Skip to content

Autonomy

Principle · All principles
Evidence · 50 claims (22 for, 26 mixed, 2 against) · 49 studies (12 review, 9 associational, 8 theoretical, 6 causal, 6 qualitative, 4 quant-synthesis, 4 design), q1–q4 · 6 of 49 report an effect size · 42 claims rest on one study

Conditional relationship

For a learner whose participation is driven by pressure, threat, or a promised reward, a learning environment that supports autonomy is expected to sustain intrinsic motivation and voluntary engagement (motivation-affect, persistence) more than one that controls them. Autonomy support here means choices that matter within a stated structure, a reason for work that is not negotiable, and informational rather than controlling language and feedback. The rationale comes from Self-Determination Theory, where autonomy is one of three basic needs. The design variable is autonomy-support, and so, on its controlling side, are rewards-incentives and praise.

The relationship weakens, or reverses, under four conditions:

  • Which outcome is measured. The evidence for motivation outcomes (free-choice persistence, interest) is stronger than the evidence for learning outcomes. Learner control in educational technology, pooled across studies, did not change achievement measurably.
  • What the learner controls and when. In one motor-learning experiment, controlling feedback helped retention and transfer only when the learner chose after seeing their own attempt. Choosing before the attempt gave no benefit. Controlling pace helped only on complex questions.
  • The learner's starting state. Prior knowledge of the task and the ability to direct one's own learning both matter. Some learners in open, self-directed courses could not direct their learning and reported demotivation. For a novice on a high-element-interactivity task, Cognitive Load Theory predicts that open choice adds search costs. No claim here tests that prediction for choice itself.
  • Horizon. Most of the motivational evidence measures engagement after a reward is withdrawn, in single sessions. Durable effects in a course over a term (term-plus) are not shown here.

Self-Regulated Learning holds the conditional model for what a learner does with the room that autonomy gives: planning, monitoring against a criterion, and choosing a next action. This page does not repeat it. Autonomy is the condition that leaves room for regulation. Whether the learner can use that room is self-regulated learning's question. No pattern page instantiates this principle yet. Learner Choice and Choice Boards are the nearest elements.

Default design, while the relationship is untested

This is the page's earlier guidance, kept as a concrete default that a designer can act on and revise. Each step is a design proposal, with its evidence stated beside it. No claim in the wiki tests autonomy-supportive teaching, as a package, against controlling teaching on learning outcomes.

  1. Fix the destination and vary the path. State the objective and the success criterion, then offer choices that change the substance of how it is reached or shown: topic, medium, order or pace. Cosmetic choices such as colour or font do not count. All options must target the same objective at the same rigour. Untested here; from the earlier page. Pace control has one experiment behind it, for complex material only (learner pacing [~M]). Keeping the criterion specific has goal-setting research behind it: a narrative review reports that specific, difficult goals outperform vague "do your best" goals when the learner is committed and able (specific goals [~M]). That review is about goals, not choice, so it supports fixing the destination and says nothing about how many paths to offer.
  2. Give a reason for required work. Open fixed activities with why the task matters, in terms the learner might value. Untested here. The suggestion comes second-hand through a theoretical article (autonomy-supportive versus controlling teaching [+W]), which describes rationale, choice and promoting the activity's value as autonomy-supportive strategies. It reports no study.
  3. Use informational, not controlling, language and incentives. Prefer "here is what would help" to "you must". Do not promise tangible rewards for doing an activity the learner already finds interesting. Feedback about competence is fine. Partly supported: expected tangible rewards reduced later free-choice engagement in one meta-analysis, a competing meta-analysis found the effect minimal, and both found that verbal praise did not undermine engagement (rewards [~M]).
  4. Bound the choice to what the learner can judge. For learners new to the task, offer two or three vetted options and model how to choose, then widen the options as competence shows on unsupported tasks. Untested here; from the earlier page. The claims on open, self-directed courses (connectivist courses [~W]) and on learner control (learner pacing [~M]) argue against unbounded choice. Neither tests bounded choice. One experiment with high-school writers found that novices given process goals (the steps of the strategy) outperformed those given outcome goals, and that a shift to outcome goals helped once the skill was in hand (process goals for novices [~M]). The goals were assigned, not chosen, so it supports telling a novice what to attend to, not withholding choice. Dose and progression: in early sessions offer the narrow choice (topic or order) and fix the process; widen to choices of method or product once the learner meets the criterion on an unsupported task; hand over the choice of goal last. Change gradually. In an 18-month case study of three one-to-one music-lesson pairs, a teacher's sudden, large increase in autonomy support did not hold and the pair reverted to a directive pattern within a few lessons, while another teacher's autonomy-supportive framing, kept up through a five-year-old's early resistance, settled into a high-engagement pattern (scaffolding and autonomy form self-reinforcing patterns [~M]). Three purposively chosen cases show what can happen, not how often.
  5. Put the learner's choice after their own performance where possible. Let the learner ask for feedback, a hint or a worked step once they have seen how an attempt went, not before. One laboratory motor-learning experiment (deciding after the trial [~M]). Do not carry it to academic tasks as a finding.
  6. Leave room to plan, monitor and reflect, and support those steps as Self-Regulated Learning describes. See that page for its evidence.
  7. Judge the design on the outcome that matters. Check a learning measure as well as a motivation measure. Reported enjoyment or satisfaction alone does not show that the design worked.

How far each neighbouring claim may be carried: the reward meta-analyses are mostly laboratory studies of already-interesting tasks. The feedback-timing study used a slider-positioning task with university volunteers. The learner-control synthesis covers educational technology in general. The connectivist report is second-hand and describes open online courses. The goal-setting claims concern assigned goals. The music-lesson study is three dyads of one-to-one instrumental teaching. The early-literacy claims come from one ERIC report on four European projects, partly the authors' argument and partly teacher report. None of them tests a classroom autonomy-support programme.

Fitting the design to a situation

Ask for three facts when the brief omits them: which outcome counts (voluntary engagement and persistence, or learning on a stated measure), how much the learners already know about the task, and what incentives are already in place (grades, points, prizes). Most rows rest on one study or a disputed meta-analytic literature, so treat them as defaults to check, not settled findings.

If the brief says… Then change… Basis
The goal is sustained voluntary engagement or persistence with the subject Offer choices that change how the objective is reached (topic, medium, order), give a reason for required work, and use informational rather than controlling language Autonomy support increases intrinsic motivation and engagement [+M]; Autonomy-supportive versus controlling teaching [+W]
The plan uses points, prizes or other tangible rewards for doing a task learners already find interesting Drop rewards for merely doing the task; keep feedback about competence, which did not undermine engagement Expected tangible rewards reduce later free-choice engagement [~M]
Praise is part of the routine (children, after success) Praise the effort or strategy used, not the learner's intelligence Praise for intelligence undermines motivation after failure [~M]; Intelligence praise against effort praise [~M]
The goal is achievement, and the plan is to give learners control of the software or sequence Do not expect control in general to raise achievement; if any control is given, give control of pace on complex, high element-interactivity material Learner pacing helped on complex material; learner control in general showed almost no effect [~M]
Learners can ask for feedback or hints during practice Let them ask after they have seen how an attempt went, not decide in advance; check whether this carries over from the motor task it was found on Deciding on feedback after the trial improved motor retention [~M]; Deciding only before the trial showed no benefit [~M]
An open, self-directed course with little structure (online-self-paced, MOOC-style) Keep the choice of topic or project, but supply a plan, milestones and a few vetted options; teach planning as in Self-Regulated Learning Not all students in connectivist courses could direct their own learning [~M]
Learners are new to the task (novice) and the brief asks for wide choice Offer two or three options that target the same objective, model how to choose, and widen the options as unsupported performance improves; state process goals (the steps to attend to) early, outcome goals once the skill holds bounded choice is an untested proposal, bounded by connectivist courses [~M]; Process goals beat outcome goals for novices [~M] (assigned goals, carried to choice untested)
Young children (child, about five) in early literacy, and the brief asks for child-chosen activities Distinguish choosing which activity from thinking for oneself within it. Keep teacher-led activities in which children make their own connections (an open shared reading of a big book, a text the children compose with the teacher as scribe), and keep explicit teaching of the spelling system; do not rely on free choice of activity to reach it Autonomy at one level may inhibit another [~W]; Orthographic mastery needs instruction [~W]; Big-book reading and cognitive autonomy [+W]; Co-construction with the teacher as scribe [+W]; all from one report, argument and teacher report, no comparison
One-to-one tuition or coaching over many sessions, and the learner resists early choices Keep offering choices and reasons in explanatory rather than directive language for several sessions before judging; raise autonomy support in steps, not in one jump; watch a learner with little wish for control, whose pair may drift into a teacher-led pattern Scaffolding and autonomy form self-reinforcing patterns [~M] (three music-lesson cases, carried to other tuition untested)
Online course in which discussion converges on the teacher (online-instructor-led, asynchronous boards) Give small groups a student-led discussion task: rotating leaders choose the topic, run the discussion and post a summary, and the task is assessed Student-led discussion groups sustained online participation [+W]; The redesigned unit engaged students on all framework elements [+W] (one 40-student unit, one week compared with another, no learning measure)
Group work where learners might pick partners, roles or start times Share the decision: the teacher forms groups and learners choose roles, the teacher sets the collaboration content, and collaboration starts when learners signal they are ready Teachers and students favoured shared over full student control [~M]; Collaboration starting when students signal readiness was preferred [~M] (stated preferences in a tool-design study, not outcomes)
Use of a tool (an AI writing assistant, a platform) is mandated and monitored Explain why the tool helps and leave learners room in how they use it; do not count on the mandate to produce deep use: in one survey, a mandate showed no detectable direct link to deep revision External mandate showed no direct relationship with deep revision [~M] (associational; no equivalence test)
A teacher is present in an AI-based or technology-rich lesson Use the teacher's support for motivation, and leave learners some unsupported problem solving Teacher support benefits depend on balance with student autonomy [~M]
Success will be judged by satisfaction or enjoyment surveys Add a learning measure at the stated horizon; enjoyment of a choice does not show that learning improved untested proposal

Observation, state and explanation

What can be observed is behaviour under stated conditions: whether a learner starts or returns to a task when nothing requires it, what they choose when given options, how long they persist, what they say about why they are doing the work, and how they respond to a reason or an incentive. "Feels autonomous", "is intrinsically motivated" and "is controlled" are inferred states. A questionnaire score of perceived autonomy is a self-report, not a measure of the state. Compliance also looks like engagement. Record what options were offered, what was said about why, what incentives were in place, and what the learner could already do.

Response Plausible explanations Observation that could change the interpretation
Works only while a reward, grade or check is present; stops when it is removed Engagement controlled by the incentive; the task holds no interest for this learner; the learner cannot do the task unaided and the check provided structure Remove the incentive but keep the structure and a stated rationale, and record free-time engagement; give a short unsupported probe of the skill; ask what the learner would do with the task if it were not graded.
Given a choice, picks the easiest option or does not choose Avoiding effort or risk; cannot tell the options apart (too little task knowledge); options are trivial or irrelevant to them; choosing is unfamiliar or culturally unexpected Ask the learner to say what each option involves; offer fewer, more distinct options with a model of choosing; check whether the choice changes when the options are tied to something the learner said they value.
Reports enjoying the choice but learns no more Motivation moved and learning did not; the choice concerned features unrelated to learning; the chosen option was weaker; the learning measure is not aligned with what was chosen Compare learning across options on the same aligned measure; check whether the options were equal in rigour; reobserve at a delayed horizon.
Disengages or becomes anxious in an open, self-directed task Missing task knowledge or regulation skill rather than missing autonomy; too many decisions; does not value the goal Supply a plan or narrow the options, keep the choice of topic, and observe whether execution follows; give a content probe with support; ask what the learner is trying to achieve.
Refuses or sighs at offered choices early in a new one-to-one or small-group relationship Unused to being asked; testing the adult; prefers to be told (low wish for control); the choices are not ones the learner cares about Keep offering choices and reasons over several sessions and record each response; compare with how the learner acts when told what to do; ask which decisions the learner would like to make. In one case study, persistence through early resistance preceded a sustained high-engagement pattern (self-reinforcing patterns [~M]).
Engages more after a rationale is given Reason accepted as personally meaningful; teacher attention or relationship; novelty Record whether the learner can restate the reason in their own terms; compare engagement on later required tasks without a fresh rationale.

These rows are proposals and have not been tested with learners. A probe shifts confidence among explanations; it does not identify a cause. Offering a choice or asking about values is itself an intervention, so record every exposure.

Evidence and qualifications

No claim in the wiki tests the full relationship as stated. None compares an autonomy-supportive classroom or course with a controlling one on both motivation and learning, with assignment and horizon reported. The claims below each test a part of it.

Taken together: controlling incentives can reduce voluntary engagement with interesting tasks, though the size is disputed. Learner control in general does not raise achievement. In two experiments, specific, well-timed control helped: pacing on complex material, and feedback chosen after an attempt. Unstructured self-direction can demotivate some learners. Where learners and teachers were asked, both preferred shared to full student control. Case and report evidence suggests that autonomy support is built over sessions rather than switched on, and that for young children autonomy of thought within a teacher-led activity may matter more than choice of activity. Learners, tasks and outcomes differ across these claims, so do not rank them by effect label. A nonsignificant difference is not evidence of no effect.

The evidence by question

Each question below links a general claim that gathers the studies behind it into one table, one row per distinct study, with each study's design and effect as recorded and no averaged verdict (eval/synthesis/autonomy.json, scripts/synthesize_claim.py). The count is of distinct studies across the claim and the narrower claims that bear on it.

Further evidence, not yet read against this model

Claims this page cited before it was rewritten as a conditional model. The page cited one claim before the 2026-10-02 rewrite, Self-determination theory holds that autonomy support raises intrinsic motivation, and a meta-analysis finds expected tangible rewards undermine it [+M], and it has been read into the model above as a core claim. No cited claim remains to list here.

Several statements on the earlier page carried evidence markers with no claim behind them: the choice meta-analysis it named, the complementarity of autonomy and structure, and the age and culture moderators. They are kept in the deprecated block below. The wiki has no claim page for them, so they are not evidence here until one is written. Read against the model on 2026-10-19. Of the eighteen claims that had gathered here, fourteen now sit in the model above: in the default design (goal-setting, the music-lesson case study), in ## Fitting the design to a situation (novices, early literacy, one-to-one tuition, online discussion, group work, mandated tools), in the observation table, and in ## Evidence and qualifications. The four below were read and left here, because they do not change a design decision on this page.

Objective and learner-valued goal

The designer's objective is a capability, measured by a stated instrument at a stated horizon. Autonomy support is a means to it, and also, sometimes, an objective in itself (sustained voluntary engagement). Elicit separately what the learner wants, and why: to pass, to make something of their own, to stop being told what to do, to be left alone. Agreement, divergence and "I don't know" are all observations. A learner's valued goal is also what makes a rationale land. A reason the learner does not share is just one more instruction.

For example, an adult apprentice electrician may value wiring a real circuit at a job site, while the designer's objective is correct calculation of load on a written test. A choice between two calculation worksheets offers autonomy over something the learner does not care about. A choice of which real installation to plan the load calculation for, with the written test kept as the criterion, ties the choice to the learner's goal. Record both the engagement and the test result, so that a gain in one is not read as a gain in the other.

What would revise this model?

The model should weaken if well-controlled comparisons of autonomy-supportive against controlling conditions, in real courses, show no difference in voluntary engagement or persistence at a delayed horizon. It should also weaken if the reward-undermining effect turns out to be confined to laboratory tasks. If choice keeps showing motivation gains with no learning gains, keep that split rather than calling both "engagement". If well-timed, performance-contingent control (as in the feedback experiment) replicates in academic tasks while open choice does not, the principle should be restated around when the learner controls what, not around choice in general. Do not protect the model by calling every failure of choice a lack of structure, or every success of direct instruction a hidden form of autonomy support.

Motivation during a task, free-choice engagement afterwards, immediate learning, delayed retention and the learner's own valued use are separate claims. The present evidence does not establish how many options to offer, how to sequence widening choice, or how culture and age moderate any of it. For young children there is one report's argument and observation, and no comparison.

Claims

Examples

Illustrative

Choice Boards — A grid of task options (e.g., demonstrate understanding via podcast, diagram, or essay) that lets learners select how to show mastery while all options target the same objective. Effective when options are genuinely equivalent in rigor and aligned to the same learning objective.

Academic Choice (planning, working, reflecting) — A Responsive Classroom routine in which learners plan what they will do, work with that plan, and then reflect on their choices — building self-regulation alongside ownership.

Project-based learning with negotiated topics — Curricula like Expeditionary Learning / EL Education give students voice in selecting expedition topics and products within teacher-defined constraints, pairing autonomy with strong structural scaffolds.

Rationale-first task framing — Rather than changing the task, the teacher opens with "This practice matters because…" before assigning required work. Reeve's research shows rationale-giving alone measurably shifts students' perceived autonomy, even for non-negotiable activities.

Duolingo — Combines a fixed skill sequence (structure) with learner control over daily goals, lesson order within unlocked units, and session length — a commercial example of autonomy-within-structure design.

Key Sources

  • Deci, E. L., & Ryan, R. M. (2000). The "what" and "why" of goal pursuits: Human needs and the self-determination of behavior. Psychological Inquiry, 11(4), 227–268. doi:10.1207/S15327965PLI1104_01
  • Patall, E. A., Cooper, H., & Robinson, J. C. (2008). The effects of choice on intrinsic motivation and related outcomes: A meta-analysis of research findings. Psychological Bulletin, 134(2), 270–300. doi:10.1037/0033-2909.134.2.270
  • Reeve, J. (2006). Teachers as facilitators: What autonomy-supportive teachers do and why their students benefit. The Elementary School Journal, 106(3), 225–236. doi:10.1086/501484
  • Katz, I., & Assor, A. (2007). When choice motivates and when it does not. Educational Psychology Review, 19(4), 429–442. doi:10.1007/s10648-006-9027-y
  • Ryan, R. M., & Deci, E. L. (2020). Intrinsic and extrinsic motivation from a self-determination theory perspective: Definitions, theory, practices, and future directions. Contemporary Educational Psychology, 61, 101860. doi:10.1016/j.cedpsych.2020.101860
  • Hanewicz, C., Platt, A., & Arendt, A. (2017). Creating a learner-centered teaching environment using student choice in assignments. Distance Education, 38(3), 273-287. https://doi.org/10.1080/01587919.2017.1369349
  • Stefanou, C. R., Perencevich, K. C., DiCintio, M., & Turner, J. C. (2004). Supporting autonomy in the classroom: Ways teachers encourage student decision making and ownership. Educational Psychologist, 39(2), 97-110. https://doi.org/10.1207/s15326985ep3902_2
  • Herbold, K. (2011). Giving student choice in online learning environments: Addressing adult learner needs. International Journal of Technology, Knowledge & Society, 7(5). doi:10.18848/1832-3669/cgp/v07i05/56230
  • Murray, G. L., & Bollinger, D. J. (2001). Developing Cross-Cultural Awareness: Learning Through the Experiences of Others. TESL Canada Journal, 19(1). https://teslcanadajournal.ca. doi:10.18806/tesl.v19i1.920
  • Lundberg, M., & Rasmussen, J. (2018). Foundational Principles and Practices to Consider in Assessing Maker Education. i-manager's Journal of Educational Technology, Vol. 14 No. 4, January - March 2018. https://eric.ed.gov/?id=EJ1179517
  • Pittaway, S. M., & Moss, T. (2014). “Initially, We Were Just Names on a Computer Screen”: Designing Engagement in Online Teacher Education. Australian Journal of Teacher Education, 39(7). http://ro.ecu.edu.au/ajte/vol39/iss7/8. doi:10.14221/ajte.2014v39n7.10
  • Kupers, E., van Dijk, M., & van Geert, P. (2017). Changing Patterns of Scaffolding and Autonomy During Individual Music Lessons: A Mixed Methods Approach. Journal of the Learning Sciences, 26(1), 131-166. https://doi.org/10.1080/10508406.2016.1259624
  • Level Creation Tools and Deep Gamification in the BOTS Educational Game. (2015). https://gameslearningsociety.org
  • Dombey, Henrietta; Formisano, Marina Pascucci. (1999). Examining the Notion of Autonomy in the Context of Early Literacy Learning in Four European Countries. https://eric.ed.gov/?id=ED434766
  • Carter, M. J., Carlsen, A. N., & Ste-Marie, D. M. (2014). Self-controlled feedback is effective if it is based on the learner's performance: a replication and extension of Chiviacowsky and Wulf (2005). Frontiers in Psychology, 5, 1325. doi:10.3389/fpsyg.2014.01325
  • Zimmerman, B. J., & Kitsantas, A. (1999). Acquiring writing revision skill: Shifting from process to outcome self-regulatory goals. Journal of Educational Psychology, 91(2), 241–252. doi:10.1037/0022-0663.91.2.241
  • Locke, E. A., & Latham, G. P. (2002). Building a practically useful theory of goal setting and task motivation: A 35-year odyssey. American Psychologist, 57(9), 705–717. doi:10.1037/0003-066X.57.9.705
  • Yang, K. B., Liu, M., Xu, L., Rummel, N., & Aleven, V. (2026). Balancing Teacher and Student Agency: Co-Orchestration Tool Design Supporting Real-Time Dynamic Pairing. doi:10.1145/3816902
  • Li, H., & Zhang, W. (2026). Empowerment over enforcement: unpacking the psychological drivers of AI-assisted deep revision in EFL writing. Frontiers in Psychology, 17, 1871022. doi:10.3389/fpsyg.2026.1871022
  • Rothes, A., Lemos, M. S., & Gonçalves, T. (2017). Motivational Profiles of Adult Learners. Adult Education Quarterly. doi:10.1177/0741713616669588
Page metadata
Field Value
type principle
id autonomy
aliases learner-choice, structure-with-student-control-cross-cultural-tasks, student-choice-builds-ownership-maker-education, design-public-social-student-directed-tasks, persistent-autonomy-support-during-resistance, balanced-authoring-constraints, autonomy-goal-and-route-early-literacy
title Autonomy
description For a learner whose engagement is controlled by pressure or reward, meaningful choice within structure, a rationale for required work and non-controlling language are expected to sustain motivation and voluntary engagement; their effect on learning outcomes is small, inconsistent and depends on what the learner controls and when.
status review
canonical True
generated by claude/unspecified on 2026-10-19
trust unverified — no one has confirmed this page's content
sources 7 — hanewicz-2017, stefanou-2004, murray-2001, lundberg-2018, pittaway-2014, level-creation-tools-and-deep-gamification-in-the-bots-educational-game-2015, dombey-1999

Frontmatter is the page's machine-readable record — see the Schema & Guide.