FRINK Linked Data Fragments server

Matches in SemOpenAlex for { <https://semopenalex.org/work/W2896767726> ?p ?o ?g. }

• W2896767726 endingPage "3392.e6" @default.
• W2896767726 startingPage "3383" @default.
• W2896767726 abstract "•We show evidence for theta phase encoding-retrieval models in human EEG data•Neural indices of memory reactivation from pattern classifiers fluctuate at 7 or 8 Hz•Times of maximal memory reactivation are preceded by a consistent theta phase•Optimal phase for encoding and retrieval is shifted by approximately 180° Computational models and in vivo studies in rodents suggest that the hippocampal system oscillates between states that are optimal for encoding and states that are optimal for retrieval. Here, we show that in humans, neural signatures of memory reactivation are modulated by the phase of a theta oscillation. Electroencephalography (EEG) was recorded while participants were cued to recall previously learned word-object associations, and time-resolved pattern classifiers were trained to detect neural reactivation of the target objects. Classifier fidelity rhythmically fluctuated at 7 or 8 Hz and was modulated by theta phase across the entire recall period. The phase of optimal classification was shifted approximately 180° between encoding and retrieval. Inspired by animal work, we then computed “classifier-locked averages” to analyze how ongoing theta oscillations behaved around the time points at which the classifier indicated memory retrieval. We found strong theta (7 or 8 Hz) phase consistency approximately 300 ms before the time points of maximal neural memory reactivation. Our findings provide important evidence that the neural signatures of memory retrieval fluctuate and are time locked to the phase of an ongoing theta oscillation. Computational models and in vivo studies in rodents suggest that the hippocampal system oscillates between states that are optimal for encoding and states that are optimal for retrieval. Here, we show that in humans, neural signatures of memory reactivation are modulated by the phase of a theta oscillation. Electroencephalography (EEG) was recorded while participants were cued to recall previously learned word-object associations, and time-resolved pattern classifiers were trained to detect neural reactivation of the target objects. Classifier fidelity rhythmically fluctuated at 7 or 8 Hz and was modulated by theta phase across the entire recall period. The phase of optimal classification was shifted approximately 180° between encoding and retrieval. Inspired by animal work, we then computed “classifier-locked averages” to analyze how ongoing theta oscillations behaved around the time points at which the classifier indicated memory retrieval. We found strong theta (7 or 8 Hz) phase consistency approximately 300 ms before the time points of maximal neural memory reactivation. Our findings provide important evidence that the neural signatures of memory retrieval fluctuate and are time locked to the phase of an ongoing theta oscillation. Our episodic memory defines us by storing a record of our past experiences and allowing us to consciously access these records. It is widely agreed that the hippocampus and neocortical areas work in conjunction during the formation and later retrieval of a memory [1McClelland J.L. McNaughton B.L. O’Reilly R.C. Why there are complementary learning systems in the hippocampus and neocortex: insights from the successes and failures of connectionist models of learning and memory.Psychol. Rev. 1995; 102: 419-457Crossref PubMed Scopus (3380) Google Scholar, 2Teyler T.J. DiScenna P. The hippocampal memory indexing theory.Behav. Neurosci. 1986; 100: 147-154Crossref PubMed Scopus (532) Google Scholar, 3Tulving E. Markowitsch H.J. Episodic and declarative memory: role of the hippocampus.Hippocampus. 1998; 8: 198-204Crossref PubMed Scopus (868) Google Scholar, 4Rolls E.T. A theory of hippocampal function in memory.Hippocampus. 1996; 6: 601-620Crossref PubMed Scopus (400) Google Scholar]. At encoding, the hippocampus is thought to continuously store a sparse and non-overlapping index that points to ongoing activity patterns in cortical space. This hippocampal index can later be reactivated by a reminder and lead to the reconstruction of a previously stored memory pattern in neocortex [1McClelland J.L. McNaughton B.L. O’Reilly R.C. Why there are complementary learning systems in the hippocampus and neocortex: insights from the successes and failures of connectionist models of learning and memory.Psychol. Rev. 1995; 102: 419-457Crossref PubMed Scopus (3380) Google Scholar, 2Teyler T.J. DiScenna P. The hippocampal memory indexing theory.Behav. Neurosci. 1986; 100: 147-154Crossref PubMed Scopus (532) Google Scholar, 5Teyler T.J. Rudy J.W. The hippocampal indexing theory and episodic memory: updating the index.Hippocampus. 2007; 17: 1158-1169Crossref PubMed Scopus (179) Google Scholar, 6Marr D. Simple memory: a theory for archicortex.Philos. Trans. R. Soc. Lond. B Biol. Sci. 1971; 262: 23-81Crossref PubMed Scopus (1910) Google Scholar, 7Alvarez P. Squire L.R. Memory consolidation and the medial temporal lobe: a simple network model.Proc. Natl. Acad. Sci. USA. 1994; 91: 7041-7045Crossref PubMed Scopus (762) Google Scholar, 8Norman K.A. O’Reilly R.C. Modeling hippocampal and neocortical contributions to recognition memory: a complementary-learning-systems approach.Psychol. Rev. 2003; 110: 611-646Crossref PubMed Scopus (875) Google Scholar]. Many recent studies have tested these computational assumptions by tracking the reinstatement of memory-related brain activity patterns during retrieval. The basic premise that content-specific neural patterns are reactivated during retrieval has been confirmed using fMRI (for reviews, see [9Danker J.F. Anderson J.R. The ghosts of brain states past: remembering reactivates the brain regions engaged during encoding.Psychol. Bull. 2010; 136: 87-102Crossref PubMed Scopus (215) Google Scholar, 10Rissman J. Wagner A.D. Distributed representations in memory: insights from functional brain imaging.Annu. Rev. Psychol. 2012; 63: 101-128Crossref PubMed Scopus (168) Google Scholar]) and more recently also electroencephalography (EEG) and magnetoencephalography (MEG) [11Johnson J.D. Price M.H. Leiker E.K. Episodic retrieval involves early and sustained effects of reactivating information from encoding.Neuroimage. 2015; 106: 300-310Crossref PubMed Scopus (24) Google Scholar, 12Waldhauser G.T. Braun V. Hanslmayr S. Episodic memory retrieval functionally relies on very rapid reactivation of sensory information.J. Neurosci. 2016; 36: 251-260Crossref PubMed Scopus (69) Google Scholar, 13Wimber M. Maaß A. Staudigl T. Richardson-Klavehn A. Hanslmayr S. Rapid memory reactivation revealed by oscillatory entrainment.Curr. Biol. 2012; 22: 1482-1486Abstract Full Text Full Text PDF PubMed Scopus (49) Google Scholar, 14Kurth-Nelson Z. Barnes G. Sejdinovic D. Dolan R. Dayan P. Temporal structure in associative retrieval.eLife. 2015; 4: e04919Crossref Scopus (39) Google Scholar, 15Staudigl T. Vollmar C. Noachtar S. Hanslmayr S. Temporal-pattern similarity analysis reveals the beneficial and detrimental effects of context reinstatement on human memory.J. Neurosci. 2015; 35: 5373-5384Crossref PubMed Scopus (39) Google Scholar, 16Jafarpour A. Fuentemilla L. Horner A.J. Penny W. Duzel E. Replay of very early encoding representations during recollection.J. Neurosci. 2014; 34: 242-248Crossref PubMed Scopus (62) Google Scholar, 17Michelmann S. Bowman H. Hanslmayr S. The temporal signature of memories: identification of a general mechanism for dynamic memory replay in humans.PLoS Biol. 2016; 14: e1002528Crossref PubMed Scopus (46) Google Scholar]. However, no study has so far investigated the temporal fluctuations of memory-related patterns in human long-term memory and whether they are systematically linked to brain oscillations. A major computational challenge for our memory system is to effectively separate the information arriving from external sensory sources from the information generated in internal circuits. In other words, if the brain constantly pattern completes, how does it make sure that the neural coding of this internally (and possibly incorrectly) generated information does not interfere with the coding of new, incoming information? One promising explanation suggests that this is accomplished by means of neural oscillations. In particular, it has been argued that the phase of the hippocampal theta oscillation supports the chunking of mnemonic information such that the neural assemblies involved in encoding and retrieval are temporally segregated [18Nyhus E. Curran T. Functional role of gamma and theta oscillations in episodic memory.Neurosci. Biobehav. Rev. 2010; 34: 1023-1035Crossref PubMed Scopus (329) Google Scholar, 19Hasselmo M.E. Bodelón C. Wyble B.P. A proposed function for hippocampal theta rhythm: separate phases of encoding and retrieval enhance reversal of prior learning.Neural Comput. 2002; 14: 793-817Crossref PubMed Scopus (581) Google Scholar]. In a seminal paper, Pavlides et al. [20Pavlides C. Greenstein Y.J. Grudman M. Winson J. Long-term potentiation in the dentate gyrus is induced preferentially on the positive phase of theta-rhythm.Brain Res. 1988; 439: 383-387Crossref PubMed Scopus (392) Google Scholar] showed that stimulating a hippocampal assembly at one phase of the theta rhythm induced long-term potentiation (LTP), whereas stimulating at the opposite phase induced long-term depression (LTD). This finding has since been replicated many times in rodents [21Huerta P.T. Lisman J.E. Heightened synaptic plasticity of hippocampal CA1 neurons during a cholinergically induced rhythmic state.Nature. 1993; 364: 723-725Crossref PubMed Scopus (476) Google Scholar, 22Hyman J.M. Wyble B.P. Goyal V. Rossi C.A. Hasselmo M.E. Stimulation in hippocampal region CA1 in behaving rats yields long-term potentiation when delivered to the peak of theta and long-term depression when delivered to the trough.J. Neurosci. 2003; 23: 11725-11731Crossref PubMed Google Scholar] and implemented in computational models of episodic memory and the hippocampus [19Hasselmo M.E. Bodelón C. Wyble B.P. A proposed function for hippocampal theta rhythm: separate phases of encoding and retrieval enhance reversal of prior learning.Neural Comput. 2002; 14: 793-817Crossref PubMed Scopus (581) Google Scholar, 23Kunec S. Hasselmo M.E. Kopell N. Encoding and retrieval in the CA3 region of the hippocampus: a model of theta-phase separation.J. Neurophysiol. 2005; 94: 70-82Crossref PubMed Scopus (85) Google Scholar, 24Buzsáki G. Theta oscillations in the hippocampus.Neuron. 2002; 33: 325-340Abstract Full Text Full Text PDF PubMed Scopus (2224) Google Scholar, 25Hasselmo M.E. Eichenbaum H. Hippocampal mechanisms for the context-dependent retrieval of episodes.Neural Netw. 2005; 18: 1172-1190Crossref PubMed Scopus (216) Google Scholar, 26Parish G. Hanslmayr S. Bowman H. The Sync/deSync model: how a synchronized hippocampus and a de-synchronized neocortex code memories.J. Neurosci. 2018; 38: 3428-3440Crossref PubMed Scopus (31) Google Scholar]. These models share the assumption that successful retrieval is most likely at one specific phase of the hippocampal theta rhythm, opposing the optimal encoding phase [19Hasselmo M.E. Bodelón C. Wyble B.P. A proposed function for hippocampal theta rhythm: separate phases of encoding and retrieval enhance reversal of prior learning.Neural Comput. 2002; 14: 793-817Crossref PubMed Scopus (581) Google Scholar, 27Hasselmo M.E. What is the function of hippocampal theta rhythm?--Linking behavioral data to phasic properties of field potential and unit recording data.Hippocampus. 2005; 15: 936-949Crossref PubMed Scopus (333) Google Scholar]. Memory retrieval should be a continuously oscillating process that is locked to the hippocampal theta phase. Direct evidence for theta phase modulation in human long-term memory still remains elusive. fMRI studies by nature are blind to the sub-second temporal dynamics that could mediate memory reinstatement, and electrophysiological studies have so far not investigated rhythmic fluctuations in memory reactivation. To our knowledge, only one previous study exists that has shown evidence for periodic reactivation, and this was during a working memory task [28Fuentemilla L. Penny W.D. Cashdollar N. Bunzeck N. Düzel E. Theta-coupled periodic replay in working memory.Curr. Biol. 2010; 20: 606-612Abstract Full Text Full Text PDF PubMed Scopus (136) Google Scholar]. In human long-term memory, it is therefore unknown whether neural signatures of memory reactivation are locked to a theta rhythm. The present study was aimed at directly testing this hypothesis. EEG data were recorded while participants encoded novel word-object associations and were later cued with the words to retrieve the objects. EEG-based pattern classifiers were trained to detect memory-related neural patterns during recall with high temporal precision. We demonstrate that, within each retrieval period, classifier fidelity fluctuates at 7 or 8 Hz within each retrieval period and that this index of memory reactivation is locked to a particular phase of the same theta rhythm. The paradigm was a simple word-object associative memory task designed to yield a high number of correct trials (Figure 1A). Participants studied associations between action verbs and objects in random pairings and were later cued with the word to retrieve the object. Two measures of memory accuracy confirmed that participants performed the task well. The first was a subjective measure where participants indicated, via a button press after cue onset, whether and when they recalled the associated object. Participants on average indicated that they remembered the object on 94.21% (SD = 5.75%) of the trials. A second, more objective measure was accuracy in response to a question about the object’s semantic category (animate versus inanimate), which appeared at the end of each retrieval trial and which participants answered correctly on 88.20% (SD = 6.57%) of the trials. These two measures were highly correlated (rSpearman = 0.60; p < 0.05). Average accuracy for perceptual detail (photograph versus line drawing) was 85.31% (SD = 6.45%). Reaction times for the first button press when retrieving animate (mean = 3.03 s; SD = 0.95 s; min = 1.28 s; max = 6.01 s) and inanimate (mean = 2.96 s; SD = 0.77 s; min = 1.47 s; max = 4.24 s) objects did not differ significantly (t(1,23) = 0.57; p = 0.58). The time window used for classification (−200ms–1,500 ms around the cue) thus only minimally overlapped with the button press window. Our primary goal was to test whether the neural signatures of memory retrieval wax and wane in a theta oscillatory rhythm. Our neural index of memory retrieval was obtained from a linear discriminant analysis (LDA) trained to detect evidence for the reactivation of the correct object category (animate versus inanimate) during retrieval (Figure 1B; see STAR Methods for details). The LDA was trained and tested independently per participant at each retrieval time point starting with the onset of the word cue, using a leave-one-out procedure. The input into the LDA was a feature vector containing the signal amplitudes from all 128 EEG channels at a given time point. The major output of interest was the fidelity (distance or d-) values available for each trial and time point. These values represent the distance from the hyperplane that optimally separates the two classes of retrieved objects (animate versus inanimate), and their time courses served as our time-resolved, parametric index of memory reactivation. For the purpose of this study, the LDA was trained and tested during cued recall in order to isolate a purely retrieval-based signature or memory retrieval, which could then (below) be compared with a purely encoding-based index of memory classification. Additional analyses using classifiers trained on encoding and tested at retrieval are reported in the supplemental materials (Figures S1 and S4). We first asked whether evidence could be found for an oscillation in these time-resolved indices of memory reactivation (Figures 2A and 2B ). Fidelity time courses from the recall task were averaged across trials per participant and subjected to a Fourier transformation. If memory reactivation fluctuates in a theta rhythm, the resulting power spectra will show a selective increase in a band-limited lower (theta) frequency band. We compared the power spectra obtained from the real classifier outputs with a bootstrapped baseline [30Stelzer J. Chen Y. Turner R. Statistical inference and multiple testing correction in classification-based multi-voxel pattern analysis (MVPA): random permutations and cluster size control.Neuroimage. 2013; 65: 69-82Crossref PubMed Scopus (218) Google Scholar], the latter using the d-value outputs from classifiers that were trained and tested on the same EEG trials but with randomly shuffled category labels (see STAR Methods section). This procedure controls for spurious power peaks that are driven by the frequency characteristics of the raw data (e.g., a dominant oscillation in the single trials). Significant power differences between the real and shuffled data were found in frequency bins at 7–9 Hz and 13 Hz, all exceeding the 95th percentile of the empirical null distribution (Figure 2C). Power at 7–9 Hz was significantly higher (t(1,23) = 1.9425; p = 0.03) when including only correctly retrieved trials than when including all trials, suggesting a relationship of the classifier fluctuation to memory success [31Siegel M. Warden M.R. Miller E.K. Phase-dependent neuronal coding of objects in short-term memory.Proc. Natl. Acad. Sci. USA. 2009; 106: 21341-21346Crossref PubMed Scopus (379) Google Scholar]. An alternative method with more stringent criteria to determine the presence of oscillations [32Watrous A.J. Miller J. Qasim S.E. Fried I. Jacobs J. Phase-tuned neuronal firing encodes human contextual representations for navigational goals.eLife. 2018; 7: e32554Crossref PubMed Scopus (50) Google Scholar] confirmed that oscillatory power in the classifier time series was increased above baseline in the 7- to 9-Hz frequency range (Figure 2D). Moreover, a similar power spectrum was found when the classifier was trained on encoding and tested on retrieval (Figure S4). The frequency characteristics of the classifier fidelity time courses thus suggest a rhythmic fluctuation in memory reactivation that was most consistent in the 7- to 9-Hz frequency range. Our next two analyses were aimed at specifically testing for coupling between neural reactivation (i.e., classifier time series) and the phase of hippocampal theta-band oscillations. For this purpose, the raw EEG trials were projected into source space using a linear constrained minimum variance (LCMV) beamforming algorithm [33Van Veen B.D. van Drongelen W. Yuchtman M. Suzuki A. Localization of brain electrical activity via linearly constrained minimum variance spatial filtering.IEEE Trans. Biomed. Eng. 1997; 44: 867-880Crossref PubMed Scopus (1640) Google Scholar, 34Gross J. Kujala J. Hamalainen M. Timmermann L. Schnitzler A. Salmelin R. Dynamic imaging of coherent sources: studying neural interactions in the human brain.Proc. Natl. Acad. Sci. USA. 2001; 98: 694-699Crossref PubMed Scopus (1266) Google Scholar], and a hippocampal mask was used to extract the 8-Hz phase of the hippocampal virtual channels for each trial and time point. We computed a phase modulation index (MI) [35Rieke F. Warland D. de Ruyter van Steveninck R. Bialek W. Spikes: Exploring the Neural Code. MIT, 1997Google Scholar] reflecting the strength of coupling between the hippocampal 8-Hz phase and the amplitude of the classifier output. Classifier fidelity as a function of hippocampal theta phase is plotted in Figure 2E (green line). This analysis revealed a significant modulation index (M = 0.0071, SD = 0.0042; baseline: M = 0.0056, SD = 0.0006; t(1,23) = 1.8191; p < 0.05; one-sided t test), indicating that fidelity of the retrieval classifier was modulated by the phase of the hippocampal 8-Hz oscillation (Figure 2E). We next directly compared the theta phase at which classifier fidelity was maximal during encoding and retrieval. All basic analysis steps were repeated for the encoding EEG data, where an LDA discriminating animate from inanimate objects was trained and tested at each time point from 200 ms before until 1,500 ms after object onset. The full time generalization matrices showing classifier performance for encoding and retrieval can be found in Figure S1. The 8-Hz phase at encoding was then extracted from hippocampal virtual channels to calculate the phase modulation index. Classifier fidelity as a function of hippocampal theta phase during encoding is shown in Figure 2E (gray line). A significant phase modulation was found also for encoding (M = 0.0068, SD = 0.0029; baseline: M = 0.0052, SD = 0.0007; t(1,23) = 2.7494; p < 0.05; one-sided t test). In order to directly compare the encoding and retrieval phases, we identified the phase at which encoding or retrieval classification was optimal in each subject. A Rayleigh circular statistic comparing the absolute phase angles at which encoding and retrieval classification was maximal revealed that these angles significantly differed from each other (z(1,23) = 5.5342; p = 0.001). Similar statistics were obtained by fitting a sine wave to the data and identifying and extracting the phase at which classification was optimal. Together, the results of the phase modulation analyses show that retrieval fluctuates as a function of hippocampal theta (8 Hz) phase and that the optimal retrieval phase is on average 188 degrees phase shifted compared with the optimal phase during encoding. Having established that the neural retrieval patterns oscillate and are coupled to an 8-Hz oscillation, we next investigated the temporal relationship between theta phase and memory reinstatement. The analysis was inspired by the use of spike-triggered averages in animal intracranial work [35Rieke F. Warland D. de Ruyter van Steveninck R. Bialek W. Spikes: Exploring the Neural Code. MIT, 1997Google Scholar, 36Douchamps V. Jeewajee A. Blundell P. Burgess N. Lever C. Evidence for encoding versus retrieval scheduling in the hippocampus by theta phase and acetylcholine.J. Neurosci. 2013; 33: 8689-8704Crossref PubMed Scopus (82) Google Scholar]. We here adopted a similar approach computing classifier-locked averages around the time points of maximal memory reactivation (see STAR Methods for details). On each single trial, those time points of maximal classifier fidelity that exceeded the 95th percentile of a bootstrapped baseline were marked as new events of interest, the corresponding time stamps were located in the raw EEG epochs, and the ongoing EEG signal surrounding these maxima was then analyzed for phase consistency across all electrodes (Figure 3A). We used a non-parametric cluster-based permutation test to compare the real data with a temporally shuffled baseline that keeps the EEG trial structure intact but produces a random temporal alignment between the classifier maxima and the ongoing phase (see STAR Methods section). Comparing the “real” times of maximum classifier fidelity with the temporally shuffled baseline revealed a cluster of significant (pcorr < 0.05) phase consistency from 500 ms to 50 ms before the classifier maxima, centered at 7 Hz (Figure 3B). Note that, in this analysis, several classifier peaks per trial can exceed the 95th percentile criterion, and many of the classifier-locked EEG epochs will thus overlap, resulting in temporal smearing of the phase-locked activity. When running the same analysis extracting only one maximum per trial (Figure S2C), we found a similar cluster of phase locking but with a more narrow temporal extent from 500 ms to 150 ms pre-maxima, suggesting that the strongest phase-consistency effect was present roughly two theta cycles (corresponding to 2 × 143 ms = 286 ms) before mnemonic information could most confidently be decoded. This finding supports our primary hypothesis that memory reinstatement shows a consistent oscillatory timing across trials and participants in the same 7- to 9-Hz frequency band at which the classifier fluctuates (Figure 2C). It might seem counterintuitive that the strongest phase consistency was observed prior to the time points of maximum classification fidelity, rather than at the maxima themselves. However, this temporal relationship is to be expected if the phase-locked signal originates from a different, upstream region in the processing hierarchy compared to the signal that the classifier’s decision is based on. Our findings are consistent with a model where the re-instantiation of a memory trace is triggered at a consistent phase of a hippocampal/medial temporal lobe (MTL) theta oscillation, followed by memory reinstatement in a broader range of neocortical regions representing the stored memory [1McClelland J.L. McNaughton B.L. O’Reilly R.C. Why there are complementary learning systems in the hippocampus and neocortex: insights from the successes and failures of connectionist models of learning and memory.Psychol. Rev. 1995; 102: 419-457Crossref PubMed Scopus (3380) Google Scholar, 24Buzsáki G. Theta oscillations in the hippocampus.Neuron. 2002; 33: 325-340Abstract Full Text Full Text PDF PubMed Scopus (2224) Google Scholar, 37O’Reilly R.C. Norman K.A. Hippocampal and neocortical contributions to memory: advances in the complementary learning systems framework.Trends Cogn. Sci. 2002; 6: 505-510Abstract Full Text Full Text PDF PubMed Scopus (241) Google Scholar, 38O’Reilly R.C. Bhattacharyya R. Howard M.D. Ketz N. Complementary learning systems.Cogn. Sci. 2014; 38: 1229-1248Crossref PubMed Scopus (125) Google Scholar]. The aim of the next analysis was to identify the brain regions involved in producing the observed clusters of theta phase consistency, with the hypothesis that the effect should be present in MTL areas [1McClelland J.L. McNaughton B.L. O’Reilly R.C. Why there are complementary learning systems in the hippocampus and neocortex: insights from the successes and failures of connectionist models of learning and memory.Psychol. Rev. 1995; 102: 419-457Crossref PubMed Scopus (3380) Google Scholar, 2Teyler T.J. DiScenna P. The hippocampal memory indexing theory.Behav. Neurosci. 1986; 100: 147-154Crossref PubMed Scopus (532) Google Scholar]. Trial time courses were projected into source space using a beamforming algorithm [29Oostenveld R. Fries P. Maris E. Schoffelen J.M. FieldTrip: open source software for advanced analysis of MEG, EEG, and invasive electrophysiological data.Comput. Intell. Neurosci. 2011; 2011: 156869Crossref PubMed Scopus (5245) Google Scholar, 34Gross J. Kujala J. Hamalainen M. Timmermann L. Schnitzler A. Salmelin R. Dynamic imaging of coherent sources: studying neural interactions in the human brain.Proc. Natl. Acad. Sci. USA. 2001; 98: 694-699Crossref PubMed Scopus (1266) Google Scholar], and we then looked for the sources showing the strongest phase consistency. Contrasting all classifier maxima with the shuffled baseline (identical to the scalp level analysis), we found an activation cluster spanning large regions of occipital, temporal, and frontal cortex, primarily in the right hemisphere (maximum at Montreal Neurological Institute [MNI] coordinates xyz = 10 −10 10, thalamus; Figure 3C). Although these sources included medial temporal lobe areas, they do not suggest a specific role of the hippocampus in producing the theta-phase-locked signal preceding the classifier maxima. We next wanted to test whether the theta phase consistency systematically varied with the strength of neural reinstatement. We hypothesized that phase consistency would be highest when the classifier correctly detected neural reactivation with high fidelity and lower when the classifier was correct but less confident. Comparing classifier maxima of higher and lower fidelity revealed a significant (pcorr < 0.05) cluster at 7 Hz, preceding the maxima by 500–200 ms (Figure 3D). This cluster highly overlapped in timing, frequency, and topography with our previous classifier-triggered average analyses. When conducting the same analysis in source space, we found sources that spanned the parietal and the right medial temporal lobes (maximum MNI coordinates xyz = 50 −30 30, inferior parietal lobule; Figure 3E), strongly reminiscent of the core recollection or memory success network typically found in fMRI studies [39Rugg M.D. Vilberg K.L. Brain networks underlying episodic memory retrieval.Curr. Opin. Neurobiol. 2013; 23: 255-260Crossref PubMed Scopus (432) Google Scholar]. Our data thus suggest that the neural signatures of memory retrieval are linked to a specific phase of a theta oscillation, and this phase relationship becomes stronger with more confident neural reactivation. The source level analysis additionally confirms our a priori assumption that the phase-locked signal that precedes memory reactivation involves the MTL and other core recollection areas. Although consistent with our hypotheses, this pattern of results could in theory also be explained by an early event-related potential (ERP) elicited by the reminder word, because ERPs are generally associated with strong phase locking in slow frequencies [40Gruber W.R. Klimesch W. Sauseng P. Doppelmayr M. Alpha phase synchronization predicts P1 and N1 latency and amplitude size.Cereb. Cortex. 2005; 15: 371-377Crossref PubMed Scopus (178) Google Scholar]. Such an explanation would assume that our classifier maxima tend to occur at a consistent time point within each retrieval trial with a delay to the reminder-elicited ERP of approximately 300 ms. Several observations speak against this alternative. First, the classifier maxima were relatively evenly distributed across the entire retrieval period and did not tend to cluster around early t" @default.
• W2896767726 created "2018-10-26" @default.
• W2896767726 creator A5012219181 @default.
• W2896767726 creator A5021753069 @default.
• W2896767726 creator A5083963466 @default.
• W2896767726 creator A5084014156 @default.
• W2896767726 date "2018-11-01" @default.
• W2896767726 modified "2023-10-16" @default.
• W2896767726 title "An Optimal Oscillatory Phase for Pattern Reactivation during Memory Retrieval" @default.
• W2896767726 cites W1449283962 @default.
• W2896767726 cites W1658884838 @default.
• W2896767726 cites W1976698619 @default.
• W2896767726 cites W1977223859 @default.
• W2896767726 cites W1987769172 @default.
• W2896767726 cites W1989666886 @default.
• W2896767726 cites W1992337477 @default.
• W2896767726 cites W1993521400 @default.
• W2896767726 cites W1997045645 @default.
• W2896767726 cites W2006051044 @default.
• W2896767726 cites W2010321974 @default.
• W2896767726 cites W2022223337 @default.
• W2896767726 cites W2022856253 @default.
• W2896767726 cites W2025933288 @default.
• W2896767726 cites W2029574629 @default.
• W2896767726 cites W2037504148 @default.
• W2896767726 cites W2047057213 @default.
• W2896767726 cites W2047976711 @default.
• W2896767726 cites W2052170508 @default.
• W2896767726 cites W2053703740 @default.
• W2896767726 cites W2056336718 @default.
• W2896767726 cites W2058242738 @default.
• W2896767726 cites W2058860479 @default.
• W2896767726 cites W2061217602 @default.
• W2896767726 cites W2061358809 @default.
• W2896767726 cites W2065096279 @default.
• W2896767726 cites W2077680606 @default.
• W2896767726 cites W2096424612 @default.
• W2896767726 cites W2100377190 @default.
• W2896767726 cites W2104939199 @default.
• W2896767726 cites W2105496689 @default.
• W2896767726 cites W2107365776 @default.
• W2896767726 cites W2112575160 @default.
• W2896767726 cites W2113031839 @default.
• W2896767726 cites W2113132656 @default.
• W2896767726 cites W2119051448 @default.
• W2896767726 cites W2119565218 @default.
• W2896767726 cites W2120078534 @default.
• W2896767726 cites W2124190705 @default.
• W2896767726 cites W2125737923 @default.
• W2896767726 cites W2127317012 @default.
• W2896767726 cites W2138414057 @default.
• W2896767726 cites W2141815738 @default.
• W2896767726 cites W2144741543 @default.
• W2896767726 cites W2146783807 @default.
• W2896767726 cites W2150209193 @default.
• W2896767726 cites W2151108396 @default.
• W2896767726 cites W2152171700 @default.
• W2896767726 cites W2152372327 @default.
• W2896767726 cites W2157308590 @default.
• W2896767726 cites W2158255900 @default.
• W2896767726 cites W2159345153 @default.
• W2896767726 cites W2160904011 @default.
• W2896767726 cites W2163909553 @default.
• W2896767726 cites W2166073443 @default.
• W2896767726 cites W2168210093 @default.
• W2896767726 cites W2169366712 @default.
• W2896767726 cites W2218418953 @default.
• W2896767726 cites W2253700311 @default.
• W2896767726 cites W2504763666 @default.
• W2896767726 cites W2516799295 @default.
• W2896767726 cites W2762427114 @default.
• W2896767726 cites W2767519445 @default.
• W2896767726 cites W2953337944 @default.
• W2896767726 cites W4238424379 @default.
• W2896767726 cites W4294214781 @default.
• W2896767726 doi "https://doi.org/10.1016/j.cub.2018.08.065" @default.
• W2896767726 hasPubMedId "https://pubmed.ncbi.nlm.nih.gov/30344116" @default.
• W2896767726 hasPublicationYear "2018" @default.
• W2896767726 type Work @default.
• W2896767726 sameAs 2896767726 @default.
• W2896767726 citedByCount "71" @default.
• W2896767726 countsByYear W28967677262018 @default.
• W2896767726 countsByYear W28967677262019 @default.
• W2896767726 countsByYear W28967677262020 @default.
• W2896767726 countsByYear W28967677262021 @default.
• W2896767726 countsByYear W28967677262022 @default.
• W2896767726 countsByYear W28967677262023 @default.
• W2896767726 crossrefType "journal-article" @default.
• W2896767726 hasAuthorship W2896767726A5012219181 @default.
• W2896767726 hasAuthorship W2896767726A5021753069 @default.
• W2896767726 hasAuthorship W2896767726A5083963466 @default.
• W2896767726 hasAuthorship W2896767726A5084014156 @default.
• W2896767726 hasBestOaLocation W28967677261 @default.
• W2896767726 hasConcept C178790620 @default.
• W2896767726 hasConcept C185592680 @default.
• W2896767726 hasConcept C44280652 @default.
• W2896767726 hasConcept C70721500 @default.
• W2896767726 hasConcept C78458016 @default.

• next