controlled processing.

The contribution of fronto-parietal regions to sentence comprehension: insights from the Moses illusion

To interpret a sentence, the reader must not only process the linguistic input, but many times has also to draw inferences about what is implicitly stated. In some cases, the generation and integration of inferred information may lead to semantic illusions. In these sentences, subjects fail to detect errors such as in "It was two animals of each kind that Moses took on the ark" despite knowing that the correct answer is Noah, not Moses. The relative inability to notice these errors raises questions about how people establish and integrate inferences and which conditions improve error detection. To unravel the neural processes underlying inference and error detection in language comprehension, we carried out an fMRI study in which participants read sentences containing true or false statements. The false statements either took the form of more obvious (i.e., clearly false) or subtle (i.e., semantic illusions) inconsistent relations. Participants had to decide if each statement was true or false. Processing semantic illusions relative to true and clearly false sentences significantly engaged the right inferior parietal lobule, suggesting higher demands in establishing coherence. Successful versus unsuccessful error detection revealed a network of regions, including right dorsolateral prefrontal cortex, orbitofrontal, insula/putamen and anterior cingulate cortex. Such activation was significantly correlated with overall response accuracy to the illusions. These results suggest that to detect the semantic conflict, people must inhibit the tendency to draw pragmatic inferences. These findings demonstrate that fronto-parietal areas are involved in inference and inhibition processes necessary for establishing semantic coherence.

Ironic effects of drawing attention to story errors

Readers learn errors embedded in fictional stories and use them to answer later general knowledge questions (Marsh, Meade, & Roediger, 2003). Suggestibility is robust and occurs even when story errors contradict well-known facts. The current study evaluated whether suggestibility is linked to participants' inability to judge story content as correct versus incorrect. Specifically, participants read stories containing correct and misleading information about the world; some information was familiar (making error discovery possible), while some was more obscure. To improve participants' monitoring ability, we highlighted (in red font) a subset of story phrases requiring evaluation; readers no longer needed to find factual information. Rather, they simply needed to evaluate its correctness. Readers were more likely to answer questions with story errors if they were highlighted in red font, even if they contradicted well-known facts. Although highlighting to-be-evaluated information freed cognitive resources for monitoring, an ironic effect occurred: Drawing attention to specific errors increased rather than decreased later suggestibility. Failure to monitor for errors, not failure to identify the information requiring evaluation, leads to suggestibility.