Ten Steps to Complex Learning

Design Process · All design processes
Evidence · 8 claims (5 for, 2 mixed, 1 against) · 14 studies (4 causal, 3 quant-synthesis, 3 theoretical, 2 review, 1 qualitative, 1 design), q1–q4 · 1 of 14 report an effect size · 4 claims rest on one study

Description

Ten Steps is the design process that produces a 4C/ID blueprint. The pattern describes what the resulting instruction looks like — learning tasks, supportive information, procedural information, part-task practice. This page describes what a designer does to arrive at it, which is a different object and a genuinely different way of working.

Its defining move is an inversion. An objectives-based process analyses a domain into constituent objectives and designs instruction for each; Ten Steps starts by specifying the professional tasks a graduate must be able to perform, translates those into learning tasks, and only then asks what theory a learner needs in order to complete them. As the model's own summary puts it, it "topples" the traditional approach of beginning with the theory and attaching practice items to it.

The unit of design is the task class: a band of whole professional tasks of equal complexity. Task classes are sequenced simple to complex, beginning with the simplest whole task a professional could encounter and ending at the complexity a recent graduate should handle. Within a class, support is deliberately faded — a modelling example, then completion tasks with parts already done, then conventional tasks with no support — and it resets at the start of the next class. Learning tasks also vary within a class, because variability of practice is what makes the skill transfer rather than the instance.

Assessment is designed simultaneously with the learning tasks rather than after them, on the grounds that assessment drives learning; and because progress is coordination across task classes rather than mastery of separate courses, the model expects a development portfolio rather than per-course pass/fail decisions.

The canonical enumeration of the ten steps is in van Merriënboer and Kirschner's book and is deliberately not reproduced here — see Key Sources. What this page records is the process's structure and its design commitments, which is what a designer choosing between processes needs.

Implications

Context

Requirements

• A skill whose difficulty is coordination, not a body of content. The model's whole argument is about integrating knowledge, skills and attitudes; applied to a topic that decomposes cleanly it is expensive overhead
• Access to authentic professional tasks, and to practitioners who can say what the real variability looks like
• A programme-length canvas. Task classes ordered simple to complex need somewhere to run; a single short course rarely has room for more than one
• Teaching staff prepared to coach, since the main activity becomes small-group work on whole tasks rather than presentation

Constraints

• It is demanding to design and to staff. The published implementation guidance is largely about organisational preconditions — small-group rooms, multidisciplinary teaching teams including practitioners, continuous progress monitoring, and preparing students for their new role — none of which is an instructional design problem
• Whole tasks are heavy for novices, which is why part-task practice exists as a component and why the fading schedule is not optional
• A competency framework is not a substitute for tasks. A programme that takes competencies as its backbone can still fail to transfer if the activities were not built from real professional tasks; the model asks for a matrix linking the two rather than a choice between them
• Fixed-length semesters with pre-planned exam weeks sit badly with portfolio-based progress monitoring

Grain Size

• Programme
• Course (one or two task classes)

Target Goals

• Professional competence — the integrated performance of whole tasks under real variability
• Transfer to the workplace, which is the outcome the model is explicitly optimised for
• Skills where coordinating constituent parts is the difficulty, rather than acquiring them

Target Learners

• Learners heading for a defined professional practice, where the tasks they will face can be characterised
• Novices, provided the fading schedule and part-task practice are actually built; the model is not a licence to hand beginners unsupported authentic work

Theory

Supporting

• Cognitive Load Theory — the model's mechanism: whole tasks raise intrinsic load, so support, sequencing and part-task practice exist to keep total load survivable while it is raised

Contradicting / Qualifying

• Behaviorism — the objectives-first tradition Ten Steps was formulated against, which decomposes a domain into separately-taught parts

Claims

• Ten Steps design of a workplace PD program is time and resource-intensive, spanning nearly two years for a five-month program [-W]
• Task complexity and task difficulty are distinct concepts [~W]

Supporting

• A training-design argument, not tested by any study recorded here, holds that practising whole complex tasks improves transfer of complex skills [+M] — the model's central bet
• Part-task practice reduces cognitive load for absolute novices during initial skill acquisition. [+M] — the fourth component, and why whole-task design does not mean whole-task only
• Contingent scaffolding improves learning more than fixed or absent support. [+M]
• Fading support promotes the transfer of responsibility from instructor to learner. [+M] — the within-class fading schedule

Contradicting

• Worked examples can become redundant or counterproductive for advanced learners. [~M] — the modelling example that opens a task class is wasted or harmful on learners who have already reached that complexity, which is what makes the reset at each new class a real design decision rather than a formality

Design

Sequence

1. Specify the professional tasks a graduate must perform, with practitioners, including their real variability
2. Link tasks to competencies — a matrix with tasks on one axis and competencies or standards on the other, so coverage is checkable in both directions. See Standards Crosswalk where the competencies come from outside
3. Group tasks into task classes of equal complexity and order the classes simple to complex
4. Design the learning tasks for each class, with variability within the class and a fading support schedule across it — modelling example, completion tasks, conventional tasks
5. Design the assessment programme alongside the tasks, not after them, and at portfolio rather than per-course granularity
6. Derive the supportive information each class needs — the theory, mental models and problem-solving strategies for the non-recurrent parts of the task
7. Derive the procedural information for the recurrent parts — step-by-step guidance presented at the moment of use
8. Add part-task practice where a routine sub-skill needs automating beyond what the whole tasks will give it

Elements Used

• Whole-Task Performance
• Supportive Information
• Part-Task Practice
• Simple-to-Complex Sequencing
• Transfer Tasks

Affordances

• Cognitive Load Management [+M] — the constraint every sequencing and support decision in the model is answering to

Personalization

• Learners may take different paths through the designed task classes based on demonstrated proficiency rather than a fixed order; development portfolios are what make that selection informed rather than arbitrary
• Where supportive and procedural information changes faster than it can be authored — a fast-moving technical domain — point learners at maintained external documentation instead of writing it, which is also more authentic
• Where a full programme is not available, design one task class properly rather than compressing several; a course that runs one class at genuine complexity is closer to the model than one that samples all of them

Related Processes

• Systematic Instructional Design — the objectives-first process this was formulated against
• Understanding by Design — also derives instruction backwards from a desired end, but from understandings rather than from professional tasks
• Successive Approximation Model — an alternative answer to the same complaint about waterfall design, iterating rather than restructuring the unit of design

Examples

• Blended Android app development course (Amsterdam University of Applied Sciences) — task classes from a single-screen dice app to multi-screen applications retrieving remote data, with modelling examples and completion tasks opening each class, and external documentation standing in for authored supportive information
• Health professions education, the setting much of the model's implementation guidance is drawn from
• Five-month workplace-based PD program blueprint for teacher differentiation skills

Key Sources

• Frerejean, J., van Merriënboer, J. J. G., Kirschner, P. A., Roex, A., Aertgeerts, B., & Marcellis, M. Designing Instruction for Complex Learning. Ch. 24 in Design for Learning. EdTech Books. https://edtechbooks.org/id/complex_learning
• van Merriënboer, J. J. G., & Kirschner, P. A. (2018). Ten steps to complex learning: A systematic approach to four-component instructional design (3rd ed.). Routledge. doi:10.4324/9781315113210
• Frerejean, J., van Merriënboer, J. J. G., Kirschner, P. A., Roex, A., Aertgeerts, B., & Marcellis, M. (2019). Designing instruction for complex learning: 4C/ID in higher education. European Journal of Education, 54(4), 513–524. https://doi.org/10.1111/ejed.12363

Page metadata

Field	Value
type	process
id	ten-steps-to-complex-learning
title	Ten Steps to Complex Learning
description	The task-centred design process that produces a 4C/ID blueprint — specify the professional tasks first, group them into task classes of equal complexity ordered simple to complex, then derive the supporting theory and part-task practice from the tasks rather than the other way round.
status	review
grain_size	programme
author	Jeroen J. G. van Merriënboer and Paul A. Kirschner
generated	by claude/unspecified on 2026-09-03
trust	unverified — no one has confirmed this page's content

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