Pith. sign in

REVIEW 4 major objections 6 minor 28 references

Stimulus-Evoked Network Dynamics in Human Cortical Organoids: From a Graph-Computational Framework to Repeated-Stimulation Depression

T0 review · 4 major / 6 minor · reviewed 2026-07-31 · grok-4.5

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗

Pith's one-line read Repeated daily stimulation, not age, collapses the evoked network in human cortical organoids from nearly the whole array to about 10%.

desk verdict Honest negative result on propagation metrics plus a real but thin control contrast showing repeated stimulation, not age, collapses the evoked network. read the letter →

arxiv 2607.28068 v1 pith:MNX3SBX2 submitted 2026-07-30 q-bio.NC cs.AI

Esmaeil S. Nadimi , Vinay C. Gogineni , Jan-Matthias Braun , Martin R{o}ssel Larsen , Victoria Blanes-Vidal , Helle Bogetofte Barnkob This is my paper · ORCID
classification q-bio.NCcs.AI
keywords corticalorganoidsHD-MEAstimulus-evokeddynamicsnetworksynchronyrepeatedstimulationhabituationfunctionalgraphsmessagepassing
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

This paper set out to measure whether stimulus-evoked activity in human cortical organoids spreads as structured multi-hop computation on a graph, using functional graphs, a graph-neural dynamical model, and a biological message-passing bound on integration depth. Once sampling rate and stimulus timing were correctly recovered, the evoked response was a fast, near-synchronous network burst with no latency-versus-distance gradient, so the propagation metrics do not apply. Reframing around response size, synchrony, and shared variability, the authors show that repeated daily stimulation progressively depresses and spatially contracts the responding population. A developmentally matched organoid stimulated for the first time only at day 7 still engaged about 93% of the array, while organoids with five prior sessions engaged about 10%, separating stimulation history from ordinary maturation. First-trial capacity stayed relatively intact while within-session endurance collapsed, and coherence among remaining responders was preserved.

What carries the argument

The developmentally matched, stimulation-naive control (same age, first stimulated only at day 7), which breaks the confound between repeated stimulation and ordinary maturation and thereby carries the causal claim about stimulation history.

What would settle it

Add more developmentally matched, never-before-stimulated organoids at day 7: if they also engage only ~10% of the array, or if repeatedly stimulated organoids fail to contract relative to multiple naive controls, the history-versus-maturation dissociation fails.

Watch

Extended reading notes

Core claim

Under focal electrical stimulation on HD-MEAs, human cortical organoids produce a near-synchronous network-wide burst rather than an outward-propagating wave, so propagation-based graph metrics (effective integration depth, reachability, dmax) are not applicable. Independently, repeated daily stimulation drives a progressive depression and spatial contraction of the evoked response that is attributable to stimulation history: at matched age, a first-ever stimulation engaged ~93% of electrodes versus ~10% after five prior sessions, with fresh first-trial capacity largely preserved while session endurance and responsive population size collapsed.

Load-bearing premise

The claim that stimulation history, not age, causes the collapse rests on a single stimulation-naive control organoid alongside two repeatedly stimulated ones.

Editorial extensions

If this is right

  • Propagation-based graph metrics (reachability depth, Deff ≥ dmax) should be applied to organoid or culture data only after an empirical latency-versus-distance check shows real outward spread.
  • Longitudinal stimulation studies that stimulate every preparation cannot attribute network change to plasticity versus maturation without a stimulation-naive age-matched control.
  • Organoid response engineering should track first-trial capacity and within-session endurance separately, because repeated drive mainly degrades endurance and recruitable population size.
  • At ~10 trials, edge-level functional connectivity graphs over thousands of electrodes are not reliably thresholdable; stable claims should rest on aggregate measures such as responsive-population size and PC1 coherence.
  • If the depression is reversible after rest, it points to homeostatic adaptation; if durable, it points to lasting reconfiguration of the recruitable pool.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Protocols that use daily electrical drive as a training or entrainment signal in organoid biocomputing may be silently shrinking the usable network unless recovery intervals are built in.
  • The preserved coherence of the remaining day-7 core suggests a shrinking recruitable periphery rather than global desynchronization, which predicts that multi-site or weaker stimuli might re-engage dropped units more than stronger single-site drive.
  • A systematic inter-stimulus-interval sweep would map the recovery time constant of the fast within-session drop and test whether the slow across-day contraction shares the same recovery kinetics.
  • If desynchronization truly tracks prior stimulation history while magnitude depression is intrinsic, timing precision could serve as a cheap longitudinal marker of cumulative drive load.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

4 major / 6 minor

Summary. The authors propose a graph-computational framework (stimulus-conditioned functional graphs, a GNN system-identification model, a biological message-passing bound Deff ≥ dmax, and metrics EIC/RI/SSI/PR/CRI) to test whether stimulus-evoked activity in human cortical organoids reflects multi-hop propagation. Applied to longitudinal HD-MEA recordings from three organoids, the framework’s propagation premises fail: after correcting sampling rate and recovering stimulus timing, the evoked response is a near-synchronous network burst (peak-latency vs. distance slope ≈ 0), so Deff, RI, and dmax are inapplicable and per-day edge graphs are unreliable at K=10. Reframing around magnitude, synchrony, responsive-population size, and PC1 coherence, they report that repeated daily stimulation progressively depresses and spatially contracts the evoked response. A developmentally matched, stimulation-naïve control (organoid 612) at day 7 engaged ~93% of the array versus ~10% in two organoids with five prior sessions, with first-trial capacity relatively preserved while within-session endurance collapsed.

Significance. If the history-versus-maturation dissociation holds, the work supplies a needed control design for organoid stimulation studies that otherwise confound plasticity with development, and it cleanly dissociates preserved first-trial capacity from degrading within-session endurance and spatial contraction of the responsive pool. The transparent negative result on propagation-based graph metrics—and the explicit recommendation to verify latency–distance structure before applying Deff/RI/dmax-type measures—is a genuine methodological contribution for HD-MEA organoid work. Strengths include full execution and then principled retirement of the original metric suite (Table 4), data-driven stimulus-time recovery, and control-supported rather than over-claimed population inference. The main limit on significance is that the load-bearing causal claim rests on a single naïve control and n=2 repeated preparations.

major comments (4)
  1. [§3.2–3.5, Tables 2–3, §4.1, §4.7] The central claim that day-7 collapse is driven by stimulation history rather than age (§3.2–3.5, §4.1; Tables 2–3) is identified by a single developmentally matched naïve control (612: ratio 1914, n_resp=3788) against two repeatedly stimulated organoids (~206/268; 384/486). Section 4.7 correctly flags n=1 control, but the manuscript still treats this contrast as load-bearing. 612’s day-7 response is the strongest in the dataset; without a second naïve control, pre-stim spontaneous baselines establishing exchangeability of 612 with 552/613, or batch/plating covariates, atypical viability or coupling in 612 remains a viable alternative. Either add ≥1 naïve control or substantially soften causal language to a single-preparation, hypothesis-generating contrast.
  2. [Table 3, §3.2, §3.7] Organoid 613’s responsive trajectory is highly non-monotonic (42%→72%→14%→100%→12%; Table 3), with day 3 lowest total spikes and day-4 recordings truncated to 8–9 trials (§3.7). The paper notes these anomalies but still pools 613 with 552 for the repeated-stimulation narrative. Intermediate-day claims and any trajectory shape (rise-then-fall) should be restricted to 552 or explicitly sensitivity-tested excluding D3/D4 of 613; the shared D7 endpoint is the only robust cross-preparation result for the repeated arm.
  3. [Theorem 1, Eq. (2), §2.4–2.5, §3.1, Table 4] Theorem 1 / Eq. (2) (Deff ≥ dmax) and the GNN depth pipeline (§2.4, Eqs. 8–10) occupy substantial Methods and Introduction space, yet §3.1 and Table 4 correctly retire them because there is no measurable propagation. The paper does not report the actual Lk-vs-k curves, cross-validated Deff values, or baseline comparisons that were computed before retirement. Either move the unused GNN/BMP apparatus to Supplementary with a brief empirical null report, or show the system-identification results that justify having run the full program—otherwise the framing overpromises relative to the retained read-outs (n_resp, M, S, PC1, Σ).
  4. [§3.4, §4.5] Within-session desynchronization is advanced as possibly history-dependent because naïve 612 showed magnitude depression without significant desynchronization while repeatedly stimulated 613 desynchronized (§3.4, §4.5). With n=1 control and mixed patterns in 552, this should remain clearly labeled exploratory; any abstract/discussion wording that pairs it with the control-validated population-size effect should be separated so the stronger n_resp/spatial-contraction result is not diluted by an underpowered secondary hypothesis.
minor comments (6)
  1. [Abstract, §1.3] Abstract and §1.3 state the control engaged ~93% vs ~10%; give electrode counts alongside percentages for consistency with Tables 2–3.
  2. [§2.2] Eq. (3)–(4): state whether the MAD threshold and δ=0.25Δ were fixed a priori or tuned; a one-sentence sensitivity note would help reproducibility.
  3. [§2.5, §4.8] SSI (Eq. 14) and multi-site stimulation are discussed as future work (§4.8) after being defined in the main metric suite; a forward reference in §2.5 would reduce the sense that unused metrics are primary deliverables.
  4. [Table 2, §2.6] Table 2 header “response/baseline ratio” should specify the exact aggregation (session sum vs mean of per-trial Mk) to match Eqs. (17)–(20).
  5. [Abstract, passim] Minor typographical issues: missing spaces in several compounds (e.g., “stimulation-naïvecontrol”, “whereatday7”, “~93%of”); standardize throughout.
  6. [§4.1, §4.3] Cite the specific prior longitudinal stimulation studies [19, 27] with a clearer statement of what design element (naïve age-matched control; capacity vs endurance split) is new versus confirmatory.

Circularity Check

0 steps flagged · score 0.0 of 10

No significant circularity: the BMP bound is set aside after an empirical null, and the depression claim is a control contrast, not a fit or definitional restatement.

full rationale

The paper’s load-bearing chain is empirical, not definitional. Sampling-rate and stimulus-onset recovery are data-driven preprocessing; the latency–distance slope≈0 is a direct measurement that falsifies the premise of Theorem 1, after which Deff, RI, and dmax are explicitly retired (Table 4, §3.1, §3.8). That is the opposite of a circular success: the framework is not forced to fit the data and is not used to underwrite the main claim. The positive result—day-7 responsive fraction ~93% in the stimulation-naïve control versus ~10% after five prior sessions, with preserved first-trial magnitude and collapsing session mean—is a between-preparation contrast on descriptive read-outs (M, nresp, Σ, PC1), not a parameter fitted on one subset and relabeled as a prediction. PC1 on linearly detrended trial matrices and FDR-thresholded responsive sets are standard summary statistics; they do not encode a generative target that the same quantities then “predict.” References are external (Shannon, graph theory, organoid MEA literature); there is no self-citation uniqueness theorem or author-imported ansatz carrying the argument. Small-n and single-control limitations affect causal strength, not circularity. Score 0 is therefore appropriate.

Assumptions & free parameters 6 free parameters · 5 assumptions · 3 invented entities

The positive claim is experimental, not a long formal derivation. It still depends on domain premises about what the control matches, on analysis thresholds that define bursts and responsive electrodes, and on invented graph quantities that were mostly retired after the synchrony finding. Free parameters are preprocessing and statistics choices; they can move n_resp and slopes but are not tuned to invent the day-7 control gap in an obvious closed loop.

free parameters (6)
  • Population burst threshold multiplier (median + 5×1.4826×MAD) = 5 × MAD rule (Eq. 3)
    Sets which rate peaks count as candidate evoked bursts for stimulus-time recovery; changes which trials enter all downstream metrics.
  • Stimulus grid match tolerance δ = 0.25Δ = 0.25 × 30 s
    Window for assigning rate peaks to the 30 s protocol grid; affects onset alignment and apparent failures.
  • Responsive-electrode FDR q and one-sided paired t across K=10 trials = q=0.05, K≈10
    Defines V_resp and thus the headline array-fraction collapse; q=0.05 and trial count directly gate who counts as responsive.
  • Response/baseline windows Wr=[0,200] ms, Wb=[-250,-50] ms = 200 ms equal-width windows
    Fixed analysis windows for magnitude, synchrony, and dimensionality; not cross-validated in text.
  • GNN depth selection and training hyperparameters = lr=1e-3, wd=1e-4, split 70/15/15
    Adam lr 1e-3, weight decay 1e-4, 70/15/15 split, ReLU/Xavier; Deff = argmin_k L_k would matter if propagation metrics applied.
  • Edge inclusion permutation p<0.05 FDR on evoked cross-correlation = p<0.05 FDR
    Governs stimulus-conditioned graphs later judged unreliable at K=10; threshold choice interacts with low trial count.
assumptions (5)
  • domain assumption Organoid 612 is developmentally matched to 552/613 at day 7 except for stimulation history (same culture timing context; shared stim geometry with 613).
    Load-bearing for attributing 93% vs 10% engagement to stimulation history rather than batch, viability, or placement; stated in §§1.3, 2.1, 4.1.
  • domain assumption Peak-latency versus distance slope ≈ 0 implies no measurable multi-hop outward propagation, so dmax/Deff/RI tests are inapplicable.
    Justifies retiring the original metric suite (§3.1, Table 4); standard for wave-vs-synchrony interpretation in cultures.
  • ad hoc to paper Theorem 1: if I(S;Xv)>0 and dist_G(vs,v)=k then Deff ≥ dmax (biological message-passing bound).
    Paper-proposed bound linking channel-capacity style preservation to graph hops (§1.2.1); premise fails empirically here when there is no distance-ordered activation.
  • domain assumption At K=10 trials, pairwise evoked correlations are not stably thresholdable (permutation |r| thresholds ~0.98), so edge graphs are non-primary.
    §3.6 statistical ceiling; standard multiple-comparison / low-N correlation issue.
  • domain assumption HD-MEA multi-unit electrode channels can stand in as graph nodes for population readouts (n_resp, PC1, magnitude).
    Throughout Methods; common MEA abstraction, not single-neuron resolved.
invented entities (3)
  • Biological message-passing theorem / bound (Deff ≥ dmax)
    purpose: Link observable propagation depth to required GNN integration depth as a testable systems-identification principle.
    Introduced in §1.2.1 as Theorem 1; not validated here because evoked activity lacks outward propagation structure.
  • Effective integration capacity (EIC), reachability index (RI), stimulus separability index (SSI), computational reconfiguration index (CRI)
    purpose: Graph-computational metric suite for integration depth, reach, encoding, and plasticity.
    Defined in §2.5; EIC/RI/SSI largely N/A or unestimated; CRI/PR retained only in reframed form (Table 4).
  • Stimulus-conditioned functional dynamic graphs Gt with precedence-constrained evoked cross-correlation weights
    purpose: Represent post-stimulus effective interactions for message-passing models.
    §2.3 construction carried out but not reliable as primary evidence at available trial count (§3.6).

how reviews work

0 comments
Cite this review

Pith. "Pith review of Stimulus-Evoked Network Dynamics in Human Cortical Organoids: From a Graph-Computational Framework to Repeated-Stimulation Depression." pith.science (2026). https://pith.science/paper/MNX3SBX2

@misc{pith2026260728068,
  author       = {Pith},
  title        = {Pith review of: Stimulus-Evoked Network Dynamics in Human Cortical Organoids: From a Graph-Computational Framework to Repeated-Stimulation Depression},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/MNX3SBX2}},
  note         = {Machine review of arXiv:2607.28068}
}
read the original abstract

Human cortical organoids provide an experimentally accessible model of early neural circuit formation, yet whether their activity reflects structured information processing rather than spontaneous synchronization is unclear. We developed a graph-computational framework to quantify stimulus-evoked propagation. This includes stimulus-conditioned functional graphs, a graph-constrained dynamical (graph-neural-network) model used as a system-identification tool, a biological message-passing principle bounding integration depth by observable propagation depth, and a suite of graph-level metrics. We carried this program out in full on longitudinal HD-MEA recordings from three organoids. Once the true acquisition sampling rate and stimulus timing were recovered, the evoked response proved to be a fast, near-synchronous network burst with no measurable outward propagation (peak-latency vs. distance slope = 0). The propagation/integration-depth metrics (Deff ,reachability index, dmax) therefore do not apply, and per-day connectivity graphs were not reliably estimable at the available trial count, a negative result with methodological consequences for applying such metrics to organoid data. Reframing around synchrony, response-population size and shared variability revealed a control-validated phenomenon, i.e., repeated daily stimulation progressively depressed and spatially contracted the evoked response. That repeated stimulation reshapes organoid networks is established, but longitudinal designs in which every preparation is stimulated cannot separate this from developmental maturation. We break that confound with a developmentally-matched, stimulation-naive control, where at day 7, an organoid receiving its first-ever stimulation engaged 93% of the array, whereas organoids with five prior sessions engaged 10%.

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

28 extracted references · 10 canonical work pages

  1. [1]

    Bullmore, E., & Sporns, O. (2009). Complex brain networks: graph theoretical analysis of structural and functional systems.Nature Reviews Neuroscience10(3), 186–198.https:// doi.org/10.1038/nrn2575

    doi:10.1038/nrn2575 2009

  2. [2]

    Shannon, C.E. (1948). A Mathematical Theory of Communication.Bell System Technical Jour- nal27(3), 379–423.https://doi.org/10.1002/j.1538-7305.1948.tb01338.x

    arXiv 1948

  3. [3]

    Kipf, T.N., & Welling, M. (2017). Semi-Supervised Classification with Graph Con- volutional Networks.5th International Conference on Learning Representations (ICLR). arXiv:1609.02907

    arXiv 2017

  4. [4]

    Gilmer, J., Schoenholz, S.S., Riley, P.F., Vinyals, O., & Dahl, G.E. (2017). Neural Message Passing for Quantum Chemistry.Proceedings of the 34th International Conference on Machine Learning (ICML), PMLR70, 1263–1272

    2017

  5. [5]

    Berdondini, L., Imfeld, K., Maccione, A., Tedesco, M., Neukom, S., Koudelka-Hep, M., & Martinoia, S. (2009). Active pixel sensor array for high spatio-temporal resolution electrophys- iological recordings from single cell to large scale neuronal networks.Lab on a Chip9(18), 2644–2651.https://doi.org/10.1039/b907394a

    work page doi:10.1039/b907394a 2009

  6. [6]

    org/10.3389/fnins.2016.00121

    Amin, H., Maccione, A., Marinaro, F., Zordan, S., Nieus, T., &Berdondini, L.(2016).Electrical responses and spontaneous activity of human iPS-derived neuronal networks characterized for 3-month culture with 4096-electrode arrays.Frontiers in Neuroscience10, 121.https://doi. org/10.3389/fnins.2016.00121

    arXiv 2016

  7. [7]

    Litwin-Kumar, A., Harris, K.D., Axel, R., Sompolinsky, H., & Abbott, L.F. (2017). Optimal Degrees of Synaptic Connectivity.Neuron93(5), 1153–1164.e7.https://doi.org/10.1016/ j.neuron.2017.01.030

    2017

  8. [8]

    Gao, P., Trautmann, E., Yu, B.M., Santhanam, G., Ryu, S., Shenoy, K., & Ganguli, S. (2017). A theory of multineuronal dimensionality, dynamics and measurement.bioRxiv214262.https: //doi.org/10.1101/214262

    doi:10.1101/214262 2017

Show all 28 references
  1. [9]

    Mazzucato, L., Fontanini, A., & La Camera, G. (2016). Stimuli Reduce the Dimensionality of Cortical Activity.Frontiers in Systems Neuroscience10, 11.https://doi.org/10.3389/ fnsys.2016.00011 16

    2016

  2. [10]

    Trujillo, C.A., Gao, R., Negraes, P.D., Gu, J., Buchanan, J., Preissl, S., Wang, A., Wu, W., Haddad, G.G., Chaim, I.A., Domissy, A., Vandenberghe, M., Devor, A., Yeo, G.W., Voytek, B., & Muotri, A.R. (2019). Complex Oscillatory Waves Emerging from Cortical Organoids Model Earl...

    2019 doi

  3. [11]

    Cai, H., Ao, Z., Tian, C., Wu, Z., Liu, H., Tchieu, J., Gu, M., Mackie, K., & Guo, F. (2023). Brain organoid reservoir computing for artificial intelligence.Nature Electronics6, 1032–1039. https://doi.org/10.1038/s41928-023-01069-w

    2023 doi

  4. [12]

    Kagan, B.J., Kitchen, A.C., Tran, N.T., Habibollahi, F., Khajehnejad, M., Parker, B.J., Bhat, A., Rollo, B., Razi, A., & Friston, K.J. (2022). In vitro neurons learn and exhibit sentience when embodied in a simulated game-world.Neuron110(23), 3952–3969.e8.https://doi.org/ 10.1...

    2022 doi

  5. [13]

    Zucker, R.S., & Regehr, W.G. (2002). Short-Term Synaptic Plasticity.Annual Review of Phys- iology64, 355–405.https://doi.org/10.1146/annurev.physiol.64.092501.114547

    2002 arXiv

  6. [14]

    Turrigiano, G.G., Leslie, K.R., Desai, N.S., Rutherford, L.C., & Nelson, S.B. (1998). Activity- dependent scaling of quantal amplitude in neocortical neurons.Nature391(6670), 892–896. https://doi.org/10.1038/36103

    1998 doi

  7. [15]

    Turrigiano, G.G., & Nelson, S.B. (2004). Homeostatic plasticity in the developing nervous system.Nature Reviews Neuroscience5(2), 97–107.https://doi.org/10.1038/nrn1327

    2004 doi

  8. [16]

    Poli, D., Pastore, V.P., & Massobrio, P. (2015). Functional connectivity in in vitro neuronal assemblies.Frontiers in Neural Circuits9, 57.https://doi.org/10.3389/fncir.2015.00057

    2015

  9. [17]

    Benjamini, Y., & Hochberg, Y. (1995). Controlling the False Discovery Rate: A Practical and Powerful Approach to Multiple Testing.Journal of the Royal Statistical Society: Series B (Methodological)57(1), 289–300.https://doi.org/10.1111/j.2517-6161.1995.tb02031.x

    1995

  10. [18]

    1968.10480934

    Sen, P.K.(1968).EstimatesoftheRegressionCoefficientBasedonKendall’sTau.Journal of the American Statistical Association63(324), 1379–1389.https://doi.org/10.1080/01621459. 1968.10480934

    1968

  11. [19]

    Osaki, T., Duenki, T., Chow, S.Y.A., Ikegami, Y., Beaubois, R., Levi, T., Nakagawa- Tamagawa, N., Hirano, Y., & Ikeuchi, Y. (2024). Complex activity and short-term plasticity of human cerebral organoids reciprocally connected with axons.Nature Communications15, 2945.https://do...

    2024 doi

  12. [20]

    AlamElDin, D.-M., Moenkemoeller, L., Loeffler, A., Habibollahi, F., Schenkman, J., Mitra, A., van der Molen, T., Ding, L., Laird, J., Schenke, M., Johnson, E.C., Kagan, B.J., Hartung, T., & Smirnova, L. (2025). Human neural organoid microphysiological systems show the building...

    2025 doi

  13. [21]

    Sharf, T., van der Molen, T., Glasauer, S.M.K., Guzman, E., Buccino, A.P., Luna, G., Cheng, Z., Audouard, M., Ranasinghe, K.G., Kudo, K., Nagarajan, S.S., Tovar, K.R., Pet- zold, L.R., Hierlemann, A., Hansma, P.K., & Kosik, K.S. (2022). Functional neuronal cir- cuitry and osci...

    2022 doi

  14. [22]

    Smirnova, L., Caffo, B.S., Gracias, D.H., Huang, Q., Morales Pantoja, I.E., Tang, B., Zack, D.J., Berlinicke, C.A., Boyd, J.L., Harris, T.D., Johnson, E.C., Kagan, B.J., Kahn, J., Muotri, A.R., Paulhamus, B.L., Schwamborn, J.C., Plotkin, J., Szalay, A.S., Vogelstein, J.T., Wor...

    2023

  15. [23]

    Smirnova, L. (2024). Biocomputing with organoid intelligence.Nature Reviews Bioengineering. https://doi.org/10.1038/s44222-024-00200-6

    2024 doi

  16. [24]

    Brown, P., & Varghese, S. (2026). Computing inspired by the brain: a journey from algo- rithmstoorganoids.Nature Computational Science6(7), 686–696.https://doi.org/10.1038/ s43588-026-01012-x

    2026

  17. [25]

    Robbins, A., Schweiger, H.E., Hernandez, S., Spaeth, A., Voitiuk, K., Parks, D.F., van der Molen, T., Geng, J., Cline, I., Kosik, K.S., Salama, S.R., Sharf, T., Mostajo-Radji, M.A., Haussler, D., & Teodorescu, M. (2026). Goal-directed learning in cortical organoids.Cell Re- po...

    2026

  18. [26]

    Duenki, T., & Ikeuchi, Y. (2026). Multi-organoid loop cerebral connectoids exhibit enhanced neuronal network dynamics and sequence-specific entrainment.Communications Biology9, 302.https://doi.org/10.1038/s42003-026-09589-9

    2026 doi

  19. [27]

    Chow, S.Y.A., Hu, H., Duenki, T., et al. (2025). Repetitive stimulation modifies network characteristics of neural organoid circuits.bioRxiv2025.01.16.633310.https://doi.org/10. 1101/2025.01.16.633310

    2025

  20. [28]

    Schröter, M., et al. (2025). Advances in large-scale electrophysiology with high-density micro- electrode arrays.Lab on a Chip25, 4844–4885.https://doi.org/10.1039/D5LC00058K 18

    2025 doi

Pith tools

Reviewed July 31, 2026 · model on record in the stance chip above.