Proactive Interference

Abstract

Proactive interference is the disruption of new learning and memory by material learned earlier, the forward-acting counterpart of retroactive interference. Underwood argued in 1957 that much of what looks like forgetting over a day is not decay but interference from lists learned before, and Keppel and Underwood showed the same buildup across trials even over a few seconds. Wickens then demonstrated that shifting the semantic category of the items produces a sharp release from proactive interference, restoring recall and revealing the dimension on which the memory was encoded. Later work relocated the effect to working memory, tying susceptibility to individual differences in capacity and to prefrontal control processes that resolve competition among candidates. This article defines proactive interference, traces its buildup and release, and reviews its mechanisms in memory and the brain.

Keywords: proactive interference, forgetting, interference theory, working memory, release from interference

Learn a friend's old phone number well enough and a new one becomes strangely hard to hold: the familiar digits keep pushing forward and crowding out the ones being learned. That everyday intrusion is proactive interference, the process by which earlier learning impairs the retention of material acquired later (Underwood, 1957). It runs forward in time, from old memories to new, which is what distinguishes it from retroactive interference, where new learning disrupts what was already stored. The two together were long offered as the mechanism behind most forgetting.

The construct matters because it overturned an intuitive picture of memory in which traces simply fade with the passage of time. If forgetting were mere decay, then a person's first list of the day and their twentieth should be equally well retained, yet they are not: the more prior material a person has learned, the worse their memory for anything new, even when the retention interval is identical (Keppel & Underwood, 1962). Forgetting, on this account, is driven less by elapsed time than by competition among memories. The sections below define proactive interference, follow its buildup across trials and its release when the material shifts, set it within classical interference theory, and trace it into working memory and the brain.

## What Proactive Interference Is

Proactive interference is the impairment of memory for recently learned material caused by material learned before it. The defining feature is the forward direction of the disruption: prior learning reaches ahead in time to compete with, and degrade, the retrieval of things learned later (Underwood, 1957). It is measured by comparing memory for a target set in people who have learned prior sets against memory in people who come to the same target fresh; the deficit in the experienced group, holding the retention interval constant, is the interference. Because the retention interval is matched, the deficit cannot be attributed to the mere passage of time.

It is worth separating the effect from its neighbours. Proactive interference is the mirror image of retroactive interference, in which later learning disrupts earlier memories; the two are defined by the direction in which the interference travels, not by different underlying stores. It is also distinct from simple decay, the hypothesised fading of a trace with disuse: interference locates forgetting in competition between memories rather than in the erosion of any one of them (Keppel & Underwood, 1962). Keeping these apart matters because they make opposite predictions. Decay says older and more numerous memories should be irrelevant to new learning, whereas proactive interference predicts precisely that they should hurt it.

## The Buildup Across Trials

The most influential evidence for proactive interference came from an unlikely place: the short-term forgetting task that had seemed to prove the opposite. Peterson and Peterson had shown that a single trigram, once rehearsal was blocked by a distracting count-back task, was largely forgotten within about eighteen seconds, a rapid loss widely read as the decay of a fragile short-term trace (Peterson & Peterson, 1959). Brown had reported the same rapid loss independently in Britain and presented it squarely as a test of decay theory, and the paradigm has carried both names since as the Brown-Peterson task (Brown, 1958). Keppel and Underwood looked again at where in the experiment that forgetting occurred and found something the decay account could not explain: on the very first trial, before any prior items existed to interfere, retention after eighteen seconds was nearly perfect. The steep forgetting curve emerged only over later trials (Keppel & Underwood, 1962).

That pattern is the signature of proactive interference. As trial follows trial, the items from earlier trials accumulate and compete at retrieval with the current one, so recall of the current item declines even though the retention interval never changes. The forgetting in the Brown-Peterson task was therefore not the decay of a trace over eighteen seconds but the buildup of interference across the session, imported from the person's own earlier trials (Keppel & Underwood, 1962). The finding generalised Underwood's earlier, larger-scale point: reviewing many long-term retention studies, he had shown that a subject's recall of a list fell steeply with the number of lists they had previously learned in the laboratory, so that most of the forgetting attributed to the passage of a day was really interference from the subject's own prior learning (Underwood, 1957).

## Release From Proactive Interference

If proactive interference builds up because successive items resemble one another and compete, then breaking the resemblance should relieve it. Wickens established exactly this with the release-from-proactive-interference technique. Using the Brown-Peterson procedure, he ran several trials with items drawn from one semantic category, watching recall decline as interference accumulated, and then, on a critical trial, switched to items from a different category. Recall on that shift trial sprang back up, often to near the level of the first trial, a rebound he called release from proactive interference (Wickens, 1970).

The release effect is more than a curiosity, because its size measures how the material was encoded. If shifting along some dimension produces a large release, the memory system must have been using that dimension to tell items apart; if a shift produces none, the system was blind to it. Wickens used this logic as an assay of encoding, mapping which attributes of a word (its taxonomic category, its sense modality, whether it named something living) the memory system registered, by seeing which shifts bought a release (Wickens, 1970). The technique built on the finding that interference in short-term memory grows with the similarity between the interfering and the target items, so that reducing similarity is what lifts the interference (Wickens, Born, & Allen, 1963). The two demonstrations below drive the buildup and then trigger the release.

## Interference Theory and Forgetting

Proactive interference was the centrepiece of interference theory, the dominant account of forgetting through the middle of the twentieth century. On the classical version, forgetting reflects two processes acting on associations between stimuli and responses: unlearning, in which a competing association is weakened while a new one is acquired, and response competition, in which the old and new responses vie at the moment of retrieval and the wrong one sometimes wins (Underwood, 1957). Proactive interference was explained mainly through competition and the spontaneous recovery of earlier responses over time, which is why a prior list hurts recall of a later one more and more as the retention interval lengthens.

The theory did not survive unamended. Postman and Underwood, reviewing its accumulating difficulties, set out where the simple two-factor account strained: the exact conditions for unlearning, the fate of recovered responses, and the role of the general set to produce list-appropriate items all resisted the original formulation, and they proposed a revised scheme with a broader role for response-set suppression (Postman & Underwood, 1973). Decades later the framework was recast again in the language of cognitive control. Anderson argued that interference is best understood not as a passive collision of associations but as a consequence of the executive processes that select among competing memories, so that retrieving one item actively inhibits its competitors and that inhibition, not decay or blind competition, produces much of the observed forgetting (Anderson, 2003). Table 1 sets out the landmark paradigms that built this picture.

## Proactive Interference in Working Memory

Interference is not confined to the long lists of the verbal-learning laboratory; it operates over the handful of items held in working memory from moment to moment. Kane and Engle made the individual-differences case: people with low working-memory span accumulate proactive interference faster than high-span people, and loading a high-span person with a divided-attention task makes them behave like a low-span one. Their reading was that resisting interference is an active, attention-demanding operation, and that the executive capacity to keep the current context distinct from prior ones is what protects memory from competition (Kane & Engle, 2000). Susceptibility to proactive interference thus became one of the clearest behavioural windows onto executive control.

The similarity principle carries over as well. Bunting showed that proactive interference in complex working-memory span tasks depends on the similarity between the to-be-remembered items and the intervening material, so that memory for numbers is harmed by other numbers but spared when the processing task uses words instead (Bunting, 2006). At the theoretical extreme, some have argued that interference is not merely one influence on the capacity limit but its source: Endress and Szabó proposed that the apparent fixed capacity of working memory falls out of interference among similar items, so that what looks like a storage limit of a few slots is really the point at which accumulated interference makes items indiscriminable (Endress & Szabó, 2017). On that view proactive interference is not a nuisance layered on top of the memory limit; it is close to being the limit itself.

## Proactive Interference in the Brain

Because resolving proactive interference is an act of control, it has a localisable neural signature. Jonides and Nee reviewed the imaging evidence and concluded that overcoming interference in working memory reliably recruits the left inferior frontal gyrus in the prefrontal cortex, a region engaged whenever a familiar but currently irrelevant item must be rejected as a response (Jonides & Nee, 2006). The workhorse task in this literature is the recent-probes procedure: a probe that appeared in the immediately preceding trial, but is not part of the current set, is especially tempting, and saying no to it demands exactly the interference resolution the prefrontal region supports.

Finer-grained work has begun to fractionate that resolution into stages. Nee, Jonides, and Berman distinguished interference arising at the point of retrieval from interference arising during the maintenance of the memory set, mapping them onto partly separable prefrontal and posterior systems rather than a single control site (Nee, Jonides, & Berman, 2007). Response-time modelling adds a complementary account of the cost. Öztekin and McElree found that proactive interference does not slow the underlying speed of memory access; instead it removes the fast, familiarity-based path to a decision, forcing reliance on a slower recollective check, so the behavioural slowing reflects a change in which process is used rather than a general sluggishness of retrieval (Öztekin & McElree, 2007). The demonstration below reproduces the recent-probes task that these studies turn on.

## Worked Example

The logic of buildup and release can be made concrete with a schematic set of recall scores. Imagine a Brown-Peterson experiment run in blocks of four trials, all items drawn from the category fruits, with recall measured after an eighteen-second filled delay. On trial 1, with no prior items to compete, recall is high, say 90%. As interference builds, trial 2 falls to about 65%, trial 3 to 48%, and trial 4 to 40% — the accumulating fruit words from earlier trials increasingly intrude on the current one (Keppel & Underwood, 1962). The drop from 90% to 40% across four same-category trials is the buildup of proactive interference.

Now, on trial 5, the category shifts to professions. Recall jumps back to about 85%, nearly the trial-1 level (Figure 1). That rebound is the release: because the new items no longer resemble the accumulated fruit words, the earlier trials stop competing, and memory is freed from the interference that had dragged trial 4 down to 40% (Wickens, 1970). The size of the release, here about 45 percentage points recovered, is the experimental measure of how strongly the memory system had been using semantic category to encode the items; a shift the system ignored would have produced no such recovery. The whole phenomenon is captured in these two numbers: the fall from 90% to 40% as like competes with like, and the return to 85% the moment the likeness is broken.

Discussion

Proactive interference reshaped the psychology of forgetting by moving its cause from the clock to the contents of memory. The decay picture, in which traces simply weaken with time, could not explain why a person's recall of new material depended so heavily on how much they had already learned, nor why forgetting over eighteen seconds should appear only after the first few trials of a session (Underwood, 1957; Keppel & Underwood, 1962). Once forgetting was seen as competition among memories, the release technique turned that competition into a measuring instrument, using the recovery on a category shift to read off the dimensions along which material had been encoded (Wickens, 1970).

The idea has proved durable because it kept finding new homes. What began as an account of long lists of nonsense syllables became an account of the rapid loss in short-term memory, then of the capacity limit of working memory, and finally of a specific control operation with an address in the prefrontal cortex (Jonides & Nee, 2006). The modern reframing in terms of inhibition and executive control did not discard the classical findings so much as explain them at a different level, recasting the collision of associations as the output of processes that actively select among competitors (Anderson, 2003). Across every version the core claim has held: we forget less because memories fade than because they get in each other's way.

## Current Directions

Current work is pursuing the mechanisms of buildup and release with sharper tools and pushing the phenomenon into new domains. Reviewing the cognitive and neural evidence, Kliegl and Bäuml describe buildup and release as two dissociable processes rather than a single continuum, with the release effect in particular tied to a contextual updating that lets the system treat the shifted material as belonging to a new episode, supported by partly distinct neural signatures (Kliegl & Bäuml, 2021). The emphasis has moved from cataloguing which shifts produce release toward specifying what the release actually does to the underlying representation.

A second line examines how long-term knowledge and ageing modulate the effect. Rhodes, Buchsbaum, and Hasher found that the influence of long-term memory on working memory cuts both ways, producing proactive facilitation as well as interference, and that the balance between them shifts with age, so that older adults are differentially vulnerable to the interfering side (Rhodes, Buchsbaum, & Hasher, 2022). Complementing this, Mercer has continued to probe how the familiarity of the interfering material governs its impact, showing that the prior exposure and familiarity of competing items shape how much proactive interference they generate in verbal memory (Mercer, 2025). Together these directions are turning a classical laboratory effect into a quantitative account of when prior learning helps, when it hurts, and for whom.

Commonly Confused With

Retroactive Interference

Both are forms of interference in which one body of learning degrades another, and both fall under classical interference theory, which is why they are easily run together; they differ only in the direction the disruption travels. Proactive interference runs forward, so older learning impairs a memory formed later; retroactive interference runs backward, so newer learning impairs a memory formed earlier. Learning an old phone number and then struggling with the new one is proactive; learning the new number and then finding the old one gone is retroactive. The two are ordinarily produced together whenever similar material is learned in sequence, and both were offered as the twin engines of everyday forgetting (Underwood, 1957; Postman & Underwood, 1973).

Common Misconceptions

Forgetting is mainly the fading of memory traces over time.

Much of it is interference, not decay. Recall of new material falls with the amount learned beforehand even when the retention interval is held constant, which pure fading cannot explain (Underwood, 1957).

The rapid forgetting in the Brown-Peterson task proves short-term traces decay in seconds.

On the first trial, before any prior items exist, retention after eighteen seconds is nearly perfect. The steep loss appears only as interference builds up across later trials (Keppel & Underwood, 1962).

Proactive interference is a fixed property of the stimuli.

Its size depends on the person and the situation. Low working-memory span and divided attention both increase it, showing that resisting interference is an active, capacity-limited operation (Kane & Engle, 2000).

Glossary

Brown-Peterson task.

A procedure in which a small set of items is recalled after a short delay filled with a distractor task that prevents rehearsal; the paradigm in which the buildup of proactive interference over seconds was demonstrated.

Buildup of proactive interference.

The progressive decline in recall across successive trials of similar material, as items from earlier trials accumulate and compete with the current one at retrieval.

Cue overload.

The loss of a retrieval cue's effectiveness as more memories become associated with it, one way of describing why accumulated similar items interfere with a target.

Decay theory.

The view that memory traces weaken and are lost through the mere passage of time; the account that interference theory was marshalled against.

Encoding.

The transformation of incoming material into a memory representation; the release technique reveals which attributes of an item were used during encoding.

Familiarity.

A fast, strength-based signal that an item has been encountered before; proactive interference makes familiarity misleading, because interfering items feel familiar without belonging to the current set.

Interference theory.

The account of forgetting as competition between memories, driven by proactive and retroactive interference, that dominated the study of forgetting through the mid-twentieth century.

Proactive inhibition.

An older synonym for proactive interference, still current in the short-term-memory literature, naming the forward-acting disruption of new memories by earlier learning.

Proactive interference.

The impairment of memory for recently learned material caused by material learned earlier; the forward-acting counterpart of retroactive interference.

Recent-probes task.

A working-memory recognition task in which a probe from the immediately preceding trial, though not in the current set, must be rejected; the standard assay of interference resolution in the brain.

Release from proactive interference.

The recovery of recall when the material shifts along a dimension the memory system was using, breaking the similarity that drove the buildup; its size indexes the dimension of encoding.

Response competition.

The vying of old and new responses at the moment of retrieval, one of the two mechanisms by which classical interference theory explained forgetting.

Retroactive interference.

The impairment of an earlier memory by material learned afterwards; the backward-acting mirror image of proactive interference.

Temporal discriminability.

The ease with which the target memory can be distinguished in time from competing ones; low discriminability among recent similar items is one account of why proactive interference builds up.

Working memory.

The limited system that holds and manipulates information over short intervals; its capacity limit and its resistance to interference are closely bound together.

Key Researchers

Michael C. Anderson (living). Cognitive neuroscientist at the MRC Cognition and Brain Sciences Unit, University of Cambridge; his 2003 reformulation recast interference theory around executive control and inhibition, arguing that retrieving a memory actively suppresses its competitors. Homepage - Google Scholar - ORCID

Karl-Heinz T. Bäuml (living). Chair of developmental and cognitive psychology at the University of Regensburg; his laboratory maps the cognitive and neural mechanisms behind the buildup and release of proactive interference and their relation to memory control. Homepage - Google Scholar

John Jonides (living). Cognitive neuroscientist at the University of Michigan; his imaging studies localised the resolution of proactive interference in working memory to the left inferior frontal gyrus of the prefrontal cortex. Wikipedia - Google Scholar - ORCID

Michael J. Kane (living). Cognitive psychologist at the University of North Carolina at Greensboro; with Randall Engle he showed that susceptibility to proactive interference scales with working-memory capacity and rises under divided attention. Homepage - Google Scholar - ORCID

Benton J. Underwood (1915-1994). Experimental psychologist at Northwestern University and a central architect of interference theory; his 1957 analysis showed that much apparent forgetting over a day is proactive interference from the learner's own prior laboratory lists. Wikipedia

Delos D. Wickens (1909-1988). Experimental psychologist at Ohio State University; his release-from-proactive-interference technique showed that shifting the semantic category of the items restores short-term recall, turning interference into a probe of encoding. Wikipedia

Frequently Asked Questions

What is proactive interference?
Proactive interference is the disruption of memory for recently learned material by material learned earlier. It acts forward in time, from old memories to new ones, so the more a person has already learned, the harder it is to retain something new, even when the retention interval is held constant (Underwood, 1957).

How is proactive interference different from retroactive interference?
The two differ in the direction of the disruption. Proactive interference is earlier learning harming a later memory, while retroactive interference is later learning harming an earlier memory. They are defined by which way the interference travels rather than by separate memory systems (Underwood, 1957).

What is release from proactive interference?
It is the sudden recovery of recall when the material shifts to a new category after several similar trials. Because the new items no longer resemble the accumulated ones, they stop competing, and recall springs back toward its starting level, which is why the size of the release indexes how the items were encoded (Wickens, 1970).

Does the Brown-Peterson task show that memory decays in seconds?
Not on closer inspection. On the first trial, before any prior items can interfere, retention after eighteen seconds is nearly perfect, and the rapid forgetting appears only over later trials as interference builds up, so the effect is proactive interference rather than pure decay (Keppel & Underwood, 1962).

Why do some people suffer more proactive interference than others?
Resisting interference is an active, capacity-limited operation. People with low working-memory span accumulate proactive interference faster, and dividing a high-span person's attention makes them just as vulnerable, which points to executive control as the protective factor (Kane & Engle, 2000).

Where in the brain is proactive interference resolved?
Overcoming proactive interference in working memory reliably engages the left inferior frontal gyrus of the prefrontal cortex. This region is recruited whenever a familiar but currently irrelevant item has to be rejected, as in the recent-probes task (Jonides & Nee, 2006).

Why does proactive interference slow recognition?
It does not slow the basic speed of memory access. Instead it removes the fast, familiarity-based route to a decision and forces reliance on a slower recollective check, so the extra time reflects a change in which process is used rather than general sluggishness (Öztekin & McElree, 2007).

Is proactive interference always harmful?
Not entirely. Prior knowledge can also help, producing proactive facilitation alongside interference, and the balance between the two shifts with age, so long-term memory both supports and disrupts working memory depending on the situation (Rhodes, Buchsbaum, & Hasher, 2022).

References