Recipe catalog / causal-attribution

Identify personal and situational explanations

Label whether an explanation attributes a specified behavior to the person, circumstances, both, or gives no cause.

You need to label whether an explanation points to the person, the situation, or both.

Explore this recipe interactively · Source and implementation guide

Use causal-attribution in TypeScript

Install with npm install jev-recipes. Requires Node.js 22.9 or newer and ES modules. Set TYPESAFE_API_KEY in your server environment for live calls, which send input to TypeSafe and use API quota. See the installation guide.

import { causalAttribution } from 'jev-recipes/causal-attribution';

const result = await causalAttribution({
  "behavior": "Alex arrived late to the meeting.",
  "explanation": "Alex arrived late because the train was cancelled."
});
console.log(result);

Input contract

Field	Type	Needed
behavior	string	Required
explanation	string	Required
context	string	Optional
minConfidence	number	Optional

Full input and result schemas

{
  "input": {
    "$schema": "https://json-schema.org/draft/2020-12/schema",
    "type": "object",
    "properties": {
      "behavior": {
        "type": "string",
        "description": "One focal behavior or outcome identifying the person or actor whose behavior is being explained."
      },
      "explanation": {
        "type": "string",
        "description": "The supplied explanation to classify, including any quoted attribution."
      },
      "context": {
        "type": "string",
        "description": "Supplied identities or references needed to interpret the explanation."
      },
      "minConfidence": {
        "type": "number",
        "minimum": 0,
        "maximum": 1
      }
    },
    "required": [
      "behavior",
      "explanation"
    ]
  },
  "result": {
    "$schema": "https://json-schema.org/draft/2020-12/schema",
    "type": "object",
    "properties": {
      "model": {
        "type": "string"
      },
      "usage": {
        "type": "object",
        "properties": {
          "input_tokens": {
            "type": "integer",
            "minimum": 0,
            "maximum": 9007199254740991
          },
          "output_tokens": {
            "type": "integer",
            "minimum": 0,
            "maximum": 9007199254740991
          }
        },
        "required": [
          "input_tokens",
          "output_tokens"
        ],
        "additionalProperties": false
      },
      "status": {
        "type": "string",
        "enum": [
          "ready",
          "review"
        ]
      },
      "verdict": {
        "type": "string",
        "enum": [
          "personal",
          "situational",
          "mixed",
          "none",
          "unclear"
        ]
      },
      "confidence": {
        "type": "number",
        "minimum": 0,
        "maximum": 1
      },
      "probabilities": {
        "type": "object",
        "propertyNames": {
          "type": "string",
          "enum": [
            "personal",
            "situational",
            "mixed",
            "none",
            "unclear"
          ]
        },
        "additionalProperties": {
          "type": "number",
          "minimum": 0,
          "maximum": 1
        },
        "required": [
          "personal",
          "situational",
          "mixed",
          "none",
          "unclear"
        ]
      }
    },
    "required": [
      "model",
      "usage",
      "status",
      "verdict",
      "confidence",
      "probabilities"
    ],
    "additionalProperties": false
  }
}

Saved example result

This hand-authored response demonstrates the contract. It is not a model accuracy measurement. Run it without an API key: npx jev-recipes demo causal-attribution.

{
  "model": "demo-fixture",
  "usage": {
    "input_tokens": 0,
    "output_tokens": 0
  },
  "status": "ready",
  "verdict": "situational",
  "confidence": 0.96,
  "probabilities": {
    "personal": 0.01,
    "situational": 0.96,
    "mixed": 0.01,
    "none": 0.01,
    "unclear": 0.01
  }
}

Evaluation evidence

Fixture only

No verified live accuracy measurement is available. Evaluate representative cases before using this decision in your workflow.

Use the evaluation guide to measure this decision on your own labeled cases.

Limitations

Labels the supplied explanation, not the actual cause, blame, or a psychological trait of the actor.
Personal and situational refer to the focal actor, not the speaker or whoever is named in a cause.
A personal attribution alone does not establish bias or a fundamental attribution error.

Related recipes

attribution-match: Use attribution-match to check who made or endorsed a statement, rather than what kind of cause it gives.
motivation-source: Use motivation-source to distinguish enjoyment of an activity from pursuing a separate outcome in a stated reason.
verify: Use verify to assess supplied evidence for a causal claim; causal-attribution does not establish causation.