Chapter 2

Cognitive Determinants of Health Behavior

Mark Conner

1 Introduction

The prevalence of health behaviors varies acrosssocial groups. For example, in the WesternWorld smoking is generally more prevalentamong those from economically disadvantagedbackgrounds. This might suggest such socio-demographic factors as the focus of interven-tions to change health behaviors. However, suchfactors are frequently impossible to change orrequire political intervention at national or inter-national levels (e.g., change in income distri-bution). This is one reason why a consider-able body of research has focused on moremodifiable factors assumed to mediate the rela-tionship between socio-demographic factors andhealth-related behaviors. One important set ofsuch factors is the thoughts and feelings theindividual associates with the particular health-related behavior. These are often referred to ashealth cognitions and are the focus of this chap-ter. Although research does examine the roleof individual health cognitions (e.g., outcomeexpectancies), most of the research in this areauses models that include sets of health cognitionsthat are assumed to combine in different waysto determine behavior. These models are collec-tively known as social cognition models (SCMs;Conner and Norman, 2005). They prominently

M. Conner (�)Institute of Psychological Sciences, University of Leeds,Leeds LS2 9JT, UKe-mail: m.t.conner@leeds.ac.uk

include the Health Belief Model (HBM; e.g.,Abraham and Sheeran, 2005; Janz and Becker,1984), Protection Motivation Theory (PMT;e.g., Maddux and Rogers, 1983; Norman et al,2005), Theory of Reasoned Action/Theory ofPlanned Behavior (TRA/TPB; e.g., Ajzen, 1991;Conner and Sparks, 2005), and Social CognitiveTheory (SCT; e.g., Bandura, 2000; Luszczynskaand Schwarzer, 2005). Stage models repre-sent a different form of SCM which doesnot assume behavior change to be linear, butrather to occur in discrete stages. Prochaska andDiClemente’s (1984) Transtheoretical Model ofChange (TTM) is the most commonly appliedstage model. Below these SCMs are describedand research using each is reviewed. There isconsiderable overlap between the models and thekey health cognitions they identify. Building onthis overlap some work has attempted to inte-grate SCMs into a unified theory of the determi-nants of health behaviors (Fishbein et al, 2001).This integrated model will also be described.Finally, this chapter overviews recent work inthis area on intention stability as an importantmediator of cognitive effects, affective expectan-cies as a highly predictive yet insufficiently con-sidered variable, and implementation intentionsas an important volitional technique to promoteaction.

2 Social Cognition Models

Social cognition models (SCMs) detail theimportant cognitions that distinguish between

A. Steptoe (ed.), Handbook of Behavioral Medicine, DOI 10.1007/978-0-387-09488-5_2, 19© Springer Science+Business Media, LLC 2010

20 M. Conner

those performing and not performing behaviors.The focus is on the cognitions or thought pro-cesses that intervene between observable stim-uli and behavior in real-world situations (Fiskeand Taylor, 1991). This approach is founded onthe assumption that behavior is best understoodas a function of people’s perceptions of real-ity, rather than objective characterizations of thestimulus environment. SCMs can be seen as onepart of self-regulation research. Self-regulationprocesses are defined as those “... mental andbehavioral processes by which people enact theirself-conceptions, revise their behavior, or alterthe environment so as to bring about outcomes init in line with their self-perceptions and personalgoals” (Fiske and Taylor, 1991, p. 181). Self-regulation research has emerged from a clinicaltradition in psychology which views the individ-ual as striving to eliminate dysfunctional patternsof thinking or behavior and engage in adap-tive patterns of thinking or behavior (Bandura,1982; Turk and Salovey, 1986). Self-regulationinvolves cognitive re-evaluation of beliefs, goalsetting, and ongoing monitoring and evaluatingof goal-directed behavior. Two phases of self-regulation activities have been defined: motiva-tional and volitional (Gollwitzer, 1990). In themotivational phase costs and benefits are con-sidered in order to choose between goals andbehaviors. This phase is assumed to concludewith a decision (or intention) concerning whichgoals and actions to pursue at a particular time.In the subsequent volitional phase, planning andaction directed toward achieving the set goal pre-dominate. The majority of SCMs focus on themotivational phase, although work with imple-mentation intentions focuses on the volitionalphase of action.

2.1 The Health Belief Model

The Health Belief Model (HBM) is the oldestand most widely used SCM (see Abraham andSheeran, 2005, for a recent review). In one ofthe earliest studies, Hochbaum (1958) reportedthat perceived susceptibility to tuberculosis and

the belief that people with the disease could beasymptomatic (so that screening would be bene-ficial) distinguished between attendees and non-attendees for chest x-rays. Haefner and Kirscht(1970) extended this research by demonstrat-ing that an intervention designed to increaseparticipants’ perceived susceptibility, perceivedseverity, and anticipated benefits resulted in agreater number of checkup visits to the doctorover an 8-month period compared to a controlcondition.

The HBM posits that health behavior is deter-mined by two cognitions: perceptions of illnessthreat and evaluation of behaviors to counteractthis threat (see Fig. 2.1). Threat perceptions arebased on two beliefs: the perceived susceptibil-ity of the individual to the illness (“How likelyam I to get ill?”) and the perceived severity ofthe consequences of the illness for the individual(“How serious would the illness be?”). Similarly,evaluation of possible responses involves con-sideration of both the potential benefits of andbarriers/costs to action. Together these fourbeliefs are believed to determine the likelihoodof the individual performing a health behavior.The specific action taken is determined by theevaluation of the available alternatives, focusingon the benefits or efficacy of the health behaviorand the perceived costs or barriers of performingthe behavior. Hence individuals are most likelyto follow a particular health action if they believethemselves to be susceptible to a particular con-dition which they also consider to be serious andbelieve that the benefits outweigh the costs of theaction taken to counteract the health threat.

Fig. 2.1 The Health Belief Model

2 Cognitive Determinants of Health Behavior 21

Two further cognitions usually included inthe model are cues to action and health moti-vation. Cues to action are assumed to include adiverse range of triggers to the individual tak-ing action which may be internal (e.g., physicalsymptom) or external (e.g., mass media cam-paign, advice from others) to the individual (Janzand Becker, 1984). An individual’s perception ofthe presence of cues to action would be expectedto prompt adoption of the health behavior ifhe/she already holds other key beliefs favoringaction. Health motivation refers to more stabledifferences between individuals in the value theyattach to their health and their propensity to bemotivated to look after their health. Individualswith a high motivation to look after their healthshould be more likely to adopt relevant healthbehaviors.

The HBM has provided a useful frameworkfor investigating health behaviors and has beenwidely used. It has been found to success-fully predict a range of behaviors. For exam-ple, Janz and Becker (1984) found that across18 prospective studies, the 4 core beliefs werenearly always significant predictors of healthbehavior (82, 65, 81, and 100% of studiesreport significant effects for susceptibility, sever-ity, benefits, and barriers, respectively). Harrisonet al (1992), in a review with more stringentinclusion criteria, reported that susceptibilityand barriers were the strongest predictors ofbehavior. Some studies have found that thesehealth beliefs mediate the effects of demographic

correlates of health behavior. For example,Orbell et al (1995) reported perceived suscepti-bility and barriers to entirely mediate the effectsof social class upon uptake of cervical screening.The HBM has also inspired a range of success-ful behavior change interventions (e.g., Joneset al, 1987).

The main strength of the HBM is thecommon-sense operationalization it uses includ-ing key beliefs related to decisions abouthealth behaviors. However, further research hasidentified other cognitions that are stronger pre-dictors of health behavior than those identi-fied by the HBM, suggesting that the modelis incomplete. This prompted a proposal toadd self-efficacy and intention to the modelto produce an “extended health belief model”(Rosenstock et al, 1988) which has generallyimproved the predictive power of the model(e.g., Hay et al, 2003).

2.2 Protection Motivation Theory

Protection Motivation Theory (PMT; Maddoxand Rodgers, 1983; see Norman et al, 2005 fora review) is a revision and extension of the HBMwhich incorporates various appraisal processesidentified by research into coping with stress.In PMT, the primary determinant of perform-ing a health behavior is protection motivationor intention to perform a health behavior (seeFig. 2.2). Protection motivation is determined

Fig. 2.2 Protection Motivation Theory

22 M. Conner

by two appraisal processes: threat appraisal andcoping appraisal. Threat appraisal is based on aconsideration of perceptions of susceptibility tothe illness and severity of the health threat in avery similar way to the HBM. Coping appraisalinvolves the process of assessing the behavioralalternatives which might diminish the threat.This coping process is itself assumed to be basedon two components: the individual’s expectancythat carrying out a behavior can remove thethreat (action-outcome efficacy) and a belief inone’s capability to successfully execute the rec-ommended courses of action (self-efficacy).

Together these two appraisal processes resultin either adaptive or maladaptive responses.Adaptive responses are those in which the indi-vidual engages in behaviors likely to reduce therisk (e.g., adopting a health behavior) whereasmaladaptive responses are those that do notdirectly tackle the threat (e.g., denial of thehealth threat). Adaptive responses are held tobe more likely if the individual perceives him-self or herself to be facing a health threat towhich he/she is susceptible and which is per-ceived to be severe and where the individual per-ceives such responses to be effective in reducingthe threat and believes that he/she can success-fully perform the adaptive response. The PMThas been successfully applied to the predictionof a number of health behaviors (for a recentreview see Norman et al, 2005). Meta-analyticreviews of PMT (Floyd et al, 2000; Milne et al,2000) indicate protection motivation (i.e., inten-tions) and self-efficacy to be the most powerful

predictors of behavior, while self-efficacy andresponse costs were most strongly associatedwith intentions.

2.3 Theory of Planned Behavior

The Theory of Planned Behavior (TPB; Ajzen,1991) was developed by social psychologists andhas been widely applied to understanding healthbehaviors (see Conner and Sparks, 2005, for areview). It specifies the factors that determinethat individual’s decision to perform a partic-ular behavior (see Fig. 2.3). Importantly thistheory added “perceived behavioral control” tothe earlier Theory of Reasoned Action (TRA;Ajzen and Fishbein, 1980). The TPB proposesthat the key determinants of behavior are inten-tion to engage in that behavior and perceivedbehavioral control over that behavior. As in thePMT, intentions in the TPB represent a person’smotivation or conscious plan or decision to exerteffort to perform the behavior. Perceived behav-ioral control (PBC) is a person’s expectancy thatperformance of the behavior is within his/hercontrol and confidence that he/she can performthe behavior and is similar to Bandura’s (1982)concept of self-efficacy.

In the TPB, intention is assumed to be deter-mined by three factors: attitudes, subjectivenorms, and PBC. Attitudes are the overall evalu-ations of the behavior by the individual as posi-tive or negative. Subjective norms are a person’s

Fig. 2.3 Theory of Planned Behavior

2 Cognitive Determinants of Health Behavior 23

beliefs about whether significant others thinkhe/she should engage in the behavior. PBC isassumed to influence both intentions and behav-ior because we rarely intend to do things weknow we cannot and because believing that wecan succeed enhances effort and persistence andso makes successful performance more likely.

Attitudes are based on behavioral beliefs (oroutcome expectancies), that is, beliefs about theperceived outcomes of a behavior. In particu-lar, they are a function of the likelihood of theoutcome occurring as a result of performing thebehavior (e.g., “How likely is this outcome?”)and the evaluation of that outcome (e.g., “Howgood or bad will this outcome be for me?”).It is assumed that an individual will have alimited number of consequences in mind whenconsidering a behavior. This expectancy-valueframework is based on Fishbein’s (1967) earliersummative model of attitudes. Subjective normis based on beliefs about salient others’ approvalor disapproval of whether one should engage ina behavior (e.g., “Would my best friend want meto do this?”) weighted by the motivation to com-ply with each salient other on this issue (e.g.,“Do I want to do what my best friend wants meto do?”). Again it is assumed that an individ-ual will only have a limited number of referentsin mind when considering a behavior. PBC isbased on control beliefs concerning whether onehas access to the necessary resources and oppor-tunities to perform the behavior successfully(e.g., “How often does this facilitator/inhibitoroccur?”), weighted by the perceived power, orimportance, of each factor to facilitate or inhibitthe action (e.g., “How much does this facilita-tor/inhibitor make it easier or more difficult toperform this behavior?”). These factors includeboth internal control factors (information, per-sonal deficiencies, skills, abilities, emotions) andexternal control factors (opportunities, depen-dence on others, barriers). As for the other typesof beliefs it is assumed that an individual willonly consider a limited number of control factorswhen considering a behavior.

The TPB has been widely tested and suc-cessfully applied to the understanding of a

variety of behaviors (for reviews see Ajzen,1991; Conner and Sparks, 2005). For example,in a meta-analysis of the TPB, Armitage andConner (2001) reported that across 154 appli-cations, attitude, subjective norms, and PBCaccounted for 39% of the variance in intention,while intentions and PBC accounted for 27% ofthe variance in behavior across 63 applications.Intentions emerged as the strongest predictorsof behavior, while attitudes were the strongestpredictors of intentions.

The TPB has also informed a number of inter-ventions designed to change behavior. For exam-ple, Hill et al (2007) employed a randomizedcontrol trial to test the effectiveness of a TPB-based leaflet compared to a control conditionin promoting physical exercise in a sample ofschool children. The leaflet condition comparedto the control condition significantly increasednot only reported exercise but also intentions,attitudes, subjective norms, and PBC. Additionalanalyses indicated that the impact on exercisewas mediated by the increases the leaflet hadproduced (compared to the control group) inintentions and PBC.

2.4 Social Cognitive Theory

In Social Cognitive Theory (SCT; Bandura,1982) behavior is held to be determined by threefactors: goals, outcome expectancies, and self-efficacy (see Fig. 2.4). Goals are plans to actand can be conceived of as intentions to performthe behavior (see Luszczynska and Schwarzer,2005). Outcome expectancies are similar tobehavioral beliefs in the TPB but here are splitinto physical, social, and self-evaluative depend-ing on the nature of the outcomes considered.Self-efficacy is the belief that a behavior is oris not within an individual’s control and is usu-ally assessed as the degree of confidence theindividual has that he/she could still performthe behavior in the face of various obstacles(and is similar to PBC in the TPB). Bandura(2000) recently added socio-structural factors to

24 M. Conner

Fig. 2.4 Social Cognitive Theory

his theory. These are factors assumed to facil-itate or inhibit the performance of a behaviorand affect behavior via changing goals. Socio-structural factors refer to the impediments oropportunities associated with particular livingconditions, health systems, political, economic,or environmental systems. They are assumed toinform goal setting and be influenced by self-efficacy. The latter relationship arises becauseself-efficacy influences the degree to which indi-viduals pay attention to opportunities or imped-iments in their life circumstances. This compo-nent of the model incorporates perceptions of theenvironment as an important influence on healthbehaviors.

SCT has been successfully applied to pre-dicting and changing various health behaviors.However, unlike a number of the other mod-els considered above, many of the applica-tions of SCT only assess one or two com-ponents of the model (usually self-efficacy)rather than all components. Self-efficacy andaction-outcome expectancies along with inten-tions have been found to be the most impor-tant predictors of a range of health behaviorsin a diverse range of studies (for reviews seeBandura, 2000; Luszczynska and Schwarzer,2005).

2.5 Stage Models of Health Behavior

The SCMs considered above assume that thecognitive determinants of health behaviors actin a similar way during initiation (e.g., quittingsmoking for the first time) and maintenance ofaction (e.g., trying to stay quit). In contrast, instage models psychological determinants maychange across such stages of behavior change(see Sutton, 2005, for a review). An importantimplication of the stages view is that differentcognitions may be important determinants at dif-ferent stages in promoting health behavior. Themost widely used stage model is Prochaska andDiClemente’s (1984) Transtheoretical Model ofchange (TTM). Their model has been widelyapplied to analyze the process of change inalcoholism treatment and smoking cessation.DiClemente et al (1991) identify five stages ofchange: pre-contemplation (not thinking aboutchange), contemplation (aware of the need tochange), preparation (intending to change in thenear future and taking action in preparation forchange), action (acting to change), and main-tenance (of the new behavior). Individuals areseen to progress sequentially from one stageto the next, with maintenance the end stage ofsuccessful change. For example, in the case of

2 Cognitive Determinants of Health Behavior 25

smoking cessation, it is argued that in the pre-contemplation stage the smoker is unaware thathis/her behavior constitutes a problem and hasno intention to quit. In the contemplation stage,the smoker starts to think about changing his/herbehavior, but is not committed to try to quit. Inthe preparation stage, the smoker has an inten-tion to quit and starts to make plans to quit. Theaction stage is characterized by active attempts toquit, and after 6 months of successful abstinencethe individual moves into the maintenance stage.This stage is characterized by attempts to preventrelapse and to consolidate the newly acquirednon-smoking status.

Although widely applied, the evidence in sup-port of stage models and different stages ismodest (see Sutton, 2000, 2005). Sutton (2000)concludes that the distinctions between TTMstages are “logically flawed” and based on “arbi-trary time periods.” The sequential movementthrough stages has not generally been supported(Sutton, 2005). In addition, it has proved diffi-cult to support the key prediction that there aredifferent determinants of behavior change in dif-ferent stages. Evidence from stage-matched ver-sus stage-mismatched intervention studies doesnot generally provide support for the TTM (seeLittell and Girvin, 2002, for a systematic reviewof the effectiveness of interventions applyingthe TTM to health-related behaviors). Thus, atpresent, research findings do not support theadded complexity and increased cost of stage-tailored interventions compared to the linearapproach advocated in other SCMs. West (2005)in reviewing stage models in relation to smok-ing has recently suggested that work on the TTMshould be abandoned.

3 Integration of Social CognitionModels

The overlap between SCMs has promptedattempts to integrate them. This may bevaluable, especially since they include someof the same cognitive determinants. Forexample, intention, self-efficacy, and outcome

expectancies appear in several models. Oneimportant attempt to integrate these modelswas that by Bandura (SCT), Becker (HBM),Fishbein (TRA), Kaufen (self-regulation), andTriandis (Theory of Interpersonal Behavior) aspart of a workshop organized by the US NationalInstitute of Mental Health in response to theneed to promote HIV-preventive behaviors. Theworkshop sought to “identify a finite set ofvariables to be considered in any behavioralanalysis” (Fishbein et al, 2001, p. 3). Theyidentified eight variables which, they argued,should account for most of the variance in any(deliberative) behavior. These were organizedinto two groups. First, those variables whichwere viewed as necessary and sufficient deter-minants of behavior. Thus, for behavior to occuran individual must (i) have a strong intention,(ii) have the necessary skills to perform thebehavior, and (iii) experience an absence ofenvironmental constraints that could preventbehavior. Second were those variables that wereseen primarily to influence intention (although adirect effect on behavior was noted as possible).Thus, a strong intention is likely to occur whenan individual (i) perceives the advantages (orbenefits) of performing the behavior to outweighthe perceived disadvantages (or costs, i.e.,outcome expectancies), (ii) perceives the social(normative) pressure to perform the behavior tobe greater than that not to perform the behavior,(iii) believes that the behavior is consistentwith his/her self-image, (iv) anticipates theemotional reaction to performing the behaviorto be more positive than negative, and (v) hashigh levels of self-efficacy. Figure 2.5 illustratesthis integrated model. This approach has beenfurther developed by Fishbein (2008) in hisintegrative model (IM) of behavioral predictionalthough this has not, as yet, been widely tested.

4 Current Directions

A clear contribution of work with SCMshas been their ability to identify key corre-lates of health behavior that can be targeted

26 M. Conner

Self-discrepancy

Advantages/Disadvantages

Environmental Constraints

Social Pressure Intention Behavior

Skills

Emotional Reaction

Self-efficacy

Fig. 2.5 The “major theorists” integrated social cognition model

in interventions to change behavior. Acrossstudies the strongest relationships with behav-ior emerge for intentions, self-efficacy, andoutcomes expectancies (Conner and Norman,2005). However, in focusing on correlates ofhealth behavior rather than examining causalrelationships research may have overestimatedthe size of relationships. For example, while cor-relational research indicates intentions to havea strong effect size on behavior (Armitage andConner, 2001), studies looking at manipulationsof intentions indicate that a medium to largechange in intentions is associated with only asmall to medium effect sized change in behavior(Webb and Sheeran, 2006). A further importantlimitation with much work on SCMs is that whilethey usefully identify cognition change targets,they commonly do not specify the best meansto change such cognitions (work on self-efficacyis an exception to this trend; Bandura, 2000).Recent work on classifying behavior change

interventions (e.g., Abraham and Michie, 2008)and the more widespread assessment of mediat-ing cognitions in intervention studies may pro-vide the basis for further insights into how bestto change cognitions and assessing their causalimpact on behavior change for health behaviors.In the remainder of this section three directionsof current research on cognitive determinants ofhealth behavior are briefly reviewed.

4.1 Intention Stability

In the vast majority of quality applications ofSCMs to predicting health behavior, a prospec-tive design is employed where the predictorsof behavior are measured by questionnaire (attime 1) and then behavior is measured at a sec-ond time point (in stronger designs behaviorchange is the focus of interest). An important

2 Cognitive Determinants of Health Behavior 27

assumption of such a design is that the measuredcognitions (e.g., attitudes) remain unchangedbetween their measurement and the opportu-nity to act. So, for example, the assumptionis that intentions do not change in betweenwhen the (time 1) questionnaire is completedand the time points at which the respondenthas the opportunity to act. This is an explicitlimiting condition of the TRA/TPB (Ajzen andFishbein, 1980). However, cognitions includ-ing intentions may indeed change in this timeperiod and such change provides one impor-tant limitation on their power to predict behav-ior. Several studies have now demonstratedthe power of intention stability to moderatethe intention-behavior relationship (see Connerand Godin, 2007, for a review). For example,Conner et al (2002) found that intentions werestrong predictors of healthy eating up to 6 yearslater, but only among those whose intentionshad remained stable over an initial period of6 months.

A number of factors have been found toinfluence the intention-behavior relationship.For example, anticipating feeling regret if onedoes not perform a behavior or perceivinga strong moral norm to act have both beenfound to significantly increase the intention-behavior relationship (see Cooke and Sheeran,2004, for a review). Sheeran and Abraham(2003) showed intention stability to moderatethe intention-behavior relationship for exercis-ing and that intention stability mediated theeffect of other moderators of the intention-behavior relationship (e.g., anticipated regret,certainty). This suggests that the mechanism bywhich a number of these other moderators mayhave their effect on intention-behavior relation-ships is through changing the temporal stabil-ity of intentions. Hence, factors that might beexpected to make individual intentions more sta-ble over time would be expected to increase theimpact that these intentions have on behaviorand so increase the intention-behavior relation-ship. Thus intention stability might be a usefulfocus of attention as a key mediating variable inintervention studies attempting to change healthbehavior.

4.2 Affective Influences

One criticism of work with SCMs has beenthe failure to explicitly consider affective influ-ences on behavior (Conner and Armitage, 1998).Outcome expectancies included in PMT, TPB,and SCT do not preclude consideration of affec-tive outcomes, although the outcomes typicallyconsidered do not focus on affective states.Over the last few years a number of studieshave examined the impact of expectations ofaffect associated with performance of a behavior.For example, studies have examined anticipatedregret as a determinant of behavior within thecontext of the TPB (see Sandberg and Conner,2008, for a review). Regret is a negative affec-tive state that can be anticipated pre-behaviorallyand so influence subsequent behavior. Studiesgenerally report that such anticipated affectivestates add significant variance to predictions ofintentions but not behavior and may be par-ticularly important in relation to certain affec-tive behaviors (e.g., condom use; Glasman andAlbarracin, 2006). Other studies have shownaffective outcomes to be better predictors ofbehavior than more instrumental outcomes (e.g.,Lawton et al, 2007). Work has also examinedthe affect that accompanies performance of thebehavior (sometimes referred to as anticipatoryaffect or affective attitudes; Loewenstein, 1996)rather than following performance of the behav-ior. Such affective attitudes have been explicitlyadded to the TPB (Conner and Sparks, 2005)and been reported to be stronger predictors ofintentions and behavior than instrumental atti-tudes (Ajzen, 2001; Lawton et al, 2009). Inaddition, some studies indicate affective atti-tudes to directly predict behavior independent ofintentions (e.g., Lawton et al, 2009). Affectiveexpectations and their influence on health behav-ior would appear to be an important and growingfocus for research in this area.

4.3 Implementation Intentions

The majority of research reviewed thus far hasfocused on motivational influences of cognitive

28 M. Conner

variables on behavior (i.e., impacting on inten-tion formation). However, other research hasbegun to focus on the volitional phase of action(Bagozzi, 1993). One volitional variable that hasbeen widely tested in relation to health behavioris implementation intentions. Gollwitzer (1993)makes the distinction between goal intentionsand implementation intentions. While the formeris concerned with intentions to perform a behav-ior or achieve a goal (i.e., “I intend to do x”), thelatter is concerned with if-then plans which spec-ify an environmental prompt or context that willdetermine when the action should be taken (i.e.,“I intend to initiate the goal-directed behavior xwhen situation y is encountered”). Importantly,the if–then plan in an implementation intentioncommits the individual to a specific course ofaction when certain environmental conditionsare met. Sheeran et al (2005) note that “toform an implementation intention, the personmust first identify a response that will lead togoal attainment and, second, anticipate a suitableoccasion to initiate that response. For exam-ple, the person might specify the behavior ‘gojogging for 20 minutes’ and specify a suitableopportunity ‘tomorrow morning before work’”(p. 280). Gollwitzer (1993) argues that by form-ing implementation intentions individuals passcontrol of intention enactment to the environ-ment. The specified environmental cue promptsthe action so that the person does not have toremember the goal intention or decide whento act.

Sheeran et al (2005) provide an in-depthreview of both basic and applied research withimplementation intentions. For example, Milneet al (2002) found that an intervention usingpersuasive text based on PMT prompted posi-tive pro-exercise cognition change but did notincrease exercise. However, when this interven-tion was combined with encouragement to formimplementation intentions, significant behav-ior change was observed (see Gollwitzer andSheeran, 2006, for a meta-analysis of suchstudies). Thus implementation intention for-mation moderates the intention-behavior rela-tionship demonstrating that two people withequally strong goal intentions may differ in their

volitional readiness depending on whether theyhave taken the additional step of forming animplementation intention. Implementation inten-tion formation has been shown to increase theperformance of a range of health behaviorswith, on average, a medium effect size.Implementation intentions appear to be particu-larly effective for those with strong goal inten-tions and in overcoming forgetting that appearsto be a common problem in enacting inten-tions. Provided effective cues are identified inthe implementation intention (i.e., ones that willbe commonly encountered and are sufficientlydistinctive) forgetting appears to be much lesslikely.

5 Conclusions

A number of social cognition models have beendeveloped to describe the key cognitive determi-nants and their relationship to behavior. Thesekey cognitions include intentions, self-efficacy,and outcome expectancies. Recent research hassought to integrate such models (Fishbein et al,2001). Current research has focused on inten-tion stability as an important mediating vari-able explaining the impact of health cognitionson behavior. Other work is examining affectiveinfluences on health behaviors and how the for-mation of implementation intentions promotesthe performance of behavior.

References

Abraham, C. and Michie, S. (2008). A taxonomy ofbehavior change techniques used in interventions.Health Psychol, 27, 379–387.

Abraham, C., and Sheeran, P. (2005). The health beliefmodel. In M. Conner & P. Norman (Eds.), PredictingHealth Behaviour: Research and Practice with SocialCognition Models, 2nd Ed (pp. 28–80). Maidenhead:Open University Press.

Ajzen, I. (1991). The theory of planned behavior. OrganizBehav Hum Dec Proc, 50, 179–211.

Ajzen, I. (2001). Nature and operation of attitudes. AnnRev Psychol, 52, 27–58.

2 Cognitive Determinants of Health Behavior 29

Ajzen, I., and Fishbein, M. (1980). UnderstandingAttitudes and Predicting Social Behavior. EnglewoodCliff, NJ: Prentice Hall.

Armitage, C. J., and Conner, M. (2001). Efficacy of thetheory of planned behaviour: a meta-analytic review.Br J Soc Psychol, 40, 471–499.

Bagozzi, R. P. (1993). On the neglect of volition inconsumer research: a critique and proposal. PsycholMarketing, 10, 215–237.

Bandura, A. (1982). Self-efficacy mechanism in humanagency. Am Psychol, 37, 122–147.

Bandura, A. (2000). Health promotion from the per-spective of social cognitive theory. In P. Norman,C. Abraham, & M. Conner (Eds.), Understandingand Changing Health Behaviour: From Healthbeliefs to Self-Regulation (pp. 229–242). Switzerland:Harwood Academic.

Conner, M., and Armitage, C. J. (1998). Extendingthe theory of planned behavior: a review andavenues for further research. J Appl Soc Psychol, 28,1430–1464.

Conner, M., and Godin, G. (2007). Temporal stabilityof behavioural intention as a moderator of intention-health behaviour relationships. Psychol Health, 22,875–896.

Conner, M., and Norman, P. (Eds.) (2005). PredictingHealth Behaviour: Research and Practice withSocial Cognition Models, 2nd Ed. Maidenhead: OpenUniversity Press.

Conner, M., Norman, P., and Bell, R. (2002). The the-ory of planned behavior and healthy eating. HealthPsychol, 21, 194–201.

Conner, M., and Sparks, P. (2005). The theory of plannedbehaviour and health behaviours. In M. Conner& P. Norman (Eds.), Predicting Health Behaviour:Research and Practice with Social Cognition Models,2nd Ed (pp. 170–222). Maidenhead: Open UniversityPress.

Cooke, R., and Sheeran, P. (2004). Moderation ofcognition-intention and cognition-behaviour rela-tions: a meta-analysis of properties of variables fromthe theory of planned behaviour. Br J Soc Psychol, 43,159–186.

DiClemente, C. C., Prochaska, J. O., Fairhurst, S. K.,Velicer, W. F., Velasquez, M. M., and Rossi, J. S.(1991). The process of smoking cessation: an anal-ysis of precontemplation, contemplation, and prepa-ration stages of change. J Consult Clin Psychol, 59,295–304.

Fishbein, M. (1967). Attitude and the prediction ofbehavior. In M. Fishbein (Ed.), Readings in AttitudeTheory and Measurement (pp. 477–492). New York:Wiley.

Fishbein, M. (2008). A reasoned action approach tohealth promotion. Med Dec Making, 28, 834–844.

Fishbein, M., Triandis, H. C., Kanfer, F. H., BeckerM., Middlestadt, S. E., and Eichler, A. (2001).Factors influencing behavior and behavior change. InA. Baum, T. A. Revenson, & J. E. Singer (Eds.),

Handbook of Health Psychology (pp. 3–17). Mahwah,NJ: Lawrence Erlbaum Associates.

Fiske, S. T., and Taylor, S. E. (1991). Social Cognition,2nd Ed. New York: McGraw-Hill.

Floyd, D. L., Prentice-Dunn, S., and Rogers, R. W.(2000). A meta-analysis of protection motivation the-ory. J Appl Soc Psychol, 30, 407–429.

Glasman, L. R., and Albarracin, D. (2006). Forming atti-tudes that predict future behavior: a meta-analysisof the attitude-behavior relation. Psychol Bull, 132,778–822.

Gollwitzer, P. M. (1990). Action phases and mind-sets. InE. T. Higgins & R. M. Sorrentino (Eds.), Handbookof Motivation and Cognition: Foundations of SocialBehavior, Vol. 2 (pp. 53–92). New York: GuilfordPress.

Gollwitzer, P. M. (1993). Goal achievement: the role ofintentions. Eur Rev Soc Psychol, 4, 142–185.

Gollwitzer, P., and Sheeran, P. (2006). Implementationintentions and goal achievement: a meta analysisof effects and processes. Adv Exp Soc Psychol, 38,69–119.

Haefner, D. P. and Kirscht, J. P. (1970). Motivationaland behavioural effects of modifying health beliefs.Public Health Rep, 85, 478–484.

Harrison, J. A., Mullen, P. D., and Green, L. W. (1992).A meta-analysis of studies of the health belief modelwith adults. Health Educ Res, 7, 107–116.

Hay, J. L., Ford, J. S., Klein, D., Primavera, L. H.,Buckley, T. R., Stein, T. R., Shike, M., and Ostroff,J. S. (2003). Adherence to colorectal cancer screen-ing in mammography-adherent older women. J BehavMed, 26, 553–576.

Hill, C., Abraham, C., and Wright, D. (2007). Cantheory-based messages in combination with cognitiveprompts promote exercise in classroom settings? SocSci Med, 65, 1049–1058.

Hochbaum, G. M. (1958). Public Participation inMedical Screening Programs: A Socio-psychologicalStudy. Public Health Service Publication No 572.Washington, DC: United States Government PrintingOffice.

Janz, N. K., and Becker, M. H. (1984). The health beliefmodel: a decade later. Health Educ Q, 11, 1–47.

Jones, P. K., Jones, S. L., and Katz, J. (1987). Improvingcompliance for asthmatic patients visiting the emer-gency department using a health belief model inter-vention. J Asthma, 24, 199–206.

Lawton, R., Conner, M., and McEachan, R. (2009).Desire or reason: predicting health behaviors fromaffective and cognitive attitudes. Health Psychol, 28,56–65.

Lawton, R., Conner, M., and Parker, D. (2007). Beyondcognition: predicting health risk behaviors frominstrumental and affective beliefs. Health Psychol, 26,259–267.

Loewenstein, G. (1996). Out of control: visceral influ-ences on behavior. Organiz Behav Hum Dec Proc, 65,272–292.

30 M. Conner

Littell, J. H., and Girvin, H. (2002). Stages of change. Acritique. Behav Modif, 26, 223–273.

Luszczynska, A., and Schwarzer, R. (2005). Social cog-nitive theory. In M. Conner & P. Norman (Eds.),Predicting Health Behaviour: Research and Practicewith Social Cognition Models, 2nd Ed (pp. 127–169).Maidenhead: Open University Press.

Maddux, J. E., and Rogers, R. W. (1983). Protectionmotivation and self-efficacy: a revised theory of fearappeals and attitude change. J Exp Social Psychol, 19,469–479.

Milne, S., Sheeran, P., and Orbell, S. (2000). Predictionand intervention in health-related behavior: a meta-analytic review of protection motivation theory. JAppl Soc Psychol, 30, 106–143.

Milne, S., Orbell, S., and Sheeran, P. (2002). Combiningmotivational and volitional interventions to promoteexercise participation: protection motivation theoryand implementation intentions. Br J Health Psychol,7, 163–184.

Norman, P., Boer, H., and Seydel, E. R. (2005).Protection motivation theory. In M. Conner &P. Norman (Eds.), Predicting Health Behaviour:Research and Practice with Social Cognition Models,2nd Ed (pp. 81–126). Maidenhead: Open UniversityPress.

Orbell, S., Crombie, I., and Johnston, G. (1995). Socialcognition and social structure in the prediction ofcervical screening uptake. Br J Health Psychol, 1,35–50.

Prochaska, J. O., and DiClemente, C. C. (1984).The Transtheoretical Approach: Crossing TraditionalBoundaries of Therapy. Homewood, IL: Dow JonesIrwin.

Rosenstock, I. M., Strecher, V. J., and Becker, M. H.(1988). Social learning theory and the health beliefmodel. Health Educ Q, 15, 175–183.

Sandberg, T., and Conner, M. (2008). Anticipated regretas an additional predictor in the theory of plannedbehaviour: a meta-analysis. Br J Soc Psychol, 47,589–606.

Sheeran, P., and Abraham, C. (2003). Mediator of moder-ators: temporal stability of intention and the intention-behavior relationship. Pers Soc Psychol Bull, 29,205–215.

Sheeran, P., Milne, S., Webb, T. L., and Gollwitzer,P. M. (2005). Implementation intentions and healthbehaviours. In M. Conner & P. Norman (Eds.),Predicting Health Behaviour: Research and Practicewith Social Cognition Models, 2nd Ed (pp. 276–323).Maidenhead: Open University Press.

Sutton, S. (2000). A critical review of the transtheoreticalmodel applied to smoking cessation. In P. Norman, C.Abraham, & M. Conner (Eds.), Understanding andChanging Health Behaviour: From Health Beliefsto Self-Regulation (pp. 207–225). Reading, England:Harwood Academic Press.

Sutton, S. (2005). Stage models of health behaviour.In M. Conner & P. Norman (Eds.), PredictingHealth Behaviour: Research and Practice withSocial Cognition Models, 2nd Ed (pp. 223–275).Maidenhead: Open University Press.

Turk, D. C., and Salovey, P. (1986). Clinical informa-tion processing: bias inoculation. In R. E. Ingham(Ed.), Information Processing Approaches to ClinicalPsychology (pp. 305–323). New York: AcademicPress.

Webb, T. L., and Sheeran, P. (2006). Does changingbehavioral intentions engender behavior change? Ameta-analysis of the experimental evidence. PsycholBull, 132, 249–268.

West, R. (2005). Time for a change: putting the transthe-oretical (stages of change) model to rest. Addiction,100, 1036–1039.

http://www.springer.com/978-0-387-09487-8