Inverted central auditory hierarchies for encoding local intervals and global temporal patterns

Inverted central auditory hierarchies for encoding local intervals and global temporal patterns Asokan Williamson Hancock Polley

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.

Inverted central auditory hierarchies for encoding local intervals and global temporal patterns 2021 Current Biology 10.1016/j.cub.2021.01.076 Asokan2021-inverted-central-auditory-hierarchies-for-en Asokan2021-inverted-central-auditory-hierarchies-for-en.txt, the project's staged copy 3ff98eb4c9173bb4aa4e3b358226a818a445f5d0 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 the timescale of neural activity changes from midbrain to cortex, and what each stage can therefore encode Quotation withheld because this record is not cleared for redistribution. whether the slow timescales that cost the cortex temporal precision are what let it encode global patterns that subcortical stations cannot Quotation withheld because this record is not cleared for redistribution. shows the auditory hierarchy is inverted between two encoding jobs - fidelity for local intervals falls from IC to A1 while sensitivity to global rhythm appears only in A1 - and that the cortical signal is carried by spike timing with no change in firing rate Quotation withheld because this record is not cleared for redistribution. existence Quotation withheld because this record is not cleared for redistribution. hierarchical temporal coding in the central auditory pathway Quotation withheld because this record is not cleared for redistribution. electrophysiology electrophysiology Quotation withheld because this record is not cleared for redistribution. mechanisms Dendrite-targeting cortical GABAergic interneurons dampen pyramidal-neuron spike decay when stimulus timing becomes predictable Quotation withheld because this record is not cleared for redistribution. mechanisms Corticothalamic feedback recruits thalamic reticular nucleus inhibition, which dampens feedforward thalamocortical drive during rhythms Quotation withheld because this record is not cleared for redistribution. mechanisms Cortex switches from a recurrent deliberative mode to a pass-through mode once incoming sound matches internal predictive models status: the paper's own words: may reflect Quotation withheld because this record is not cleared for redistribution. the cortex reads regularity by compressing its own spike decay constant, not by changing how much it fires Quotation withheld because this record is not cleared for redistribution. the timescale that changes is about 40x faster than the stimulus feature driving it, so the sensitivity must be inherited from other circuits rather than computed by the unit itself Quotation withheld because this record is not cleared for redistribution. candidate source: dendrite-targeting somatostatin or NDNF interneurons that dampen recurrent excitation and track slow state variables Quotation withheld because this record is not cleared for redistribution. alternative source: corticothalamic feedback damping thalamocortical input through the thalamic reticular nucleus, suggested by the weaker MGB effect Quotation withheld because this record is not cleared for redistribution. stimulus-specific adaptation, the nearest neighbouring phenomenon, works instead by synaptic fatigue for standard inputs while deviant inputs stay unadapted Quotation withheld because this record is not cleared for redistribution. peripheral speed is a design constraint: the morphologies and synapses that give the early stages sub-millisecond precision are what make them unsuited to slow structure Quotation withheld because this record is not cleared for redistribution. Cortical sensitivity to regularity did not extend to longer or more complex cycles, so the integration window demonstrated is bounded at roughly 0.8-1.6 s. what this limits: the regularity result, with its ceiling measured: SINGLE UNIT SENSITIVITY TO REGULARITY IS LIMITED TO STIMULUS TIMESCALES ON THE ORDER OF 4-8 REPEATING ELEMENTS OR 0.8-1.6s Quotation withheld because this record is not cleared for redistribution. Only A1 in passively listening mice was studied; higher-order fields and behaviourally engaged listening remain to be tested. what this limits: the generality beyond A1 and beyond passive listening: it would be important to test the CANDIDATE HIGHER-ORDER AREAS ... PARTICULARLY WHEN MICE ARE BEHAVIORALLY ENGAGED Quotation withheld because this record is not cleared for redistribution. The circuit that actually confers sensitivity to the ~800 ms regularity is not identified; candidate local and long-range mechanisms are only conjectured. what this limits: any mechanistic reading: the timescale dynamics ARE LIKELY THE PRODUCT OF OTHER LOCAL OR LONG-RANGE CIRCUITS THAT RESPOND MORE DIRECTLY TO THE CHANGE IN PREDICTABILITY - the circuit that confers the sensitivity is not identified Quotation withheld because this record is not cleared for redistribution. the cortical sensitivity has a measured ceiling - it holds for cycles of four to eight intervals and is gone at twelve what this limits: the same scope as limitations[0] - the effect holds for cycles of four to eight and is gone beyond Quotation withheld because this record is not cleared for redistribution. passive listening only; whether engagement in a task extends the integration window is untested duplicate of: limitations[1] · what this limits: the same scope as limitations[1], for engagement specifically - only passive listening was tested Quotation withheld because this record is not cleared for redistribution. only A1 was recorded, so the higher-order fields where longer regularities might be encoded were not sampled what this limits: the same scope as limitations[1], for anatomy specifically - only A1 was recorded, so the higher-order fields where longer regularities might be encoded were not sampled Quotation withheld because this record is not cleared for redistribution. cell class is inferred from spike waveform width, not from a molecular marker what this limits: every result reported by cell class: REGULAR and FAST SPIKING are assigned by THE TROUGH TO PEAK INTERVAL ... EXCEEDING 0.6ms, a waveform proxy rather than a molecular identification Quotation withheld because this record is not cleared for redistribution. the neural-timescale metric (decay constant of an exponential fit to the spike-count autocorrelation), developed for hierarchies of visual cortical areas and here applied to the auditory pathway Quotation withheld because this record is not cleared for redistribution. the autocorrelation-decay measure of neural timescale, and the expectation that it lengthens up a hierarchy Quotation withheld because this record is not cleared for redistribution. the paradigm of contrasting repeating with random acoustic patterns Quotation withheld because this record is not cleared for redistribution. the binary support vector machine, used here as a readout of ensemble state Quotation withheld because this record is not cleared for redistribution. a random-random stimulus, keeping the stimulus random for all 50 post-baseline cycles instead of switching from rhythmic to random (Figure S4 - three sentences describe this one control, recorded once) eliminates: the ordinal arrangement of random-following-rhythmic, rather than the change in regularity, as the cause of the spike-patterning differences · outcome (stated by the tagger, no quotation): held Quotation withheld because this record is not cleared for redistribution. repeating the whole analysis on cycle-order-shuffled data, for every recording in all three brain regions - the exclusion is carried by the word shuffling rather than spelled out, hence the lower weight eliminates: structure that would appear in the analysis regardless of the true ordering of cycles · outcome (stated by the tagger, no quotation): held Quotation withheld because this record is not cleared for redistribution. neural timescale = the decay constant of a single unit's spike-count autocorrelation, about 20ms in A1 - NOT the window over which the unit integrates, which is 800ms; the paper states the 40x gap itself Quotation withheld because this record is not cleared for redistribution. regular, sense 1 = a property of the stimulus sequence (intervals repeating in a fixed order), the paper's independent variable Quotation withheld because this record is not cleared for redistribution. regular, sense 2 = a waveform class of neuron (trough-to-peak > 0.6ms, putative pyramidal); same word, unrelated referent, both used throughout Quotation withheld because this record is not cleared for redistribution. context = the global statistics of the stimulus sequence (random vs rhythmic), not behavioural or task context Quotation withheld because this record is not cleared for redistribution. tonotopy here is an instrument, not an object: the frequency gradient and its reversal are used to certify that the probe is in A1 Quotation withheld because this record is not cleared for redistribution. hierarchy = the ascending anatomical pathway IC to MGB to A1, ordered by synaptic distance from the cochlea Quotation withheld because this record is not cleared for redistribution. claim_type: existence a second reading reaching the same value. Recorded: existence — 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: simultaneous three-probe extracellular single-unit recording from IC, MGB and A1 of awake head-fixed mice, with template-matching interval decoding and an SVM on principal components of ensemble spiking 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: emergent cortical sensitivity to the regularity of a sequence of intervals, expressed as compression of the spike decay constant a second reading DIFFERING from the recorded value. Recorded: hierarchical temporal coding in the central auditory pathway — 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. replicates → previously documented progressive loss of temporal fidelity along the ascending auditory pathway (reviews cited as refs 1, 2, 4) Quotation withheld because this record is not cleared for redistribution. extends → prior animal work on stimulus-specific adaptation (SSA) and contrast gain control (refs 17, 20-26) Quotation withheld because this record is not cleared for redistribution. builds-on → the neural-timescale measure (autocorrelation decay constant) established across visual cortical hierarchies (refs 6, 7) Quotation withheld because this record is not cleared for redistribution. builds-on → human work on cortical activity and perceptual awareness of complex repeating acoustic patterns (refs 15, 16) Quotation withheld because this record is not cleared for redistribution. distinguishes-from → stimulus-specific adaptation and contrast gain control Quotation withheld because this record is not cleared for redistribution. distinguishes-from → SSA and contrast gain control, on anatomical distribution differs in: THE AXIS IS ANATOMICAL DISTRIBUTION. 'while SSA, contrast gain control or other neural signatures of stimulus context changes are robustly expressed in the A1, Quotation withheld.' The known adaptation phenomena appear at every level; this one does not - so a cortical result that resembles them is distinguished by WHERE ELSE IT CAN BE FOUND rather than by what it looks like. Quotation withheld because this record is not cleared for redistribution. distinguishes-from → rate-based accounts of context-dependent adaptation differs in: THE AXIS IS WHAT CARRIES THE SIGNAL - RATE OR TEMPORAL PATTERN. 'Quotation withheld.' The existing accounts are not denied their data; they are denied coverage of a signal their measure cannot see. Quotation withheld because this record is not cleared for redistribution. extends → human work on cortical responses to repeating acoustic patterns Quotation withheld because this record is not cleared for redistribution. builds-on → visual-cortex measurements of neural timescale by autocorrelation decay Quotation withheld because this record is not cleared for redistribution. confirms → prior reports of declining temporal fidelity along the ascending auditory pathway Quotation withheld because this record is not cleared for redistribution. neural timescale —operationalised-as→ decay time constant neural timescale decay time constant strength of the relation: 1 · note: imported from visual-cortex hierarchy work; the quote names the source field as well as the measure Quotation withheld because this record is not cleared for redistribution. stimulus regularity —operationalised-as→ inter-burst intervals repeated across cycles stimulus regularity inter-burst intervals repeated across cycles strength of the relation: 1 Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but the span writes 'regular contexts' / intervals that 'repeated across cycles', not 'stimulus regularity'. Quotation withheld because this record is not cleared for redistribution. A1 —operationalised-as→ best frequency gradient A1 best frequency gradient strength of the relation: 0.9 · note: tonotopy as the criterion for having reached A1 - the same word is the object of study elsewhere in this corpus Quotation withheld because this record is not cleared for redistribution. regular spiking —operationalised-as→ trough to peak interval regular spiking trough to peak interval strength of the relation: 1 · note: cut at 0.6ms; fast spiking is the same measure on the other side of it Quotation withheld because this record is not cleared for redistribution. fast spiking —same-as→ putative parvalbumin-expressing interneuron fast spiking putative parvalbumin-expressing interneuron strength of the relation: 0.6 — the record marks this weak · note: an identity asserted by apposition and hedged by 'putative'; the evidence for it is waveform width, not a marker Quotation withheld because this record is not cleared for redistribution. context sensitivity —distinct-from→ SSA context sensitivity SSA strength of the relation: 1 · note: discriminated by anatomical distribution: SSA is expressed in IC and MGB too, this is absent in IC Quotation withheld because this record is not cleared for redistribution. context sensitivity —distinct-from→ contrast gain control context sensitivity contrast gain control strength of the relation: 1 · note: same demarcation paragraph; its fourth point, rate versus timing, is carried in claim relations Quotation withheld because this record is not cleared for redistribution. SSA —is-a→ spike rate adaptation SSA spike rate adaptation strength of the relation: 1 Quotation withheld because this record is not cleared for redistribution. decay constant —distinct-from→ mean firing rate decay constant mean firing rate strength of the relation: 1 · note: dissociated empirically in the same units and the same 100ms window, not merely conceptually Quotation withheld because this record is not cleared for redistribution. A1 —contrasted-with→ IC A1 IC strength of the relation: 1 · note: on rhythm classification the cortex succeeds and the midbrain fails - the reverse of the interval-decoding ordering Quotation withheld because this record is not cleared for redistribution. A1 —contrasted-with→ MGB A1 MGB strength of the relation: 1 Quotation withheld because this record is not cleared for redistribution. slow neural timescales —causes→ interfere with precise temporal coding slow neural timescales interfere with precise temporal coding strength of the relation: 0.9 · note: the cost. Same property and same sentence as the edge below, opposite valence - the sign lives in the target, not in a note Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but slow timescales 'MAY interfere' with precise temporal coding; causes is hedged. Quotation withheld because this record is not cleared for redistribution. slow neural timescales —causes→ cortical advantage slow neural timescales cortical advantage strength of the relation: 0.9 · note: the capability: an advantage over subcortical stations for encoding slowly evolving features Audit 2026-09-28: DOUBTFUL - read against its full span and kept, but the sluggish processing 'COULD SUGGEST' a cortical advantage; causes is hedged. Quotation withheld because this record is not cleared for redistribution. rhythmic patterns —causes→ shorter decay constants rhythmic patterns shorter decay constants strength of the relation: 1 · note: the headline effect, stated in the same sentence that rules out any change in mean firing rate Quotation withheld because this record is not cleared for redistribution. A1 —part-of→ IC-MGB-A1 hierarchy A1 IC-MGB-A1 hierarchy strength of the relation: 1 Quotation withheld because this record is not cleared for redistribution. IC —part-of→ ascending auditory pathway IC ascending auditory pathway strength of the relation: 1 Quotation withheld because this record is not cleared for redistribution. Kruskal-Wallis test Kruskal-Wallis test Quotation withheld because this record is not cleared for redistribution. ANOVA ANOVA Quotation withheld because this record is not cleared for redistribution. ANOVA ANOVA 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. effect size effect size Quotation withheld because this record is not cleared for redistribution. normality test normality test Quotation withheld because this record is not cleared for redistribution. multiple-comparison correction multiple-comparison correction Quotation withheld because this record is not cleared for redistribution. principal component analysis principal component analysis Quotation withheld because this record is not cleared for redistribution. support vector machine classification support vector machine classification Quotation withheld because this record is not cleared for redistribution. cross-validation cross-validation Quotation withheld because this record is not cleared for redistribution. Kruskal-Wallis Kruskal-Wallis Quotation withheld because this record is not cleared for redistribution. Cliff's delta Cliff's delta Quotation withheld because this record is not cleared for redistribution. Bonferroni correction Bonferroni correction Quotation withheld because this record is not cleared for redistribution. one-sample t-test one-sample t-test Quotation withheld because this record is not cleared for redistribution. Cohen's d Cohen's d Quotation withheld because this record is not cleared for redistribution. two-way repeated measures ANOVA two-way repeated measures ANOVA 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. support vector machine support vector machine Quotation withheld because this record is not cleared for redistribution. Neural timescales, measured as the autocorrelation decay constant, roughly double from auditory midbrain to cortex. Quotation withheld because this record is not cleared for redistribution. Accuracy of decoding the interval between two noise bursts falls progressively from midbrain to thalamus to cortex. Quotation withheld because this record is not cleared for redistribution. Only cortical ensemble activity, not thalamic or midbrain activity, allowed a classifier to tell rhythmic from random interval sequences. Quotation withheld because this record is not cleared for redistribution. Rhythmic context shortened neural timescales strongly and reliably in A1, weakly in MGB, and not at all in IC. Quotation withheld because this record is not cleared for redistribution. Regularity was encoded by the arrangement of spikes in time rather than by any change in mean firing rate. Quotation withheld because this record is not cleared for redistribution. Cortical sensitivity to regularity was bounded, disappearing for rhythmic cycles longer than roughly 4 to 8 repeating intervals. Quotation withheld because this record is not cleared for redistribution. The timescale compression occurred in putative pyramidal (regular-spiking) neurons across layers but not in putative parvalbumin interneurons. Quotation withheld because this record is not cleared for redistribution. A random-random control sequence produced chance-level classification and no timescale change, ruling out simple stimulus-order effects. Quotation withheld because this record is not cleared for redistribution. neural timescales roughly double from midbrain to cortex Quotation withheld because this record is not cleared for redistribution. the median interval-decoding threshold rises from 1.4ms in IC to 16ms in MGB to 64ms in A1 Quotation withheld because this record is not cleared for redistribution. ensemble spiking distinguishes rhythmic from random sequences in A1 but not in MGB or IC Quotation withheld because this record is not cleared for redistribution. the discriminating signal is spike arrangement in time; mean firing rate does not change Quotation withheld because this record is not cleared for redistribution. in A1, decay constants were slightly reduced for rhythms of eight repeating intervals and did not differ between random and rhythmic epochs with twelve Quotation withheld because this record is not cleared for redistribution. the timescale compression appears in regular-spiking units across all A1 layers and not at all in fast-spiking units Quotation withheld because this record is not cleared for redistribution. under regular input the cortex comes to resemble the thalamus on the two timing measures Quotation withheld because this record is not cleared for redistribution. a random-to-random control produced near-chance classification, ruling out the ordinal position of the epoch Quotation withheld because this record is not cleared for redistribution. decoder accuracy threshold decoder accuracy threshold Quotation withheld because this record is not cleared for redistribution. neural timescale asymmetry index neural timescale asymmetry index Quotation withheld because this record is not cleared for redistribution. spike waveform classification spike waveform classification Quotation withheld because this record is not cleared for redistribution. SSA SSA Quotation withheld because this record is not cleared for redistribution. neural timescale neural timescale Quotation withheld because this record is not cleared for redistribution. neural timescale asymmetry index neural timescale asymmetry index Quotation withheld because this record is not cleared for redistribution. template matching template matching Quotation withheld because this record is not cleared for redistribution. regular spiking regular spiking Quotation withheld because this record is not cleared for redistribution. hierarchical stage hierarchical stage Quotation withheld because this record is not cleared for redistribution. cycle size cycle size Quotation withheld because this record is not cleared for redistribution. neural timescale neural timescale Quotation withheld because this record is not cleared for redistribution. decoder threshold decoder threshold Quotation withheld because this record is not cleared for redistribution. mean firing rate mean firing rate Quotation withheld because this record is not cleared for redistribution. random-random stimulus random-random stimulus Quotation withheld because this record is not cleared for redistribution. stimulus context stimulus context Quotation withheld because this record is not cleared for redistribution. spike decay time constant spike decay time constant Quotation withheld because this record is not cleared for redistribution. sound-evoked firing rate sound-evoked firing rate Quotation withheld because this record is not cleared for redistribution. 1.4ms, 16ms, 64ms — median interval-decoding threshold in inferior colliculus, medial geniculate body and primary auditory cortex respectively Quotation withheld because this record is not cleared for redistribution. 0.8s — duration of one complete stimulus cycle, the minimum integration window needed to detect regularity Quotation withheld because this record is not cleared for redistribution. 40x — how much slower the stimulus feature (800ms rhythm) is than the cortical spiking timescale (20ms) it modulates Quotation withheld because this record is not cleared for redistribution. F = 20.95, p = 6x10 −9 — main effect of brain structure on interval classification accuracy in the mixed-design ANOVA Quotation withheld because this record is not cleared for redistribution. 5/164 — number of mice and single units contributing A1 timescale measurements in the random versus rhythmic comparison Quotation withheld because this record is not cleared for redistribution. 1.4ms, 16ms, 64ms — median interval-decoding threshold in IC, MGB and A1 Quotation withheld because this record is not cleared for redistribution. 0.8 - 1.6s — the span of stimulus regularity A1 single units are sensitive to Quotation withheld because this record is not cleared for redistribution. approximately 40x — how much slower the driving stimulus feature is than the timescale that reports it (800ms vs 20ms) Quotation withheld because this record is not cleared for redistribution. 0.6ms — trough-to-peak cut between regular- and fast-spiking units Quotation withheld because this record is not cleared for redistribution. 200ms — mean inter-burst interval, with no interval below 100ms so that forward suppression cannot contribute Quotation withheld because this record is not cleared for redistribution. 20 ms, 70 dB SPL, 1-64 kHz — duration, level and bandwidth of the frozen noise burst used throughout Quotation withheld because this record is not cleared for redistribution. less than 5% — permitted refractory-period violations for a cluster to count as a single unit Quotation withheld because this record is not cleared for redistribution. 6-8 weeks — age of the C57BL6 mice at recording 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/10.1016/j.cub.2021.01.076.