The student sits down with a plan.
Twenty minutes later, the plan has collided with reality.
The task was harder than expected. Attention wandered. One strategy worked; another did not. The deadline is closer. Confidence has changed. The learner now has a choice: repeat the original plan, abandon the task, or revise what happens next.
That movement between intention, action and reflection is central to the work of Barry J. Zimmerman, one of the major figures in modern research on self-regulated learning.
Zimmerman, a Distinguished Professor Emeritus at the Graduate Center of the City University of New York, died on 17 February 2025 at the age of 82. CUNY's memorial account describes a career spanning decades in which he developed influential models of how learners proactively regulate cognition, motivation and behavior rather than being treated as passive recipients of instruction.
His work matters because “independent learner” can sound like a personality type.
Zimmerman turned it into a process that could be studied.
Self-regulated learning is broader than metacognition
Metacognition is essential to self-regulated learning, but the two are not synonyms.
A learner may monitor understanding, notice confusion and choose a strategy. Those are metacognitive processes. But self-regulated learning also includes goal setting, motivation, self-efficacy, persistence, management of time and environment, help seeking and behavioral choices.
Ernesto Panadero's 2017 review of six major self-regulated learning models describes SRL as an umbrella involving cognitive, metacognitive, behavioral, motivational and emotional or affective dimensions.
This breadth is one reason the field is both useful and difficult to measure.
A questionnaire that asks whether a learner usually plans ahead is not equivalent to observing how that learner regulates a particular task. A strong model therefore needs to specify processes, phases and context rather than treating “self-regulation” as one amount a person possesses.
Zimmerman's model is cyclical
One of Zimmerman's most influential contributions is the cyclical phase model.
In simplified form, self-regulation unfolds across three recurring phases: forethought, performance, and self-reflection.
Forethought includes task analysis, goal setting, strategic planning and motivational beliefs. The learner forms an approach before action begins.
Performance includes self-control and self-observation during the task. Attention, strategy use, imagery, self-instruction and monitoring operate while the work is happening.
Self-reflection includes self-judgment and self-reaction after performance. The learner interprets outcomes, attributes causes, experiences satisfaction or dissatisfaction, and carries those reactions into the next cycle.
The important word is cycle.
Reflection is not the end of learning. It changes the next round of forethought.
A disappointing result can produce better planning, defensive withdrawal, more effort, strategy change or loss of confidence depending on how it is interpreted.
The model is social-cognitive, not purely internal
Zimmerman worked within a social-cognitive tradition influenced by Albert Bandura.
This means self-regulation is not imagined as a sealed process inside the learner. Personal factors, behavior and environment interact.
A student may change the environment by moving to a quieter room. A teacher's feedback changes the learner's interpretation of performance. A peer provides a strategy that alters future planning. A deadline changes motivation and time allocation. Prior success changes self-efficacy, which changes the difficulty of goals attempted next.
This matters politically and ethically in education because the language of self-regulation can otherwise become individualising.
If a student struggles, institutions may say they need better self-management while ignoring inaccessible materials, unstable housing, hostile classrooms, disability barriers or poorly designed instruction.
Zimmerman's framework does not require that mistake. Environment is part of the regulation system.
Self-efficacy is a mechanism, not motivational decoration
Zimmerman's work repeatedly examined self-efficacy: beliefs about one's capability to perform particular actions or tasks.
Self-efficacy can influence goal choice, effort, persistence and response to difficulty. But it should not be confused with generic confidence or self-esteem.
It is domain- and task-related.
A learner may have high self-efficacy for analysing literature and low self-efficacy for statistics. Those beliefs may be more or less calibrated to performance.
The important scientific question is not whether positive belief is always good. Overconfidence can impair regulation if it suppresses preparation or help seeking. Underconfidence can narrow participation despite sufficient capability.
For self-regulation, belief matters because it helps determine what actions a learner is willing to attempt and sustain.
Zimmerman also described how self-regulation develops
His work did not treat expert self-regulation as something learners simply decide to possess.
Zimmerman proposed developmental levels in which skills can move from observation to emulation, then self-control, and finally self-regulation. A learner first sees a model, then imitates key patterns with support, then performs in structured conditions, and eventually adapts strategies flexibly to changing contexts.
This sequence is important because it protects self-regulated learning from a common pedagogical contradiction: telling novices to “take ownership” while withholding the modelling and scaffolding needed to learn how.
Autonomy is not the absence of teaching.
It can be an outcome of teaching that gradually makes strategic control more available to the learner.
Measurement has always been a difficult part of the field
Self-regulated learning is dynamic, yet many of its most convenient measures are static self-reports.
A student may endorse “I plan my study time carefully” without doing so in the task being studied. Retrospective reports are vulnerable to memory, self-presentation and general beliefs about oneself.
Zimmerman and later researchers therefore developed and used more task-proximal approaches, including think-aloud methods, structured interviews and microanalytic measures that ask about forethought, strategy use or reflection at specific points in an activity.
No method is perfect.
Observation may miss internal regulation. Self-report may capture belief rather than behavior. Trace data from digital systems can record clicks and timing without revealing the reasons behind them.
A strong evidence base often triangulates across methods.
The model became influential across domains
Zimmerman's self-regulated learning framework travelled beyond conventional classroom study.
Researchers applied related principles in sport, music, health, writing and professional education. This breadth reflects a genuine feature of regulation: goal-directed cycles of planning, performance and reflection occur in many human activities.
But cross-domain use does not imply that the same strategies work identically everywhere.
The task structures differ. Feedback differs. Stakes differ. Expertise changes what needs monitoring. Emotional demands differ.
The general architecture may travel more easily than specific interventions.
This is one reason profiles should distinguish the influence of a model from universal validation of every application carrying its name.
European research has tested the architecture against alternatives
Self-regulated learning is not a one-model field.
Panadero's 2017 review, from the Autonomous University of Madrid, compared six influential SRL models and identified broad areas of convergence as well as differences in phase structure, motivation, emotion and metacognitive emphasis.
This European synthesis is useful because it resists founder-centred history. Zimmerman is highly influential, but Pintrich, Winne and Hadwin, Boekaerts, Efklides and others developed distinct frameworks.
No single model should be treated as the final ontology of self-regulation.
The field is stronger when models are compared by the phenomena they explain, the measurements they enable and the predictions they make.
The learner can regulate badly
The term self-regulation sounds intrinsically positive.
But regulation can be ineffective. A learner can set unrealistic goals, choose weak strategies, monitor the wrong cues, attribute failure inaccurately or persist when strategy change would be wiser.
This matters because self-regulation is not equivalent to self-control in the moral sense.
It is a process architecture. The quality of regulation depends on the quality of information and strategy inside the loop.
A highly organised learner using ineffective methods may regulate consistently toward poor outcomes. A flexible learner who revises plans may appear less orderly while learning more.
That makes feedback and metacognitive calibration central.
AI can regulate parts of learning on the learner's behalf
Digital systems increasingly plan, prompt, schedule, summarise and recommend next actions.
An AI tutor can choose which problem appears next, detect an error, remind the learner of a goal and suggest a strategy. These functions overlap with parts of self-regulation.
This can be supportive. It can also alter what the learner practises.
If the system always sets goals, does the learner learn to set them? If it detects every knowledge gap, does the learner become better at monitoring? If it schedules every review, does the learner develop strategy knowledge or simply comply with reminders?
The evidence does not yet justify a universal answer.
The Institute frames this as a regulation allocation question: which regulatory functions should be externalised, shared or deliberately retained for human practice? That is Institute synthesis, not a standard Zimmerman construct.
The answer may differ by age, expertise and purpose.
Self-regulation is not abandonment
There is a cultural danger in celebrating self-regulated learners at a time when institutions are eager to reduce support.
A school can call asynchronous materials “learner autonomy” when it has simply removed teachers. A workplace can call unmanaged reskilling “self-directed development” while transferring all responsibility for adaptation to employees.
Zimmerman's framework should not be used to baptise neglect.
Self-regulation develops through models, feedback, opportunities, environments and social support. Help seeking itself can be a regulatory strategy.
The mature question is not how to make the individual carry everything alone.
It is how to design environments in which people progressively gain more accurate and flexible control over their own learning.
Goals are not neutral containers
Self-regulation begins before a strategy is chosen because goals determine what counts as progress. A learner trying to finish a chapter quickly will regulate differently from one trying to be able to explain the argument tomorrow. A musician trying to avoid mistakes during rehearsal may choose different practice from one preparing for robust performance under pressure.
Zimmerman's framework therefore links goal properties with strategy and self-evaluation. Specific, proximal goals can make progress easier to monitor, while long-term goals can organise persistence and meaning. But no goal form is universally optimal. Overly narrow goals can produce local success at the expense of broader learning.
This point matters for educational technology. If the system's goal is completion, it may optimise prompts for completion. If the goal is durable transfer, a different sequence of feedback and testing may be appropriate. Regulation is always regulation toward something.
The Institute treats goal ownership as an applied design question: who defines the criterion by which the learner regulates? This is not a distinct Zimmerman construct, but it becomes important when platforms increasingly choose tasks, pacing and success metrics automatically.
Emotion is part of the loop, not noise around it
Self-regulated learning models developed over time to give more explicit attention to emotion as well as cognition and motivation. Failure can produce frustration, shame, curiosity or determination. Those responses affect whether the learner persists, seeks help, changes strategy or disengages.
A purely cognitive account can misread withdrawal as lack of planning when the regulatory problem is emotional cost. Conversely, a learner may persist because anxiety prevents them from stopping, even when strategy change would be wiser.
Panadero's comparative review is useful here because it shows how different SRL models distribute emphasis across cognition, motivation, affect and context. The field is not converged on one fine-grained process map.
For applied work, the implication is restraint. A dashboard that infers “poor self-regulation” from late submissions may be measuring workload, care responsibilities, disability, stress or task design. Trace data can reveal behavior; interpretation needs context.
Feedback must become usable information
Feedback does not regulate learning by existing. The learner has to perceive it, interpret it and convert it into a change in goals, strategy or effort. This links Zimmerman's work to feedback literacy and metacognitive control.
A comment such as “be more analytical” may be evaluative without being regulatory. It tells the learner something is lacking but may not reveal what to do differently. More specific feedback can still fail if the learner cannot diagnose when the advice applies.
The self-regulated learner is therefore not simply someone who receives less help. It is someone increasingly able to turn information about performance into a next action. That capacity can be scaffolded socially before it becomes more independent.
What the evidence does not show
Zimmerman's work does not show that self-regulated learning is one stable trait.
It does not show that metacognition and self-regulated learning are interchangeable terms.
It does not show that self-report questionnaires directly measure actual regulation in every task.
It does not show that learners should regulate without guidance.
It does not show that self-efficacy should simply be maximised regardless of accuracy.
It does not show that one cyclical model is the only valid representation of learning regulation.
And it does not justify institutional policies that individualise failures caused by poor environments or unequal resources.
A legacy of turning agency into process
Zimmerman's lasting contribution is not the idea that learners should be responsible.
It is the scientific decision to ask what learner agency consists of in action.
What happens before the task? What happens while difficulty unfolds? What does the learner notice? How do they interpret the result? Which belief, strategy or environmental choice changes the next attempt?
Those questions turn “independent learning” from a character judgment into a sequence that can be taught, observed, supported and revised.
The student at the desk will not follow the original plan perfectly.
Self-regulation begins where the plan meets evidence.
Research references
Zimmerman BJ. A social cognitive view of self-regulated academic learning. Journal of Educational Psychology. 1989;81(3):329–339. https://doi.org/10.1037/0022-0663.81.3.329
Zimmerman BJ. Becoming a Self-Regulated Learner: An Overview. Theory Into Practice. 2002;41(2):64–70. https://doi.org/10.1207/S15430421TIP4102_2
Panadero E. A Review of Self-regulated Learning: Six Models and Four Directions for Research. Frontiers in Psychology. 2017;8:422. https://doi.org/10.3389/fpsyg.2017.00422
The Graduate Center, CUNY. In Memoriam: Distinguished Professor Emeritus Barry Zimmerman. 2025. https://www.gc.cuny.edu/news/memoriam-distinguished-professor-emeritus-barry-j-zimmerman
Educational Psychology Review. Barry J. Zimmerman’s Enduring Legacy to Educational Psychology. 2025. https://doi.org/10.1007/s10648-025-10052-0