Falsely remembering something by accidentally combining features from separate real memories into one.
Imagine you played with a red truck and a blue car yesterday. Today, someone asks if you played with a red car. You might say yes, because you really did see something red and you really did see a car — your brain just mixed up which color went with which toy.
Memory conjunction errors occur when the brain stores individual features of experienced stimuli somewhat independently, and upon retrieval, incorrectly reassembles features from separate experiences into a composite that was never actually encountered. For example, after seeing the words 'blackmail' and 'jailbird,' a person may later falsely recognize 'blackbird' as having been seen, because both constituent syllables are genuinely familiar. This type of error is particularly insidious because the component elements are authentic — only their combination is false — which makes these fabricated memories feel highly convincing and are often endorsed with strong confidence. The phenomenon extends beyond words to faces, sentences, and even autobiographical events, revealing that episodic memory is fundamentally constructive rather than reproductive.
The same glitch looks different depending on the terrain. Finance, medicine, a relationship, a team — same mechanism, different costume.
Investors may misattribute performance characteristics from one fund to another, recombining a strong return figure from one quarter with a different fund's risk profile, leading to distorted portfolio assessments based on composite memories that blend separate data points.
Clinicians who see many patients in sequence may inadvertently transfer a symptom reported by one patient into their memory of another patient's case, creating a hybrid clinical picture that leads to incorrect diagnoses or treatment plans.
Students studying multiple topics in rapid succession may blend facts across subjects — attributing a date from one historical event to another, or mixing up which theory belongs to which theorist — producing confident but incorrect answers on exams.
People in long-term relationships may merge details from different arguments or conversations, insisting their partner said something that actually combines statements from separate occasions, fueling conflicts over what was 'really said.'
Users testing multiple product prototypes may give feedback that inadvertently blends features from different versions, reporting that they liked a specific combination of interface elements that never actually appeared together in any single prototype.
During performance reviews, managers may construct a composite memory of an employee's behavior that merges positive actions from one quarter with negative incidents from another, creating a blended narrative that doesn't accurately represent any specific time period.
Consumers of news may merge details from separate stories — attributing a quote from one politician to another, or combining the location of one event with the casualties of a different one — creating confidently held but factually hybrid accounts of current events.
Mark Tippens Reinitz, William J. Lammers, and Barbara Pitts Cochran, 1992 — formalized the memory conjunction error paradigm in their paper 'Memory-conjunction errors: Miscombination of stored stimulus features can produce illusions of memory' published in Memory & Cognition.
A constructive, flexible memory system that stores and recombines features was adaptive because it enabled generalization from past experiences, creative problem-solving, and future simulation. The ability to extract patterns and recombine elements helped ancestors anticipate novel threats and opportunities rather than being limited to exact replays of prior experience.
Retrieval-augmented generation (RAG) systems and large language models can produce conjunction-like errors by blending factual details from separate training documents or retrieved passages, generating plausible-sounding but factually hybrid statements where each component is grounded in real data but the combination is false — a form of hallucination driven by feature recombination across sources.
50 cards are free to preview. Buyers unlock the rest of the deck plus the interactive training — Spot-the-Bias Quiz unlimited, Swipe Deck with spaced repetition, My Blindspots, Decision Pre-Flight, the Printable Deck + Cheat Sheets, and the Field Guide e-book. $29.50$59.
