COM-B model and Michie et al.'s Behavior Change Wheel intervention design tool
The Behavior Change Wheel (BCW), developed by Susan Michie, Lou Atkins, and Robert West (Michie et al., 2011, Implementation Science; and the accompanying book The Behaviour Change Wheel: A Guide to Designing Interventions) synthesizes 19 pre-existing behavior-change frameworks into a single systematic methodology, addressing a well-documented problem in intervention design: many behavior-change programs were historically built ad hoc, without an explicit theoretical basis connecting the chosen intervention components to the specific behavioral barriers they were meant to address.
Before any diagnostic or design work begins, the Behavior Change Wheel methodology insists on precisely specifying the target behavior using an explicit framework — commonly the "who, what, when, where, how often, with whom" specification popularized alongside the BCW literature — rather than working from a vague goal statement ("improve medication adherence") that leaves the actual target behavior ambiguous.
This precision matters because COM-B diagnosis and subsequent intervention-function selection (Stages 2–3) are behavior-specific: the capability, opportunity, and motivation barriers relevant to "a specific patient population taking a specific medication at a specific time of day" can differ substantially from the barriers relevant to a superficially similar but actually distinct behavior ("attending a monthly clinic appointment"), even though both might loosely fall under a vague umbrella goal like "improve adherence." A properly specified target behavior for a digital health context might read: "Adults aged 18–65 recently prescribed an SSRI take their prescribed dose at the same time each day, for the first 90 days of treatment, without prompting from another person" — specific enough that the subsequent diagnostic and intervention-design stages have a well-defined target to work against.
This emphasis on precise behavioral specification reflects the BCW's broader design philosophy, distinguishing it from less systematic intervention-design approaches: every subsequent step in the methodology (diagnosis, function selection, technique selection, evaluation) is explicitly and traceably linked back to this initial specification, producing an intervention whose components can each be justified by reference to a specific identified barrier for a specific defined behavior, rather than a bundle of generically "good practice" components assembled without a clear underlying behavioral logic.
At the center of the Behavior Change Wheel sits the COM-B model: the proposition that any behavior occurs as the result of an interaction between Capability (psychological and physical capacity to enact the behavior), Opportunity (external factors that make the behavior possible or prompt it), and Motivation (conscious and automatic mental processes that direct behavior) — and that a systematic behavioral diagnosis across these three components is the necessary precursor to selecting appropriate intervention content.
Each of COM-B's three core components splits into two sub-components, giving six distinct diagnostic dimensions:
• Psychological Capability: does the person have the necessary knowledge, cognitive skills, or psychological capacity (e.g., understanding why the medication matters, ability to plan and remember)? • Physical Capability: does the person have the necessary physical skill, strength, or stamina (e.g., dexterity to use an inhaler correctly)? • Physical Opportunity: does the environment provide the necessary time, resources, triggers, or physical infrastructure (e.g., is the medication accessible where and when it needs to be taken)? • Social Opportunity: do social norms, cultural context, or interpersonal relationships support or hinder the behavior (e.g., stigma around a mental-health treatment, family support or lack thereof)? • Reflective Motivation: does the person consciously plan, evaluate, and intend to perform the behavior (e.g., genuine belief the medication is worthwhile, formed intention to take it)? • Automatic Motivation: do the person's habits, emotional reactions, and impulses support the behavior (e.g., an established automatic routine, absence of an emotional aversion to the act)?
Diagnosis is typically conducted through a structured behavioral analysis — interviews, surveys, or validated instruments (e.g., the COM-B-based Theoretical Domains Framework, a complementary 14-domain elaboration frequently used alongside COM-B for finer-grained diagnosis) applied to the specific target population and behavior defined in Stage 1 — systematically identifying which of the six sub-components represent genuine barriers for this specific behavior and population, rather than assuming (as many less systematic intervention designs implicitly do) that motivation alone is the relevant barrier, when in many cases capability or opportunity deficits are the actual limiting factor and would render a purely motivation-focused intervention ineffective regardless of how well-designed it is.
Once the COM-B diagnosis identifies which specific capability, opportunity, or motivation sub-components represent genuine barriers, the Behavior Change Wheel's middle ring provides nine intervention functions — broad categories of intervention strategy — with an explicit, validated mapping (published in the original Michie et al. 2011 paper) specifying which functions are theoretically appropriate for which COM-B deficits, replacing intuition-based intervention selection with an evidence-linked matching process.
The Behavior Change Wheel's nine intervention functions, each targeting particular COM-B deficits according to the published mapping matrix:
• Education: increasing knowledge or understanding (targets Psychological Capability, Reflective Motivation) • Persuasion: using communication to induce positive or negative feelings or stimulate action (targets Reflective and Automatic Motivation) • Incentivisation: creating an expectation of reward (targets Reflective and Automatic Motivation) • Coercion: creating an expectation of punishment or cost (targets Reflective and Automatic Motivation) • Training: imparting skills (targets Physical and Psychological Capability) • Restriction: using rules to reduce opportunity for competing behaviors, or increase the target behavior (targets Physical Opportunity) • Environmental Restructuring: changing the physical or social context (targets Physical and Social Opportunity) • Modelling: providing an example for people to aspire to or imitate (targets Social Opportunity, Reflective and Automatic Motivation) • Enablement: increasing means or reducing barriers to increase capability or opportunity beyond education/training alone (targets all components broadly, e.g., through technology, medication, or practical support)
For a diagnosed capability deficit (a digital health example: patients don't understand why consistent medication timing matters — a Psychological Capability and Reflective Motivation gap), the mapping matrix directs the designer toward Education, Persuasion, and Training as the theoretically appropriate functions, while ruling out functions like Restriction or Coercion as poor theoretical fits for this specific diagnosed barrier profile.
Because most real-world target behaviors present barriers across multiple COM-B sub-components simultaneously (rarely a single, isolated deficit), most well-designed interventions select and combine multiple intervention functions (typically 3–5) rather than relying on a single function — directly reflecting the multi-component diagnosis from Stage 2 rather than a one-size-fits-all intervention strategy.
Intervention functions (Education, Persuasion, Training, etc.) are still broad strategic categories, not yet actionable content. The Behavior Change Wheel's outermost operational layer translates each selected function into specific, granular Behavior Change Techniques (BCTs) drawn from the standardized BCT Taxonomy v1 (Michie et al., 2013, Annals of Behavioral Medicine) — the actual, implementable content components of the eventual intervention.
Prior to the BCT Taxonomy's publication, behavior-change intervention research suffered from a well-documented "active ingredients" problem: published intervention descriptions used inconsistent, overlapping, and ambiguous terminology ("motivational interviewing," "goal setting," "self-monitoring") that made it genuinely difficult to know whether two studies claiming similar techniques were actually implementing the same underlying behavioral mechanism, hampering both intervention replication and evidence synthesis across studies.
The BCT Taxonomy v1 addresses this by defining 93 standardized, clustered techniques (grouped into 16 higher-level categories: Goals and planning, Feedback and monitoring, Social support, Shaping knowledge, Natural consequences, Comparison of behavior, Associations, Repetition and substitution, Comparison of outcomes, Reward and threat, Regulation, Antecedents, Identity, Scheduled consequences, Self-belief, Covert learning) — each BCT defined precisely enough, with explicit inclusion/exclusion examples, to be reliably and consistently identified, coded, and replicated across different interventions and research teams, functioning analogously to a standardized pharmacopeia for behavioral rather than pharmacological active ingredients.
For a digital health app implementing an Education + Persuasion + Training intervention-function combination targeting medication-timing consistency, appropriate BCTs drawn from the taxonomy might include: "Information about health consequences" (Shaping knowledge cluster), "Credible source" (Persuasion-aligned), "Self-monitoring of behavior" (Feedback and monitoring cluster), "Habit formation" (Repetition and substitution cluster), and "Action planning" (Goals and planning cluster) — each individually traceable back through the intervention-function selection (Stage 3) to the specific diagnosed COM-B deficit (Stage 2) it is intended to address, producing the fully theory-linked chain of justification that is the Behavior Change Wheel methodology's central design contribution.
A BCT can be theoretically well-matched to a diagnosed COM-B deficit and still be a poor practical choice for a specific real-world implementation context. The Behavior Change Wheel's final filtering stage applies the APEASE criteria — Affordability, Practicability, Effectiveness/cost-effectiveness, Acceptability, Side-effects/safety, and Equity — to narrow the theoretically-justified candidate techniques down to a deliverable final intervention package.
The six APEASE criteria, applied by the design team (ideally with target-population input) to each candidate intervention function and BCT:
• Affordability: can the intervention be delivered within the available budget, at the scale intended? A BCT requiring extensive one-on-one clinician time may be theoretically ideal but unaffordable at population scale, motivating a digital-delivery adaptation instead. • Practicability: can the intervention actually be delivered as designed, given real-world constraints (staff training, technology infrastructure, user technical literacy)? A sophisticated AI-driven personalization feature may be impracticable for a resource-constrained health system to build and maintain. • Effectiveness and cost-effectiveness: is there evidence this specific technique works for this behavior and population, and does its benefit justify its cost relative to alternatives? This criterion explicitly pulls in outcome evidence, not just theoretical COM-B fit. • Acceptability: will the target population and other stakeholders (clinicians, caregivers, regulators) find the intervention acceptable? A coercion-based function (e.g., financial penalties) may be theoretically effective for some behaviors but unacceptable, and potentially unethical, in a mental-health treatment-adherence context. • Side-effects/safety: could the intervention produce unintended negative consequences? This directly parallels concerns raised elsewhere in this simulation gallery — for instance, gamification mechanics that are theoretically effective for Motivation deficits but risk producing streak anxiety or compulsive-use side effects. • Equity: does the intervention reduce or exacerbate health inequalities across the target population? A smartphone-app-only intervention may inadvertently exclude lower-income or older patients with less digital access, widening rather than narrowing an existing disparity — a criterion explicitly built into the APEASE framework specifically because many popular intervention-evaluation frameworks omit equity considerations entirely.
Applying APEASE typically eliminates a substantial fraction (often 30–50%) of the theoretically-justified candidate techniques identified in Stages 3–4, reflecting the framework's explicit acknowledgment that theoretical appropriateness (a good COM-B-to-BCT match) is necessary but not sufficient for a deliverable, ethical, real-world intervention — the final output of the full BCW process is therefore not simply "the most theoretically justified intervention" but "the most theoretically justified intervention that also survives practical, ethical, and equity scrutiny."
The Behavior Change Wheel has become one of the most widely cited and applied intervention-design frameworks in public health and digital health intervention development specifically because it provides a fully traceable chain — from precisely defined target behavior, through COM-B diagnosis, through theoretically-matched intervention functions, down to specific standardized BCTs, filtered by explicit practical and ethical criteria — addressing the "black box" problem where earlier intervention designs could not clearly articulate which components were doing the actual behavioral work, or why they were selected over plausible alternatives.