Accessible vs. inaccessible attitudes

According to Fazio, attitudes often influence behaviour through a spontaneous process. Effects of attitudes can occur quickly, but only for people whose attitude is accessible (easy to retrieve). When attitudes are accessible, they come to mind instantly when we see the attitude object. The attitude then influences how we behave towards the object. If the attitude is less accessible, it doesn’t come to mind, and so it doesn’t influence our behaviour.

For example, suppose you are walking by an ice cream seller. You may spontaneously recall your passion for ice cream, and this attitude may motivate a decision to buy some. But if you don’t spontaneously recall your attitude (because it is inaccessible – perhaps you are distracted by a more pressing thought at the time you walk past the ice cream seller), it will lie dormant and not elicit the decision to buy. Indeed, there’s a great deal of evidence that attitudes do exert a stronger influence on behaviour when they are accessible than when they are difficult to retrieve.


Incentive for change

To understand how attitudes can be changed, it is first important to understand attitude functions – the psychological needs that attitudes fulfil. Early theories proposed a number of important attitude functions. For example, people may have a positive attitude towards objects that help them become popular among people they like, but not objects that make them estranged from those people. This is the social adjustment function, which provides the basis for the entire fashion industry: people tend to like clothing that is popular among people they like.


In the earliest model of attitude change, Hovland, Janis and Kelley suggested that persuasive messages change people’s attitudes when they highlight some incentive for this change. For example, an advertisement might describe the utilitarian benefits of buying a particular model of car (e.g. good fuel economy) or the social-adjustment benefits (e.g. a sporty look). The incentives must seem important if the message recipients are to change their attitude.

Hovland et al.’s theory also suggests that processing of any message must occur in stages if it is to be successful. The intended audience must:

1. pay attention to the message,

2. comprehend the message, and

3. accept the message’s conclusions.


Seminal theories of attitude function


McGuire’s information-processing approach to persuasion McGuire extended this theory further. According to his model, a message will elicit the desired behaviour only if it succeeds at six stages. People must:

1. encounter the message (presentation stage);

2. attend to it (attention stage);

3. understand it (comprehension stage);

4. change their attitude (yielding stage);

5. remember their new attitude at a later time (retention stage);and

6. the new attitude must influence their behaviour (behavior stage).

Interestingly, even if the odds of passing each stage are quite good, the chances of completing all the stages can be low. For example, we might optimistically assume that a Nike running shoe ad has an 80 per cent chance of success at each stage. If this were the case, the laws of probability indicate that the odds of successfully completing all of the stages would be only 0.26 (0.8 ×0.8 × 0.8 × 0.8 × 0.8 × 0.8). In other words, the ad would have a 26 per cent chance of getting someone to buy the running shoes.

In reality, the odds of completion of each stage (especially yielding and behaviour) may be far lower, creating even lower chances of success (possibly less than 1 per cent). For this reason, modern marketing initiatives take steps to compel completion of each stage, where this is possible. So advertisers will present the message many times, make it attention-grabbing and memorable, and make the message content as powerful as they can.

Motivation and ability

Two newer models of persuasion, the ‘elaboration likelihood model’ and the ‘heuristic–systematic model’predict that the effects of persuasive messages depend on people’s motivation and ability to think carefully about them. If someone is highly motivated and able to process a persuasive message, they should be heavily influenced by the strength of the arguments in the message. But if they are less motivated or able to process the message, then they should be strongly affected by simple cues within the message, such as the presenter’s attractiveness or expertise. Many variables influence motivation and ability. Motivation is high when the message is relevant to personal goals and there is a fear of being wrong. Ability is high when people are not distracted and when they possess high cognitive skills. Although all of these variables have been studied in connection with both models of persuasion, most of this research has focused on the personal relevance of  the message.

For example, Petty et al. found that the  attractiveness of the spokesperson presenting a message influences attitudes when the issue is not personally relevant, but has no effect when the issue is personally relevant. In contrast, the strength of the argument within the message influences attitudes when the issue is personally relevant, but not when the issue is not personally relevant. These findings support the predictions of the elaboration likelihood model and the heuristic–systematic model.

Although many experiments have revealed similar effects, the heuristic–systematic model suggests that high personal relevance should not always lead to the lower use of cues such as the presenter’s attributes. For example, when a personally relevant message contains ambiguous arguments (i.e. it has strengths and weaknesses), people may be more persuaded by a message from an expert source than from an inexpert source.

According to this model, high personal relevance causes people to use environmental cues when the message arguments themselves provide no clear conclusions. This prediction has received some experimental support.

Early theories of attribution

We said earlier that attributions are explanations for events and behaviour. Heider differentiated between two types of causal attribution – personal and situational. Personal attributions refer to factors within the person, such as their personality characteristics, motivation, ability and effort. Situational attributions refer to factors within the environment that are external to the person. For example, if we were discussing why a particular student has failed an important university examination, we would consider personal factors (such as her academic ability and how much effort she invested in preparing for the exam). But we might also look at situational attributions (such as whether she had good tuition, access to library facilities and sufficient time to study).

Heider noted that we tend to overestimate internal or personal factors and underestimate situational factors when explaining behaviour. This tendency has become known as the fundamental attribution error, which we’ll return to in the next section.

In a similar vein, Jones and Davis found that we tend to make a correspondent inference about another person when we are looking for the cause of their behaviour. In other words, we tend to infer that the behaviour, and the intention that produced it, correspond to some underlying stable quality. For example, a correspondent inference would be to attribute someone’s aggressive behaviour to an internal and stable trait within the person – in this case, aggressiveness. Jones and Davis argued that this tendency is motivated by our need to view people’s behaviour as intentional and predictable, reflecting their underlying personality traits. But in reality, making correspondent inferences is not always a straightforward business. The information we need in order to make the inferences can be ambiguous, requiring us to draw on additional cues in the environment, such as the social desirability of the behaviour, how much choice the person had, or role requirements.

Like Heider, Kelley likened ordinary onlookers to naive scientists who weigh up several factors when attributing causality. Kelley’s covariation model of attribution states that, before two events can be accepted as causally linked, they must co-occur.

The covariation of events and behaviour was assessed across three important dimensions:

1. consistency – does the person respond in the same way to the same stimuli over time?

2. distinctiveness – do they behave in the same way to other different stimuli, or is the behaviour distinctively linked to specific stimuli?

3. consensus – do observers of the same stimuli respond in a similar way?

Kelley argued that we systematically analyse people- and environment-related information, and that different combinations of information lead to different causal attributions. For example, while attributing causality for behaviour like ‘John laughed at the comedian’, we would run through the following considerations:

1. If John always laughs at this comedian, then his behavior is highly consistent.

2. If John is easily amused by comedians, then his behavior has low distinctiveness.

3. If practically no one else in the audience laughed at the comedian, then his behaviour has low consensus.

A combination of high consistency, low distinctiveness and low consensus would lead to a dispositional (internal) attribution for John’s laughter, such as ‘John has a peculiar tendency to laugh at all comedians; he must be very easily amused.’ In contrast, a combination of high consistency, high distinctiveness and high consensus would lead to an external attribution, such as ‘John likes this comedian, but he doesn’t like many other comedians, and other people like this comedian too; this comedian must be funny.

The effects of bias

Both the Jones–Davis and the Kelley models of attribution view the social perceiver as a rational person who uses logical principles of thinking when attributing causality. But empirical research has discovered persistent biases in the attributional processes. According to Fiske and Taylor, bias occurs if the social  perceiver systematically distorts (over-uses or under-uses) what are thought to be correct and logical procedures. We will now look in more detail at four of the most pervasive biases: the fundamental attribution error, the actor–observer effect, the self-serving bias and the ultimate attribution error.

The fundamental attribution error Ross defined the fundamental attribution error (FAE) as the tendency to underestimate the role of situational or external factors, and to overestimate the role of dispositional or internal factors, in assessing behaviour.

The earliest demonstration of the FAE was an experiment by Jones and Harris, in which American college students were presented with another student’s written essay that was either for or against the Castro government in Cuba. Half the participants were told that the essay writer had freely chosen whether to write a ‘pro’ or ‘anti’ Castro essay (choice condition), and the other half were told that the essay writer was told which position to take (no-choice condition). After reading the essay, participants were asked what the essay writer’s ‘true’ attitude was towards Castro’s Cuba. The participants tended to view the writer’s attitude as consistent with the views expressed in the essay, regardless of the choice/no-choice condition. While they didn’t totally disregard that the no-choice writers had been told what position  to take, they viewed this as less important than their attitudinal disposition. In other words, they underestimated the impact of the no-choice condition. In another classic study, Ross, Amabile and Steinmetz randomly assigned pairs of participants in  a quiz game to act as contestant and questioner. Questioners were instructed to set ten difficult general knowledge questions of their own choosing. Despite the relative situational advantage of the questioners, both the contestants and observers of the quiz game rated the questioners as significantly more knowledgeable than the contestants. Heider put forward a largely cognitive explanation for the FAE. He suggested that behaviour has such salient properties  that it tends to dominate our perceptions. In other words, what we notice most in (a) behaviour and (b) communication is (c) the person who is central to both. People are dynamic actors – they move, talk and interact, and these features come to dominate  our perceptual field. Supporting this cognitive explanation, Fiske and Taylor argued that situational factors such as social context, roles and situational pressures are ‘relatively  pallid and dull’ in comparison with the charisma of the dynamic actor. While this is a commonsense and intuitive explanation, we discuss later in this chapter how this bias is only pervasive in Western individualistic cultures. So the FAE turns out to be not so fundamental after all!

The actor–observer effect

While we tend to attribute other people’s behaviour to dispositional factors, we tend to attribute our own behaviour to situational factors. This is called the actor–observer effect (AOE).

Consider how easily we explain our own socially undesirable behaviour (such as angry outbursts) to extenuating, stressful circumstances, and yet we are less sympathetic when others behave in this way. Instead, we often conclude that the person is intolerant, impatient, unreasonable, selfish, etc. This bias has been found  in both laboratory experiment and applied clinical settings. For example, psychologists and psychiatrists are more likely to attribute their clients’ problems to internal stable dispositions, whereas the clients are more likely to attribute their own problems to situational factors. There are several competing explanations for the AOE, but we will outline just two of them here.

1 Perceptual salience As for the FAE, one explanation is perceptual and essentially argues that actors and observers quite  literally have ‘different points of view’. As actors, we can’t see ourselves acting. From an actor’s point of view,  what is most salient and available are the situational influences  on behaviour – the objects, the people, the role requirements  and the social setting. But from an observer’s point of view, other people’s behaviour is more dynamic and salient than the situation or context. These different vantage points for actors an observers appear to lead to different attributional tendencies, i.e. situational attributions for actors and dispositional attributions for observers.

Taylor and Fiske attempted to test the perceptual salience hypothesis by placing observers at three different vantage points around two male confederates who sat facing each other engaged in conversation. Observers sat either behind confederate

A with confederate B in their direct visual field, or behind B, watching A, or to the side, between A and B with both in sight.

A schematic figure of a study that attempted to test the perceptual salience hypothesis. Two confederates sat facing each other and were engaged in conversation. They were observed from three different vantage points – from behind Confederate A, from behind Confederate B, and from midway between A  and B. Consistent with the perceptual salience hypothesis, the results showed that observers sitting behind A, watching B, rated B as more causal, while those sitting behind B, watching A, saw A as more causal. The observers watching from midway between A and B perceived both as equally influential.

After A and B had interacted for five minutes, each observer was asked to rate each confederate on various trait dimensions, and the extent to which their behaviour was caused by dispositional and situational factors. They also rated how much each confederate (a) set the tone of the conversation, (b) determined the kind of information exchanged and (c) caused the other’s behaviour.

Consistent with the perceptual salience hypothesis, Taylor and Fiske found that the two observers sitting behind A, watching B, rated B as more causal, while those sitting behind B, watching A, saw A as more causal. The observers sitting in between A and B perceived both confederates as equally influential.

In a similar vein, McArthur and Post manipulated the salience of two people engaged in conversation through the use of lighting. When one participant was made more salient than the other by being illuminated by bright light, observers rated the behaviour of the illuminated person as more dispositionally and less situationally caused.

2 Situational information Another explanation for the AOE focuses on information. Actors have more information about the situational and contextual influences on their behaviour, including its variability and flexibility across time and place. But observers are unlikely to have such detailed information about the actors unless they know them very well, and have observed their behavior over time and in many different situations. It therefore seems that observers assume more consistency in other people’s behavior compared to their own, and so make dispositional attributions for others, while making situational attributions for their ownbehaviour.

The self-serving bias

It is well known that people tend to accept credit for success and deny responsibility for failure. More generally, we also tend to attribute our success to internal factors such as ability, but attribute failure to external factors such as bad luck or task difficulty. This is known as the self-serving bias. How often have we heard governments taking credit when there is national economic growth and prosperity, attributing it to their economic policies and prudent financial management?

And yet, in times of economic hardship, they are quick to blame external causes, such as the international money markets or worldwide recession. Although the strength of the self-serving bias varies across cultures, it has been found to occur cross-culturally. The usual explanation is motivational factors: that is, the need for individuals to enhance their self-esteem when they succeed and protect their self-esteem when they fail. Attributing success to internal causes has been referred to as the self-enhancing bias, and attributing failure to external causes as the self-protection bias. But Miller and Ross argue that there is only clear support for the self-enhancing bias, and that people do often accept personal responsibility for failure. They also claim that the self-enhancing bias can be explained by cognitive factors without recourse to motivational explanations. For example, we are more likely to make self-attributions for expected than unexpected outcomes, and most of us expect to succeed rather than fail. Even so, it is difficult to argue against the motivational hypothesis, and the prevailing consensus is that both motivational and cognitive factors have a part in the self-serving bias.

The motivation for self-enhancement is also linked to achievement attributions. According to Weiner’s attributional theory of motivation and emotion, the attributions people make for success and failure elicit different emotional consequences, and are characterized by three underlying dimensions – locus,  stability and control (see table).

Achievement attributions for success and failure, and their characteristics on the three underlying dimensions of locus, stability and control


The locus dimension refers to whether we attribute successand failure internally or externally. Consistent with the self enhancement bias, we are more likely to feel happier and better about ourselves if we attribute our success internally (to factors such as ability and effort) rather than externally (to good luck or an easy task). In contrast, attributing failure internally is less likely to make us feel good about ourselves than attributing it externally.

The stability dimension refers to whether the cause is perceived as something fixed and stable (like personality or ability) or something changing and unstable (such as motivation or effort).

The controllability dimension refers to whether we feel we have any control over the cause.

The tendency to attribute negative outcomes and failure to internal, stable and uncontrollable causes is strongly associated with clinical depression and has been referred to as a depressive attributional style. The reformulated learned helplessness model of depression views this attributional style as directly causing depression. But others have argued that it is merely a symptom, reflecting the affective state of the depressed individual. Whether it is a cause  or symptom, attributional retraining programmes , in which people are taught to make more self-enhancing attributions, are widely accepted as an important therapeutic process for recovery from depression.

The ultimate attribution error

The self-serving bias also operates at the group level. So we tend to make attributions that protect the group to which we belong. This is perhaps most clearly demonstrated in what Pettigrew called the ultimate attribution error (UAE).

By extending the fundamental attribution error to the group context, Pettigrew demonstrated how the nature of intergroup relations shapes the attributions that group members make for the same behaviour by those who are in-group and out-group members. So prejudicial attitudes and stereotypes of disliked out-groups lead to derogating attributions, whereas the need for positive enhancement and protection of the in-group leads to group-serving attributions.

People are therefore more likely to make internal attributions  for their group’s positive and socially desirable behaviour, and external attributions for the same positive behaviour displayed by out-groups. In contrast, negative or socially undesirable in-group behaviour is usually explained externally, whereas negative outgroup behaviour is more frequently explained internally.

This intergroup bias has been found in a number of contexts. Taylor and Jaggi found it among Hindus in southern India, who gave different attributions for exactly the same behaviour performed by Hindu and Muslim actors. Duncan found that white American college students categorized the same pushing behaviour as ‘violent’ if perpetrated by a black actor but as ‘just playing around’ when perpetrated by a white actor.

The most dramatic illustration of the UAE is an investigation by Hunter, Stringer and Watson of how real instances of violence are explained by Protestants and Catholics in Northern Ireland. Catholic students made predominantly external attributions for their own group’s violence but internal, dispositional attributions for Protestant violence. Similarly, Protestant students attributed their own group’s violence to external causes and Catholic violence to internal causes.

There is also substantial evidence of the tendency to make more favourable attributions for male success and failure. Studies have found that both men and women are more likely to attribute male success to ability and female success to effort and luck, especially in tasks that are perceived to be ‘male’.

The same bias is found for failure attributions – male failure is explained by lack of effort, whereas female failure is attributed to lack of ability. Bear in mind though that most of these studies were conducted in the seventies and eighties, and relatively few have been published more recently. Given the social and attitudinal changes associated with women’s roles over this time, and the fact that the effects were relatively small, it is possible that these biases have now diminished in Western societies.

Cultural differences

There is now strong evidence that people in non-Western cultures do not make the same kinds of attributions as people in Western individualistic societies. The fundamental attribution error, which was originally thought to be a universal cognitive bias, is not found in collectivist cultures. Instead, many non-Western people place less emphasis on internal dispositional explanations, and more emphasis on external and situational explanations.

Miller was among the first social psychologists to suggest that such differences arise from different cultural representations of the person that are learned during social development, rather than from cognitive and perceptual factors. Western notions of the person are predominantly individualistic, emphasizing the central importance and autonomy of the person, whereas non Western notions tend to be holistic, stressing the interdependence between the person and their social relationships, role obligations and situational norms.

Miller conducted a cross-cultural study to compare the attributions made for prosocial and deviant behaviours by a sample of Americans and Indian Hindus of three different age groups (eight, eleven and fifteen years) and an adult group with a mean age of 40. Miller found that the older Americans made significantly more dispositional attributions than the older Hindus, and Hindus made significantly more situational attributions. There were few significant differences between the American and Hindu children aged eight and eleven. But Miller found a significant linear age increase in dispositional attributions among Americans, and a similar linear age increase in situational attributions for the Indian sample. It therefore appears that the FAE is very culture specific, and the cognitive and perceptual explanations originally advanced for the FAE need to be reconsidered in light of Miller’s findings.

Moscovici and Hewstone proposed that attributions are not only cognitive, but also social and cultural phenomena that are based on social representations – consensually shared knowledge, beliefs and meaning systems that are learned and socially communicated through language. Every society has its own stock of common sense and culturally agreed explanations for a wide range of phenomena, such as health and illness, success and failure, wealth and poverty, prosocial and deviant behaviour. People do not necessarily engage in an exhaustive cognitive analysis to explain events around them, as some of the early models of attribution suggest. Instead, they draw on socially shared and readily culturally available explanations.

Social schemas

It would be very difficult to function if we went about our everyday lives without prior knowledge or expectations about the  people, roles, norms and events in our community. Social cognition research suggests that our behaviour and interactions in the social world are facilitated by cognitive representations in our minds called schemas – mental or cognitive structures that contain general expectations and knowledge of the world.

A schema contains both abstract knowledge and specific examples about a particular social object. It ‘provides hypotheses about incoming stimuli, which includes plans for interpreting and gathering schema-related information’.

Schemas therefore give us some sense of prediction and control of the social world. They guide what we attend to, what we perceive, what we remember and what we infer. All schemas appear to serve similar functions – they all influence the encoding (taking in and interpretation) of new information, memory for old information and inferences about missing information.

Not only are schemas functional, but they are also essential to our well-being.

A dominant theme in social cognition research is that we are  cognitive misers, economizing as much as we can on the effort we need to expend when processing information. Many judge ments, evaluations and inferences we make in the hustle and bustle of everyday life are said to be ‘top of the head’phenomena, made with little thought and considered deliberation.

So schemas are a kind of mental short-hand used to simplify reality and facilitate processing. Schema research has been applied to four main areas: person schemas,  self schemas, role schemas and event schemas

Person schemas

Person schemas – often referred to as person prototypes – are configurations of personality traits that we use to categorize people and to make inferences about their behaviour. (The prototype is the ‘central tendency’, or average, of the category members.) In most Western cultures we tend to categorize individuals in terms of their dominant personality traits. We may infer from our observations and interactions with A that he is shy, or that B is opinionated. Most people would agree that Robin Williams is a prototypical extrovert and Woody Allen is a prototypical  neurotic.

Trait or person schemas enable us to answer the question: ‘what kind of person is he or she?. In so doing, they help us to anticipate the nature of our social interactions with individuals, giving us a sense of control and predictability.

Self schemas

Just as we represent and store information about others, we do the same about ourselves, developing complex and varied schemas that define our self-concept based on past experiences. Self schemas are cognitive representations about ourselves  that organize and process all related information.

They develop from self-descriptions and traits that are salient and important to our self-concept. Indeed, they can be described as components of self-concept that are central to our identity and self-definition. For example, people who value independence highly are said to be self-schematic along this dimension. People for whom dependence–independence is not centrally important are said to be aschematic on this dimension. Different self schemas become activated depending on the changing situations and contexts in which we find ourselves. For example, yourself schema as fun-loving and frivolous when you are with your friends may be quite different from your self schema as serious and dutiful when you are with your family. You will have schemas for your real self and also for your ‘ideal’ and ‘ought’ selves.

Role schemas

The norms and expected behaviours of specific roles in society are structured into role schemas. They will include both achieved roles – including occupational and professional roles, such as doctor or teacher – and ascribed roles, over which we have little control – such as age, gender and race.

The roles and expectations associated with these categories are commonly referred to as stereotypes – mental representations  of social groups and their members that are widely shared. Prolific empirical research on stereotypes views the process of categorizing individuals into their respective social groups as highly functional in that it simplifies the inherent complexity of social information.

Social categories such as male/female, black/white, old/young are viewed as highly salient and prior to any other kind of person categorization. Fiske refers to age, gender and race as the ‘top three’ because they are the most central and visually accessible categories. So when we meet someone for the first time, we attend to obvious and salient physical cues in guiding our interactions with them. With increased familiarity, the notion is that stereotypes based on physical cues become less important, and we may subsequently employ trait-based or person schemas.

Event schemas

Commonly referred to as cognitive scripts, event schemas describe behavioural and event sequences in everyday activities. They provide the basis for anticipating the future, setting goals and making plans. We know, for example, that the appropriate behavioural sequence for eating at a restaurant is to enter, wait to be seated, order a drink, look at the menu, order the meal, eat, pay the bill and leave. The key idea here is that our commonsense understanding of what constitutes appropriate behaviour in specific situations is stored in long-term memory, and it is activated unconsciously whenever we need it.