{
  "version": 1,
  "updatedAt": "2026-09-05",
  "canonicalUrl": "https://cognitive-biases.github.io/research/lab/",
  "interpretationRule": "Never infer a later stage from external papers alone. External evidence can motivate a track, but project stages advance only when the corresponding project artifact exists.",
  "resultGate": "Project result stage requires methods, raw or summary data, reproducible scoring, uncertainty, limitations and documented protocol deviations. External papers alone do not advance a project track.",
  "stageOrder": [
    "idea",
    "protocol",
    "experiment",
    "result",
    "replication"
  ],
  "stageDefinitions": {
    "idea": "A falsifiable research question with related established concepts, but no project protocol yet.",
    "protocol": "A reproducible comparison protocol is published. No project result is implied.",
    "experiment": "A preregistered or otherwise documented run exists with model/task/sample metadata, but results are not yet promoted to a finding.",
    "result": "A project result is published with methods, raw or summary data, uncertainty and limitations.",
    "replication": "The result has been repeated across a meaningfully different sample, task, model or independent group."
  },
  "tracks": [
    {
      "track": "ai-anchoring-loop",
      "name": "AI Anchoring Loop",
      "stage": "protocol",
      "whyItMatters": "AI can become the first numerical reference before a person has formed an independent estimate.",
      "nextMilestone": "Collect 40 complete adult pilot sessions under the preregistered AI Advice Order v1 protocol, then publish anonymized data, scoring output, limitations and any deviations before considering result stage.",
      "protocolSlugs": [
        "ai-advice-order"
      ],
      "studyArtifacts": [
        "/research/ai-advice-order-v1/",
        "/experiments/ai-advice-order-v1/",
        "/data/studies/ai-advice-order-v1.json"
      ],
      "studyIds": [
        "ai-advice-order-v1"
      ]
    },
    {
      "track": "evaluation-history-contamination",
      "name": "Evaluation History Contamination",
      "stage": "protocol",
      "whyItMatters": "Repeated AI reviews can look independent even when later evaluators are partly inheriting earlier scores.",
      "nextMilestone": "Run the complete blind, history-framing and prior-score matrix on at least three fully specified models, publish raw result JSON plus scorer output, and document deviations before considering experiment or result stage.",
      "protocolSlugs": [
        "blind-vs-history-aware-ai-judging"
      ],
      "studyArtifacts": [
        "/research/ai-judge-history-v1/",
        "/experiments/ai-judge-history-v1/",
        "/data/studies/ai-judge-history-v1.json",
        "/data/studies/ai-judge-history-prompt-pack-v1.json",
        "/schemas/ai-judge-history-results.schema.json"
      ],
      "studyIds": [
        "ai-judge-history-v1"
      ]
    },
    {
      "track": "sycophancy-reinforcement-loop",
      "name": "Sycophancy Reinforcement Loop",
      "stage": "protocol",
      "whyItMatters": "An agreeable answer can be mistaken for independent support for a view the user already signalled.",
      "nextMilestone": "Separate model agreement from human confidence and test neutral, preference-signalling and counterevidence conditions.",
      "protocolSlugs": [
        "agreement-pressure"
      ],
      "studyArtifacts": [],
      "studyIds": []
    },
    {
      "track": "cognitive-offloading-debt",
      "name": "Cognitive Offloading Debt",
      "stage": "protocol",
      "whyItMatters": "Faster completion and durable understanding are different outcomes.",
      "nextMilestone": "Run an immediate-quality plus delayed-retention comparison rather than measuring task completion alone.",
      "protocolSlugs": [
        "offloading-retention"
      ],
      "studyArtifacts": [],
      "studyIds": []
    },
    {
      "track": "source-memory-blur",
      "name": "Source-Memory Blur",
      "stage": "protocol",
      "whyItMatters": "Mixed workflows may make it harder to remember whether an idea came from the person, a source or the model.",
      "nextMilestone": "Measure delayed source attribution for human-only, AI-only and mixed creation conditions.",
      "protocolSlugs": [
        "source-memory"
      ],
      "studyArtifacts": [],
      "studyIds": []
    },
    {
      "track": "synthetic-consensus-illusion",
      "name": "Synthetic Consensus Illusion",
      "stage": "idea",
      "whyItMatters": "Several generated answers may feel like independent confirmation even when their information lineage overlaps.",
      "nextMilestone": "Publish a protocol comparing several apparently independent AI answers with the same answers plus source-lineage disclosure.",
      "protocolSlugs": [],
      "studyArtifacts": [],
      "studyIds": []
    },
    {
      "track": "confidence-transfer",
      "name": "Confidence Transfer",
      "stage": "idea",
      "whyItMatters": "Confident wording may change human reliance even when the underlying evidence does not improve.",
      "nextMilestone": "Publish a protocol holding answer quality constant while varying certainty language and objective calibration information.",
      "protocolSlugs": [],
      "studyArtifacts": [],
      "studyIds": []
    }
  ],
  "studies": [
    {
      "studyId": "ai-advice-order-v1",
      "title": "Does AI Advice Order Change Numerical Judgment?",
      "status": "preregistered-pilot",
      "track": "ai-anchoring-loop",
      "researchQuestion": "When the same standardized AI-labelled numerical advice is shown before rather than after an independent estimate, does the final human estimate align more strongly with that advice?",
      "hypothesis": "Within participants, final estimates will show greater normalized alignment with the standardized AI advice in advice-first trials than in independent-estimate-first trials.",
      "canonicalUrl": "https://cognitive-biases.github.io/research/ai-advice-order-v1/",
      "instrumentUrl": "https://cognitive-biases.github.io/experiments/ai-advice-order-v1/"
    },
    {
      "studyId": "ai-judge-history-v1",
      "title": "Does Evaluation History Change an AI Judge's Verdict?",
      "status": "preregistered-benchmark",
      "track": "evaluation-history-contamination",
      "researchQuestion": "When the same fixed answer and rubric are evaluated by the same model, does showing an earlier score change the new judgment compared with a blind evaluation?",
      "hypothesis": "History-aware judgments will move toward the displayed prior score more than blind or history-framing-only judgments, even though the prior score is experimentally uninformative.",
      "canonicalUrl": "https://cognitive-biases.github.io/research/ai-judge-history-v1/",
      "instrumentUrl": "https://cognitive-biases.github.io/experiments/ai-judge-history-v1/"
    }
  ]
}