Oscillatory patterns in temporal lobe reveal context reinstatement during memory search

Oscillatory patterns in temporal lobe reveal context reinstatement during memory search Manning Polyn Baltuch Litt Kahana

Tag layer of the DHSS Meta-Tagging Project. The article text is reproduced from the staged copy named in sourceDesc and is not altered; the tags are held separately in standOff so that overlapping and nested spans are preserved.

Oscillatory patterns in temporal lobe reveal context reinstatement during memory search 2011 Proceedings of the National Academy of Sciences 10.1073/pnas.1015174108 Manning2011-oscillatory-patterns-in-temporal-lobe-reveal Manning2011-oscillatory-patterns-in-temporal-lobe-reveal.txt, the project's staged copy f51b67e9c4a7e3e0839e7c963daa928adaea9715 not-redistributable neuroscience

the paper's own words

the tag layer; a sentence written by the tagger about the paper, always anchored to a quotation but not itself a quotation

0 spans located by searching the stored text for the quote's exact words, allowing any whitespace between them whether the gradually drifting neural state present while a word is studied returns when that word is recalled, and returns with it the states of its list neighbours Quotation withheld because this record is not cleared for redistribution. is there direct neurophysiological evidence for contextual reinstatement, which behavioural work has only supported indirectly Quotation withheld because this record is not cleared for redistribution. finds a neural signature of temporal context reinstatement — recall-time activity resembles study-time activity for the recalled item and, in graded fashion, for its neighbours — and shows its strength predicts each patient's behavioural contiguity effect Quotation withheld because this record is not cleared for redistribution. existence Quotation withheld because this record is not cleared for redistribution. temporal context reinstatement during memory search Quotation withheld because this record is not cleared for redistribution. electrophysiology electrophysiology Quotation withheld because this record is not cleared for redistribution. mechanisms Recalling an item revives the contextual state bound to it at study, and that retrieved context then cues other items encoded in similar contexts, which is what produces the contiguity effect. Quotation withheld because this record is not cleared for redistribution. mechanisms A slowly drifting contextual state becomes associated with each event as it is experienced, so that later recall of the event reinstates that state. Quotation withheld because this record is not cleared for redistribution. context is operationalised by a property the theory predicts — slow drift — so the features analysed are exactly those that could carry a context signal Quotation withheld because this record is not cleared for redistribution. the circularity that would create is blocked by splitting selection from test: the filter is fitted on study events only, and the hypothesis is tested on recall events Quotation withheld because this record is not cleared for redistribution. the filter's cost is stated rather than hidden — removing content information also removes the forward asymmetry the behavioural effect shows, so a known feature of the phenomenon is invisible by construction Quotation withheld because this record is not cleared for redistribution. the discriminating prediction is bidirectional decay: content reinstatement alone predicts a forward boost, autocorrelated noise predicts no lag structure, and only context predicts symmetric decay Quotation withheld because this record is not cleared for redistribution. rival accounts are built and run rather than argued against — three neural network simulations using the same presented and recalled items as the real analysis Quotation withheld because this record is not cleared for redistribution. the rehearsal alternative is tested through a consequence it would have elsewhere: if rehearsal drove the effect, neural contiguity should track the primacy effect, and it does not Quotation withheld because this record is not cleared for redistribution. individual differences carry the inference: the same patients supply both the neural measure and the behavioural one, so the two can be correlated across people rather than compared as group means Quotation withheld because this record is not cleared for redistribution. a gradual-drift result is treated as insufficient on its own, because ordinary autocorrelated processes would produce it too Quotation withheld because this record is not cleared for redistribution. only negative lags are used for the reinstatement measure, because forward lags cannot separate content from context Quotation withheld because this record is not cleared for redistribution. the inclusion threshold is justified as an explicit trade-off between dimensionality and sample size rather than presented as a natural cutoff Quotation withheld because this record is not cleared for redistribution. Parietal and occipital results were inconclusive because of insufficient data. what this limits: the parietal and occipital results: OUR FINDINGS IN THE PARIETAL AND OCCIPITAL LOBES WERE INCONCLUSIVE DUE TO INSUFFICIENT DATA Quotation withheld because this record is not cleared for redistribution. Null results outside temporal lobe do not establish that other regions are uninvolved. what this limits: the extratemporal null: OUR INABILITY TO FIND NEURAL SIGNATURES OF CONTEXT REINSTATEMENT IN EXTRATEMPORAL BRAIN REGIONS DOES NOT NECESSARILY MEAN THAT THOSE REGIONS ARE NOT INVOLVED IN CONTEXT REINSTATEMENT Quotation withheld because this record is not cleared for redistribution. The analysis cannot resolve structures within the coarse regions of interest. what this limits: any structure-level claim: OUR ANALYSIS DOES NOT DISTINGUISH BETWEEN STRUCTURES CONTAINED WITHIN THE REGIONS OF INTEREST WE EXAMINED Quotation withheld because this record is not cleared for redistribution. A large share of recording sessions was discarded for having too few candidate context features. what this limits: the analysed sample: WE FURTHER EXCLUDED AN ADDITIONAL 68 SESSIONS IN WHICH FEWER THAN FIVE CANDIDATE CONTEXT FEATURES WERE IDENTIFIED Quotation withheld because this record is not cleared for redistribution. a null in other regions is not read as evidence of absence duplicate of: limitations[1] · what this limits: the same scope as limitations[1] - absence of evidence is not read as evidence of absence Quotation withheld because this record is not cleared for redistribution. the regions of interest are too coarse to identify a structure what this limits: the same scope as limitations[2] - the regions of interest are too coarse to name a structure Quotation withheld because this record is not cleared for redistribution. two of the four regions could not be assessed what this limits: the same scope as limitations[0] - two of the four lobes could not be assessed Quotation withheld because this record is not cleared for redistribution. substantial attrition: features were found in 132 of 144 sessions, then 68 further sessions were dropped for having fewer than five what this limits: the full attrition behind every result: FEATURES ... WERE IDENTIFIED IN 132 (OF 144) RECORDING SESSIONS, then 68 MORE WERE EXCLUDED for having fewer than five features Quotation withheld because this record is not cleared for redistribution. rehearsal-based and context-based accounts are conceded to be hard to tell apart in principle what this limits: any claim that separates the two accounts: REHEARSAL-BASED MODELS HAVE BEEN SHOWN TO BE DIFFICULT TO DISTINGUISH FROM CONTEXT-BASED MODELS Quotation withheld because this record is not cleared for redistribution. the primacy control rests on rehearsal being the mechanism of primacy, which the paper acknowledges is not settled what this limits: the primacy-based control: ALTHOUGH REHEARSAL IS ONE OF THE HYPOTHESIZED MECHANISMS UNDERLYING PRIMACY, WE RECOGNIZE THAT OTHER FACTORS, SUCH AS ENHANCED ATTENTION TO EARLY LIST ITEMS, MAY ALSO CONTRIBUTE TO PRIMACY Quotation withheld because this record is not cleared for redistribution. electrode placement was set by clinical need, not by the hypothesis what this limits: which regions could be sampled at all: Quotation withheld Quotation withheld because this record is not cleared for redistribution. principal components analysis and the Kaiser criterion, used to turn thousands of correlated frequency-electrode measurements into a small orthogonal basis in which event-to-event similarity is preserved Quotation withheld because this record is not cleared for redistribution. a 20-neuron simulated network, used not to model the brain but to derive what each competing account would predict for the lag profile Quotation withheld because this record is not cleared for redistribution. the Hilbert transform on band-passed signals, with power computed across the whole session before segmentation specifically to avoid edge artifacts Quotation withheld because this record is not cleared for redistribution. the analysis checked against rehearsal during study as an alternative source of the temporal context effect eliminates: rehearsal, per se, as the explanation of the neural context-reinstatement finding - the paper reports the check came back against it · outcome (stated by the tagger, no quotation): held Quotation withheld because this record is not cleared for redistribution. the feature vectors given enough dimensions that a neural signature of context reinstatement could be observed at all if it were there eliminates: a null that means the measure could not have detected the effect rather than that the effect is absent · outcome (stated by the tagger, no quotation): held Quotation withheld because this record is not cleared for redistribution. feature vectors feature vectors Quotation withheld because this record is not cleared for redistribution. Rehearsal-based models → context-based models THE TWO CANNOT BE SEPARATED IN THIS STUDY - the DESIGN form / It bounds what the paper's own results can mean, and it lives in a methods paragraph where nothing downstream will find it. TWO MODEL CLASSES THAT CANNOT BE TOLD APART BY THE DATA, with the reason given: a context-based mechanism is needed to explain the rehearsal pattern, so the two are entangled by construction. Quotation withheld because this record is not cleared for redistribution. context — as external environmental conditions versus as an internal, ever-changing mental state; the paper's theory requires both Quotation withheld because this record is not cleared for redistribution. context — operationalised: the principal components whose activity is positively autocorrelated during study, and nothing else Quotation withheld because this record is not cleared for redistribution. reinstatement — of content (the item's own representation) versus of context (the states of its neighbours); the whole design exists to separate these Quotation withheld because this record is not cleared for redistribution. contiguity effect — behavioural (transitions between neighbouring items at recall) versus neural (similarity declining with absolute lag); the paper coins the second and correlates the two Quotation withheld because this record is not cleared for redistribution. contextual features — as environmental cues, as recently studied items, as internal mental states, or as a process evolving randomly; four readings the founding theories left open Quotation withheld because this record is not cleared for redistribution. gradual change — as evidence for context versus as the signature of any autocorrelated process; the paper insists the first requires more than the second Quotation withheld because this record is not cleared for redistribution. method: electrocorticography in epilepsy patients during delayed free recall; oscillatory power in five bands per electrode, reduced by PCA, with components selected for slow autocorrelation during study and tested for similarity to recall-time activity as a function of lag a second reading DIFFERING from the recorded value. Recorded: electrophysiology — note: a second reading, verbatim-verified, disagreeing with the value on the record. Not applied - a person decides. Quotation withheld because this record is not cleared for redistribution. phenomenon: temporal context reinstatement during episodic memory search a second reading DIFFERING from the recorded value. Recorded: temporal context reinstatement during memory search — note: a second reading, verbatim-verified, disagreeing with the value on the record. Not applied - a person decides. Quotation withheld because this record is not cleared for redistribution. builds-on → retrieved context theories of episodic memory (refs 1, 5, 6, 24) Quotation withheld because this record is not cleared for redistribution. builds-on → Tulving (ref 2) - episodic retrieval as mental time travel Quotation withheld because this record is not cleared for redistribution. supports → retrieved context theories of episodic memory (refs 1, 5, 6, 24), by supplying the neurophysiological evidence they lacked Quotation withheld because this record is not cleared for redistribution. extends → prior demonstrations that an item's own representation is reinstated at recall (refs 25, 26), by showing the neighbours' representations return too Quotation withheld because this record is not cleared for redistribution. contradicts → theories deriving contiguity from direct item-to-item associations formed within seconds (refs 11–14) Quotation withheld because this record is not cleared for redistribution. critiques → rehearsal-based accounts of the neural contiguity effect, via the absent primacy correlation Quotation withheld because this record is not cleared for redistribution. supports → Tulving's characterisation of episodic retrieval as mental time travel Quotation withheld because this record is not cleared for redistribution. builds-on → the hypothesis that medial temporal lobe and prefrontal cortex maintain and update temporal context Quotation withheld because this record is not cleared for redistribution. context reinstatement —causes→ contiguity effect context reinstatement contiguity effect strength of the relation: 0.8 Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but models 'SUGGEST' that contextual reinstatement underlies the contiguity effect; causes is hedged. Quotation withheld because this record is not cleared for redistribution. contiguity effect —is-a→ temporal clustering contiguity effect temporal clustering strength of the relation: 0.95 Quotation withheld because this record is not cleared for redistribution. episodic memory —contrasted-with→ semantic memory episodic memory semantic memory strength of the relation: 0.95 Quotation withheld because this record is not cleared for redistribution. content reinstatement —distinct-from→ context reinstatement content reinstatement context reinstatement strength of the relation: 1 Quotation withheld because this record is not cleared for redistribution. temporal context —operationalised-as→ autocorrelation temporal context autocorrelation strength of the relation: 1 Quotation withheld because this record is not cleared for redistribution. temporal context —operationalised-as→ feature vectors temporal context feature vectors strength of the relation: 0.95 Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but endpoint by anaphora: autocorrelation is named 'a defining feature of temporal context' (previous sentence); operationalised-as is loose. Quotation withheld because this record is not cleared for redistribution. context reinstatement —operationalised-as→ neural similarity context reinstatement neural similarity strength of the relation: 0.95 Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but only the context-reinstatement model PREDICTS the observed neural-similarity pattern; operationalised-as is loose. Quotation withheld because this record is not cleared for redistribution. free recall —contrasted-with→ cued recall free recall cued recall strength of the relation: 0.9 Quotation withheld because this record is not cleared for redistribution. medial temporal lobe —part-of→ temporal lobe medial temporal lobe temporal lobe strength of the relation: 0.95 Quotation withheld because this record is not cleared for redistribution. regression regression Quotation withheld because this record is not cleared for redistribution. t-test t-test Quotation withheld because this record is not cleared for redistribution. t-test t-test Quotation withheld because this record is not cleared for redistribution. correlation correlation Quotation withheld because this record is not cleared for redistribution. principal components analysis principal components analysis Quotation withheld because this record is not cleared for redistribution. Kaiser criterion for component retention Kaiser criterion for component retention Quotation withheld because this record is not cleared for redistribution. autocorrelation autocorrelation Quotation withheld because this record is not cleared for redistribution. z-score standardisation z-score standardisation Quotation withheld because this record is not cleared for redistribution. standard error of the mean standard error of the mean Quotation withheld because this record is not cleared for redistribution. t test t test Quotation withheld because this record is not cleared for redistribution. paired-sample t test paired-sample t test Quotation withheld because this record is not cleared for redistribution. β coefficient β coefficient Quotation withheld because this record is not cleared for redistribution. principal components analysis principal components analysis Quotation withheld because this record is not cleared for redistribution. Kaiser criterion Kaiser criterion Quotation withheld because this record is not cleared for redistribution. z -transformed z -transformed Quotation withheld because this record is not cleared for redistribution. SEM SEM Quotation withheld because this record is not cleared for redistribution. normalized dot product normalized dot product Quotation withheld because this record is not cleared for redistribution. Neural similarity between pairs of studied words fell off as the distance between their serial positions grew, showing that the ECoG signal drifts gradually across a list. Quotation withheld because this record is not cleared for redistribution. Brain activity recorded just before a word was recalled was more similar to activity during study of that same word than to activity during study of any other list word. Quotation withheld because this record is not cleared for redistribution. Similarity between recall activity and study activity declined with absolute lag in both the forward and backward directions, the pattern predicted by context reinstatement rather than item-content reinstatement. Quotation withheld because this record is not cleared for redistribution. Participants whose neural reinstatement effect was stronger also recalled neighbouring list items successively more often, linking the neural signature to the behavioural contiguity effect. Quotation withheld because this record is not cleared for redistribution. Restricting the analysis to temporal lobe electrodes still produced the graded decrease of similarity with lag in both directions. Quotation withheld because this record is not cleared for redistribution. The frontal-lobe reinstatement signature showed no relationship to how strongly participants clustered their recalls in time, unlike the temporal-lobe signature. Quotation withheld because this record is not cleared for redistribution. Primacy strength was unrelated to neural contiguity, arguing against rehearsal during study as the source of the reinstatement effect. Quotation withheld because this record is not cleared for redistribution. The forward asymmetry that characterises behavioural contiguity was absent from the neural similarity data and from the context-reinstatement simulation. Quotation withheld because this record is not cleared for redistribution. Among the autocorrelated-noise, content-reinstatement and context-reinstatement simulations, only the context-reinstatement network reproduced the observed symmetric decline of similarity with lag. Quotation withheld because this record is not cleared for redistribution. neural similarity between two study events fell with the positional distance between them, so the recording drifts gradually across a list Quotation withheld because this record is not cleared for redistribution. recall-time activity was most similar to study of the same word, at lag zero Quotation withheld because this record is not cleared for redistribution. similarity also declined with absolute lag in both directions — the signature that separates context from content Quotation withheld because this record is not cleared for redistribution. patients with a stronger neural reinstatement signature showed a stronger behavioural contiguity effect Quotation withheld because this record is not cleared for redistribution. the temporal-lobe signature of context reinstatement correlated with the temporal clustering of recalls (r = 0.48) Quotation withheld because this record is not cleared for redistribution. the frontal lobe signature did not track behavioural clustering Quotation withheld because this record is not cleared for redistribution. neural contiguity did not correlate with the primacy effect, arguing against a rehearsal account Quotation withheld because this record is not cleared for redistribution. the forward asymmetry of the behavioural contiguity effect was absent from the neural data and from the context simulation alike Quotation withheld because this record is not cleared for redistribution. only the context reinstatement simulation reproduced bidirectional decay; noise and content models did not Quotation withheld because this record is not cleared for redistribution. participants recalled 22.7% of studied words, with about a third of all responses being repetitions or intrusions Quotation withheld because this record is not cleared for redistribution. of temporal lobe electrodes, 13.9% were in hippocampus and 27.0% in other medial temporal structures Quotation withheld because this record is not cleared for redistribution. contiguity effect contiguity effect Quotation withheld because this record is not cleared for redistribution. feature vectors feature vectors Quotation withheld because this record is not cleared for redistribution. lag lag Quotation withheld because this record is not cleared for redistribution. temporal clustering score temporal clustering score Quotation withheld because this record is not cleared for redistribution. context context Quotation withheld because this record is not cleared for redistribution. contiguity effect contiguity effect Quotation withheld because this record is not cleared for redistribution. lag lag Quotation withheld because this record is not cleared for redistribution. similarity similarity Quotation withheld because this record is not cleared for redistribution. feature vectors feature vectors Quotation withheld because this record is not cleared for redistribution. temporal clustering score temporal clustering score Quotation withheld because this record is not cleared for redistribution. context representation context representation Quotation withheld because this record is not cleared for redistribution. oscillatory power oscillatory power Quotation withheld because this record is not cleared for redistribution. single-digit addition problem single-digit addition problem Quotation withheld because this record is not cleared for redistribution. lag lag Quotation withheld because this record is not cleared for redistribution. neural similarity neural similarity Quotation withheld because this record is not cleared for redistribution. temporal clustering score temporal clustering score Quotation withheld because this record is not cleared for redistribution. mean oscillatory power mean oscillatory power Quotation withheld because this record is not cleared for redistribution. serial position serial position Quotation withheld because this record is not cleared for redistribution. regions of interest regions of interest Quotation withheld because this record is not cleared for redistribution. 69 — number of neurosurgical epilepsy patients whose intracranial recordings were analysed Quotation withheld because this record is not cleared for redistribution. 22.7 ± 1.0% — proportion of studied words that participants recalled, mean ± SEM Quotation withheld because this record is not cleared for redistribution. r = 0.42, P = 0.007 — across-participant correlation between the whole-brain neural reinstatement measure and behavioural temporal clustering Quotation withheld because this record is not cleared for redistribution. r = 0.48, P = 0.03 — correlation between the temporal-lobe reinstatement signature and behavioural temporal clustering Quotation withheld because this record is not cleared for redistribution. 1,815 / 1,737 / 512 / 138 — electrodes assigned to the temporal, frontal, parietal and occipital regions of interest Quotation withheld because this record is not cleared for redistribution. 132 (of 144) — recording sessions in which gradually changing candidate context features could be identified Quotation withheld because this record is not cleared for redistribution. 13.9% / 27.0% / 59.1% — share of temporal-lobe electrodes in hippocampus, in other medial temporal lobe structures, and elsewhere in temporal lobe Quotation withheld because this record is not cleared for redistribution. 69 patients — neurosurgical patients recorded Quotation withheld because this record is not cleared for redistribution. 39 participants — those with at least five autocorrelated features, on whom the main tests were run Quotation withheld because this record is not cleared for redistribution. 132 of 144 sessions — recording sessions in which context features were found at all Quotation withheld because this record is not cleared for redistribution. 68 sessions — further sessions excluded for having fewer than five candidate features Quotation withheld because this record is not cleared for redistribution. δ: 2–4 Hz, θ: 4–8 Hz, α: 8–12 Hz, β: 12–30 Hz, and γ: 30–99 Hz — the five frequency bands in which power was measured Quotation withheld because this record is not cleared for redistribution. 200–1,600 ms — the study-event window relative to word onset Quotation withheld because this record is not cleared for redistribution. −600 to 200 ms — the recall-event window relative to vocalization — mostly before the word is spoken Quotation withheld because this record is not cleared for redistribution. 22.7 ± 1.0% — proportion of studied words recalled Quotation withheld because this record is not cleared for redistribution. 32.4 ± 2.6% — share of all responses that were repetitions or incorrect Quotation withheld because this record is not cleared for redistribution. r = 0.42, P = 0.007 — correlation between neural reinstatement and behavioural contiguity across patients Quotation withheld because this record is not cleared for redistribution. r = 0.48, P = 0.03 — the same correlation restricted to temporal lobe electrodes Quotation withheld because this record is not cleared for redistribution. 1,815 / 1,737 / 512 / 138 electrodes — electrode counts in temporal, frontal, parietal and occipital regions Quotation withheld because this record is not cleared for redistribution. 15 or 20 nouns — list length in the free recall task Quotation withheld because this record is not cleared for redistribution. 1,600 ms / 800- to 1,200-ms — word presentation duration and interstimulus interval Quotation withheld because this record is not cleared for redistribution. 18 s / 45 s — arithmetic distractor duration and the recall period Quotation withheld because this record is not cleared for redistribution. 1 to 4 wk — duration electrodes remained implanted Quotation withheld because this record is not cleared for redistribution.

The full text of this article is not redistributable, so it is not included here. The evidence sentences are withheld with it; the tag layer and its grounding status remain. The article itself is at https://doi.org/10.1073/pnas.1015174108.