Memory is not a filing cabinet. A filing cabinet stores records that remain stable until deliberately removed or modified. Memory is a living, reconstructive process — one in which later experiences actively reach back and alter what came before. One of the most reliable demonstrations of this is retroactive interference: the phenomenon in which new information impairs the ability to recall previously learned material.

Understanding retroactive interference is not merely academically interesting. It has direct implications for how students should structure their study schedules, how educators should design curricula, and how professionals can protect their skill memory when learning new procedures. It also illuminates one of the fundamental realities about human memory: that the most recently learned version of any skill or piece of information tends to dominate earlier versions, whether we want it to or not.

What Is Retroactive Interference?

Retroactive interference (sometimes called retroactive inhibition) is a form of forgetting in which newly acquired information interferes with the retrieval of previously learned material. The new learning reaches backward in time to disrupt the older memory trace, making the earlier information harder to access.

The word "retroactive" signals the temporal direction: the disrupting learning comes after the target memory, and it acts on the past. If you learn a new password today, tomorrow you may find it harder to recall last month's password. If you learn a new dance step that is superficially similar to one you practiced last year, the new version may overwrite or compete with the old one.

A Formal Definition

In cognitive psychology, retroactive interference is defined as a decrease in the ability to recall previously learned information, caused by the learning of subsequent information — particularly when the old and new information are similar in content, context, or structure.

Two conditions maximize the effect:

1. High similarity between old and new material (learning Spanish after French interferes more than learning chemistry after French)
2. Short temporal gap between original learning and new learning (the older memory has had less time to consolidate before the new learning arrives)

The Discovery: Muller and Pilzecker, 1900

Retroactive interference was first systematically demonstrated by German psychologists Georg Elias Muller and Alfons Pilzecker in 1900 — making it one of the oldest experimentally documented memory phenomena. Their work, published in a landmark monograph on experimental contributions to memory theory, established the empirical foundations that subsequent generations of cognitive psychologists would build upon.

Muller and Pilzecker showed participants lists of nonsense syllables, then tested recall after varying delays. A critical manipulation: some participants engaged in a second learning task (a new list of syllables) immediately after studying the first list. Others rested or engaged in non-learning activity during the delay interval.

The results were clear: participants who learned new material immediately after the first list recalled significantly less of the original list than those who had rested. Muller and Pilzecker called this "retroactive inhibition" and proposed that it occurred because memories need time to consolidate — to become stable — and that mental activity during this window disrupts consolidation.

Their perseveration-consolidation theory was the first systematic account of why sleep might be good for memory and why "cramming" — learning one set of material immediately followed by another — produces inferior long-term retention. More than a century later, this core insight remains among the most replicated and practically important findings in memory research.

Early Experimental Refinements

The decades following Muller and Pilzecker's work saw extensive laboratory investigation into the boundary conditions of retroactive interference. John McGeoch (1932) at the University of Missouri conducted systematic parametric studies varying the similarity between original and interpolated learning, the amount of interpolated learning, and the temporal spacing between learning episodes. His work established what became known as the similarity gradient of interference: the more similar the interpolated material to the original learning, the greater the retroactive interference produced.

McGeoch also challenged the decay theory of forgetting that was dominant in early twentieth-century psychology, arguing that forgetting should more properly be understood as a consequence of interference from competing memories rather than passive fading of memory traces over time. This theoretical stance proved enormously productive for subsequent research.

Proactive vs. Retroactive Interference: Understanding the Difference

Retroactive interference is often discussed alongside its temporal mirror image: proactive interference.

Feature	Proactive Interference	Retroactive Interference
Direction of disruption	Old to New (old disrupts new)	New to Old (new disrupts old)
When it operates	During new learning	When recalling old learning
Mechanism	Old associations compete during encoding of new material	New associations compete during retrieval of old material
Example	Struggling to remember a new phone number because the old one is so familiar	Forgetting the old phone number because the new one is now dominant
Strongest when	Old material is highly practiced and automatized	New material is similar to old and was learned recently

Both forms of interference reflect the same fundamental mechanism: competition between memory traces. The human memory system does not store each experience in a completely separate location. Related memories share neural representations and compete for retrieval. Whatever is most recently strengthened tends to win this competition.

Benton Underwood (1957, Psychological Review) made the influential theoretical argument that proactive interference — not retroactive interference — was the primary cause of long-term forgetting in everyday life. His reasoning: over a lifetime, the accumulation of prior learning creates an ever-growing pool of competing associations that interfere with the retrieval of any specific target memory. While retroactive interference from recent learning is powerful in the short term, it is the chronic background noise of a lifetime's worth of prior learning that accounts for the gradual deterioration of specific memories over years.

Why Similar Content Creates Maximum Interference

The degree of interference depends strongly on the similarity between old and new learning. This is not arbitrary — it reflects how memory is organized.

Memories are stored and retrieved via associative networks: a memory is a pattern of connections between concepts, contexts, and sensory features. When new learning activates many of the same connections as existing memories, the two sets of connections compete for dominance. The new, more recently reinforced connections tend to win, suppressing access to the older pattern.

High similarity examples (maximum interference):

• Learning French after studying Spanish (shared vocabulary structure, similar grammar)
• Learning a new software workflow that follows the same interface conventions as the old but with different key bindings
• Memorizing a new apartment's address when you have lived at a numerically similar address for years
• Learning a second instrument that requires similar hand positions to your first

Low similarity examples (minimal interference):

• Learning Spanish vocabulary after studying calculus
• Learning to cook after studying history
• Any two domains with minimal structural overlap

This explains why students who study two related subjects back-to-back (say, two foreign languages in the same evening study session) typically perform worse on both than students who interleave those subjects with unrelated material.

The Neural Basis: Overlapping Representations

Modern neuroscience offers a mechanistic account of why similarity drives interference. Research using functional neuroimaging has established that memories for similar items activate overlapping populations of neurons in the hippocampus and associated cortical areas. When new learning strengthens some of these overlapping representations, the pattern completion processes that normally enable retrieval of the original memory are disrupted — they now complete toward the new pattern rather than the old.

James McClelland, Bruce McNaughton, and Randall O'Reilly (1995, Psychological Review) developed an influential computational framework — the Complementary Learning Systems theory — that formalized why the hippocampus is particularly vulnerable to interference from similar new learning. Their model correctly predicts that retroactive interference will be greatest when new learning is most similar to original learning, and that the temporal gradient of interference (stronger with shorter gaps between original and interfering learning) reflects the dynamics of hippocampal-to-neocortical memory consolidation.

Sleep as Protection Against Retroactive Interference

Perhaps the most practically important finding in retroactive interference research is the protective role of sleep. Sleep does not just consolidate memories — it provides a period during which no new learning is occurring, protecting recently encoded memories from competitive interference.

The Jenkins and Dallenbach Study (1924)

One of the most cited early demonstrations of sleep's protective effect came from John Jenkins and Karl Dallenbach at Cornell University in 1924. They had participants learn lists of nonsense syllables, then tested recall after either 1, 2, 4, or 8 hours — spent either awake (going about normal activities) or asleep.

Recall was consistently and substantially better after sleep than after equivalent time spent awake. After 8 hours, participants who had slept recalled approximately 56% of the material; those who had been awake recalled only about 9%. The differences were not due to sleep's metabolic recovery effects — they were due to the absence of interference during sleep.

This study has been criticized on methodological grounds (the comparison confounds time of day effects with sleep effects; participants who slept learned in the evening while those who stayed awake learned in the morning), but its core finding has been replicated with more controlled designs repeatedly across the subsequent century.

Modern Neuroscience of Sleep and Memory

Contemporary neuroscience has clarified the mechanisms. Matthew Walker's laboratory at UC Berkeley and Jan Born's group at the University of Tubingen have been particularly productive in documenting the cellular and systems-level processes involved.

Slow-wave sleep (deep non-REM sleep, particularly in the first half of the night) is associated with the reactivation and transfer of hippocampus-dependent declarative memories to the neocortex. During this process, the hippocampus "replays" recently encoded experiences, strengthening their neocortical representations. Born's group demonstrated using targeted memory reactivation — playing soft sounds associated with learned spatial locations during slow-wave sleep — that this replay is a causal mechanism of memory consolidation, not merely a correlate.

REM sleep (concentrated in the second half of the night) plays a role in procedural memory consolidation and in integrating new learning with existing semantic knowledge. Research by Matthew Walker and Robert Stickgold (2004, Neuron) has linked REM sleep to creative insight — the ability to find connections between distantly related concepts — and to the overnight improvement in performance of motor sequence tasks that has been observed repeatedly since the 1990s.

Critically, both types of consolidation happen during a period of offline processing — no competing new information is being encoded. The learning you do at 10 PM is protected from interference by the 7 or 8 hours of sleep that follow it.

Practical Implication: Study Before Sleep

For learners with flexibility in their schedules, studying important material shortly before sleep (and then sleeping rather than watching television or scrolling, both of which involve new information processing) significantly improves long-term retention compared to studying earlier in the day and remaining awake for many more hours.

A study by Ullrich Wagner and colleagues (2004, Nature) found that participants who slept between initial learning and a problem-solving test were nearly three times as likely to discover an elegant hidden solution to a numerical task as those who remained awake — demonstrating that sleep's benefits extend beyond rote memory to insight and flexible problem-solving.

Encoding Specificity and Context Effects

Retroactive interference is also influenced by encoding specificity: the principle, articulated by Endel Tulving and Donald Thomson (1973, Psychological Review), that memory is stored with contextual cues, and retrieval is best when the retrieval context matches the encoding context.

When new learning occurs in the same physical environment, emotional state, or cognitive context as old learning, interference is greater — because the contextual cues that would normally help retrieve the old memory now also trigger retrieval of the new, competing memory.

This is why athletes who learn a new technique in the same gym where they practiced the old one may find the new technique particularly disruptive to their retrieval of the original. The gym context triggers both memories simultaneously, creating maximum competition.

The practical implication is counterintuitive: deliberately varying the study context can actually reduce retroactive interference by differentiating the retrieval cues for different bodies of learning. Studying your second foreign language in a different location, at a different time of day, and with different study materials from your first creates contextual distinctiveness that helps the memory system separate the two overlapping knowledge structures.

Retroactive Interference in Real-World Settings

Professional Skills Update

One of the most significant real-world contexts for retroactive interference is professional retraining. When a company transitions to a new software system, new safety procedure, or new protocol, employees must learn new behaviors that directly compete with highly automatized old behaviors.

This explains a counterintuitive finding in human factors research: experienced workers sometimes perform worse than novices on newly introduced systems. Novices have no competing old habits. Experienced workers must overcome strong proactive interference from their old habits (the old behavior tries to intrude) while simultaneously avoiding retroactive interference from the new learning suppressing their access to old skills they still need in transitional periods.

Research in aviation safety has documented this phenomenon clearly. Earl Wiener and Renwick Curry (1980) identified what they called the "automation surprise" — the tendency of experienced pilots learning glass-cockpit aircraft to make errors on tasks that are superficially similar to tasks they had performed thousands of times on analog instruments. The new system used the same conceptual categories (altitude, heading, speed) but different physical procedures, creating maximum conditions for bidirectional interference.

Healthcare provides another rich domain. Studies of surgical skill retraining find that surgeons learning new laparoscopic techniques that involve similar movements to previously learned techniques show greater error rates than surgeons for whom the technique is entirely novel. The solution in medical education has been the deliberate use of variability practice — exposing trainees to multiple variants of a procedure early in training to prevent excessive automatization of a single version that would later interfere with learning modifications.

Language Learning

Language acquisition provides some of the richest demonstrations of retroactive interference. Learners of multiple similar languages (Spanish and Italian; Norwegian and Danish; Mandarin and Cantonese) consistently report difficulty keeping the two systems separate. Speaking one language immediately after studying another produces more cross-language interference than spacing the sessions.

Research on second language learning has found that interleaving study across different aspects of a language (vocabulary, grammar, pronunciation) on separate occasions, rather than massing sessions on a single aspect, reduces interference and improves retention.

Ellen Bialystok's longitudinal research on bilingualism (2007 and subsequent) added a fascinating dimension to this picture: despite the interference challenges of maintaining two languages simultaneously, lifelong bilinguals show cognitive advantages in executive function — particularly in tasks requiring the selective suppression of competing responses. The constant practice of managing cross-language interference appears to train the inhibitory control systems involved in all competitive memory retrieval. The cost of bilingualism (interference) appears to come with a compensating benefit (stronger interference management generally).

Memory for Sequential Events

Retroactive interference explains why the final items in a sequence of events are often best remembered (the recency effect in serial recall), while the middle items — those with the most interference from items both before and after — are worst remembered. This is the psychological basis of the serial position curve: a reliable finding that memory performance varies predictably depending on where in a sequence an item was learned.

Bennet Murdock's (1962) systematic studies of free recall established the precise mathematical form of the serial position curve. The primacy effect (good recall for early items) reflects extra encoding time and rehearsal for those items. The recency effect reflects the absence of retroactive interference — nothing has been learned after the last few items. The inferior recall for middle items reflects both forward and backward interference from the surrounding items.

This has direct implications for presentation design, teaching, and meeting structure. The most important points should be placed at the beginning or end of a presentation or lesson, not buried in the middle — where retroactive interference from subsequent content will most heavily impair their retention.

Study Strategies That Minimize Retroactive Interference

Research on interference theory supports several specific study strategies:

Spacing: Distributing study sessions over time, with intervals between them, allows each session's material to consolidate before new interfering material arrives. Spaced practice is one of the most consistently supported findings in learning science, replicated across hundreds of studies and dozens of domains. A landmark review by Harold Pashler and colleagues (2007, Psychological Science in the Public Interest) found that the spacing effect was among the most robust and generalizable findings in all of cognitive psychology, with effect sizes large enough to have substantial practical significance for educational design.

Avoid studying similar subjects back-to-back: If you must study two related subjects in one day, insert a substantial period of unrelated activity or sleep between them.

Retrieval practice: Self-testing (rather than passive re-reading) strengthens memory traces and makes them more resistant to interference. A memory that has been successfully retrieved multiple times is more robustly encoded and harder to displace. Henry Roediger and Jeffrey Karpicke (2006, Science) demonstrated in a landmark study that students who studied a text passage and then took a recall test retained 50% more information one week later than students who simply re-studied the passage an equivalent number of times. The protection against interference that retrieval practice confers appears to operate by strengthening the specific retrieval pathways to a memory, making them less susceptible to competition.

Interleaving: Mixing different types of problems or topics within a study session, rather than blocking all practice on one type, reduces the buildup of competing associations and promotes more discriminated retrieval cues.

Sleep strategically: For particularly important material, study the most critical content last in a session, immediately before sleep, to maximize the consolidation window and minimize the number of waking hours during which new learning can interfere.

Contextual variation: Studying material in multiple contexts (different rooms, times of day, study formats) diversifies the retrieval cues and reduces the dependence on any single context that might later be contaminated by competing associations.

The Relationship Between Interference and Forgetting

Retroactive and proactive interference together constitute the interference theory of forgetting — one of the major theoretical accounts of why we forget. The rival account is decay theory, which holds that memories fade over time independently of intervening experiences.

The evidence has consistently favored interference theory for most types of forgetting. Studies that hold all other variables constant and vary only the amount and similarity of intervening learning show that forgetting rate is much more strongly predicted by interference than by elapsed time alone.

However, the two theories are not mutually exclusive. Very long-term forgetting (over years and decades) may involve both processes. The hippocampal consolidation research suggests that some memory traces do physically degrade if they are not reactivated — but even this degradation appears to be accelerated by competing memories that take over the hippocampal representations.

A third theoretical framework — retrieval inhibition — has gained traction through work by Michael Anderson and colleagues (2001, Nature Neuroscience). Anderson's research showed that actively trying not to think about certain memories (suppression) reduces their subsequent accessibility, suggesting that the memory system has active inhibitory mechanisms that deliberately suppress competing memories during retrieval. This implies that what looks like retroactive interference may partly involve not just competition from new memories but active inhibition of old ones — a distinction with implications for therapeutic work on intrusive memories.

Retroactive Interference and the Reliability of Eyewitness Memory

One of the most consequentially studied applications of retroactive interference is in eyewitness testimony. Research pioneered by Elizabeth Loftus at the University of Washington and later UC Irvine has demonstrated conclusively that post-event information — information encountered after witnessing an event — can retroactively alter the content of an eyewitness's memory for that event.

In Loftus and Palmer's (1974) classic study, participants who watched a film of a car accident and were then asked "How fast were the cars going when they smashed into each other?" gave higher speed estimates and were more likely to report (falsely) seeing broken glass than participants asked the same question using the word "contacted." The single word in the question retroactively altered the memory of the accident itself.

Subsequent research by Loftus established the misinformation effect: exposure to misleading post-event information reliably changes eyewitness reports, with participants incorporating the misinformation into their memory and often being unable to distinguish original from contaminated details. By the 1990s, this research had influenced legal standards for eyewitness testimony and contributed to post-conviction DNA exonerations of hundreds of individuals wrongfully convicted based on eyewitness accounts.

"The single most important variable in promoting long-term retention and transfer is practice at retrieval." — Henry Roediger, memory researcher, summarizing the practical implication of interference and memory consolidation research

The misinformation effect is retroactive interference operating in one of its most practically significant contexts. Each piece of post-event information — a news report, a leading question from a detective, a conversation with another witness — is a source of potentially interfering new learning that competes with and can overwrite the original encoded experience.

Conclusion

Retroactive interference is a humbling reminder that the human brain did not evolve to be a perfect recording device. It evolved to prioritize current, actionable information over historical records — which means the most recently learned version of anything tends to dominate what came before. This is often useful: you want your current password, your current route to work, your current job title to come to mind most easily. But it means that deliberate learning strategies are required if you need to preserve both old and new knowledge simultaneously.

The practical implications are clear and well-supported. Space your study sessions. Sleep before important tests and after important learning. Avoid massing similar subjects together. Practice retrieval actively rather than passively re-reading. Vary the contexts in which you study important material. And understand that the feeling of having learned something thoroughly in a single intense session is often an illusion — without the protective interval of sleep and time, that learning is far more vulnerable to displacement than it feels.

Memory is not a fixed archive. Managing it well requires working with its limitations rather than against them.

Frequently Asked Questions

What is retroactive interference?

Retroactive interference is a memory phenomenon in which newly acquired information impairs the recall of previously learned material. The new learning 'reaches back' to disrupt the older memory trace, making earlier information harder to retrieve. For example, learning a new phone number may make it harder to recall your old one. The effect is strongest when old and new information are similar in nature, which increases competition during retrieval.

How is retroactive interference different from proactive interference?

Retroactive interference involves new information disrupting old memories — the disruption works backward in time. Proactive interference is the reverse: old, previously learned information interferes with the acquisition or recall of new information. For instance, if you learned to drive on a manual transmission and then switch to an automatic, reaching for the clutch pedal out of habit is proactive interference. Both are forms of interference forgetting, but they operate in opposite temporal directions.

Who first identified retroactive interference?

German psychologists Georg Elias Muller and Alfons Pilzecker first systematically demonstrated retroactive interference in 1900. In their experiments, participants who engaged in a new learning task immediately after studying a list of nonsense syllables recalled significantly less of the original list than participants who rested quietly between study and test. They called this 'perseveration-consolidation theory,' arguing that memories need time to consolidate and that mental activity during this window disrupts consolidation.

How does sleep protect against retroactive interference?

Sleep plays a critical role in memory consolidation — the process by which fragile short-term memory traces are stabilized into durable long-term memories. During sleep, particularly during slow-wave and REM stages, the hippocampus replays newly encoded information and transfers it to the neocortex for long-term storage. This consolidation process happens during a period when no new information is being encoded, protecting memories from retroactive interference. Studies consistently show that sleeping shortly after learning dramatically improves later recall compared to staying awake.

What study strategies minimize retroactive interference?

The most effective strategies include: spacing study sessions rather than massing them (spaced practice allows consolidation between sessions), avoiding studying similar subjects back to back (studying French immediately after Spanish creates maximum interference), using sleep strategically by reviewing important material shortly before sleep, interleaving different subjects or problem types to build robust retrieval cues, and using retrieval practice (testing yourself) rather than passive re-reading to strengthen memory traces against competitive interference.