A Brain System for Auditory Working Memory

A Brain System for Auditory Working Memory Kumar Joseph Gander Barascud Halpern Griffiths

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.

A Brain System for Auditory Working Memory 2016 The Journal of Neuroscience 10.1523/jneurosci.4341-14.2016 Kumar2016-a-brain-system-for-auditory-working-memory Kumar2016-a-brain-system-for-auditory-working-memory.txt, the project's staged copy f7daed0986d1aeb959ad77dc63a3cbf8c366e19c 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 how a single tone is held in mind for sixteen seconds, and which areas carry the held content Quotation withheld because this record is not cleared for redistribution. whether auditory cortex is involved in auditory working memory at all, given limited and conflicting evidence Quotation withheld because this record is not cleared for redistribution. a system in which sound-specific representations in auditory cortex are kept active by long-range connections from hippocampus and frontal cortex Quotation withheld because this record is not cleared for redistribution. mechanism Quotation withheld because this record is not cleared for redistribution. auditory working memory Quotation withheld because this record is not cleared for redistribution. fmri fmri Quotation withheld because this record is not cleared for redistribution. mechanisms Tone-specific representations in left inferior frontal gyrus drive subvocal rehearsal that re-activates auditory cortex, forming a closed loop that keeps the sound representation alive during the delay duplicate of: mechanisms[4] Quotation withheld because this record is not cleared for redistribution. mechanisms Subjects who lose the tone from the focus of attention retrieve it from activated long-term memory, engaging hippocampus but importing proactive interference from tones of previous trials, which is why hippocampal activity tracks poorer accuracy Quotation withheld because this record is not cleared for redistribution. mechanisms Default-mode-network engagement during the delay indexes mind wandering, which withdraws resources from maintenance and lowers accuracy Quotation withheld because this record is not cleared for redistribution. hippocampal activity here may form part of the activated LTM component of WM, because tones changed across trials and were maintained for long intervals of time (16 s) Quotation withheld because this record is not cleared for redistribution. left inferior frontal gyrus and HG may form a closed loop in which tone-specific frontal representations drive subvocal rehearsal that reactivates auditory cortex Quotation withheld because this record is not cleared for redistribution. a retro-cue protects the cued item from decay, which is why cueing one of two tones gives activation where holding both uncued gave suppression Quotation withheld because this record is not cleared for redistribution. PT may hold symbolic templates further from the sensory pattern, which would make its content harder to decode from blood flow Quotation withheld because this record is not cleared for redistribution. posterior hippocampus may be more involved in real-time analysis of auditory stimuli, which predicts stronger connectivity with auditory cortex Quotation withheld because this record is not cleared for redistribution. Retrieval-phase activity in the dorsal-anterior right IFG cannot be attributed to retrieval rather than motor response. what this limits: the retrieval interpretation of the dorsal-anterior right IFG: WE CANNOT DISSOCIATE WHETHER THE ACTIVITY IS RELATED TO MOTOR RESPONSE OR TO RETRIEVAL OF WM INFORMATION Quotation withheld because this record is not cleared for redistribution. The rehearsal interpretation of motor-area activity during maintenance is contested in the wider literature. what this limits: the rehearsal reading of motor activity: THE ROLE OF MOTOR AREAS IN WORKING MEMORY IS STILL CONTROVERSIAL AS STUDIES USING NONREHEARSABLE STIMULI ALSO SHOW ACTIVITY IN MOTOR AREAS Quotation withheld because this record is not cleared for redistribution. Subjects were not instructed to use any particular strategy, so the subvocal-rehearsal account of the LIFG activity is inferred rather than controlled. what this limits: the subvocal-rehearsal account of the LIFG activity: WE DID NOT EXPLICITLY INSTRUCT THE SUBJECTS TO FOLLOW ANY PARTICULAR STRATEGY, so rehearsal is inferred and not controlled Quotation withheld because this record is not cleared for redistribution. for the frontal focus active only at retrieval, motor response and memory retrieval cannot be dissociated duplicate of: limitations[0] · what this limits: the same scope as limitations[0] - at retrieval, motor and memory cannot be separated Quotation withheld because this record is not cleared for redistribution. no strategy was instructed, so rehearsal is inferred rather than controlled, and the role of motor areas in working memory is itself contested what this limits: the same scope as limitations[2], with the standing dispute attached Quotation withheld because this record is not cleared for redistribution. classification was above chance in only two of five regions, and in left inferior frontal gyrus five of thirteen subjects fell below chance what this limits: the group-level classification result: FOR HG, 9 OF 13 SUBJECTS PERFORMED ABOVE THE CHANCE LEVEL ... FOR THE LEFT IFG, EIGHT SUBJECTS WERE ABOVE THE CHANCE LEVEL AND FIVE PERFORMED BELOW Quotation withheld because this record is not cleared for redistribution. three subjects were dropped from the pattern analysis for poor performance but kept in the univariate analysis, because the correlation with behaviour needed them what this limits: the sample size, which differs by analysis on purpose: three poor performers WERE DROPPED FROM THE ANALYSIS ... WE DID NOT DROP THESE SUBJECTS FROM THE UNIVARIATE ANALYSIS BECAUSE WE WANTED TO ANALYZE THE CORRELATION OF BRAIN ACTIVITY WITH BEHAVIORAL PERFORMANCE Quotation withheld because this record is not cleared for redistribution. multivoxel pattern analysis of the delay period, imported to ask whether auditory delay activity is content-specific Quotation withheld because this record is not cleared for redistribution. the retro-cue, whose protective effect is used to explain why two studies of the same question disagree Quotation withheld because this record is not cleared for redistribution. only the middle 10.5 seconds of the maintenance period were modelled, discarding 4 seconds at each end eliminates: that the maintenance signal is contaminated by the haemodynamic response to stimulus onset or to the probe at the end · outcome (stated by the tagger, no quotation): held Quotation withheld because this record is not cleared for redistribution. subjects performed a pitch discrimination task outside the scanner before taking part eliminates: that a failure of memory is really a failure of PERCEPTION - a subject who cannot hear the pitch difference cannot remember it either · outcome (stated by the tagger, no quotation): held Quotation withheld because this record is not cleared for redistribution. DMN DMN Quotation withheld because this record is not cleared for redistribution. behavioral performance behavioral performance Quotation withheld because this record is not cleared for redistribution. LTM → hippocampus COMMON CAUSE, NOT A CHAIN. Subjects who cannot sustain attention rely more on LTM; LTM reliance predicts BOTH greater hippocampal activation AND poorer performance, the latter via proactive interference from previously presented tones. The hippocampal signal and the performance drop are sibling effects. Stated as a PREDICTION, not a result. Quotation withheld because this record is not cleared for redistribution. SUSTAINED ACTIVITY IS NOT WORKING MEMORY: a region can hold information in subthreshold patterns without elevated activity, and elevated activity can reflect sustained attention instead Quotation withheld because this record is not cleared for redistribution. the maintenance regressor is only interpretable because its correlation with the encoding regressor was MEASURED at 0.04 - the phases are adjacent in time and the BOLD response is lagged Quotation withheld because this record is not cleared for redistribution. the classifier sees only the middle 10.5 s of the delay, with 4 s trimmed at each end, because the probe tone belonged to the same category and could otherwise leak into the decoding Quotation withheld because this record is not cleared for redistribution. connectivity is never compared with rest, because heart rate and other physiological variables differ between task and rest and are known to confound functional connectivity Quotation withheld because this record is not cleared for redistribution. the inclusion criteria differ BY ANALYSIS and each difference is justified: poor performers are excluded from classification to reduce noise and retained in the univariate analysis because the brain-behaviour correlation requires the spread Quotation withheld because this record is not cleared for redistribution. d-PRIME is d', the signal-detection sensitivity index, and not a priming prime despite carrying the word: it is the per-subject detection measure (0.89, 0.99, 1.16) used as the exclusion criterion for the pattern analysis Quotation withheld because this record is not cleared for redistribution. claim_type: mechanism a second reading reaching the same value. Recorded: mechanism — note: a second reading reaching the same value from a different sentence. Not applied - a person decides. Quotation withheld because this record is not cleared for redistribution. method: event-related fMRI with separate encoding, maintenance and retrieval regressors, multivoxel pattern classification, and psychophysiological-interaction connectivity a second reading DIFFERING from the recorded value. Recorded: fmri — 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: sustained activity in auditory cortex, hippocampus and inferior frontal gyrus across a sixteen-second maintenance period a second reading DIFFERING from the recorded value. Recorded: auditory working memory — 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. contradicts → Linke et al. (2011) Quotation withheld because this record is not cleared for redistribution. replicates → Linke and Cusack (2015) Quotation withheld because this record is not cleared for redistribution. builds-on → Jeneson and Squire (2011) Quotation withheld because this record is not cleared for redistribution. contradicts → Linke 2011's observation of suppression in auditory cortex during maintenance of two tones Quotation withheld because this record is not cleared for redistribution. reconciles → the conflicting reports of increased and decreased auditory cortical activity during auditory working memory Quotation withheld because this record is not cleared for redistribution. confirms → Linke and Cusack 2015's activation during maintenance of a single sound Quotation withheld because this record is not cleared for redistribution. supports → Jeneson and Squire's account in which hippocampal activity during working memory reflects recourse to long-term memory Quotation withheld because this record is not cleared for redistribution. builds-on → visual working memory studies showing content-specific subthreshold activity during the delay Quotation withheld because this record is not cleared for redistribution. sustained activity —distinct-from→ WM sustained activity WM strength of the relation: 1 · note: TWO WAYS THE OLD INDEX FAILS: a region can carry content without elevated activity, and elevated activity can be sustained attention instead Quotation withheld because this record is not cleared for redistribution. encoding —distinct-from→ maintenance encoding maintenance strength of the relation: 1 · note: CONFOUND MEASURED, NOT ASSUMED: the two regressors correlate at 0.04, so maintenance activity is not encoding leaking forward through the lagged BOLD response Audit 2026-09-28: DOUBTFUL - read against its span and kept, but the span discusses separating encoding and maintenance regressors; distinct-from is loose. Quotation withheld because this record is not cleared for redistribution. retrieval —distinct-from→ maintenance retrieval maintenance strength of the relation: 1 · note: LEAKAGE THE OTHER WAY: the probe was of the cued category, so 4 s is trimmed from each end of the delay before decoding Audit 2026-09-28: DOUBTFUL - read against its span and kept, but the span says the probe likely re-cued the maintained category; distinct-from is loose. Quotation withheld because this record is not cleared for redistribution. HG —contrasted-with→ PT HG PT strength of the relation: 1 · note: PT carries 84.42% of the surviving maintenance voxels and yet HG is the one whose PATTERNS decode the tone - activation and content dissociate Quotation withheld because this record is not cleared for redistribution. LIFG —contrasted-with→ RIFG LIFG RIFG strength of the relation: 1 · note: CRITERION: both are active throughout maintenance and only the LEFT carries decodable content - activation and content dissociate here too Quotation withheld because this record is not cleared for redistribution. RETRACTED (TWO EFFECTS OF A COMMON CAUSE, TAGGED AS CAUSING EACH OTHER - a class I have not named before) · hippocampus —causes→ behavioral performance hippocampus behavioral performance RETRACTED: TWO EFFECTS OF A COMMON CAUSE, TAGGED AS CAUSING EACH OTHER - a class I have not named before. The edge is 'hippocampus --causes--> behavioral performance'. The span is a chain of PREDICTIONS: '…on LTM, and, THEREFORE, hippocampus WOULD BE ACTIVATED…'. The upstream cause is failure to sustain attention, which produces LTM reliance; LTM reliance produces BOTH greater hippocampal activation AND poor performance. Hippocampal activation is a MARKER of the same upstream process, not a cause of the performance. Reading the two consequences as one causing the other is the classic confound, and here the paper itself lays the structure out. WHAT THE SPAN DOES SAY: housed below as the common-cause structure, since no type can record 'these two share a cause'. — strength of the relation: 0.6 — the record marks this weak · note: SIGN NEGATIVE, AND PREDICTED IN ADVANCE FROM THE RIVAL'S MODEL: more hippocampal activity should mean worse performance if it reflects falling back on long-term memory Audit 2026-09-28: DOUBTFUL - read against its span and kept, but the span states a prediction; causes states it as a finding. Quotation withheld because this record is not cleared for redistribution. DMN —causes→ behavioral performance DMN behavioral performance status: CORRELATIONAL, NOT INTERVENTIONAL: the paper reports a negative CORRELATION between DMN activity and performance; nothing was manipulated, and the mind-wandering account that follows is offered as an interpretation · direction on the measure: reduces · strength of the relation: 0.7 · note: SIGN NEGATIVE: read as mind wandering during the delay Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but DMN activity was NEGATIVELY CORRELATED with performance, interpreted as mind wandering; causes overstates a correlation. Quotation withheld because this record is not cleared for redistribution. PPI —operationalised-as→ connectivity PPI connectivity strength of the relation: 1 · note: and the rest BASELINE IS REFUSED for it: heart rate differs between task and rest, so every contrast is task-against-task Quotation withheld because this record is not cleared for redistribution. t-test t-test Quotation withheld because this record is not cleared for redistribution. ANOVA ANOVA Quotation withheld because this record is not cleared for redistribution. general linear model general linear model Quotation withheld because this record is not cleared for redistribution. regression regression Quotation withheld because this record is not cleared for redistribution. multiple-comparison correction (false discovery rate) multiple-comparison correction (false discovery rate) Quotation withheld because this record is not cleared for redistribution. permutation test permutation test Quotation withheld because this record is not cleared for redistribution. cross-validation cross-validation Quotation withheld because this record is not cleared for redistribution. small-volume correction small-volume correction Quotation withheld because this record is not cleared for redistribution. psychophysiological interaction analysis psychophysiological interaction analysis Quotation withheld because this record is not cleared for redistribution. support vector machine support vector machine Quotation withheld because this record is not cleared for redistribution. false discovery rate false discovery rate Quotation withheld because this record is not cleared for redistribution. one-tailed t tests one-tailed t tests Quotation withheld because this record is not cleared for redistribution. During the 16 s maintenance period, sustained auditory-cortex activity was concentrated in PT, with 84.42% of suprathreshold voxels lying there. Quotation withheld because this record is not cleared for redistribution. No suppression of auditory cortex activity was observed in any of the three task phases. Quotation withheld because this record is not cleared for redistribution. Only HG (55.61%) and left inferior frontal gyrus (54.25%) carried information about whether a low or high tone was being held; PT, hippocampus and right inferior frontal gyrus classified at chance. Quotation withheld because this record is not cleared for redistribution. Univariate analysis found no maintenance-period activity difference between holding a low and a high tone. Quotation withheld because this record is not cleared for redistribution. Hippocampus was significantly active in all three phases of the task - encoding, maintenance and retrieval. Quotation withheld because this record is not cleared for redistribution. Along the hippocampal long axis, encoding activity was stronger anteriorly and retrieval activity stronger posteriorly. Quotation withheld because this record is not cleared for redistribution. Hippocampal activity during maintenance correlated negatively with task accuracy: subjects with more hippocampal activity performed worse. Quotation withheld because this record is not cleared for redistribution. Auditory cortex was strongly connected to the right hippocampus, both posterior and anterior. Quotation withheld because this record is not cleared for redistribution. no suppression of auditory cortical activity was seen in any phase of the task Quotation withheld because this record is not cleared for redistribution. during maintenance 84.42 percent of surviving auditory voxels lay in PT Quotation withheld because this record is not cleared for redistribution. hippocampal activity was constant along the anterior-posterior axis during maintenance but rose posteriorly during encoding and retrieval Quotation withheld because this record is not cleared for redistribution. THE UNIVARIATE NULL THAT MOTIVATED THE PATTERN ANALYSIS: mean activity did not differ between holding a low and a high tone Quotation withheld because this record is not cleared for redistribution. only HG at 55.61 percent and left inferior frontal gyrus at 54.25 percent classified the held tone above chance Quotation withheld because this record is not cleared for redistribution. AND THE PER-SUBJECT SPLIT IS REPORTED: 9 of 13 above chance in HG, and only 8 of 13 in left inferior frontal gyrus Quotation withheld because this record is not cleared for redistribution. PARAMETRIC ASSUMPTION CHECKED: a permutation test over 100,000 resamples gave p values almost identical to the t tests Quotation withheld because this record is not cleared for redistribution. THE REGRESSOR CONFOUND MEASURED: encoding and maintenance regressors correlated at only 0.04 Quotation withheld because this record is not cleared for redistribution. and the analysis was repeated on only the middle of the delay period, with very similar results Quotation withheld because this record is not cleared for redistribution. DECODING LEAKAGE PREVENTED: the probe belonged to the cued category, so 4 s was trimmed from each end before classification Quotation withheld because this record is not cleared for redistribution. THE REST BASELINE WAS REFUSED for connectivity, because physiological variables differ between task and rest and confound functional connectivity Quotation withheld because this record is not cleared for redistribution. INCLUSION CRITERIA DIFFER BY ANALYSIS, each justified: three poor performers dropped from classification and kept in the univariate analysis Quotation withheld because this record is not cleared for redistribution. A PREDICTION DERIVED FROM THE RIVAL'S MODEL: if hippocampal activity means falling back on long-term memory, it should correlate NEGATIVELY with performance Quotation withheld because this record is not cleared for redistribution. and that is what was found - activity in both hippocampi increased as performance worsened Quotation withheld because this record is not cleared for redistribution. default mode network activity also correlated negatively with performance, read as mind wandering Quotation withheld because this record is not cleared for redistribution. as did right inferior frontal junction activity, read as updating working memory from long-term memory Quotation withheld because this record is not cleared for redistribution. parietal activity correlated POSITIVELY with performance, read as sustained attention Quotation withheld because this record is not cleared for redistribution. A CONTRADICTION RESOLVED BY A DESIGN VARIABLE: the earlier suppression finding used two uncued tones, this study cued one of two after presentation Quotation withheld because this record is not cleared for redistribution. and a later study holding a single sound found activation, agreeing with this one Quotation withheld because this record is not cleared for redistribution. in that comparison auditory cortex showed no long-range connectivity, only connections between HG and PT within it Quotation withheld because this record is not cleared for redistribution. during maintenance auditory cortex connected to right hippocampus and to left inferior frontal gyrus Quotation withheld because this record is not cleared for redistribution. accuracy ranged from 67 to 98 percent and was better for the first tone than the second Quotation withheld because this record is not cleared for redistribution. seventeen adults with normal hearing and no musical training took part Quotation withheld because this record is not cleared for redistribution. auditory working memory (AWM) auditory working memory (AWM) Quotation withheld because this record is not cleared for redistribution. multivoxel pattern analysis (MVPA) multivoxel pattern analysis (MVPA) Quotation withheld because this record is not cleared for redistribution. activated LTM component activated LTM component Quotation withheld because this record is not cleared for redistribution. AWM AWM Quotation withheld because this record is not cleared for redistribution. MVPA MVPA Quotation withheld because this record is not cleared for redistribution. activated LTM activated LTM Quotation withheld because this record is not cleared for redistribution. BOLD activity BOLD activity Quotation withheld because this record is not cleared for redistribution. accuracy scores accuracy scores Quotation withheld because this record is not cleared for redistribution. age of the subjects age of the subjects Quotation withheld because this record is not cleared for redistribution. maintenance maintenance Quotation withheld because this record is not cleared for redistribution. sustained activity sustained activity WHOSE RESULT THIS IS: Serences et al., 2009 ; Riggall and Postle, 2012 ) have shown - THE EVIDENCE FOR THIS ROW ATTRIBUTES THE WORK TO SOMEBODY ELSE. The cited name is the subject of a result verb and the sentence carries no first-person marker, so the finding, quantity or definition is the CITED author's rather than this record's. The claim is true and the quote is verbatim - what is wrong is whose result the corpus says it is. Nothing edited, nothing withdrawn: the row is marked so a page can name the source. Where the record is a commentary or review, its claim relations layer usually already names the paper being discussed - the record knows at one layer and forgets at another. Quotation withheld because this record is not cleared for redistribution. 16 s — duration for which a single cued tone had to be held in working memory Quotation withheld because this record is not cleared for redistribution. Seventeen healthy adults — number of participants recruited for the fMRI experiment Quotation withheld because this record is not cleared for redistribution. 67% to 98% (mean, 83.5%; SD, 10%) — range and mean of participants' overall task accuracy Quotation withheld because this record is not cleared for redistribution. first tone, 85.7%; second tone, 81.3% — accuracy advantage for retaining the first rather than the second tone of the pair Quotation withheld because this record is not cleared for redistribution. 0.04 — correlation between the encoding and maintenance regressors in the design matrix, used to argue maintenance activity is not contaminated by encoding Quotation withheld because this record is not cleared for redistribution. 100,000 — number of resamples used to build the null distribution for group classification accuracy Quotation withheld because this record is not cleared for redistribution. 0.0013 and 0.015 — nonparametric p values for above-chance decoding in HG and left inferior frontal gyrus, matching the parametric t test values Quotation withheld because this record is not cleared for redistribution. 0.04 — correlation between encoding and maintenance regressors, the measured bound on contamination Quotation withheld because this record is not cleared for redistribution. 55.61% and 54.25% — above-chance classification accuracy in HG and left IFG, where chance is 50% Quotation withheld because this record is not cleared for redistribution. 9 of 13 and 8 of 13 — subjects above chance in the two decodable regions Quotation withheld because this record is not cleared for redistribution. 84.42% — share of maintenance-surviving auditory voxels lying in PT Quotation withheld because this record is not cleared for redistribution. 67% to 98% — range of behavioural accuracy across subjects, the spread the correlations depend on 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/10.1523/jneurosci.4341-14.2016.