Version history

{
  "scope": "spiking neural network model of L2/3 cortex, neuromorphic hardware (TrueNorth)",
  "claim_text": "A biologically constrained spiking model of layer 2/3 incorporating four canonical interneuron classes (PV, SST, VIP, LAMP5) implements soft winner-take-all dynamics, with VIP transiently disinhibiting SST and LAMP5 providing global gain normalization.",
  "raw_fields": {
    "n": null,
    "id": "cluster_11_finding_01",
    "doi": "10.1073/pnas.2504164122",
    "pmid": "41055996",
    "year": "2025",
    "claim": "A biologically constrained spiking model of layer 2/3 incorporating four canonical interneuron classes (PV, SST, VIP, LAMP5) implements soft winner-take-all dynamics, with VIP transiently disinhibiting SST and LAMP5 providing global gain normalization.",
    "pmcid": "PMC12541343",
    "title": "Biologically grounded neocortex computational primitives implemented on neuromorphic hardware improve vision transformer performance.",
    "authors": "Iqbal A, Mahmood H, Stuart GJ, Fishell G, Honnuraiah S.",
    "journal": "Proceedings of the National Academy of Sciences of the United States of America",
    "cite_key": "Iqbal2025",
    "evidence": "Authors built a 4-interneuron-type spiking network and embedded sWTA, gain control, disinhibition, and normalization motifs onto neuromorphic hardware to improve vision-transformer performance.",
    "effect_size": null,
    "text_access": "fulltext",
    "study_system": "spiking neural network model of L2/3 cortex, neuromorphic hardware (TrueNorth)",
    "_source_cluster": "cluster_11_computational_models",
    "replication_status": "replicated_independent",
    "_source_cluster_index": 0,
    "claim_source_sentence": "In sensory cortex, pyramidal neurons integrate bottom–up and top–down signals under the modulatory influence of key interneuron classes: parvalbumin (PV), somatostatin (SST), vasoactive intestinal peptide (VIP), and LAMP5-expressing neurogliaform cells each of which mediates distinct computational roles including gain control, disinhibition, and normalization respectively ( 6 – 12 ).",
    "replication_evidence_dois": [
      "10.1371/journal.pcbi.1013469",
      "10.1371/journal.pcbi.1012036",
      "10.1093/cercor/bhae378",
      "10.1007/s00422-021-00894-6",
      "10.7554/elife.77594"
    ],
    "effect_size_source_sentence": null
  },
  "section_id": "section_12",
  "source_url": "https://github.com/AllenNeuralDynamics/ComputationalReviewVIP/blob/95e761177f7d2ec565983d3307c14ec238f9677c/evidence/section_12_evidence_package.json",
  "effect_size": null,
  "review_repo": "ComputationalReviewVIP",
  "section_ref": "wiki_page:computationalreviewvip-12-computational-models",
  "source_kind": "review_finding",
  "source_path": "evidence/section_12_evidence_package.json",
  "source_refs": [
    "paper:paper-733c0535ffcd"
  ],
  "source_span": "In sensory cortex, pyramidal neurons integrate bottom–up and top–down signals under the modulatory influence of key interneuron classes: parvalbumin (PV), somatostatin (SST), vasoactive intestinal peptide (VIP), and LAMP5-expressing neurogliaform cells each of which mediates distinct computational roles including gain control, disinhibition, and normalization respectively ( 6 – 12 ).",
  "study_system": "spiking neural network model of L2/3 cortex, neuromorphic hardware (TrueNorth)",
  "evidence_refs": [
    {
      "ref": "paper:paper-733c0535ffcd"
    }
  ],
  "section_title": "Computational Models of VIP Circuit Function",
  "source_policy": {
    "mode": "public_source_pointer_with_short_context",
    "notes": [
      "Local review repositories are read-only inputs.",
      "SciDEX stores paper metadata, structured evidence, file pointers, and short citation contexts; it does not copy full review prose."
    ],
    "source_commit_sha": "95e761177f7d2ec565983d3307c14ec238f9677c",
    "source_repository_url": "https://github.com/AllenNeuralDynamics/ComputationalReviewVIP"
  },
  "evidence_summary": "Authors built a 4-interneuron-type spiking network and embedded sWTA, gain control, disinhibition, and normalization motifs onto neuromorphic hardware to improve vision-transformer performance.",
  "review_bundle_ref": "analysis_bundle:ab-2ce40c33e827",
  "replication_status": "replicated_independent",
  "review_package_ref": "analysis_bundle:ab-2ce40c33e827",
  "source_artifact_ref": "wiki_page:computationalreviewvip-12-computational-models",
  "origin_url": "https://github.com/AllenNeuralDynamics/ComputationalReviewVIP/blob/95e761177f7d2ec565983d3307c14ec238f9677c/evidence/section_12_evidence_package.json",
  "commit_sha": "95e761177f7d2ec565983d3307c14ec238f9677c",
  "created_by": "persona-jerome-lecoq-gbo-neuroscience",
  "repository_url": "https://github.com/AllenNeuralDynamics/ComputationalReviewVIP"
}