Explain how social learning contributes to health problems such as obesity and smoking, evaluate network, survey and experimental evidence, and apply the theory to health interventions.
Each is in the Psychology glossary, with every other term in the subject.
The tendency of people to form ties with others who are already similar to them, which can make friends look alike without any influence between them.
A form of modelling in which the model is a peer similar to the observer in age or status, used in interventions to change behaviours such as eating, smoking or studying.
The spread of a behaviour, emotion or health outcome from person to person through social ties, as when people adopt the eating or smoking habits of their friends.
The tendency of people to use others’ eating as a guide to what and how much to eat, so that their food choice and intake move towards those of the people around them.
Explain the role of social learning in understanding health problems, with obesity and smoking as examples.
Explain how eating behaviour is modelled by family, peers and the media, using experimental evidence.
Describe and evaluate Christakis and Fowler (2007) on the spread of obesity in a social network.
Discuss research on the social learning of smoking, including media models and peer influence.
Explain how social learning theory is used in interventions to prevent or reduce health problems.
Analyse how biological, cognitive and sociocultural factors interact with social learning in health behaviour.
Evaluate research on social learning and health with reference to causality, measurement, bias, responsibility and change.
Many of the most serious health problems in the world, including obesity, type 2 diabetes and diseases caused by smoking, depend on everyday behaviours: what people eat, how much they move, whether they smoke or drink. These behaviours are learned, and much of that learning is social. Children eat what they see their families eat, adolescents try cigarettes when their friends and film heroes smoke, and adults change their eating habits when the people around them do.
Social learning offers one explanation of why health problems cluster in families, friendship groups and communities, and a basis for interventions that use models to change health behaviour. It draws on the sociocultural approach (behaviour learned from others and from cultural norms) and the cognitive approach (expectations and self-efficacy), and it interacts with biological factors such as appetite and nicotine dependence. Research in this area raises questions of
With a free account, Augment remembers where you stopped and what you have practiced.
Social learning theory, developed by Albert Bandura, holds that people learn new behaviour by observing models and the consequences the models receive, without needing direct reinforcement. Four mediational processes (attention, retention, reproduction and motivation) decide whether an observed behaviour is imitated, and motivation depends on expected consequences, including vicarious reinforcement. Models who are similar, liked or high in status are imitated most.
For health, the theory makes three predictions. First, people will adopt the health behaviours they see modelled by those around them and in the media. Second, the consequences that are visible matter more than those that are not: the immediate pleasure of a snack or the social approval of smoking is easy to observe, while heart disease decades later is not. Third, people will attempt a healthier behaviour only if they believe they can carry it out, so self-efficacy is central to changing health behaviour.
Eating is one of the most social of all behaviours, and food choice and the amount eaten are strongly influenced by what other people do. Psychologists call this the social modelling of eating.
Birch (1980) studied 39 preschool children in the United States. Each child, who preferred one vegetable to another, sat at lunch over several days with peers who preferred the other vegetable and chose it first. The target children increasingly chose the vegetable their peers preferred, and their own preferences shifted towards it. Preferences for food, which seem personal and even biological, were changed by observing peers.
More recent research has found the same effect outside Western countries. Kim, Chen and Cheon (2019) used a design based on Asch’s conformity studies with 89 preschool children in Singapore, who watched videos of other children choosing between healthy and unhealthy foods. After watching peers choose fruit and vegetables, the children changed their own initially unhealthy choices to match the peers on 29% of trials. This form of peer modelling worked even though the peers were on a screen and the children had never met them.
Laboratory experiments with adults show that people also copy how much others eat. Hermans, Larsen, Herman and Engels (2012) gave 85 young women in the Netherlands a meal during which a female confederate ate a small, standard or large amount. Participants ate more when their companion ate more, and they also ate more when offered a larger portion; the two influences were independent of each other and added together. Social norms about how much is appropriate to eat guided their intake.
Cruwys, Bevelander and Hermans (2015) reviewed 69 experiments, involving over 5,800 participants, that manipulated how much a model ate. Sixty-four of them found a significant modelling effect, across different foods, settings and participant groups. Modelling was stronger when people wanted to be liked by the model or saw themselves as similar to the model, which matches the attention process of social learning theory, and it was partly due to unconscious mimicry of the model’s eating.
In a further study, Hermans and colleagues (2012) showed young women a video of a woman eating sweets and found that participants did not adjust their intake to the video model. The researchers suggested that modelling of intake may need clear information about how much others have eaten in the same situation. This is a useful reminder that social learning depends on what the observer can attend to: a model whose behaviour is unclear or seems irrelevant may not be copied.
Christakis and Fowler aimed to find out whether obesity spreads from person to person through social ties, and whether the type of relationship (friend, sibling, spouse, neighbour) affects how strongly one person’s weight gain is associated with another’s. They proposed that if it does, social influence could be one reason for the rapid rise in obesity in the United States over the previous 30 years.
The researchers used data from the Framingham Heart Study, a long-running health study in a town in Massachusetts. They reconstructed a densely connected social network of 12,067 people who were assessed repeatedly between 1971 and 2003. When participants attended a health check-up, they had named friends, relatives and neighbours as contacts, many of whom were also study participants, which allowed the researchers to map who was connected to whom over 32 years.
Body-mass index (BMI) was available for everyone at each check-up, and obesity was defined as a BMI of 30 or more. The study was a longitudinal design using correlational research methods: the researchers used statistical models to test whether a person was more likely to become obese in a given period if one of their contacts became obese during the same period, while controlling for both people’s weight in the previous period and characteristics such as age, sex and education.
Clusters of obese people were present in the network at every time point, and the clusters extended to three degrees of separation (to a friend’s friend’s friend). The researchers reported that these clusters did not seem to be explained only by obese people choosing each other as friends.
A person’s chance of becoming obese increased by 57% if a friend became obese in the same period, by 40% if an adult sibling did and by 37% if a spouse did. No effect was found among neighbours living close by, and people of the same sex had a greater influence on each other than people of the opposite sex. The spread of smoking cessation in the network did not account for the spread of obesity.
Christakis and Fowler concluded that obesity appears to spread through social ties. Because geographical neighbours had no effect but distant friends did, they argued that the mechanism is social rather than a shared local environment such as nearby shops or parks. They suggested that people learn from those close to them what an acceptable body size is and how much it is normal to eat and exercise, which is a social learning explanation based on changing norms. They called this kind of spread social contagion.
YouTubeThe study is correlational, so it cannot show that one person’s weight gain caused another’s. Two alternative explanations are especially important. The first is homophily, the tendency of people to choose friends who are already similar to them, for example in their eating habits or their genetic tendency to gain weight. The second is a shared environment: friends may eat at the same restaurants or have the same kind of jobs, so they gain weight together without influencing each other.
Cohen-Cole and Fletcher (2008) showed why this matters. Using a large survey of US adolescents and the same kind of statistical model, they found apparent “contagion” between friends for acne, headaches and height, conditions that cannot plausibly spread socially. Once they controlled for shared environmental factors, these effects became small and non-significant. In a separate paper the same year, Cohen-Cole and Fletcher re-ran Christakis and Fowler’s analysis with a complementary dataset and reported that the estimated social network effect for obesity also became statistically indistinguishable from zero once standard controls for shared context were added.
BMI is an objective and reliable measure, which is a strength compared with self-reported diet. However, BMI does not distinguish fat from muscle, and the social ties were recorded from contact lists people gave for administrative purposes, not from detailed measures of how close relationships were or how often people ate together. The study measured outcomes (weight) but not the behaviours (eating, exercise) through which social learning would work, so the mechanism is inferred rather than observed.
The Framingham sample came from one town in the north-eastern United States and was mostly white. Norms about body size, food and hospitality differ across cultures and over time, so the strength of social influence on weight found in this sample may not generalise to other populations.
Describing obesity as “contagious” risks increasing stigma against people with obesity, for example if people are encouraged to avoid friends who gain weight. The authors argued for the opposite implication: that interventions could spread healthy behaviour through the same networks. Communicating correlational findings to the public responsibly means being clear about what they do and do not show.
Sources: Christakis, N. A., & Fowler, J. H. (2007). The spread of obesity in a large social network over 32 years. _New England Journal of Medicine, 357(4), 370–379. Cohen-Cole, E., & Fletcher, J. M. (2008). Detecting implausible social network effects in acne, height, and headaches. BMJ, 337, a2533. Cohen-Cole, E., & Fletcher, J. M. (2008). Is obesity contagious? Journal of Health Economics, 27(5), 1382–1387._
Smoking is a useful second example because it is clearly learned: nobody is born a smoker, and most smokers start in adolescence, when peers and media models are especially influential. Once established, smoking becomes an addiction maintained by nicotine dependence, so it shows how a behaviour that begins through social learning can be maintained by biological factors.
Akers and Lee (1996) tested Akers’s version of social learning theory, which builds on Bandura’s ideas, with 454 students in grades 7 to 12 (about ages 12 to 18) in a community in Iowa, in the United States, followed for five years.
The researchers measured whether the students’ friends and family smoked (who they associated with), how smoking was rewarded or punished in their social world, their attitudes to smoking, and whether they had seen admired people smoke (modelling). These social learning variables predicted which adolescents started or continued smoking, and smoking in turn had some effect on the social learning variables, such as the adolescents’ attitudes and the friends they associated with.
The finding that the influence runs in both directions is an example of bidirectional ambiguity: adolescents who start smoking may choose friends who smoke, and friends who smoke make smoking more likely. Social learning and the choice of social environment reinforce each other, as Bandura’s idea of reciprocal determinism predicts.
Films provide adolescents with powerful symbolic models: attractive, high-status characters who smoke without visible harm. Sargent et al. (2005) conducted a telephone survey of a nationally representative sample of 6,522 US adolescents aged 10 to 14. Each was asked which of a random selection of films, drawn from 532 recent box-office hits, they had seen, and the researchers had counted the smoking scenes in every film. This allowed them to estimate each adolescent’s exposure to smoking on screen.
After controlling for other risk factors, including whether friends, siblings and parents smoked, school performance, personality and parenting style, the odds of having tried smoking were still about 2.6 times as high for adolescents in the highest quartile of exposure as for those in the lowest. The researchers estimated that around 38% of smoking initiation in this age group could be attributed to exposure to smoking in films. Because the data were collected at one point in time, however, adolescents who were already curious about smoking may have chosen films that contained more of it.
YouTubeSocial learning also applies to giving up. In a second analysis of the Framingham network, Christakis and Fowler (2008) found that when a spouse stopped smoking, a person’s chance of smoking fell by 67%, and when a friend stopped, by 36%. Whole connected groups of people appeared to quit together, and the smokers who remained were increasingly found at the edges of the network. This suggests that social learning can spread healthy change as well as unhealthy behaviour, although the same causal limitations apply as for obesity.
If health behaviours are learned from models, health promotion can provide new models, change the visible consequences of behaviour and build self-efficacy. Interventions based on social learning range from school programmes to national media campaigns.
The Food Dudes programme, developed at Bangor University in Wales, uses peer modelling and rewards to increase children’s fruit and vegetable consumption. Horne et al. (2004) evaluated it in two inner-city London primary schools with 749 children aged 5 to 11. In the intervention school, children watched video adventures over 16 days featuring heroic peers, the Food Dudes, who enjoy eating fruit and vegetables. Children also received small rewards for eating fruit and vegetables themselves.
Compared with the control school, lunchtime consumption of fruit and vegetables in the intervention school rose substantially and was still higher at a four-month follow-up, with the largest increases among children who had eaten very little at the start. Consumption at home, measured through parents’ recall, also increased. An independent evaluation in 15 primary schools in the West Midlands of England (Upton, Taylor and Upton, 2015) found a smaller effect: consumption increased at three months only among children eating school-provided lunches, and the programme did not reduce their intake of high-fat, high-sugar foods.
A smaller independent study by Taylor et al. (2013), which used photographic food diaries, found no change in the children’s fruit and vegetable eating at home.
The difference between the developers’ evaluation and the independent ones is a reminder of possible researcher bias when programmes are evaluated by those who designed them. It also raises a question of measurement: parents’ recall of what children ate at home suggested a benefit, while photographed meals did not.
Entertainment-education uses dramas in which characters model health behaviours. In Tanzania, the radio soap opera Twende na Wakati (“Let’s go with the times”) was first broadcast in 1993 by seven of the eight mainland stations of Radio Tanzania; the eighth broadcast other programmes until 1995, so its listening area served as a comparison. Rogers et al. (1999) used annual surveys of about 2,750 households in the two areas and data from 79 health clinics. They found that the drama increased the adoption of family planning, raised listeners’ self-efficacy about family planning and encouraged them to talk about contraception with partners and friends.
This design was a field experiment at the level of whole regions, which gives more confidence in a causal effect than a correlational survey, although the regions were not randomly assigned and may have differed in other ways.
Social learning on its own cannot explain why some people who observe the same models develop a health problem and others do not. Health behaviour results from the interaction of biological, cognitive and sociocultural factors, and social learning works within this interaction.
Approach | Factor | How it interacts with social learning |
|---|---|---|
Biological | Genetic differences in appetite and weight gain; nicotine dependence | Two people learn the same eating norms but gain weight at different rates; a smoker who started through peers continues because of dependence |
Cognitive | Self-efficacy; expectations about consequences | Observing a model quit raises self-efficacy only if the observer sees the model as similar; expectations of pleasure or stress relief motivate imitation |
Sociocultural | Norms of body size and hospitality; cost and availability of food and tobacco | Cultures differ in which body sizes are admired and which foods are offered, which changes what is modelled and rewarded |
An explanation that combines these factors is more complete than one based on social learning alone, which would be reductionist. For example, the obesogenic environment, with cheap high-calorie food and little need for physical activity, makes overeating easy to reproduce, while social models make it seem normal. Changing models without changing the environment, or the reverse, is likely to have a limited effect.
The strongest real-world evidence on social learning and health, such as network and survey studies, is correlational, so homophily, shared environments and reverse causation are always possible explanations. Laboratory experiments on eating establish that modelling can change intake in a single meal, and field experiments such as the Food Dudes and Twende na Wakati show that modelling-based interventions can change behaviour, but each has its own limits. Causal conclusions are strongest when experimental mechanisms and real-world patterns point in the same direction.
Health outcomes such as BMI can be measured objectively, but the social learning processes that are supposed to produce them are usually inferred. Food intake measured in the laboratory is precise but artificial; intake measured at home relies on parents’ recall or self-report, which is vulnerable to social desirability bias. Smoking is often self-reported, and adolescents may under-report it to adults.
Much of the research comes from the United States and Western Europe. The Singapore study by Kim and colleagues and the Tanzanian radio drama show that social learning of health behaviour occurs in other cultures, but which models are respected, which foods are offered and what body size is admired vary widely, so interventions need cultural adaptation.
Social learning research has led to practical health interventions that can reach many people at low cost per person. Their effects are often modest and may fade without continued modelling, so barriers to change such as cost, access and existing habits remain important. Researchers and health campaigners also carry a responsibility not to stigmatise people with health problems while trying to change the norms around them.
Concept | Strength of the evidence | Limitation |
|---|---|---|
Causality | Experiments show modelling changes eating; field experiments show interventions change behaviour | Network and survey studies are correlational; homophily and shared environment |
Measurement | BMI and observed intake are objective | Social learning processes are inferred; diet and smoking often self-reported |
Bias | Effects found in Singapore and Tanzania as well as the West | Most research from WEIRD samples; developer-led evaluations |
Change | Peer modelling and entertainment-education increase healthy behaviour | Effects are often modest and may fade |
Responsibility | Interventions can spread healthy norms through networks | Calling obesity “contagious” risks stigma |
In the health and well-being context, social learning and health can appear in Paper 1 Section B as a scenario about a health behaviour, and in Section C as a concept-based essay on the role of social learning in understanding a health problem.
Health behaviours such as eating and smoking are learned partly by observing models in the family, peer group and media.
Visible short-term consequences of unhealthy behaviour (pleasure, belonging) act as vicarious reinforcement, while long-term harm is rarely visible.
Experiments show that people copy what and how much models eat; a review by Cruwys and colleagues found modelling effects in 64 of 69 experiments.
Christakis and Fowler (2007): a person’s chance of becoming obese rose 57% if a friend became obese, 40% for a sibling and 37% for a spouse.
Homophily and shared environments are alternative explanations: Cohen-Cole and Fletcher (2008) used similar methods to detect “contagion” of acne and height.
Sargent et al. (2005) found that exposure to smoking in films was associated with trying smoking: 2% in the lowest and 22% in the highest exposure quartile.
Interventions using peer models and entertainment-education, such as the Food Dudes and Twende na Wakati, can change health behaviour.
Social learning interacts with biological and cognitive factors and with the environment; on its own it is a reductionist explanation.
I can explain the role of social learning in understanding obesity and smoking.
I can explain how eating behaviour is modelled, using experimental evidence.
I can describe and evaluate Christakis and Fowler (2007), including the problem of homophily.
I can discuss research on the social learning of smoking.
I can explain how social learning theory is used in health interventions.
I can analyse how biological, cognitive and sociocultural factors interact with social learning.
I can evaluate research on social learning and health with reference to causality, measurement, bias, responsibility and change.