Version history

{
  "kind": "infographic",
  "prompt": "Models disagree on whether SST neurons carry the prediction signal (top-down expectations mediated through dendritic inhibition) or contribute to computing the prediction error. This reveals a fundamental unresolved question about SST's role in cortical inference.",
  "provider": "other",
  "raw_fields": {
    "papers": [
      {
        "doi": "10.7554/elife.57541",
        "value": "SST carries prediction signal — inhibitory plasticity learns to predict expected input",
        "method": "inhibitory plasticity learning rule",
        "metric": "SST role in predictive coding",
        "cite_key": "Hertag2020",
        "condition": "sensory prediction",
        "study_system": "mean-field canonical circuit",
        "value_source_sentence": "Learning prediction error neurons in a canonical interneuron circuit."
      },
      {
        "doi": "10.7554/elife.95127",
        "value": "SST signals predictions with uncertainty modulation of prediction error magnitude",
        "method": "uncertainty estimation",
        "metric": "SST role in predictive coding",
        "cite_key": "Wilmes2025",
        "condition": "Bayesian inference",
        "study_system": "microcircuit model",
        "value_source_sentence": "Uncertainty-modulated prediction errors in cortical microcircuits."
      },
      {
        "doi": "10.1101/2025.11.01.686040",
        "value": "SOM provides compartment-specific inhibition for sign-specific PE without learning",
        "method": "fixed-weight spiking dynamics",
        "metric": "SST role in predictive coding",
        "cite_key": "Nemati2025",
        "condition": "V1 L2/3 predictive coding",
        "study_system": "spiking network, 2-compartment pyramidal",
        "value_source_sentence": "A spiking network model is presented here in which two-compartment excitatory pyramidal neurons interact with three inhibitory subtypes to compute sign-specific prediction errors."
      },
      {
        "doi": "10.1371/journal.pcbi.1011921",
        "value": "SST-mediated dendritic inhibition carries top-down prediction during visuomotor mismatch",
        "method": "visuomotor prediction model",
        "metric": "SST role in predictive coding",
        "cite_key": "GalvanFraile2024",
        "condition": "visual flow mismatch",
        "study_system": "circuit model",
        "value_source_sentence": "Modeling circuit mechanisms of opposing cortical responses to visual flow perturbations."
      },
      {
        "doi": "10.1371/journal.pcbi.1013469",
        "value": "SST and PV generate distinct oscillatory patterns during predictive coding phases",
        "method": "oscillatory analysis",
        "metric": "SST role in predictive coding",
        "cite_key": "Lee2025",
        "condition": "sensory prediction",
        "study_system": "spiking network",
        "value_source_sentence": "Cortical networks with multiple interneuron types generate oscillatory patterns during predictive coding."
      }
    ],
    "comparison_id": "predictive-coding-sst-models",
    "comparison_name": "Competing predictive coding models: SST role in prediction vs error computation",
    "comparison_type": "cross-study conflict",
    "what_it_reveals": "Models disagree on whether SST neurons carry the prediction signal (top-down expectations mediated through dendritic inhibition) or contribute to computing the prediction error. This reveals a fundamental unresolved question about SST's role in cortical inference.",
    "homogeneity_check": {
      "caveats": "Different model architectures (mean-field vs spiking vs rate), different definitions of prediction error, different brain regions. The disagreement may partly reflect different levels of abstraction rather than genuine conflict.",
      "comparable": false
    },
    "suggested_plot_type": "grouped bar"
  },
  "section_id": "section_12_evidence_package",
  "source_url": "https://github.com/AllenNeuralDynamics/ComputationalReviewSST/blob/89b7e9787cd90e942b0adb531d549af3ddad30f1/evidence/section_12_evidence_package.json",
  "target_ref": "wiki_page:computationalreviewsst-12",
  "review_repo": "ComputationalReviewSST",
  "section_ref": "wiki_page:computationalreviewsst-12",
  "source_path": "evidence/section_12_evidence_package.json",
  "source_refs": [
    "paper:paper-193dfb3c1e40",
    "paper:paper-3a51e8e844f7",
    "paper:paper-7d015d30f267",
    "paper:paper-af1a62e9cd4a",
    "paper:paper-d67d30eba36a"
  ],
  "section_title": "Computational Models",
  "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": "89b7e9787cd90e942b0adb531d549af3ddad30f1",
    "source_repository_url": "https://github.com/AllenNeuralDynamics/ComputationalReviewSST"
  },
  "generation_status": "complete",
  "review_bundle_ref": "analysis_bundle:ab-8466d095488a",
  "origin_url": "https://github.com/AllenNeuralDynamics/ComputationalReviewSST/blob/89b7e9787cd90e942b0adb531d549af3ddad30f1/evidence/section_12_evidence_package.json",
  "commit_sha": "89b7e9787cd90e942b0adb531d549af3ddad30f1",
  "created_by": "persona-jerome-lecoq-gbo-neuroscience",
  "repository_url": "https://github.com/AllenNeuralDynamics/ComputationalReviewSST"
}