# Metacognition Examples and Exercises for Workplace Training

Metacognition explained through a worked example and practical exercises. Compare two attempts and adapt a worksheet for professional training.

- Canonical URL: https://www.harmate.com/en/blog/metacognition-examples-and-exercises-for-workplace-training
- Author: Harmate Team
- Published: 2026-10-08
- Updated: 2026-10-08T02:05:32.640024+00:00
- Language: en

## Content

# Metacognition Examples and Exercises for Workplace Training

Metacognition becomes practical when it is tied to a real task: predict what may be difficult, try the task, compare your expectation with visible criteria, explain a gap, adjust your approach, and try a comparable case again. Asking “Did you understand?” does not show what someone can actually do.

This guide walks through a complete exercise for adult professional training. The scenario and its rules are fictional. The activity is not a score of a person and does not diagnose motivation, intelligence, or potential.

## Metacognition: definition and difference from cognition

Metacognition means knowledge about and regulation of your own cognitive activity. Cognition is involved when you understand a request, remember a rule, or write a reply. Metacognition is involved when you examine that process: “Did I actually verify this information? Why is this approach not working here? What should I change?” The [French Scientific Council for National Education’s glossary (in French)](https://www.csen.education.gouv.fr/publications/glossaire-metacognition/) distinguishes those two components. Knowing a strategy does not mean using it at the right time.

In training, learning how to learn does not mean discovering a fixed “visual” or “auditory” learner identity. The starting point here is more concrete: a task, an explicit strategy, and a check of what that strategy enables someone to do.

## Why confidence alone is not enough

“I feel comfortable with this” describes a present impression, not the quality of a response in context. Hesitation does not prove inability either: the prompt may be incomplete, the rule unfamiliar, or a work tool unavailable.

Confidence becomes useful evidence for reflection when it is compared with an observable work product and criteria shared in advance. Ask, “Which parts do you expect to handle independently? What evidence would let us check?” The question is about this task, not “Are you competent?”

The Education Endowment Foundation’s 2025 update on metacognition and self-regulated learning advises schools and teachers to make planning, monitoring, and evaluating learning explicit and to embed these approaches in subject lessons. Its guidance is about pupils in schools; it does not establish that the same approach has the same effects in adult workplace training. [Read the EEF guidance](https://educationendowmentfoundation.org.uk/education-evidence/guidance-reports/metacognition).

Dunlosky and Rawson (2012) studied college students learning definitions, judging their own answers, and taking a recall test two days later. In that specific setup, more accurate self-evaluations were associated with better retention. The study does not show that a confidence exercise predicts an employee’s job performance. [ERIC record EJ964388](https://eric.ed.gov/?id=EJ964388).

## A fictional case: handling an unclear refund request

A customer-service trainee receives this message: “My subscription just renewed. I didn’t know it renewed automatically. Please refund me and stop future charges.”

For this exercise, the fictional company has three rules:

- in this workshop, a refund is authorized if the history confirms a first renewal no more than 14 days before the request; otherwise, refer the case for review without promising a refund;
- the agent must check billing history before confirming eligibility;
- cancellation of future charges and a refund are separate actions to verify.

These rules describe neither applicable law nor a real commercial policy. The fictional case record is available: it shows a first renewal eight days before the request, with no cancellation recorded. The workshop simulator lets learners record a refund request and a cancellation separately. Requesting a refund does not mean that money has been paid back.

Success criteria are shared before the attempt: consult the date and history; distinguish a request from a decision; verify refund and cancellation separately; explain the next step without promising an unconfirmed outcome.

Before writing, the learner reads these criteria and makes a prediction: “I expect to answer without help; my main check will be that the message is polite.” After the attempt, this expectation will be compared with the checks actually performed.

### First attempt

The learner replies: “Your refund is approved and your subscription has been canceled.” They believe the answer is clear and quick. Comparing it with the criteria reveals two observable gaps: the refund was confirmed without checking billing history, and a cancellation was claimed but not recorded.

The first case was eligible under the fictional rules: the eligibility conclusion happened to be right, but no check or refund request had been recorded. That fortunate result hides an approach that will fail when the history changes. After checking the facts and recording both actions, a correct first-case reply would be: “Your first renewal was eight days ago, so your case meets our rule. I have recorded the refund request; payment remains to be confirmed. The cancellation of future charges has been recorded.”

What did the learner assume? Which information did they not consult? In real situations, also check whether rules and tools were accessible: a gap may come from a learning need, a procedure, or an unclear prompt.

### Adjust, then try again

For the next attempt, the learner uses three prompts: “confirmed fact,” “fact to check,” and “next step I can promise.” Before replying, they will consult the history, place information in these categories, and verify recorded actions.

The trainer then offers a comparable case: the customer makes the same request, but the history changes. The elapsed time remains eight days; this is now a **third renewal**. The learner can record the cancellation but must refer the refund request for review: elapsed time alone is not enough.

Here is the complete second attempt, after consulting the history and recording the cancellation in the simulator:

> You have requested a refund and an end to future charges. I checked your record: the renewal was eight days ago and it was your third renewal. Our refund rule applies only to the first renewal, so I am referring your request for review without confirming approval. Cancellation of future charges has been recorded. The refund remains a separate request awaiting a decision.

The comparison can now be checked:

| Stated criterion | First attempt | Second attempt |
|---|---|---|
| Consult the date and history | No check performed | Date and third renewal verified |
| Separate request from decision | Approval claimed without checking | Review requested; approval not confirmed |
| Treat the two actions separately | Two outcomes claimed without action | Cancellation recorded; refund pending |
| Explain the next step | No step explained | Case referred for review |

The debrief concerns the changed approach: “In the first attempt, I thought a polite message was enough. In the second, I checked both facts and separated the actions. I still want to practice explaining a referral for review without implying approval.” This account identifies a strategy and a next difficulty; it does not establish every reason for the improvement.

The second attempt shows what this learner does in a nearby case, with the available resources and time. The limits of that observation are discussed below.

## Preparing and running a metacognition exercise

### 1. Choose a task that resembles the work

Select an action the learner may actually need to perform: answer a request, verify information, prepare a report, or communicate a decision. Avoid a broad question such as “What do you know about customer service?” It does not reveal how the person works through a case.

Give enough context to act, while leaving a real decision to make. Separate the information already supplied, the information to look up, and the information that cannot yet be confirmed.

### 2. Make the criteria visible before the attempt

Write a few criteria tied to the task, not to personality. In the fictional refund case, check whether the response:

1. identifies the date or history that must be verified;
2. distinguishes a request, a check, and a confirmation;
3. treats refund and cancellation as separate actions;
4. explains a next step without promising an unverified outcome.

If a criterion is unclear, improve it before reviewing the answer. An ambiguous rubric does not become fair just because everyone receives the same ambiguity.

### 3. Ask for a prediction tied to the criteria

Before the task, ask: “Which criterion feels easiest to meet? Which one might you miss? What makes you think so?” The learner can answer in a sentence or annotate the criteria.

This step does not need a number. A confidence scale may be useful within a defined protocol, but a standalone score does not explain the source of a gap or what to try next.

### 4. Produce a work trace

Let the learner write, calculate, verify, or explain their approach. Keep the evidence needed to compare expectation with outcome: a response, steps taken, source consulted, or rationale.

An open-ended explanation can surface an unexpected assumption or a question the trainer did not anticipate. It does not guarantee that the account is accurate or representative. Keep the learner’s wording available and distinguish it from your interpretation.

### 5. Compare and explain the gap

Review each criterion and ask: “What in the response meets it? What is missing? What did you expect?” Look for an actionable explanation: a rule was misunderstood, information was missing, the sequence was wrong, a tool was unavailable, or the prompt was ambiguous.

Avoid labels such as “not rigorous” or “lacks confidence.” They turn one observation into a judgment about the person and do not suggest a useful next attempt.

### 6. Pick an adjustment and try again

Match the adjustment to the gap. If the rule was forgotten, make it available during practice. If information was missing, practice asking for it. If the prompt was ambiguous, rewrite it. Then offer a comparable case, not an identical copy: the goal is to see whether the approach can be reused when one detail changes.

Compare both attempts against the same criteria. Note what became visible, what still required support, and what remains to be checked at work. Do not attribute every change to the exercise alone: case difficulty, time, tools, and help can also matter.

## A reusable exercise worksheet

Copy this outline and adapt it to a specific task. Criteria come from the work and are discussed with the learner, not inferred from their answers.

| Stage | Evidence to retain | Useful question |
|---|---|---|
| Before | Task, resources, criteria, and prediction | Which check might you overlook, and why? |
| During | Work product and checks actually performed | What is confirmed? What is still assumed? |
| After | Prediction and a specific gap in the work | What did you predict for this criterion, and what does the work show? |
| Adjustment | One precise change | Which check or aid will you add? |
| Retry | Different case and new work product | Is the change still useful when one detail changes? |

Two adaptations let learners practice the same loop without repeating the refund scenario. For learning to summarize a meeting, compare the first summary with visible criteria: are decisions, reservations, and responsible people distinguished? Then use a meeting where a proposal was not accepted. For learning a technical procedure, ask for a first execution with the usual resources, followed by a case containing an exception covered in the documentation. In both examples, difficulty should come from the work, not a trap designed to unsettle the learner.

## Analyze an error without judging a person

An error is a difference between a response and a defined criterion. By itself, it is not proof of a personal flaw. Separate four questions when planning what to do next:

| What you observe | What to check | Possible adjustment |
|---|---|---|
| A required step is missing | Was the rule available and understood? | Review it or provide a job aid |
| The answer asserts an unknown fact | What verification was missing? | Add a check before confirming |
| The reasoning changes across cases | Which detail changed the decision? | Compare two nearby cases under the same rule |
| The learner does not know what to write | Does the prompt specify the audience and expected outcome? | Clarify the format and retry |

Useful feedback cites part of the work, names the criterion, and suggests a next action. “You’re not confident enough” is not task feedback. “You confirmed the refund before checking the date; let’s practice that check in another case” is more actionable.

## What a trainer can conclude—and what still needs checking

A short activity may show that someone applied a stated step or rule in one case. It cannot by itself establish that they are now independent, will perform under pressure, or can handle every situation in the role.

To make the observation more useful, adapt the criteria to the real work, use more than one case, state what support is available, and ask the learner to explain choices that are not visible in the work product. If the criteria are disputed or reviewers disagree, keep that disagreement visible and clarify the standard instead of hiding it in an average.

The loop proposed here remains a way to design and check a fictional practice exercise—not a validated protocol or a guaranteed-effectiveness claim.

## Gather feedback without rating people

After the exercise, a few open questions can help uncover what was missing: “What did you check before answering?”, “When did you hesitate?”, or “What information would have helped?” These answers add context; they do not replace the work product or the criteria. For a related approach, see [formative assessment in training: observe, adapt, check](/en/blog/formative-assessment-in-training-observe-adapt-check).

A questionnaire can gather accounts and explanations from multiple learners. A tool such as Harmate can help organize that material while keeping source responses available for analysis; it does not decide whether an employee is “ready,” verify competence, or replace a trainer’s judgment.

The practical sequence is simple: ask for a prediction, observe a task, compare it with visible criteria, examine the gap, choose an adjustment, and check it in a new attempt. The work trace—and the next decision it supports—gives reflection something concrete to work with.