Skip to content

THINKING_CURRICULUM.md

Latest commit

 

History

History
178 lines (138 loc) · 5.28 KB

THINKING_CURRICULUM.md

File metadata and controls

178 lines (138 loc) · 5.28 KB
Raw

Thinking Curriculum Engine

This package adds a full Grade 1 through Grade 4 thinking, reasoning, and problem-solving layer for the kid-focused learning gate app.

The design goal is:

  • practical thinking training, not generic SEL
  • puzzle, mystery, clue, and bug-hunt framing instead of school worksheets
  • cross-subject transfer into math, reading, and coding
  • strong habits around checking, explaining, and asking for help honestly

What Is Included

  • THINKING_CURRICULUM.md
    • curriculum overview and integration notes
  • curriculum/thinking-curriculum.js
    • full browser-loadable thinking curriculum object
  • curriculum/diagnostics/thinking-placement.js
    • short adaptive thinking placement diagnostic
  • data/thinking/thinking-skill-graph.json
    • prerequisite-linked reasoning skill graph
  • data/thinking/thinking-lessons.json
    • seeded Grade 1 to Grade 4 lesson metadata
  • data/thinking/thinking-tasks.json
    • seeded task bank with 8 tasks per lesson
  • data/thinking/thinking-mastery.json
    • mastery states, scoring signals, and movement rules
  • data/thinking/thinking-rewards.json
    • deterministic companion reward hooks
  • data/thinking/thinking-parent-reporting.json
    • parent summary model and example output

Skill Graph Shape

The thinking roadmap is a graph, not a chapter list. Skills build from noticing and following directions into evidence, logic, debugging, planning, and clarifying questions.

Current domain coverage:

  • attention and careful noticing
  • visual patterns
  • number and shape patterns
  • similarities and differences
  • sorting and classification
  • sequencing
  • cause and effect
  • analogies
  • logic clues and logic grids
  • memory strategies and working memory
  • following 2-step and 3-step directions
  • checking answers and finding contradictions
  • explaining how you know
  • evidence from pictures and sentences
  • flexible thinking and multiple solutions
  • planning, estimation, and reasonableness
  • debugging your own thinking
  • asking clarifying questions
  • confidence calibration

Seeded Scope

  • Grade 1 lessons: 8
  • Grade 2 lessons: 8
  • Grade 3 lessons: 6
  • Grade 4 lessons: 6
  • Tasks per lesson: 8
  • Total lesson tasks: 224
  • Placement diagnostic tasks: 16 seeded with adaptive stop rules

Daily Thinking Loop

The thinking layer supports two modes:

  • standalone 5 to 10 minute thinking session
  • 2 to 3 minute warmup before math, reading, or coding

Daily flow:

  1. Careful noticing task
  2. Puzzle or reasoning task
  3. Teach-Noble explanation prompt
  4. Quick reflection: easy, tricky, or needed help

Mastery Model

Statuses:

  • introduced
  • practicing
  • almost_mastered
  • mastered
  • needs_review
  • blocked

Signals considered:

  • accuracy
  • hint usage
  • retries
  • time as a light signal only
  • explanation quality
  • checking behavior
  • confidence calibration
  • transfer across puzzle types

Placement Output

The thinking placement returns:

  • strongestThinkingSkills
  • shakyThinkingSkills
  • recommendedDailyThinkingWarmup
  • needsSlowDownAndCheckPractice
  • needsExplainHowYouKnowPractice
  • confidenceCalibration

It is intentionally shorter than reading or math placement and is designed to feel like a puzzle trail, not a test.

Reward Hooks

The light companion celebrates:

  • noticing carefully
  • explaining clearly
  • finding a mistake
  • checking work
  • trying again
  • saying "I'm not sure" honestly
  • solving a tricky pattern
  • using evidence

Rewards are deterministic and effort-based.

Parent Summary Model

The parent-facing summary captures:

  • today completed or not
  • thinking skills practiced
  • accuracy
  • hints used
  • explanation quality
  • rush or guess signals
  • confidence calibration
  • recommended next session
  • connection to math, reading, and coding

UX Recommendations

  • one puzzle at a time
  • large answer buttons
  • read-aloud always present
  • teach-Noble explanation prompts
  • no visible failure score
  • reward checking, not speed
  • parent detail hidden behind parent mode

Connection To Other Subjects

  • Math: pattern rules, reasonableness, step order, error checking, and multiple strategies
  • Reading: evidence, sequencing, cause and effect, contradiction detection, and explanation quality
  • Coding: following directions, debugging, logic clues, planning, and accepting multiple valid paths

Developer Wiring Order

  1. Load curriculum/thinking-curriculum.js or the JSON files into parent-mode planning code.
  2. Wire curriculum/diagnostics/thinking-placement.js into onboarding or pre-subject warmup.
  3. Add one daily thinking warmup slot before math or reading.
  4. Track explanation quality, hint usage, and checking behavior in the mastery store.
  5. Connect reward events from thinking-rewards.json to the light companion.
  6. Surface parent summary fields in parent mode only.

Open Questions For Bailey

  • Should thinking placement run during onboarding for every child, or only after reading and math placement?
  • Should the daily thinking loop always be a warmup first, or should it sometimes be a standalone gate session?
  • Do we want voice capture for explanation quality now, or only button choices plus parent-observed notes for phase one?
  • Should multi-step directions use tappable picture boards only, or do we also want drag-and-drop ordering?
  • How much of the confidence calibration should be visible to Nova versus stored quietly for parent mode?