CoChat
Sign Up
Cited by 118Memory and Neural MechanismsMemory Processes and InfluencesCognitive Functions and Memory

Does mental context drift or shift?

Open access
DuBrow, Sarah, Rouhani, Nina, Niv, Yael, Norman, Kenneth A·2017·Current Opinion in Behavioral Sciences
DOI 10.1016/j.cobeha.2017.08.003PMC PMC5766042
Open this paper in CoChat — free account

Abstract

Theories of episodic memory have generally proposed that individual memory traces are linked together by a representation of context that drifts slowly over time. Recent data challenge the notion that contextual drift is always slow and passive. In particular, changes in one’s external environment or internal model induce discontinuities in memory that are reflected in sudden changes in neural activity, suggesting that context can shift abruptly. Furthermore, context change effects are sensitive to top-down goals, suggesting that contextual drift may be an active process. These findings call for revising models of the role of context in memory, in order to account for abrupt contextual shifts and the controllable nature of context change.

AI snapshot

Written from the paper’s full text.

TL;DR

This review argues that mental context is not only a slow, passive drift; it can also shift abruptly when situation models change or prediction errors signal a new latent cause. It proposes that contextual drift and shift may coexist and can be adaptively controlled by goals, while calling for a mechanistic model that integrates both.

Key findings

  • Traditional temporal context models account for recency and contiguity through a gradually changing context representation; supporting neural evidence includes hippocampal time cells changing over seconds, place-cell changes over days to weeks, and activity-pattern change relating to temporal order and distance.
  • Event boundaries induced by changes in stimuli, task goals, or narrative structure disrupt memory across the boundary; posterior medial network regions show relatively stable patterns within events and abrupt shifts at boundaries on the order of minutes, while the hippocampus tracks coarser narrative context.
  • Prediction errors and latent-cause inference can explain abrupt context shifts: large errors signal that the current situation has changed, segment experience, and reduce interference; cited findings include better memory for old line segments followed by prediction error and memory boosts from large reward prediction errors.
  • Context change is sensitive to top-down goals: segmentation tasks can improve memory up to one month later, associative encoding goals may slow drift, and directed-forgetting cues reduce fMRI activity linked to the prior scene context, with greater reduction predicting poorer recall.

Methods

  • Narrative review synthesizing prior behavioral, neuroimaging, computational, and animal studies; no new experiment, dataset, or primary analysis is reported.
  • The paper organizes evidence around slow contextual drift, abrupt shifts at event boundaries, latent-cause inference, and goal-directed control of context.

Limitations

  • The paper states that the main unresolved challenge is developing a mechanistic model that accounts for both shifting and drifting context; it also raises open questions about whether latent causes are mutually exclusive or can coexist, how conflicting predictions from multiple causes are reconciled, and how credit is assigned when updating causes.
  • Evident from the text: as a review, it does not provide a systematic search strategy, meta-analysis, or new empirical test, so its conclusions depend on the cited studies and remain theoretical.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1016/j.cobeha.2017.08.003

Cites

66 of 66 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 67 cited sources checked · 0 supported · 0 flagged

  • Premisecitation

    This association between neural pattern change and temporal memory has been found in the same regions that exhibit slow drift.

    “Critically, this relationship has been observed in the same regions that show slow drift [20*,21].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [20*,21]

      Not checked in the free run.

  • Premisecitation

    The recency effect can be explained by recall-period context being more similar to end-of-list context than to beginning-of-list context.

    “The recency effect, or the tendency to recall end-of-list items especially well, can be explained by the idea that context in the recall period is most similar to context at the end (vs. beginning) of the list.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [2] Howard MW Kahana MJ A distributed representation of temporal context J Math Psychol 2002 46 269 299 source

      Not checked in the free run.

  • Premisecitation

    Baldassano and colleagues found that activity in posterior medial network regions is better captured as alternating stable states that last minutes.

    “By fitting a model with this property to movie-viewing fMRI data, Baldassano and colleagues [35**] showed that regions in the PMN are well-described by shifts between relatively stable patterns, lasting on the order of minutes.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [35**]

      Not checked in the free run.

  • Premisecitation

    The contiguity effect can be explained by neighboring study items sharing context and thereby priming each other for recall.

    “The contiguity effect, or the tendency to transition in recall between items that were close to each other at study, can be explained by the idea that neighboring items from the study period share context, and therefore prime each other for recall.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [2] Howard MW Kahana MJ A distributed representation of temporal context J Math Psychol 2002 46 269 299 source

      Not checked in the free run.

  • Premisecitation

    The amount of change in neural activity patterns across events is associated with memory for temporal order and temporal distance.

    “Furthermore, the degree to which neural activity patterns change across events has been associated with memory for temporal order and distance [16–19].”
    Quote in the paperSubstantial claim0 of 4 sources checked
    • cites [16] DuBrow S Davachi L Temporal memory is shaped by encoding stability and intervening item reactivation J Neurosci 2014 34 PMC4198540 25319696

      Not checked in the free run.

    • cites [17] Jenkins LJ Ranganath C Distinct neural mechanisms for remembering when an event occurred Hippocampus 2016 26 554 559 26845069 PMC6467645 source

      Not checked in the free run.

    • cites [18] Jenkins LJ Ranganath C Prefrontal and medial temporal lobe activity at encoding predicts temporal context memory J Neurosci 2010 30 15558 15565 21084610 PMC3842495 source

      Not checked in the free run.

    • cites [19] Ezzyat Y Davachi L Similarity Breeds Proximity: Pattern Similarity within and across Contexts Is Related to Later Mnemonic Judgments of Temporal Proximity Neuron 2014 81 1179 89 24607235 PMC3983791 source

      Not checked in the free run.

  • May combine two claims. 11
  • Classified as probably not a substantial claim. 8

  • May combine two claims.

    Premisecitation

    Hippocampal time cells exhibit gradually changing activity over seconds during unfilled delays even when the animal remains in a fixed location, which is consistent with drift.

    “Hippocampal “time cells” show gradually changing activity patterns on the order of seconds during unfilled delays even when the animal’s location is fixed [6–9], consistent with drift.”
    Quote in the paperSubstantial claim0 of 4 sources checked
    • cites [6] Macdonald CJ Carrow S Place R Eichenbaum H Distinct hippocampal time cell sequences represent odor memories in immobilized rats J Neurosci 2013 33 14607 16 24005311 PMC3761059 source

      Not checked in the free run.

    • cites [7] Macdonald CJ Lepage KQ Eden UT Eichenbaum H Hippocampal “time cells” bridge the gap in memory for discontiguous events Neuron 2011 71 737 749 21867888 PMC3163062 source

      Not checked in the free run.

    • cites [8] Naya Y Suzuki WA Integrating what and when across the primate medial temporal lobe Science 2011 333 773 776 21817056 source

      Not checked in the free run.

    • cites [9] Eichenbaum H Time cells in the hippocampus: a new dimension for mapping memories Nat Rev Neurosci 2014 15 732 44 25269553 PMC4348090 source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Hippocampal place cells show sensitivity to timescales of days to weeks, and such sensitivity has also been observed independently of place coding.

    “Sensitivity to longer timescales on the order of days to weeks has also been observed in hippocampal place cells [10–12] and independently of place coding [13,14].”
    Quote in the paperSubstantial claim0 of 5 sources checked
    • cites [10] Mankin EA Sparks FT Slayyeh B Sutherland RJ Leutgeb S Leutgeb JK Neuronal code for extended time in the hippocampus Proc Natl Acad Sci U S A 2012 109 19462 7 23132944 PMC3511087 source

      Not checked in the free run.

    • cites [11] Ziv Y Burns LD Cocker ED Hamel EO Ghosh KK Kitch LJ El Gamal A Schnitzer MJ Long-term dynamics of CA1 hippocampal place codes Nat Neurosci 2013 16 264 6 23396101 PMC3784308 source

      Not checked in the free run.

    • cites [12] Lever C Wills T Cacucci F Burgess N O’Keefe J Long-term plasticity in hippocampal place-cell representation of environmental geometry Nature 2002 416 90 94 11882899 source

      Not checked in the free run.

    • cites [13] Rubin A Geva N Sheintuch L Ziv Y Hippocampal ensemble dynamics timestamp events in long-term memory Elife 2015 4 PMC4749549 26682652

      Not checked in the free run.

    • cites [14] Mankin EA Diehl GW Sparks FT Leutgeb S Leutgeb JK Hippocampal CA2 Activity Patterns Change over Time to a Larger Extent than between Spatial Contexts Neuron 2015 85 190 201 25569350 PMC4392894 source

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Premisecitation

    Computational models of temporal context propose that sensory inputs are gradually integrated into a slowly changing context representation that can be associated with individual item representations.

    “This property has been captured in computational models of temporal context [2–5], which posit that gradual integration of sensory inputs creates a slowly drifting context representation that can then be linked to representations of individual items.”
    Quote in the paperSubstantial claim0 of 4 sources checked
    • cites [2] Howard MW Kahana MJ A distributed representation of temporal context J Math Psychol 2002 46 269 299 source

      Not checked in the free run.

    • cites [3] Howard MW Fotedar MS Datey AV Hasselmo ME The temporal context model in spatial navigation and relational learning: toward a common explanation of medial temporal lobe function across domains Psychol Rev 2005 112 75 116 15631589 PMC1421376 source

      Not checked in the free run.

    • cites [4] Howard MW Shankar KH Aue WR Criss AH A distributed representation of internal time Psychol Rev 2015 122 24 53 25330329 source

      Not checked in the free run.

    • cites [5] Estes WK Statistical theory of spontaneous recovery and regression Psychol Rev 1955 62 145 154 14371893 source

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Premisecitation

    A posterior medial network comprising parahippocampal cortex, retrosplenial cortex, and other default network regions integrates internal and external information to represent current event features.

    “Researchers have argued that a posterior medial network (PMN), including the parahippocampal cortex, retrosplenial cortex, and other regions of the default network, integrates internal and external information in order to represent the features of the current event [32].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [32] Ranganath C Ritchey M Two cortical systems for memory-guided behaviour Nat Rev Neurosci 2012 13 713 26 22992647 source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    In free recall of random word lists, the recency and contiguity effects are strong and reliable, and both can be accounted for by a slowly drifting mental context bound to items.

    “When freely recalling random lists of words, two effects have been identified as particularly strong and reliable – the recency effect and the contiguity effect – both of which can be explained by a slowly drifting mental context representation bound to items in memory [2].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [2] Howard MW Kahana MJ A distributed representation of temporal context J Math Psychol 2002 46 269 299 source

      Not checked in the free run.

  • Classified as probably not a substantial claim. May combine two claims.

    Premisecitation

    Rouhani and colleagues reported that surprising rewards enhanced multiple memory tests, which fits the view that segmentation rose and interference fell.

    “Rouhani and colleagues [48] further showed that large reward prediction errors boosted performance on multiple memory measures, consistent with the idea that prediction errors led to increased segmentation and reduced interference.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [48] Rouhani N Norman KA Niv Y Dissociable effects of surprising rewards on learning and memory bioRxiv 2017 10.1101/111070 PMC6117220 29553767

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    A neuroimaging study found that a forget cue for the first list reduced fMRI signals tied to the scene context, and larger reductions predicted poorer later recall of that list.

    “When instructed to forget the first list, fMRI activity associated with the scene context dropped, and the extent of that drop was related to participants’ subsequent failure to recall items from that list [61*, Figure 2].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [61*, Figure 2]

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    The hippocampus responds to story-level context more broadly than the posterior medial network, keeping two interwoven narratives distinct.

    “In a different study, the hippocampus was found to be sensitive to narrative context at an even coarser level than the PMN, separately representing two interleaved story lines [36*].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [36*]

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Interpretationcitation

    Mismatches between expected and actual outcomes might produce boundaries and later shifts in context, aligning with findings that such mismatches promote segmentation in learning.

    “One possibility is that prediction errors, or discrepancies between expected and actual outcomes, generate event boundaries and ensuing context shifts [38–40], consistent with evidence that prediction errors can drive segmentation in learning [41,42].”
    Quote in the paperSubstantial claim0 of 5 sources checked
    • cites [38] Reynolds JR Zacks JM Braver TS A computational model of event segmentation from perceptual prediction Cogn Sci 2007 31 613 643 21635310 source

      Not checked in the free run.

    • cites [39] Zacks JM Kurby CA Eisenberg ML Haroutunian N Prediction Error Associated with the Perceptual Segmentation of Naturalistic Events J Cogn Neurosci 2011 23 4057 4066 21671745 PMC8653780 source

      Not checked in the free run.

    • cites [40] Zacks JM Speer NK Swallow KM Braver TS Reynolds JR Event Perception: A Mind/Brain Perspective Psychol Bull 2007 133 273 293 17338600 PMC2852534 source

      Not checked in the free run.

    • cites [41] Redish AD Jensen S Johnson A Kurth-Nelson Z Reconciling reinforcement learning models with behavioral extinction and renewal: implications for addiction, relapse, and problem gambling Psychol Rev 2007 114 784 805 17638506 source

      Not checked in the free run.

    • cites [42] Gershman SJ Niv Y Learning latent structure: Carving nature at its joints Curr Opin Neurobiol 2010 20 251 256 20227271 PMC2862793 source

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Premisecitation

    The temporal context framework can explain boundary effects on memory by positing a rapid rise in drift rate.

    “Within the temporal context framework, an abrupt increase in contextual drift rate can explain boundary effects on memory [37], but what might drive this increase?”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [37] Polyn SM Norman KA Kahana MJ A context maintenance and retrieval model of organizational processes in free recall Psychol Rev 2009 116 129 156 19159151 PMC2630591 source

      Not checked in the free run.

  • May combine two claims.

    Interpretationcitation

    If sensory input changes suddenly while the task goal stays fixed, hippocampus and lateral prefrontal cortex may link information across that change, possibly by holding internal context steady.

    “When sensory cues change abruptly but the goal stays the same, the hippocampus and lateral prefrontal cortex may help bind information that spans the change [56], perhaps by maintaining a stable internal context [57].”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [56] DuBrow S Davachi L Temporal binding within and across events Neurobiol Learn Mem 2016 134 107 114 27422018 PMC5018468 source

      Not checked in the free run.

    • cites [57] Polyn SM Kahana MJ Memory search and the neural representation of context Trends Cogn Sci 2008 12 24 30 18069046 PMC2839453 source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Orbitofrontal cortex tracks hidden states and encodes how likely each candidate latent cause is at a given moment in inference.

    “Neurally, the orbitofrontal cortex, which is sensitive to unobservable states [51], has been shown to represent the probability that each latent cause is currently active during the inference process [52].”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [51] Schuck NW Cai MB Wilson RC Niv Y Human Orbitofrontal Cortex Represents a Cognitive Map of State Space Neuron 2016 91 1402 1412 27657452 PMC5044873 source

      Not checked in the free run.

    • cites [52] Chan SCY Niv Y Norman KA A Probability Distribution over Latent Causes, in the Orbitofrontal Cortex J Neurosci 2016 36 7817 28 27466328 PMC4961772 source

      Not checked in the free run.

  • May combine two claims.

    Interpretationreasoning

    The hippocampus and posterior medial network may encode broad event structure, maintaining integration during stable periods and changing quickly at new situations.

    “Thus, the hippocampus and PMN may support high-level structure, integrating within stable events and rapidly shifting when a new situation is encountered.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Interpretationreasoning

    In passive tasks with unstructured material, gradual drift might come from inferring structure over random, unrelated items.

    “Indeed, it is possible that the slow, steady drift observed in unstructured passive encoding tasks is the result of an inference process over arbitrary, unrelated stimuli.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Interpretationcitation

    Both gradual and abrupt context changes might operate at once: certain context codes could change slowly and survive boundaries, whereas others track the inferred situation and can switch quickly.

    “Having said this, it is important to note that drifting and shifting contexts may exist simultaneously in the brain; there are likely to be some contextual representations that drift passively over time and persist across event boundaries [22,56] and other contextual representations that are more sensitive to the inferred situation and thus more likely to shift suddenly.”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [22] Chan SCY Applegate MC Morton NW Lingering representations of stimuli influence recall organization Neuropsychologia 2017 97 72 82 28132858 PMC5354630 source

      Not checked in the free run.

    • cites [56] DuBrow S Davachi L Temporal binding within and across events Neurobiol Learn Mem 2016 134 107 114 27422018 PMC5018468 source

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Premisecitation

    Latent cause inference theory proposes that individuals store many statistical models of the world and determine which unseen cause best explains what they currently sense.

    “According to a recent theory of latent cause inference, people build up a library of statistical models of their environment, and then infer which is the hidden, or “latent,” cause of their current sensory observations [43–46].”
    Quote in the paperSubstantial claim0 of 4 sources checked
    • cites [43] Gershman SJ Blei DM Niv Y Context, learning, and extinction Psychol Rev 2010 117 197 209 20063968 source

      Not checked in the free run.

    • cites [44]

      Not checked in the free run.

    • cites [45] Soto FA Gershman SJ Niv Y Explaining compound generalization in associative and causal learning through rational principles of dimensional generalization Psychol Rev 2014 121 526 558 25090430 PMC4165620 source

      Not checked in the free run.

    • cites [46] Gershman SJ Niv Y Exploring a latent cause theory of classical conditioning Learn Behav 2012 40 255 268 22927000 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S2 (checkable).

    Interpretationreasoning

    A better proposal may be that several latent causes can be active together, not just one at a time.

    “For instance, Figure 1 depicts one latent cause switching to another, suggesting that latent causes are mutually exclusive, but it seems more reasonable to posit that multiple latent causes can coexist (i.e., you may continue to think about, or even discuss work-related issues at the surprise party, making both the “work” and “party” latent causes relevant).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    The evidence reviewed shows that both detecting situational change and intentionally altering one's context strongly influence memory and its neural basis.

    “We have reviewed evidence that inferring changes in the current situation and deliberately changing our mental context have substantial effects on memory behavior and the brain.”
    Quote in the paperSubstantial claim
18 of 18

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • The new variant of TCM differs from original TCM in that learning is error-driven rather than purely Hebbian, so association strength only grows if there is a discrepancy between predicted and observed stimuli.

    “First, learning is error-driven rather than purely Hebbian, which means that association strength only grows if there is a discrepancy between predicted and observed stimuli”

    From Predictive Representations: Building Blocks of Intelligence (2024) · cites this paper as DuBrow et al., 2017

    The section “Inferring change in the state of the world” says prediction errors may cause context shifts and segment memories, but it does not describe an error-driven association-strength rule. Searches for Hebbian learning and temporal-context-model association updates found no passage comparing a new TCM variant with original TCM or saying associations grow only when predictions differ from observations. Page numbers are unavailable in our copy. (LLM, conf=0.76)

    Experimental — accuracy not yet measured

Cites

Flagged first, then by citations

Cited by

Sorted by citations

Related

Authors
DuBrow, S., Rouhani, N., Niv, Y., & Norman, K. A.
Affiliations
Princeton University (US)
Corresponding author
Norman, Kenneth A
ORCID
Venue
Current Opinion in Behavioral Sciences
Publisher
Elsevier BV
Published
2017-10
Volume
17
Pages
141-146
ISSN
2352-1546 (ISSN-L)
DOI
10.1016/j.cobeha.2017.08.003
PMC
PMC5766042
OpenAlex
W2750869205
Semantic Scholar
fc68452aa5815b5fe22670ee18f73d3f1ed52ec5
Cited by
118
References
65
Citations by year
2026: 10 · 2025: 20 · 2024: 18 · 2023: 14 · 2022: 16 · 2021: 14 · 2020: 13 · 2019: 8 · 2018: 5
FWCI
0.79
Type
journal article
Language
en
Open access
Licence
www.elsevier.com/tdm/userlicense/1.0/ (tdm)
Locations
  • Current Opinion in Behavioral Sciences (publishedVersion)
  • PubMed (publishedVersion)
  • Europe PMC (PubMed Central) (acceptedVersion)
  • PubMed Central (acceptedVersion)
Funders
  • National Institutes of Health (5r01mh112357-05)
  • National Institute of Mental Health (R01-MH112357)
  • National Institute on Drug Abuse (R01DA042065)
Topics
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Memory and Neural Mechanisms
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Memory Processes and Influences
  • Social Sciences › Psychology › Experimental and Cognitive Psychology › Cognitive Functions and Memory
Keywords
episodic memory, memory context, memory traces

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-05.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-05.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review