Illusory Correlation

Two things can end up feeling linked even when they genuinely aren't. Whatever example fits the wrong idea simply gets remembered more easily than the ones that don't.

This bias infers a non-existent or exaggerated association between two variables through selective attention and memory. It shows up in social cognition and judgment whenever a salient or distinctive event gets overweighted relative to its actual base rate.

A person notices that every time they forget their umbrella, it rains. They remember those dramatic, wet days vividly and conclude the two events are connected — ignoring the many dry days they also forgot their umbrella, and the rainy days they did carry it.

In a clinical judgment study, a psychologist reviews case files and notes that patients who drew unusual figures on a projective test also reported paranoid symptoms. Because both features stand out, they get remembered together far more easily than they should. A proper contingency analysis later shows the actual co-occurrence rate is near chance, but the clinician's selective memory of the striking paired cases — and neglect of the much larger group of patients with neither feature — inflates the perceived link tenfold. It mirrors a classic finding where minority-group members performing an undesirable behavior form a doubly distinctive pairing that gets overestimated, despite identical actual rates across every group.

A rare or dramatic event draws more attention and gets remembered better. That vivid memory is exactly what makes the connection to another event feel real.

Distinctive event representations get remembered together far more readily than ordinary ones do, inflating the perceived co-occurrence between them. A standout pairing picks up amplified associative strength in memory, skewing the sense of how often the two things actually go together.

Actively look for all examples, not just the striking ones. Counting how often things actually happen together checks whether the link is real.

Systematic sampling paired with a genuine base-rate comparison counteracts the salience-driven memory bias directly. Blind or randomized observation reduces how much a standout example can distort the perceived correlation.

Overgeneralization from limited samples; Ignoring base rate information; Mistaking salience for causation

An adversarial actor can deliberately stage or amplify salient, distinctive co-occurrences—such as pairing a stigmatized group with negative events in media coverage—to manufacture a perceived correlation in the audience's mind that has no statistical basis. By controlling which exemplars are made memorable and repeatedly surfacing them, propagandists can build durable false associations that resist correction even when base-rate data is presented. This technique is especially potent in disinformation campaigns where the goal is to implant stereotypic associations without exposing explicit causal claims to scrutiny.

Train observers to record and compare base rates systematically using contingency tables or co-occurrence matrices to make actual frequencies visible. Introduce structured adversarial review where reviewers explicitly ask "what cases disconfirm this association?" to counteract selective encoding by forcing retrieval of non-salient instances. Implement randomized sampling protocols and blind coding of observations to prevent distinctive exemplar nodes from receiving disproportionate weight during encoding and retrieval.