Review article
Buildup and release from proactive interference – Cognitive and neural mechanisms

https://doi.org/10.1016/j.neubiorev.2020.10.028Get rights and content

Highlights

  • Proactive interference (PI) means that prior learning can impede new learning. [79]

  • Several methods have emerged in the literature that can induce PI release. [74]

  • Results from behavioral and imaging studies on PI buildup and release are reviewed. [83]

  • Release from PI can be due both to improved encoding and improved retrieval. [76]

  • Issues that require further behavioral or imaging work are specified. [69]

Abstract

Interference from related memories is generally considered one of the major causes of forgetting in human memory. The most prevalent form of interference may be proactive interference (PI), which refers to the finding that memory of more recently studied information can be impaired by the previous study of other information. PI is a fairly persistent effect, but numerous studies have shown that there can also be release from PI. PI buildup and release have primarily been studied using paired-associate learning, the Brown-Peterson task, or multiple-list learning. The review first introduces the three experimental tasks and, for each task, summarizes critical findings on PI buildup and release, from both behavioral and imaging work. Then, an overview is provided of suggested cognitive mechanisms operating on the encoding and retrieval stages as well as of neural correlates of these mechanisms. The results indicate that, in general, both encoding and retrieval processes contribute to PI buildup and release. Finally, empirical gaps in the current work are emphasized and suggestions for future studies are provided.

Introduction

In today's society, we almost constantly have to process a plethora of information that is directed at us, be it at work or when we try to stay up-to-date on current political and societal topics. The continuous addition of new information to our long-term memory poses a major challenge for the targeted recall of information relevant to the accomplishment of a current task. Indeed, when two or more memories are related but we want to access only one of them, interference can arise. Suppose you are trying to remember a politician's current stance on a given topic, like tax policy. Even if you tend to follow the news very closely, you may have difficulty with this task if the politician has previously flip-flopped on the issue, and you remember both an interview in which they favored tax increases as well as a statement where they argued in favor of tax breaks. If you cannot clearly distinguish which statement came first and which came last, you will likely experience difficulty at recalling the politician's most recent position on the issue. However, if you are able to separate the two statements on a temporal basis, you may resolve the interference and produce the correct response.

Interference effects like those encountered in our everyday lives directly relate to the two most prominent forms of interference studied in the memory literature: retroactive interference and proactive interference. Retroactive interference was first reported by Müller and his student Pilzecker in their monograph in 1900. Müller and Pilzecker (1900) showed that retention of some originally studied (target) information – for instance, the information that a politician initially voiced support for tax increases – was worse when study of that information was followed by the study of interpolated (nontarget) information – e.g., the information that the same politician later supported tax breaks. Retroactive interference had been widely accepted as the major factor underlying episodic forgetting (e.g., Jenkins and Dallenbach, 1924; Skaggs, 1925), when Underwood (1957) published a seminal paper in which he demonstrated that memory failure is not always due to the detrimental effects of subsequent nontarget learning. Indeed, to stay with our example, correctly remembering that a politician currently favors tax breaks can be complicated when the politician earlier expressed support for tax increases. More generally speaking, forgetting can be caused by the detrimental effects of nontarget information studied prior to the study of the target information, i.e., proactive interference [PI]. Since the late 1950s, PI has been extensively studied, with results indicating that PI arises over a wide range of materials and settings and may indeed reflect one of the major causes of forgetting in human memory (for reviews, see Anderson and Neely, 1996; Crowder, 1976).

Given that the additional learning of nontarget information – be it prior to or subsequent to the study of some target information – can impair the targeted use of our memory in many situations, it would be desirable to have a range of tools available to reduce such interference. Memory research over the past decades has identified such tools. For instance, Ekstrand (1967) provided a classic demonstration that sleep can help to reduce retroactive interference. This researcher showed that the detrimental effects of subsequent learning of (nontarget) material on initially studied (target) material can be reduced when the interval prior to test of the target material was filled with sleep rather than wakefulness. A classic example of PI release comes from Tulving and Watkins (1974), who showed that the detrimental effects of prior nontarget learning can be reduced when the nontarget material is tested prior to study of the subsequently learned target material.

The specific goal of the present review is to focus on PI as a central form of forgetting, and provide an overview of both the classic and more recent findings on (i) buildup of PI and (ii) methods that can enable release from PI. Results from several lines of research have demonstrated PI buildup particularly in three types of memory tasks: in paired-associate learning, the Brown-Peterson task, and multiple-list learning. In paired-associate learning, participants may initially study a first (nontarget) list of stimulus-response word pairs (e.g., house-rent, or A–B) and then a second (target) list of pairs where the same stimulus word is presented as in the first list but a new response word is connected to each stimulus word (e.g., house-lease, or A–D). On the final test, the ‘A’ stimulus word is shown as a cue (e.g., house-?) and participants are asked to recall the response word of the second (target) list (lease, or the ‘D’ response). Recall of the ‘D’ response is typically impaired when compared to a control condition in which the ‘B’ and ‘D’ responses are linked to different stimulus words (e.g., earth-round; table-cook, or A–B, C–D), thus reflecting PI buildup (e.g., Greeno, 1964; Postman and Underwood, 1973). In the Brown-Peterson task, participants study multiple lists of items which, for instance, may all belong to a single semantic category (e.g., sports). After study of each list and a short distractor task, they are tested on the immediately preceding list (e.g., Wickens, 1970, 1973). Recall performance typically declines across lists, reflecting buildup of PI. Finally, in multiple-list learning PI designs, participants may study a target list of unrelated items (e.g., nose, wind, mouse, etc.) and are then tested on it. PI buildup in this task is reflected in the finding that target list recall is typically worse when, prior to study of the target list, additional nontarget lists were studied compared to when subjects engaged in unrelated distractor activities for the same duration of time (e.g., Postman et al., 1968).

For each of the three tasks used to induce PI buildup, multiple ways have been identified by which PI buildup can be released. In paired-associated learning, for instance, PI can be reduced as a result of prior experience with PI (e.g., Wahlheim and Jacoby, 2011) or when participants are reminded of the nontarget material during subsequent study of the target material (Wahlheim and Jacoby, 2013). In the Brown-Peterson task, participants can show a recovery from PI when the target material is dissimilar in content from the previously studied nontarget material (e.g., Gardiner et al., 1972) or when the time interval between study of the nontarget material and the subsequent study of the target material is increased (e.g., Kincaid and Wickens, 1970). Finally, in multiple-list learning, a PI reduction can be achieved by directing participants prior to study of the target material to forget the just studied nontarget material (e.g., Bjork et al., 1968) – for instance, by emphasizing that it would not be relevant for the later memory test – or when there is a change in context between the prior encoding of the nontarget material and the subsequent encoding of the target material (e.g., Sahakyan and Kelley, 2002). Interpolated tests of the nontarget material prior to study of the target material can also release PI (e.g. Szpunar et al., 2008).

Theoretical explanations of PI buildup and release assume that both encoding and retrieval processes can critically contribute to the two types of PI effects. Retrieval processes have been argued to be critically involved in PI buildup because the prior study of nontarget material makes it more difficult at the time of test to focus the memory search on the target information. Regarding PI release, an improved ability to differentiate between the nontarget and target material has been suggested to underlie PI release, although the proposals about the nature of the cognitive processes enabling such enhanced discrimination vary across experimental tasks. For instance, enhanced discrimination may be induced due to a greater reliance on the ability to recollect the target material (Jacoby et al., 2010), the use of more effective retrieval cues (Wixted and Rohrer, 1993), or by making the nontarget material more distinctive, so that on a posthoc basis, it can be easier filtered out from the mental search set (Thomas and McDaniel, 2013).

Encoding processes have also been assumed to contribute to PI buildup and release. Regarding PI buildup, the prior study of nontarget material has been suggested to impair subsequent encoding of the target material, because attentional resources can deteriorate with amount of encoded information and thus impair target encoding (Crowder, 1976; Pastötter et al., 2011). Several processes have been argued to induce PI release at the encoding level, and these processes seem to vary with the single tasks. For instance, the encoding problem may be prevented through a reset process that makes the encoding of the target material again as effective as the encoding of the initially studied nontarget material (e.g., Pastötter et al., 2008); or the encoding problem may be compensated by the use of more effective strategies to encode the target material, relative to the prior encoding of the nontarget material (e.g., Sahakyan and Delaney, 2003). PI release may also result from a mixture of encoding and retrieval processes. This may occur when, during encoding of the target material, individuals are reminded of the nontarget material, which may result in an integrated memory representation – including both the target and the nontarget material as well as information on the order in which the two types of material were provided – and improved recall of the target information (e.g., Wahlheim and Jacoby, 2013).

In this review, we first provide an overview of the three experimental tasks that have traditionally been used to induce PI buildup, before we report for each task the various methods that have been applied to induce PI release. For both PI buildup and each of the single PI release methods we report (a) the basic procedure and main findings, (b) the suggested cognitive mechanisms operating on the encoding and retrieval stages, and (c) the current knowledge on neural processes operating on each of the two stages. A final summary section will discuss results on PI buildup and release methods, emphasize empirical gaps in the current work, and offer suggestions for future studies.

Section snippets

PI buildup

Paired-associate learning was first introduced in the late 19th century by American philosopher and psychologist Mary Calkins (Calkins, 1894). In many respects, this type of PI-buildup task is representative of the stimulus-response associationist analysis of learning that dominated experimental psychology in the first half of the 20th century. In a typical A–B, A–D paired-associate learning task, participants initially study a first (nontarget) list consisting, for instance, of

PI buildup

A second important class of PI tasks is the Brown-Peterson task, which was introduced independently in the late 1950s by Brown (1958) and Peterson and Peterson (1959). In the Brown-Peterson task, participants study multiple lists of items that may consist of strings of letters, words, or numbers. After study of each list and a short distractor task, participants are tested on the immediately preceding list (e.g., Wickens, 1970, Wickens, 1973). Recall performance typically declines across lists,

PI buildup

The third prominent task used to study PI is multiple-list learning. PI buildup in this task is typically examined by having participants study a target list – consisting, for instance, of unrelated nouns (e.g., nose, wind, mouse, etc.) – which is either preceded by the study of additional nontarget lists that consist of the same type of study material (e.g., unrelated nouns), or is preceded by unrelated distractor activities for the same duration of time (e.g., simple arithmetic tasks; see

Principal findings and explanations of PI buildup and release

Research from the past eight decades has demonstrated that PI buildup as well as release from PI are very robust findings that arise across a wide variety of experimental tasks, learning conditions, and study materials. This research has provided a number of indices of PI buildup, at both the behavioral and the neural level. Behaviorally, the studies have shown that the preceding encoding of other material can reduce recall of the target material, lead to intrusions from the preceding material

Conclusions

The research reviewed here suggests that PI release methods can oppose the problems that prior encoding of nontarget material induces for both the encoding and retrieval of target information. Prior nontarget encoding typically increases inattention, and PI release methods can neutralize such tendency, for instance, by inducing a switch to a superior encoding strategy or by resetting the encoding process. Prior nontarget encoding can also lead to a less focused memory search at test, but PI

References (129)

  • R.N.A. Henson et al.

    Functional magnetic resonance imaging of proactive interference during spoken cued recall

    Neuroimage

    (2002)
  • M.W. Howard et al.

    A distributed representation of temporal context

    J. Math. Psychol.

    (2002)
  • R.R. Hunt et al.

    How does distinctive processing reduce false recall?

    J. Mem. Lang.

    (2011)
  • W. Klimesch

    Alph A–B and oscillations, attention, and controlled access to stored information

    Trends Cogn. Sci.

    (2012)
  • R.J. Koppenaal

    Time changes in the strengths of A–B, A–C lists: spontaneous recovery?

    J. Verb. Learn. Verb. Behav.

    (1963)
  • A. Mecklinger et al.

    Dissociable brain mechanisms for inhibitory control: effects of interference content and working memory capacity

    Cogn. Brain Res.

    (2003)
  • J. Onton et al.

    Frontal midline EEG dynamics during working memory

    Neuroimage

    (2005)
  • S. Palva et al.

    New vistas for frequency band oscillations

    Trends Neurosci.

    (2007)
  • B. Pastötter et al.

    Oscillatory brain activity before and after an internal context change – evidence for a reset of encoding processes

    NeuroImage

    (2008)
  • L. Postman et al.

    Temporal changes in interference

    J. Verb. Learn. Verb. Behav.

    (1968)
  • F.R. Richter et al.

    Predicting the integration of overlapping memories by decoding mnemonic processing states during learning

    Neuroimage

    (2016)
  • L. Sahakyan et al.

    Can encoding differences explain the benefits of directed forgetting in the list method paradigm?

    J. Mem. Lang.

    (2003)
  • W.H. Saufley et al.

    Cue-selection interference in paired-associate learning

    J. Verb. Learn. Verb. Behav.

    (1964)
  • P.B. Sederberg et al.

    Oscillatory correlates of the primacy effect in episodic memory

    NeuroImage

    (2006)
  • J.R. Anderson

    A spreading activation theory of memory

    J. Verb. Learn. Verb. Behav.

    (1983)
  • S.K. Atwater

    Proactive inhibition and associative facilitation as affected by degree of prior learning

    J. Exp. Psychol.

    (1953)
  • A. Baddeley

    Human Memory. Theory and Practice

    (1990)
  • D. Badre et al.

    Frontal lobe mechanisms that resolve proactive interference

    Cereb. Cortex

    (2005)
  • A.J. Barnier et al.

    Directed forgetting of recently recalled autobiographical memories

    J. Exp. Psychol. Gen.

    (2007)
  • B.H. Basden et al.

    Directed forgetting: further comparisons of the item and list methods

    Memory

    (1996)
  • R.K. Biss et al.

    Interference from previous distraction disrupts older adults’ memory

    J. Gerontol. B

    (2013)
  • R.A. Bjork et al.

    The modification of short-term memory through instructions to forget

    Psychonom. Sci.

    (1968)
  • G.B. Blumenthal et al.

    Delayed release from proactive interference with meaningful material: how much do we remember after reading brief prose passages?

    J. Exp. Psychol. Hum. Learn. Mem.

    (1977)
  • G.H. Bower

    Mental imagery and associative learning

  • J. Brown

    Some tests of the decay theory of immediate memory

    Q. J. Exp. Psychol.

    (1958)
  • M.W. Calkins

    Association studies from the Harvard Psychological Laboratory

    Psychol. Rev.

    (1894)
  • A.J. Chanales et al.

    Interference between overlapping memories is predicted by neural states during learning

    Nat. Commun.

    (2019)
  • R.G. Crowder

    Principles of Learning and Memory

    (1976)
  • K.M. Dallett

    Proactive and retroactive inhibition in the AB, AB’ paradigm

    J. Exp. Psychol.

    (1964)
  • K.M. Dallett et al.

    Contextual stimuli and proactive inhibition

    J. Exp. Psychol.

    (1968)
  • F.J. DaPolito

    Proactive Effects with Independent Retrieval of Competing Responses

    (1966)
  • C.F. Darley et al.

    Effects of prior free recall testing on final recall and recognition

    J. Exp. Psychol.

    (1971)
  • P.F. Delaney et al.

    Remembering to forget: the amnesic effect of daydreaming

    Psychol. Sci.

    (2010)
  • M. D’Esposito et al.

    The neural substrate and temporal dynamics of interference effects in working memory as revealed by event-related functional MRI

    Proc. Natl. Acad. Sci.

    (1999)
  • K.M. Divis et al.

    Retrieval speeds context fluctuation: why semantic generation enhances later learning but hinders prior learning

    Mem. Cogn.

    (2014)
  • R.J. Dolan et al.

    Dissociating prefrontal and hippocampal function in episodic memory encoding

    Nature

    (1997)
  • E. Eckert et al.

    Memory for frequency as a function of environmental context

    Bull. Psychonom. Soc.

    (1984)
  • B.R. Ekstrand

    Effect of sleep on memory

    J. Exp. Psychol.

    (1967)
  • N.R. Ellis et al.

    Short-term memory for pictures and words by mentally retarded and nonretarded persons

    Am. J. Ment. Defic.

    (1985)
  • R. Engle

    Pupillary measurement and release from proactive inhibition

    Percept. Motor Skills

    (1975)
  • Cited by (17)

    • Contextual inference in learning and memory

      2023, Trends in Cognitive Sciences
    • The association between control of interference and white-matter integrity: A cross-sectional and longitudinal investigation

      2022, Neurobiology of Aging
      Citation Excerpt :

      Healthy human aging is associated with an average decline in memory functions (Koen & Yonelinas, 2014; Nyberg et al., 2020), including a decreased ability to control proactive interference (PI) in long-term memory (Healey et al., 2013; Wahlheim, 2014), and working memory (Loosli et al., 2016; Samrani & Persson, 2021). PI is a process where old and/or irrelevant information interferes with new, goal-relevant information (Hasher & Zacks, 1988; Jonides & Nee, 2006), and has been proposed as a major source of forgetting in both long-term (Kliegl & Bäuml, 2021) and working memory (Oberauer & Lewandowsky, 2008). Increased PI in working memory has been found to correlate with worse performance in several other cognitive domains including episodic memory, block design, fluency, and processing speed.

    • Automatic and effortful control of interference in working memory can be distinguished by unique behavioral and functional brain representations

      2022, NeuroImage
      Citation Excerpt :

      PI arises when past goal-relevant information interferes with current goal directed cognition and is commonly tested in experimental tasks by evoking familiarity from a task-irrelevant probe after recent exposure on a previous trial (Hasher and Zacks, 1988; Jonides and Nee, 2006). While PI has primarily been studied in the context of long-term memory (LTM; Kliegl and Bäuml, 2021), there is substantial evidence for PI also in working memory (WM; Jonides and Nee, 2006). WM is typically characterized by two distinct component processes: on-line maintenance of information within the focus of attention (FOA) and retrieval of information from long term memory (LTM).

    • Context differentiation and remindings in episodic memory updating

      2021, Psychology of Learning and Motivation - Advances in Research and Theory
      Citation Excerpt :

      Bjork (1978) and Finn (2017) in this series have discussed studies of mechanisms that differentiate or integrate contexts associated with episodic memories. Here, we extend on those reviews by synthesizing classic and contemporary studies that characterize the operation of both updating mechanisms (for a related approach, see Kliegl & Bäuml, 2021). Studies examining how relationships among memories affect subsequent retrieval accuracy have spanned broadly across domains in human memory research (for reviews, see Murayama, Miyatsu, Buchli, & Storm, 2014; Pastötter, Tempel, & Bäuml, 2017).

    View all citing articles on Scopus
    View full text