Learners reason, they do not just recall. They analyze, evaluate, and create. They justify their thinking with evidence and persevere through complex problems. Teachers raise cognitive demand for every student and protect the time and the struggle that deep thinking requires.
The arc has two movements. Thinking gets deeper, from applying to creating. And ownership shifts, from the teacher driving the thinking to the learner driving it. A practice reaches Sustaining when both are true and they hold for every student, not just the confident few.
Evidence base: revised Bloom's taxonomy, the surface-deep-transfer model, desirable difficulty and productive struggle, and cognitive load theory.
Start with the goal
Know the difference between recall and analysis, evaluation, and synthesis, and what counts as evidence.
Understand that strong thinking is reasoned and defensible, not just correct.
Do analyze, evaluate, and create, and justify their reasoning with evidence.
Equity and BelongingWatch who gets asked to think hard. When rigor is rationed to a few students, it stops being rigor and becomes tracking. This thread runs through every practice below.
Technology and AIAI enhances learning only through critical use and evaluation. Each practice names what that looks like. Treat outputs as drafts, check for accuracy, bias, and false agreement, decide when not to use the tool, protect your own thinking, and disclose how you used it.
Paired practices
1Cognitive demand of questioning
Emerging
Developing
Embedded
Sustaining
Teacher
Asks mostly closed, recall questions, with a few open-ended ones.
Asks a mix of closed and open-ended questions.
Asks some closed and mostly open-ended questions.
Asks primarily open-ended questions and has students generate their own.
Student
Answer recall questions with one- or two-word responses.
Explain their thinking in a sentence when asked.
Analyze and justify answers with evidence in most responses.
Pose their own higher-order questions during the lesson.
AI in practice critical use and evaluation
Teacher
- Treat an AI-generated question bank as a first draft, cutting leading and low-demand items before using any.
- Have students interrogate an AI-suggested strong question, naming what it assumes or leaves out.
Student
- Ask AI to challenge an answer, then judge whether the pushback is real reasoning or just agreeable hedging.
- Form a position first, then use AI to surface a counter-question, not to supply the answer.
How to measureTally open vs. closed questions over a 10-minute segment, and note whether students pose their own; the open-question share locates the level.
2Equitable thinking participation
Emerging
Developing
Embedded
Sustaining
Teacher
Directs higher-order questions to fewer than half of students.
Directs higher-order questions to about half of students.
Directs higher-order questions to most students.
Directs higher-order questions to every student during the lesson.
Student
Fewer than half of students offer reasoning aloud.
About half of students offer reasoning aloud.
Most students offer reasoning, supported by routines.
Nearly all students reason aloud or in writing each lesson.
AI in practice critical use and evaluation
Teacher
- Check whether AI-suggested prompts assume background knowledge some students lack, and rewrite for access.
- Watch that AI scaffolds extend thinking for striving students rather than think for them.
Student
- Rehearse reasoning privately with AI, then bring their own words to the group, not the AI's.
- Multilingual learners reason in their home language with AI, then verify the shared idea is theirs.
How to measureTrack which students receive higher-order questions; the share reached, from fewer than half to every student, sets the level.
3Think time and productive struggle
Emerging
Developing
Embedded
Sustaining
Teacher
Calls on students within 1-2 seconds and steps in at the first sign of struggle.
Uses wait time with some students and allows brief struggle before guiding.
Uses 3 or more seconds of wait time with most students and prompts rather than rescues.
Consistently protects wait time and productive struggle for all students.
Student
Stop and wait for help when a task gets hard.
Attempt one strategy before asking for help.
Try two or more strategies before seeking help.
Persist and self-prompt through difficulty without waiting to be rescued.
AI in practice critical use and evaluation
Teacher
- Set tools to give hints, not answers, and name when reaching for AI would skip the struggle that builds skill.
- Spot students offloading the hard thinking to AI and redirect them back into the problem.
Student
- Sit with a hard problem before opening AI, noticing when a quick answer would replace their own learning.
- Ask AI why an attempt failed, then retry, rather than copying its solution.
How to measureTime the wait after questions and note whether the teacher prompts or rescues; longer protected wait and student self-prompting raise the level.
4Depth of thinking
Emerging
Developing
Embedded
Sustaining
Teacher
Assigns tasks that ask students to recall and apply.
Assigns tasks that ask students to compare and explain relationships.
Assigns tasks that ask students to evaluate and justify.
Assigns tasks that ask students to synthesize sources and create original work.
Student
Produce answers that restate or apply a procedure.
Produce work that compares ideas or shows relationships.
Produce work that evaluates and cites evidence.
Produce original work that synthesizes multiple sources.
AI in practice critical use and evaluation
Teacher
- Generate an AI exemplar, then have students find where it is shallow, generic, or wrong.
- Use AI to draft a higher-demand task, then verify the rigor is real and not just longer wording.
Student
- Brainstorm angles with AI, then do the analysis themselves; a fluent answer is not a correct one.
- Compare their own thinking against an AI draft to push past the obvious, not to replace it.
How to measureCode a task and its student work by thinking level; recall and apply are lower, synthesize and create are Sustaining.
5Evidence and justification
Emerging
Developing
Embedded
Sustaining
Teacher
Accepts answers without asking for reasons.
Asks for a reason on some answers.
Requires evidence on most answers as a routine.
Has students weigh and critique the strength of evidence.
Student
State answers without reasons.
Give a reason when prompted.
Cite evidence for claims without prompting.
Critique the strength of their own and peers' evidence.
AI in practice critical use and evaluation
Teacher
- Feed students a confident but flawed AI argument and have them locate the weak or fabricated evidence.
- Model checking an AI claim against a primary source before accepting it.
Student
- Verify that sources an AI cites actually exist and say what they claim; treat citations as unproven until checked.
- Ask AI for counterevidence, then weigh it themselves rather than adopting it.
How to measureCount how often students cite evidence unprompted in a sample of responses; routine, unprompted justification places the level.
6Working with multiple sources
Emerging
Developing
Embedded
Sustaining
Teacher
Provides one source to summarize.
Provides two sources to compare.
Provides three or more sources to integrate.
Has students locate and synthesize their own sources.
Student
Summarize one source.
Compare points across two sources.
Integrate ideas from three or more sources.
Curate and synthesize self-selected sources into an argument.
AI in practice critical use and evaluation
Teacher
- Show how an AI summary can flatten or distort a source, then require checking against the original.
- Have students compare an AI synthesis to their own reading and name what it missed.
Student
- Treat an AI synthesis as a draft map, verifying each claim against the real sources before using it.
- Notice when AI over-agrees with their thesis, and seek sources that complicate it.
How to measureCount the sources students must integrate; one source is lower, self-selected synthesis is Sustaining.
7Critical evaluation of claims
Emerging
Developing
Embedded
Sustaining
Teacher
Presents information without questioning the source.
Models checking who made a claim and why.
Has students evaluate credibility and bias on most sources.
Has students independently interrogate claims, including AI output.
Student
Accept claims as given.
Ask where a claim came from when prompted.
Check the credibility and bias of sources routinely.
Independently verify claims across sources before using them.
AI in practice critical use and evaluation
Teacher
- Have students fact-check an AI response and name the sycophancy, bias, or overconfidence they find.
- Test a claim you know is wrong together, and discuss why the tool went along with it.
Student
- Test the tool with a claim they know is false to see whether it pushes back or caves.
- Cross-check AI output against independent sources, and ask whose perspective it leaves out.
How to measureNote whether students check source, bias, and accuracy and on how many sources; routine independent checking is Sustaining.
8Transfer to new contexts
Emerging
Developing
Embedded
Sustaining
Teacher
Assigns practice in the same context taught.
Assigns practice with changed surface details.
Assigns practice in a new, unfamiliar context.
Assigns practice in an authentic real-world context.
Student
Solve problems that mirror the example.
Solve problems with changed details.
Apply learning to an unfamiliar problem.
Transfer learning to a real-world situation.
AI in practice critical use and evaluation
Teacher
- Ask first whether AI is the right tool, or whether using it short-circuits the transfer students need to practice.
- Use AI to draft a real-world scenario, then verify it is accurate and genuinely novel.
Student
- Stress-test a solution in a new context with AI, then verify the response rather than trust it.
- Decide when a transfer task is theirs to work through without AI.
How to measureJudge the distance between the practice context and the taught example; same context is lower, authentic real-world transfer is Sustaining.
9Perseverance on complex tasks
Emerging
Developing
Embedded
Sustaining
Teacher
Plans mostly single-step tasks with at most one multi-step problem.
Plans 1-2 multi-step problems reached only by fast finishers.
Plans multi-step tasks with time for most students to attempt them.
Centers the lesson on a multi-step problem all students work through.
Student
Complete single-step tasks and disengage as complexity rises.
Some students persist through multi-step tasks, trying another approach.
Most students persist through multi-step tasks, trying multiple approaches.
Nearly all students plan, monitor, and adjust to persist independently.
AI in practice critical use and evaluation
Teacher
- Make explicit where AI help is allowed and where the struggle is the learning, so tools support not replace it.
- Use AI to map checkpoints for a complex task without handing over the solution path.
Student
- Use AI for a single checkpoint hint, naming the point where more help would replace their thinking.
- Ask AI for a different strategy when stuck, then evaluate whether it actually fits.
How to measureCount multi-step problems in the plan and the share of students who persist through them; more multi-step work and broader persistence raise the level.
10Original creation and idea generation
Emerging
Developing
Embedded
Sustaining
Teacher
Assigns one correct product that matches a model.
Allows minor variation on a set product.
Invites students to design an original product toward a goal.
Positions students to define problems and create for real audiences.
Student
Reproduce the modeled product.
Add personal variation to a set product.
Design an original product toward the goal.
Frame their own problem and create original work for a real audience.
AI in practice critical use and evaluation
Teacher
- Have students think first, use AI only to critique a draft, and disclose how it was used.
- Resist AI's first, average answer, prompting students to go beyond what the tool produces.
Student
- Draft original work without AI, then use it to pressure-test the draft, and disclose its role honestly.
- Keep their own voice, rejecting AI phrasing that flattens it.
How to measureJudge how much of the product students originate; reproducing a model is lower, framing a problem and creating for a real audience is Sustaining.
Planning and setting
Design moves that set up the practices above.
Clarity of the learning goal
Every lesson opens from a clear goal. Emerging: the objective is implicit. Developing: the objective is stated. Embedded: a clear know, understand, and do is shared with success criteria. Sustaining: students can state the goal and criteria in their own words.
Learning objective rigor
The objective signals cognitive demand. Emerging objectives use verbs like apply, use, identify. Developing uses analyze, compare, differentiate. Embedded uses evaluate, critique, justify. Sustaining uses create, design, develop, invent, and targets transfer.
Thinking made visible
Reasoning supports are designed into the room. At Emerging, thinking stays in students' heads. By Sustaining, thinking routines and reasoning frames are posted, used consistently, and selected independently by students.
Evidence and measurement
Each practice above carries its own measure. These are the primary evidence sources for the component as a whole.
Evidence for this component
- A tally of teacher question types and which students are asked to reason.
- Student work showing analysis, evidence, and original creation.
- Notes or transcripts of students reasoning and justifying aloud.