Version history

{
  "kind": "infographic",
  "prompt": "Across mouse motor-cortex studies, latent dimensionality is small but stability and criticality regimes are layer- and method-dependent — bears on whether E→E recurrence enforces a single canonical low-dimensional manifold.",
  "provider": "other",
  "raw_fields": {
    "papers": [
      {
        "n": null,
        "doi": "10.64898/2026.01.30.702162",
        "value": "3-7",
        "method": "supervised latent model",
        "metric": "latent variance captured (%)",
        "n_analyzed": null,
        "ci_or_error": null,
        "text_access": "abstract_only",
        "n_definition": "animals (count not in abstract)",
        "scope_region": "motor cortex",
        "study_system": "motor cortex (2D reaching)",
        "taxonomic_level": "broad category",
        "scope_population": "all units",
        "value_source_sentence": "Despite only capturing 3-7% of total variance and spanning two dimensions, this supervised method outperforms other common methods at an offline decoding task, and explains the long-term stability of neural manifolds observed in previous literature.",
        "experimental_conditions": "chronic recordings"
      },
      {
        "n": null,
        "doi": "10.1016/j.celrep.2023.112574",
        "value": ">10,000 neurons; 30,000,000 synapses",
        "method": "biophysical simulation",
        "metric": "model scale (neurons; synapses)",
        "n_analyzed": null,
        "ci_or_error": null,
        "text_access": "abstract_only",
        "n_definition": "model neurons",
        "scope_region": "M1",
        "study_system": "mouse M1 model",
        "taxonomic_level": "subcategory",
        "scope_population": "all cortical cell types",
        "value_source_sentence": "We develop a multiscale, biophysically detailed model of mouse primary motor cortex (M1) with over 10,000 neurons and 30 million synapses.",
        "experimental_conditions": "model"
      },
      {
        "n": null,
        "doi": "10.1152/jn.00061.2020",
        "value": "14",
        "method": "chronic two-photon imaging",
        "metric": "days of recording at criticality",
        "n_analyzed": null,
        "ci_or_error": null,
        "text_access": "abstract_only",
        "n_definition": "days",
        "scope_region": "motor and premotor cortex",
        "study_system": "mouse forelimb motor/premotor cortex",
        "taxonomic_level": "subcategory",
        "scope_population": "L2/3 and L5 pyramidal",
        "value_source_sentence": "In contrast, layer 5 circuits operated away from the critical network state for all 14 days of recording and learning.",
        "experimental_conditions": "lever-press learning"
      }
    ],
    "audit_issues": [
      {
        "dimension": "metric_definition",
        "description": "Row 1: latent variance captured (3–7%). Row 2: model scale (>10,000 neurons; 30,000,000 synapses). Row 3: number of days at criticality (14). These are three unrelated quantities and cannot be plotted on a shared y-axis as a 'grouped bar.'",
        "entries_affected": [
          "10.64898/2026.01.30.702162",
          "10.1016/j.celrep.2023.112574",
          "10.1152/jn.00061.2020"
        ]
      },
      {
        "dimension": "study_system",
        "description": "Row 1 is an analysis of recorded motor-cortex data; row 2 is a biophysical model; row 3 is chronic two-photon imaging of lever-press learning. The studies are about different aspects of motor cortex (decoding, simulation, plasticity timecourse).",
        "entries_affected": [
          "10.64898/2026.01.30.702162",
          "10.1016/j.celrep.2023.112574",
          "10.1152/jn.00061.2020"
        ]
      }
    ],
    "audit_verdict": "REDESIGN",
    "comparison_id": "low_dim_motor_cortex",
    "comparison_name": "Dimensionality and stability of motor-cortex latent manifolds",
    "comparison_type": "convergent evidence",
    "what_it_reveals": "Across mouse motor-cortex studies, latent dimensionality is small but stability and criticality regimes are layer- and method-dependent — bears on whether E→E recurrence enforces a single canonical low-dimensional manifold.",
    "homogeneity_check": {
      "caveats": [
        "Mix of recording, modelling and reaching-task studies; metrics not directly commensurable"
      ],
      "n_definition_uniform": "false",
      "scope_region_uniform": "true",
      "taxonomic_level_uniform": "false",
      "scope_population_uniform": "false"
    },
    "suggested_plot_type": "grouped bar",
    "mandatory_caption_caveats": [
      "Rows report three unrelated quantities (% latent variance, model scale, days at criticality) and do not share a common metric.",
      "Phase 7 writer: REDESIGN verdict — implement as: Either rebuild the figure around a single common metric (e.g., latent dimensionality of M1 across studies) by re-extracting from each paper, or split into three separate panels with distinct y-axes. As currently configured, the comparison is not meaningful."
    ]
  },
  "section_id": "section_14",
  "source_url": "https://github.com/AllenNeuralDynamics/ComputationalReviewRecurrence/blob/79ce062d54a924ce05953ec90aa9d26044d2b48f/evidence/section_14_evidence_package.json",
  "target_ref": "wiki_page:computationalreviewrecurrence-14-predictive-coding",
  "review_repo": "ComputationalReviewRecurrence",
  "section_ref": "wiki_page:computationalreviewrecurrence-14-predictive-coding",
  "source_path": "evidence/section_14_evidence_package.json",
  "source_refs": [
    "paper:paper-28cc9cc5cd02",
    "paper:paper-39da47e849b2",
    "paper:paper-6a004d3b2777"
  ],
  "section_title": "14. Predictive-coding and dynamical-systems accounts — the role of recurrent excitatory feedback in error signalling, state estimation, and reservoir computing, evaluated against mouse data",
  "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": "79ce062d54a924ce05953ec90aa9d26044d2b48f",
    "source_repository_url": "https://github.com/AllenNeuralDynamics/ComputationalReviewRecurrence"
  },
  "generation_status": "complete",
  "review_bundle_ref": "analysis_bundle:ab-d9c479db9be9",
  "origin_url": "https://github.com/AllenNeuralDynamics/ComputationalReviewRecurrence/blob/79ce062d54a924ce05953ec90aa9d26044d2b48f/evidence/section_14_evidence_package.json",
  "commit_sha": "79ce062d54a924ce05953ec90aa9d26044d2b48f",
  "created_by": "persona-jerome-lecoq-gbo-neuroscience",
  "repository_url": "https://github.com/AllenNeuralDynamics/ComputationalReviewRecurrence"
}