Loading field map
Why do people smoke, exercise, or skip vaccinations—and what can be done to change those patterns? Health behavior theory emerged to answer these questions, but the answers have shifted dramatically over the past seventy years. Early frameworks assumed that behavior flowed from rational, individual calculations of risk and benefit. Later models argued that social context, environmental structure, and the quality of motivation matter at least as much. The most recent approaches ask whether behavior is best understood not as a product of discrete factors but as an emergent property of complex systems. Each framework in this sequence extended, challenged, or tried to unify what came before, and several remain in productive tension today.
The first generation of health behavior theory grew out of mid-twentieth-century public health campaigns. The Health Belief Model (HBM), developed in the 1950s to explain why people failed to participate in tuberculosis screening, proposed that behavior depends on perceived susceptibility, perceived severity, perceived benefits of action, and perceived barriers. It also included cues to action—a reminder or prompt. The HBM treated behavior as a rational weighing of threat and payoff, and it ignored social influence and habit. It was a practical tool for designing one-shot prevention messages, but its predictive power proved limited for behaviors that require sustained change.
In the 1970s, social psychologists extended the rational-choice logic. The Theory of Reasoned Action (TRA) argued that intention is the immediate cause of behavior and that intention itself is shaped by two factors: attitude toward the behavior (personal evaluation) and subjective norm (perceived social pressure). Unlike the HBM, the TRA explicitly incorporated social influence, but it assumed that most behaviors are under volitional control. When researchers applied the TRA to behaviors such as quitting smoking or using condoms, they found that perceived control often mattered as much as intention.
The Theory of Planned Behavior (TPB), introduced in 1985, addressed this gap by adding perceived behavioral control—the person's sense of how easy or difficult performing the behavior would be. The TPB kept the TRA's attitude and subjective norm components but positioned perceived control as both a direct predictor of behavior and a modifier of intention. This made the model more useful for behaviors that require resources or skills, but it still assumed a linear, conscious decision process. The HBM, TRA, and TPB together formed a cognitive-rational tradition that dominated intervention design through the 1990s.
At nearly the same time, Albert Bandura's Social Cognitive Theory (SCT, 1977) offered a fundamentally different picture. SCT introduced reciprocal determinism: behavior, personal factors (cognition, affect), and environmental influences all interact bidirectionally, not in a one-way chain from belief to action. Bandura emphasized observational learning, self-efficacy (the belief that one can execute the behavior), and outcome expectations. Self-efficacy quickly became the most powerful single predictor in health behavior research, surpassing constructs from the TRA/TPB lineage. Unlike the rational models, SCT treated the environment as something people both influence and are influenced by, and it gave a central role to modeling and social learning.
Shortly after SCT, the Transtheoretical Model (TTM, 1983) took a different approach by focusing on when people change rather than why. The TTM proposed that behavior change unfolds through five stages: precontemplation, contemplation, preparation, action, and maintenance. Each stage requires different processes of change—cognitive strategies for early stages, behavioral strategies for later ones. The TTM was appealing for its practical staging algorithms and for matching interventions to a person's readiness. However, critics argued that the stages are arbitrary boundaries drawn on a continuous process, that people often skip stages or relapse unpredictably, and that the model conflates time with genuine qualitative shifts. Compared with SCT, which explained change through a single set of mechanisms, the TTM offered a sequence of discrete phases but lacked a unified causal theory.
While SCT and the TTM debated change processes, another line of research asked whether the quality of motivation matters. Self-Determination Theory (SDT, 1985) distinguished between intrinsic motivation (doing something because it is inherently satisfying) and extrinsic motivation (doing something for an external reward or to avoid punishment). SDT argued that lasting behavior change requires the internalization of extrinsic motives, which depends on satisfying three basic psychological needs: autonomy, competence, and relatedness. This directly challenged the rational-intention models, which treated all motivation as equally effective if it produced a strong intention. SDT predicted that controlled motivation (e.g., from pressure or guilt) would be less stable than autonomous motivation, and hundreds of studies have supported that claim.
Also in the late 1980s, researchers began to argue that individual-level theories were insufficient because behavior is embedded in social and physical environments. The Social Ecological Model (SEM, 1990) formalized this by distinguishing multiple levels of influence: intrapersonal (individual beliefs), interpersonal (social networks), organizational (workplaces, schools), community (norms, resources), and policy (laws, regulations). The SEM did not propose specific causal mechanisms; instead, it provided a framework for thinking about interventions at every level. It coexisted with SCT, which also addressed environment, but SCT focused on how individuals learn from and act on their environment, whereas the SEM emphasized structural determinants that operate independently of individual perception. The SEM became especially influential for obesity prevention and tobacco control, where changing the environment (e.g., banning smoking in public places) often proved more effective than trying to change individual beliefs.
By the early 2000s, the proliferation of models created a problem: which one should an intervention designer use? The Integrated Behavioral Model (IBM, 2000) attempted to synthesize the strongest predictors from the TRA/TPB tradition, the HBM, and SCT. It kept intention as the central driver but added knowledge and skills as enabling factors, salience as a moderator, and environmental constraints as direct barriers. The IBM also incorporated attitude (experiential and instrumental), perceived norm (injunctive and descriptive), and personal agency (self-efficacy and perceived control). Despite its comprehensiveness, the IBM never gained wide adoption. It was criticized as a laundry list of predictors without a clear theoretical core, and its complexity made it hard to apply in practice. The IBM illustrates a recurring tension: more complete models often become too cumbersome for real-world use.
A different kind of simplification came from the Fogg Behavior Model (FBM, 2009), developed in the field of persuasive technology. The FBM states that behavior occurs when three factors converge: motivation (sufficient desire), ability (ease of performing the behavior), and a prompt (a trigger or call to action). The FBM is deliberately parsimonious—it collapses dozens of constructs into three dimensions—and it is designed for digital interventions where designers can manipulate prompts and ability through interface design. Unlike the IBM, the FBM does not aim to explain variance in naturalistic behavior; it aims to guide design. It has been criticized for oversimplifying motivation and for neglecting social and cultural context.
Around 2010, health behavior theory encountered a challenge from complexity science. Complexity Science and Systems Thinking (2010) argued that health behaviors arise from feedback loops, nonlinear interactions, and emergent properties that cannot be captured by linear predictive models. For example, smoking rates in a population are influenced by advertising, peer norms, addiction biology, policy changes, and economic factors—all interacting dynamically. Complexity researchers use system dynamics modeling, network analysis, and agent-based models to simulate how interventions ripple through a system. This approach rejects the idea that behavior can be understood by listing a fixed set of predictors; instead, it treats behaviors as emergent properties of interconnected systems. The practical implication is that interventions should aim to shift system structures (e.g., changing feedback loops) rather than change individual cognitions. Complexity science remains a minority approach in health behavior research, but it is growing, especially in obesity and chronic disease prevention.
In the same period, the Behavior Change Wheel (BCW, 2011) offered a comprehensive framework for designing interventions and policies. At its core is the COM-B model, which proposes that behavior occurs when an individual has Capability (psychological and physical), Opportunity (social and physical environment), and Motivation (automatic and reflective). These three components interact: for example, increasing opportunity can boost motivation. The BCW then maps COM-B to nine intervention functions (e.g., education, persuasion, environmental restructuring) and seven policy categories (e.g., guidelines, regulation). Unlike the IBM, which was a synthesis of existing constructs, the BCW was built from a systematic review of behavior change techniques and a consensus process. It has been widely adopted in public health and implementation science for its practical step-by-step guidance. The BCW aligns with complexity science in recognizing that behavior is shaped by multiple interacting factors, but it treats those factors as modifiable through planned interventions rather than as emergent system properties.
Current health behavior theory is pluralistic. The most active frameworks are Social Cognitive Theory, Self-Determination Theory, the Social Ecological Model, the Behavior Change Wheel, and Complexity Science and Systems Thinking. These frameworks agree that behavior is multiply determined—no single factor suffices—and that interventions must address context, not just individual decisions. They also agree that intentions are necessary but not sufficient; self-efficacy, motivation quality, environmental opportunity, and habit all matter.
Yet deep disagreements remain. The first concerns the locus of causality: cognitive-rational models and SDT locate the main drivers inside the individual (beliefs, needs), while the SEM and complexity science emphasize external structures and emergent dynamics. A second disagreement is about linearity: the BCW and most traditional models assume that changing a determinant (e.g., capability) will predictably change behavior, whereas complexity science argues that relationships are nonlinear and often produce counterintuitive effects. A third debate concerns parsimony versus comprehensiveness: the FBM champions simplicity for design, while the BCW accepts complexity to cover all possible influences. No single framework has won, and perhaps none will. Instead, researchers increasingly match frameworks to the question at hand—using SCT to understand self-efficacy in a specific behavior, SDT to design autonomy-supportive interventions, the SEM to plan multi-level policies, the BCW to systematically select intervention functions, and complexity science to model long-term population dynamics.