Executive Summary
Table of Contents

Introduction

• Teenage drivers have by far the highest fatal crash involvement rate of any age group.
• One significant factor related to the elevated crash rate of young drivers is the frequency with which they engage in risky-driving behaviors.
• In an effort to reduce the crash propensity of young drivers, the National Highway Traffic Safety Administration (NHTSA) has begun a program of research designed to understand better the factors related to the high crash rate, in particular risk taking, for drivers under 25 years of age. A special focus of this program is to develop a conceptual framework for understanding young driver risk taking for use in the development of more effective traffic safety programs and messages.
• NHTSA has recognized that a conceptual framework for understanding risky-taking behaviors must include both internal (e.g., memory, attitudes, and risk perception) and external factors (e.g., family, peers, and environment). The focus of this review is the internal factors.
• As a way to define terms and to conceptualize risk taking driving behaviors from a cognitive perspective for this project, a cognitive model of risky-driving behaviors is presented that includes areas where traffic safety messages and programs (interventions) might be applied to increase the likelihood of safe driving. The model conceptualizes risky and safe driving behaviors as the outcome of a decision making process in which risky driving may be chosen over behaviors that are less risky because the risky driving affords the person greater perceived benefit.
• This project has two main goals. The first is to gain a better understanding of the internal factors in the risk taking conceptual framework. The second goal is to develop preliminary guidelines for constructing traffic safety messages that are based upon how young people process information. Both goals are achieved by reviewing the literature on several topics related to how young people think and process information, and through discussion with subject-matter experts. The guidelines are still under development.

Memory

• Traffic safety messages that cannot be remembered or recalled effectively when necessary will have little or no impact on a person's driving behaviors. It is, therefore, critical for readers interested in constructing traffic safety messages for young people to understand how human memory develops and functions.
• Short-term memory (STM) has been described as working memory because it is the type of memory used for ongoing cognitive activities. It has also been characterized as the conscious part of memory where activities such as decision making, reasoning, symbol manipulation, and problem solving take place.
• Studies on STM capacity have shown that children can recall less information from STM than adults, suggesting that STM capacity may increase up to about 14 years of age. Thus, until this age, children cannot think about as many items at the same time as can adults. In addition, the speed at which information in STM can be processed increases up until about age 17. Therefore, any cognitive process requiring short-term memory, such as decision making, reasoning, or understanding a traffic safety message will proceed at a slower rate for younger people.
• Long-term memory (LTM) stores our experiences and knowledge. LTM capacity seems to be unlimited. The main difference between children and adults in LTM is that adults, through education and experience, have more knowledge about the world and specific strategies for dealing with potentially dangerous situations such as the situations that lead to risky driving. This specific knowledge allows adult drivers to more effectively problem solve and make decisions, assess the actual risk of certain behaviors, process the information in messages and programs, predict the outcomes of their behaviors, and think about the consequences of their behaviors in a more global way than younger drivers.

Attention

• Attention is a process of concentrating or focusing limited cognitive resources to facilitate perception or mental activity. Thus, attention is a process that is necessary for information processing--the information will not get into memory if it has not been attended to. In order to perceive, interpret, and understand a message, such as a traffic safety message, children or adults must be able to shift their attention to the appropriate message while filtering out other stimuli (called selective attention). Further, since they might be able to devote only part of their attention to the message (called divided attention), they must be able to devote enough attention to the appropriate message. They must also be able to focus their attention long enough to receive the entire message (called sustained attention). A lack of capacity in any of these attentional processes could lead to either a misunderstanding of traffic safety messages or a complete failure to receive them.
• The ability to filter out irrelevant and to focus on appropriate information increases up to about 17 to 20 years of age.
• The ability to divide one's attention between two information sources is quite poor for people of any age, but improves up to about 11 years of age.
• The ability to concentrate one's attention for a long period of time increases up to at least 16 years of age, with females having somewhat better ability for sustained attention.

Learning

• Learning has been defined as any relatively permanent change in behavior or thinking that results from past experiences. An understanding of learning processes is important for those interested in developing messages and programs that attempt to improve safe driving practices. Most traffic safety messages are designed to either change how people think about a traffic safety issue or to change people's safety behaviors. In other words, they are designed to educate people. Therefore, a comprehensive understanding of learning processes is central in the development of effective traffic safety messages and programs. Three learning processes are particularly relevant: classical conditioning, operant conditioning, and observational learning.
• The simplest kind of associative learning, classical conditioning, involves the association between reflexive responses (such as many emotional responses) and a stimulus (such as food or a person). Three main factors influence the chance of whether classical conditioning will take place: 1) The probability of classical conditioning occurring increases with the number of pairings of the unconditioned and the conditioned stimulus; 2) The probability of conditioning taking place decreases with increases in the time between presentation of the unconditioned and the conditioned stimulus; and 3) The probability of conditioning a response increases with the intensity of the unconditioned response.
• In operant conditioning, an action occurs that is followed by some outcome. If the outcome is positive, then the action is likely to be repeated. If the outcome is punishing, the action that led to it will become less likely to be repeated. Thus, through both reinforcement and punishment, new behaviors are learned and others are extinguished. Several factors affect whether and how a behavior is operantly conditioned: 1) The effectiveness of the reinforcement or punishment to change behavior decreases as the amount of time between the behavior and consequence increases; 2) The effectiveness of the reinforcement or punishment to change behavior increases with the intensity of the reinforcer or punisher; and 3) Behavior does not have to be reinforced or punished each time it occurs in order for that behavior to be conditioned.
• Humans can benefit from the experiences of others in order to learn behaviors and their consequences. Such learning is called observational learning. Four processes that influence observational learning have been identified: 1) Attention. Observational learning will not occur unless the person is paying reasonably close attention to the person or people performing the behavior; 2) Memory. Observational learning will not occur unless the person can remember the actions and consequences at a later time; 3) Ability to Reproduce the Action. Observational learning will not occur if the person cannot reproduce the action; and 4) Motivation. Observational learning will not occur unless the person has some reason for performing the behavior.

Reasoning

• A prominent feature of human cognition is our ability to draw conclusions based upon the things that we have learned. The process by which people draw conclusions from their knowledge of the world is called reasoning. When people reason, they are generating a belief (i.e., conclusion) that has been inferred from what they know. Thus reasoning is intimately related to most other cognitive abilities including learning, moral development, verbal ability, memory, attitude formation, and problem solving. Many traffic safety messages and programs use logic to both teach and convince drivers to drive safely. Further, much of what people learn about the driving task and driving situations is based upon reasoning processes. Therefore, a thorough understanding of how reasoning develops and the problems people have with reasoning is necessary for the development of effective traffic safety messages.

• One common type of reasoning is called class-inclusion reasoning; that is, determining what "things" go with what category. Generally, the ability to draw logical conclusions from given premises in class-inclusion reasoning improves with age. However, people of all age groups have difficulty with class-inclusion reasoning, particularly when they are faced with conclusions that do not follow from the given information. Several causes for poor class-inclusion reasoning are reviewed.
• Another common type of reasoning involves people drawing conclusions based upon contingency relationships usually described in if-then premises, called conditional reasoning. Conditional reasoning ability improves with age. However, even at older ages, the ability is quite poor overall. As in class-inclusion reasoning, people especially have difficulty when conclusions do not follow logically from the given information. Several causes for poor conditional reasoning are reviewed.
• Another reasoning process in which people frequently engage is reasoning by analogy. In this type of reasoning, people draw a conclusion about something new by noting the similarities between the new item and something else for which they are already quite familiar. The ability to effectively reason by analogy improves up to at least 19 years of age with no differences between the sexes. Overall, people are relatively good at this type of reasoning. Several causes for poor analogical reasoning are reviewed.
• A final important reasoning process for people is hypothesis formation and testing. In this type of reasoning, a person forms a belief and then seeks information related to the belief, similar to a scientist trying to understand a phenomenon. Ability for this type of reasoning is poor for people of all ages. Several causes for poor hypothesis formation and testing are reviewed.

Motivation

• Motivation is the set of influences that account for the initiation, direction, intensity, and persistence of behavior. Since traffic safety messages and programs are designed to change unsafe driving behaviors or enhance safe ones, understanding the reasons why these behaviors occur is necessary for constructing appropriate messages and programs. Moreover, an understanding of motivations is important because in order to change behaviors and ways of thinking, the young drivers must have a motivation for changing their behavior. If the driver has no motivation for learning the message or participating in the traffic safety program, then learning and behavior change will generally not occur. The review focuses upon two motives that are thought to have a great impact on the traffic safety behaviors of young drivers and the construction of traffic safety messages and programs for young drivers: sexual motivation and arousal/sensation seeking motivation.
• A powerful behavioral motivator is the need for sexual fulfillment or the sex drive. While the sex drive produces clear motivational effects on behavior, its influence on driving behaviors, particularly unsafe driving behaviors, is poorly documented. However, a few studies suggest that the sex drive, at least for males, may negatively influence traffic safety. At least one focus group study with college students (18 to 22 years of age) has shown that, by self report, one reason for drinking and driving is to show off in order to attract the attention of the opposite sex. In the population of young male drivers, drinking and driving activities may be motivated, in part, by the sex drive.

• A large body of work has documented the fact that the arousal motive can have negative traffic safety consequences for people who need to increase their levels of arousal, those commonly called sensation seekers. Accurate identification of sensation seekers is important both for the study of this motivation and for appropriately targeting traffic safety messages and programs.
• Zuckerman and his colleagues have developed a test (the Sensation Seeking Scale, SSS) in which behaviors related to sensation seeking are self reported. This test has been used extensively to define the demographics of sensation seeking and its relationship to unsafe driving behaviors. Numerous studies utilizing the SSS have shown that males score higher on the total SSS than females. Studies have also documented that scores on the SSS tend to increase with age, up until the late teens. Several studies have shown that high sensation seeking is related to drinking and driving, speeding, lack of safety belt use, and traffic violations.

Risk Perception

• An unavoidable component of a person's life is risk and uncertainty. Our thoughts about these risks and how we assess them have been termed risk perception. The perception of risk and uncertainty is integral to many cognitive tasks. People appraise risk when making decisions about uncertain events and situations. When reasoning about probabilistic information and the likelihood of negative outcomes, both children and adults must think about risk. Thus, an understanding of risk perception, and how people assess risk, is essential for understanding both risky-taking behavior and developing traffic safety messages to reduce risky driving.
• A number of studies have investigated perceptions of traffic crash and injury risk by age. The majority of these studies have found that young drivers tend to perceive less risk in specific crash scenarios and general driving than do older drivers, and are poorer at identifying hazards when driving. Young drivers also tend to see themselves as less likely to be in a crash than others in their own age group.
• Because of the elevated crash rate of males over females, many studies of risk perception have restricted their subjects to males. However, the few studies that have investigated crash risk perception for females have shown mixed results. Some studies found that males rated crash scenarios as less risky than did females. Other studies, however, have found no sex differences in crash involvement risk perception. It is clear, however, that females, like males, tend to consider themselves less likely to be in a crash and to be better drivers than others in their peer group, although to a lesser degree than males.
• The chances of a crash are judged to be greater when the subject is a passenger than when he or she is driving the vehicle, showing that perceived control of a vehicle is an important factor in the assessment of risk for younger drivers. In addition, younger drivers using safety belts tend to rate the risk of crash involvement as higher than those not using belts. This effect may be due to safety belt use sensitizing drivers to the possibility of a crash.
• Several factors may be involved in the misperception of risk. One is the optimism bias. Several studies have shown that young drivers tend to think of themselves as less likely to be involved in a crash than the average driver, primarily because of the belief that their driving skills are above average. Another is the availability heuristic. Instead of relying upon well-established facts and recognizing that their experiences, short-term memory capacity, and retrieval abilities are limited, people attempt to discern how frequently something occurs by trying to recall examples. People using the availability heuristic to assess traffic crash risk search memory and retrieve many trips that were crash free, leading even poor drivers to underestimate their risks. Another potential source of error in risk perception is the failure of young drivers to understand the effects of cumulative risk. A final factor that has been shown to bias judgment of outcome likelihood is the desirability of the outcome. People have a tendency to judge outcomes that they want as being more likely than outcomes that they do not want.

Problem Solving and Decision Making

• Undoubtedly the most complex cognitive activity that humans engage in is attempting to find solutions to problems; this includes the activity of decision making. Problem solving is an ubiquitous activity in youthful driving and traffic safety. Much of high-risk driving can be conceptualized as a problem faced by a driver. Several traffic safety programs have attempted to teach young drivers new strategies for dealing with the problems that lead to high risk driving. Understanding the way young drivers go about finding solutions to problems (and making decisions) and the deficiencies they have in this process is important in the development of traffic safety messages.
• Efficient problem solving involves developed memory and attention processes, good verbal ability, proficient reasoning, and knowledge and experience of the world. As such, the ability to problem solve effectively develops along with these other cognitive processes. As children grow older, they adopt increasingly sophisticated strategies for solving problems. Included in the developmental changes that occur in the first 18 years or so is a tremendous increase in domain-general and domain-specific knowledge, the ability to generate several potential solutions to a problem, the ability to ignore irrelevant problem information, the ability to think about more than one dimension of the problem simultaneously, and the ability to think about relationships among events in a bidirectional way.

Social Cognition

• Social cognition has to do with how people make sense of other people and themselves. The study of social cognition is the study of how people think about others and how others influence a person's thoughts. Because driving typically occurs in a social setting or involves social thinking, such as attitudes, an understanding of social cognition is important in developing messages or programs to reinforce or change people's traffic safety related behavior. This section provides a brief overview of several key elements of social cognition that have implications for the development of traffic safety messages and programs including attributions, social schemata, and scripts.
• Attribution theory is concerned with how people go about assigning causes to the events they observe. It focuses on how people use information in the social environment to arrive at causal explanations for events. Thus, it is often described as the study of perceived causality, with attribution referring to the perception of inference of cause. An important theme of attribution theory is that in making attributions, people act as naive scientists; that is, people intuitively, or in a common sense way, infer or deduct causes of events around them. A second theme is that there are several tendencies in attribution making. In general, people tend to be more likely to view two events as causally related if the events are similar to one another or if they occur near to one another in time or space, to attribute behavior to a single cause rather than multiple causes, and to attribute causes of other peoples' behaviors to internal factors while they attribute causes of their own behavior to external factors. A third theme of attribution theory has to do with how people make inferences about others' intentions and dispositions.
• Social schemata are cognitive structures representing organized knowledge about objects, people, and past situations, and are helpful in organizing, making sense of, and remembering details. Researchers divide social schemata into several types including person schemata, self-schemata, role schemata, and scripts. Because the last type of social schemata, scripts, represents a person's knowledge structure for a sequence of events, it is particularly important for driving. Many of the events in the "driving" script include traffic safety related behavior. Thus, an understanding of scripts is especially useful for the design of traffic safety messages and programs for young drivers.
• Scripts are structures that describe appropriate sequences of events in well-known situations. They are comprised of several sequential steps and serve to organize information about the sequence of predictable actions, locations, roles, and props that constitute events. Scripts are learned throughout a person's lifetime, both by participation in and observation of events and are activated automatically whenever a similar event is encountered in the real world or referred to verbally. In general, the more often a script is activated, the more abstract and complex it becomes. The more recently and the more frequently the script has been activated in the past, the greater the likelihood of activation in the future.
• There is evidence that scripts are not subject to much change and may in fact resist change (known as the perseverance effect), even when there is information that is inconsistent with or contradicts the script. Thus, it may be difficult for traffic safety messages and programs to effect change in a young driver's script. The perseverance effect appears to be reduced, not when people are simply told about the contradictions and requested to be unbiased, but rather, when people are asked to think carefully about how they evaluate the evidence and when people are cautioned to be aware of their biases as they interpret information. The perseverance effect may also be lessened when people are forced to counter argue their scripts; that is, to explain why their scripts might be wrong.

Attitude Formation and Change

• Attitudes can be thought of as relatively stable mental positions held toward ideas, objects, or people. While there is no universally agreed upon definition of attitudes, there is widespread consensus that: 1) evaluation constitutes a central and possibly predominant aspect of attitudes, 2) attitudes are represented in memory, and 3) both behavioral antecedents and consequences of attitudes have affective, cognitive, and behavioral domains. Because appropriate traffic safety behaviors may be influenced by attitudes towards driving and traffic safety, knowledge about how attitudes develop, endure, and change is necessary for constructing effective messages and programs.
• Attitudes are learned. Thus, the processes discussed in the section on learning apply generally in attitude formation. Attitudes may be formed directly through questioning, personal experience, and operant conditioning, or indirectly through classical conditioning, social learning, and observation.
• Attitudes are believed to predict behavior in a complex way. A popular theory states that attitudes and cultural norms combine to determine behavioral intentions, which in turn produce a voluntary behavior. A number of studies have found that differences in the extent to which attitudes guide behavior result from differences in how easily or quickly a person can retrieve the attitude from memory. Highly accessible attitudes have been found to be more predictive of behavior than less accessible attitudes. It has been found that people holding highly accessible attitudes toward an object are more likely than those holding less accessible attitudes to evaluate information relating to the attitude object in a biased manner, and thus to shape their behavior in a direction consistent with their attitudes. Several studies have found a relationship between self-monitoring and attitude-behavior consistency. High self-monitors are individuals who monitor their behavioral choices on the basis of situational information, while low self-monitors guide their choices on the basis of salient information from relevant inner states such as attitudes, feelings, and dispositions. Other factors that have been found to mediate the relationship between attitudes and behavior include habit or past behavior, stability of attitudes over time, volitional control of behavior, and degree of direct experience with the attitude object.
• Attitudes not only affect behavior; they are also influenced by behavior. Two major explanations of the influence of behavior on attitudes have been advanced. The first is dissonance reduction. That is, because we have a strong need for cognitive consistency, we change our attitudes to make them more consistent with our behavior. The second explanation of the influence of behavior on attitudes is self-perception theory, which posits that when internal states (e.g., attitudes) are weak or ambiguous, people must infer them from knowledge about their overt behavior and the circumstances in which the behavior occurred. That is, we look to our behavior when our attitudes are not completely clear in order to figure out our attitudes.
• Because attitudes are learned rather than innate, they are susceptible to change through persuasion. Persuasion refers to the intentional attempt to influence or change the attitudes of other people. Thus, persuasion is a process that involves three components: communicator or source, message, and audience or target population. Factors related to each of these components affect the chances and degree of attitude change resulting from the persuasion process. 1) Communicator or Source: Source credibility, to a large extent, is characterized by expertise and trustworthiness. In general, communications will be more persuasive if they are perceived to come from a highly credible and respected source, the person or source states an opinion that is contrary to what would be expected, the person or source is attractive, and the person or source is seen as similar to the recipient. 2) Message: Studies show that one-sided messages are more persuasive when audiences already favor the source's position; two-sided messages (both sides of issue) are more persuasive when they oppose it. Messages that include appeals to fear are generally effective only when the presented threat is severe, the likelihood of it occurring is high, and the audience is able to do something to prevent or eliminate it. 3) Audience or Target Population: There is evidence that audience involvement with an issue (often measured by personal importance) moderates the effects of communicator and message factors, as well as the persistence of attitude change. The motivation level and ability of recipients influences the cues they will be most likely to attend to during the persuasion process.

Verbal Ability

• The term verbal ability refers to all use of language, including oral communication, oral comprehension, reading, and writing. Traffic safety messages, of course, use verbal means for conveying information. If the verbal ability of the recipient is not accounted for in the messages, then the message will not be processed and will have no chance of improving traffic safety behaviors. Thus, an understanding of verbal ability is necessary for effective traffic safety messages and programs.
• Oral communication and comprehension increase with age. Phonology, which is the way in which the sounds of language are produced, begins to develop in infancy and continues through age five or six. The understanding of word meaning also begins in infancy and continues at least through 11 years of age, when children begin to master abstract word phrasings such as metaphors. Along the same lines, size of vocabulary increases dramatically during the first five years, approximately doubling in size each year and continues at a much slower rate throughout the lifetime. Grammar tends to appear around 1.5-to-2 years of age when children begin making two-word sentences and continues to improve up to at least 15 years of age.
• Reading and writing abilities increase with age. Reading skills begin to develop before 1 year. Fluent reading and simple comprehension appears around 9 years of age. Reading for complex comprehension of written material is achieved between 16 and 19 years of age.
• Females tend to be superior to males in oral language, reading, and writing.

Moral Development

• Moral development refers to the changes that occur with age and experience in how individuals deal with moral issues. A major influence on all driving behaviors, in particular high risk driving, is the moral principles or rules by which a person lives. Moral principles determine the motivation for many social behaviors, such as driving. Traffic safety programs such as victim impact panels attempt to appeal to peoples' morality in an attempt to get them to change their high risk driving behaviors. As such, it is important to understand the acquisition of moral thinking so that appropriate programs and messages that rely on moral thinking can be produced.
• The predominant approach to understanding moral development for the past 30 years builds on the social cognition work of Piaget, and is characterized by his ideas that individuals play an active role in their own development and that cognition is of central importance in social development. Piaget's early work included a preliminary examination of children's development of moral judgments, although he did not pursue these investigations. Research on moral development was expanded upon and refined by Kohlberg.
• Kohlberg viewed moral development as a process in which children form their own values and moral concepts out of active efforts to organize and understand social experiences. Kohlberg proposed six stages of moral development. Stages 1 and 2 make up the preconventional level, in which rules and social expectations are conceived as being external to the self. It is primarily children who are found at this level of moral development. Stages 3 and 4 make up the conventional level, in which individuals identify with or have internalized the rules and social conventions of others, including authorities. It is primarily adolescents and adults who are found at this level. Stages 5 and 6 make up the postconventional or principled level and are characterized by the ability of individuals to separate themselves from the rules and expectations of others and think in terms of self-chosen principles. Relatively few people attain this level of moral thinking.
• Kohlberg's moral stages point to, by definition, age differences in moral development. However, the relationship between age and moral development is not a simple one. Movement from stage to stage does not occur merely as individuals age, and movement through every developmental stage is not inevitable. Still, rough age patterns can be discerned.
• One of the more thorough reviews of the literature found that sex differences in moral reasoning are exceedingly rare. Of 108 studies reviewed, only eight clearly indicated significant differences favoring males and many of these were confounded by occupational level or educational status. It is concluded that the moral reasoning of males and females is more similar than different.
• Moral development is stimulated by the provision of role-taking opportunities and these opportunities arise from participation in school and peer life, and interaction with the political and social institutions of the larger society, as well as from family participation. All of these types of participation converge to stimulate moral development. The more social stimulation (through these role-taking opportunities), the faster the rate of moral development.

Piaget's Theory of Cognitive Development

• Jean Piaget devoted much of his life to studying children's cognitive development--the way in which mental processes such as thinking, reasoning, and perceiving the world evolve. Piaget's interest was in the qualitative rather than quantitative characteristics of development. That is, he was concerned, not with how much children know, but how they come to know it.
• Piaget's theory has several core assumptions. Foremost among these is the idea of constructivism--that children are active thinkers, constantly trying to construct more advanced understandings of the world. According to Piaget, children learn by doing--they not only observe and imitate the world around them, they interpret it as well. At the same time, Piaget recognized the role of experience in cognitive development, noting that the physical and social contexts in which children act help to give shape to their constructions of the world. In Piaget's view, adaptation, and thus, intellectual or cognitive growth, comes about through the dual processes of assimilation and accommodation. Assimilation involves modifying or changing new information to fit into what is already known and accommodation involves restructuring or modifying what is already known so that new information will fit in better. It is the constant balancing of these two processes that leads to adaptation to the environment and underlies the process of cognitive development.
• Piaget proposed four stages of cognitive development, each qualitatively different from one another. These stages include the sensorimotor stage, the preoperational stage, the concrete operational stage, and the formal operational stage. During each stage, distinctive styles of thinking emerge. Piaget's introduction of qualitatively different stages of thinking was in marked contrast to the prevailing view of the time that children's cognitive activity was identical to adults' cognitive activity, only less efficient. 1)Sensorimotor Stage: The sensorimotor stage is generally characterized by actions, movements, and perceptions that occur in the absence of language. These actions are coordinated through schemata, viewed by Piaget as simple mental images or patterns of action that individuals use to organize information and interpret the things they see, hear, smell, and touch. The main trend during this overall sensorimotor period is the development of object permanence. 2) Preoperational Stage: The preoperational stage is characterized by the development of symbolic or representational thinking; that is, thinking dependent on symbols rather than on sensorimotor relationships. Important characteristics of preoperational thought are children's lack of understanding of the concept of conservation and their growing ability to overcome egocentrism. 3) Concrete Operational Stage: At about age seven, children begin to understand the concept of conservation. In Piaget's view, this understanding is essential for the acquisition and subsequent development of logical thought. The main characteristics of the concrete operational stage include: the ability to use operations and mentally reverse actions; attainment of conservation skills; use of logical reasoning instead of intuitive reasoning, but only in concrete circumstances; the inability to engage in abstract thought; and the capability to classify or divide things into sets or subsets and to consider their interrelations. 4) The Formal Operational Stage: During this stage, logical operations are no longer tied to concrete problems; children can now understand abstractions, consider hypothetical questions, and design formal ways to test abstract ideas. There is evidence that some presumably normal people never attain formal operations.
• The empirical support for Piaget's cognitive stages implies, by definition, that there are age differences in cognitive development. At the same time, Piaget, himself, made it clear that the ages associated with different stages are always average and approximate. Roughly, the sensorimotor stage extends from birth to age 1.5-to-2; the preoperational stage from about age 1.5-to-2 to 6-to-7; the concrete operational stage from age 7-to-8 to 11-to-12; and the formal operational stage from age 11-to-12 to 14-to-15.
• The effects of sex on cognitive development have not been widely studied. Thus, there is little support for sex differences in Piaget's theory.
• There have been a number of studies exploring whether children can be trained to display advanced cognitive skills. Based on several reviews of the conservation literature, it was concluded that conservation can apparently be taught, although even highly individualized training is only successful about half the time. Studies focusing on children's attainment of formal operational thinking have also shown that certain cognitive skills can be taught through training.

Introduction
Table of Contents

Despite the fact that motor vehicle death rates have declined significantly since 1975, motor vehicle crashes continue to be the major cause of death and serious disability for adolescents and young adults. On a per population basis, drivers under age 25 in the United States (U.S.) had the highest rate of involvement in fatal crashes of any age group in 1996 and their fatality rate based on vehicle miles traveled was four times greater than the comparable rate for drivers age 25 to 65 (National Highway Traffic Safety Administration, NHTSA, 1997a). Teenage drivers have by far the highest fatal crash involvement rate of any age group based on number of licensed drivers (61.26 per 100,000 licensed drivers in 1993), and most teenage passenger deaths (67 percent) occur when a teenage driver is at the wheel (Insurance Institute for Highway Safety, 1995). Motor vehicle injury rates also show that teenagers continue to have vastly higher rates than the population in general.

Risky-driving behaviors may contribute heavily to the high crash and injury rates for drivers under the age of 25 years. For example, drinking and driving is a major factor in young driver fatal crashes. In spite of the fact that the proportion of fatally injured young drivers (21-to-24 years of age) with blood alcohol concentrations greater than or equal to 0.10 percent has declined steadily from 1982 to 1996, from 40 to 27 percent (NHTSA, 1997a), this age group has consistently had the highest proportions of any age group. A study by the University of Michigan Transportation Research Institute (UMTRI) found that Michigan drivers under the age of 21 accounted for 14 percent of drunk driving convictions when this same age group makes up only 8 percent of the licensed driving population. The study also discovered that of all alcohol-involved crashes in Michigan leading to a felony drunk driving conviction, 23 percent were for drivers under the age of 21 (Eby, 1995a; Eby, Hopp, & Streff, 1996). There is also evidence that young people, more frequently than others, speed (e.g., Jonah, 1986; Koneni, Ebbesen, & Koneni, 1976; Soliday, 1974; Wasielewski, 1984), travel with shorter headways (e.g., Wasielewski, 1984), run yellow lights (e.g., Koneni, Ebbesen, & Koneni, 1976), and fail to use safety belts (e.g., Eby & Hopp, 1997; NHTSA, 1997b). These facts underscore the need for effective traffic safety programs and messages designed specifically for adolescents and young adults.

In an effort to reduce the crash propensity and resulting injuries of young drivers, NHTSA has begun a program of research designed to better understand the factors related to the high crash rate, in particular, risk taking, for drivers under 25 years of age (see, e.g., NHTSA, 1995a, 1995b). A special focus of this program is to develop a conceptual framework for adolescent risk-taking behaviors that can assist in the development of measures, such as public information and education programs, to increase safe driving behaviors. NHTSA recognizes that a comprehensive framework for understanding risk- taking behaviors must include not only external factors such as social interactions or family life, but also internal factors such as the information processing capabilities and strategies of youth. NHTSA's framework is illustrated in Figure 1. As shown in this figure, internal and external factors mutually influence each other and can collectively give rise to risky-taking behaviors. The focus of the project reported here is limited to gaining a better understanding of the internal factors. An additional focus of the project is to use the synthesized information to generate a set of guidelines for the development of effective traffic safety messages and programs for young people. The guidelines are currently being developed.

Figure 1: A conceptual framework for understanding risk-taking behavior. The internal factors are the focus of this literature review.

A Model of Risky-Driving Behaviors

An unavoidable component of a person's life is risk and uncertainty. As a matter of everyday living, we engage in activities and are exposed to situations that have some element of risk. Risk is particularly prevalent in motor vehicle travel and is influenced by a multitude of factors including the decisions that people make about how they drive, who they drive with, under what conditions they drive, and why they are driving. For example, in a survey of high school students, Summala (1987) found that about 60 percent of male students and 33 percent of female students reported that they at least occasionally engaged in high risk driving for fun. As NHTSA (1995a, 1995b) and others (e.g., Hodgdon, Bragg, & Finn, 1981; Jonah, 1986) have pointed out, decisions made by young drivers that result in risky-driving behaviors may strongly contribute to the elevated crash risk of young drivers.

Despite the prevalence of risk, and the abundance of research on risk and risk taking, there is disagreement about what constitutes risk and risky-taking behaviors (e.g., Fischhoff, 1985; Yates & Stone, 1992a). The large literature on young driver risk taking has been extensively reviewed and will not be reviewed here (see, e.g., COMSIS Corp. & the Johns Hopkins University, 1995; Hodgdon, Bragg, & Finn, 1981; Jonah, 1986, 1997). However, as a way to define terms and to conceptualize risk taking driving behaviors from a cognitive perspective for this project, we present a cognitive model of risky-driving behaviors (Figure 2) that includes areas where traffic safety messages and programs (interventions) might be applied to increase the likelihood of safe driving. While the form of this model is unique to this project, it draws heavily upon concepts presented by others (Fishburn, 1968; Furby & Bayth-Marom, 1992; Hodgdon, Bragg, & Finn, 1981; Jonah, 1986; Olk & Waller, 1998; Wilde, 1976; Yates, 1990; Yates & Stone, 1992a, 1992b).

Figure 2. A decision making model of risky-driving behavior, showing where traffic safety messages and programs (interventions) might be applied to increase the likelihood of safe driving.

According to Yates (1990; Yates & Stone, 1992a), risky-taking behavior is the result of a decision process in which risk is just one component of a set of factors that are considered in the decision. Thus, the model presented here conceptualizes risky and safe driving behaviors as the outcome of a decision making process in which risky driving may be chosen over behaviors that are less risky because the risky driving affords the person greater perceived benefit. It is important to note that the decision process is not characterized by an intensive review of information and courses of action. The process may happen quite rapidly and the person may consider only partial information when making a decision. Further, the driver may not be aware of the decision process, either because it occurs rapidly or because the process is nonconscious. If at least two courses of action are considered, then the person chooses which action to take. The model applies only to a single decision made at a certain time. During the course of an automobile trip, the driver may make hundreds of decisions, some of which lead to risky-driving behaviors and some of which do not.

The model is divided into two parts: subjective and objective. The subjective component of the model, shown enclosed by a dashed line, represents the cognitive factors involved in the decision making process, including the driver's memories, attentional capacities, perceptions of risk, attitudes, motivations, moral influences, and learning, reasoning, and problem solving abilities. The objective component, shown enclosed by a dotted line, constitutes the driving behaviors; that is, those actions that we observe on the road. For this model, we define all driving behaviors as either safe or risky. Risky-driving behaviors are those actions that increase the objective likelihood of a crash or the severity of injury should a crash occur (e.g., Olk & Waller, 1998; Simpson, 1996; Williams, 1997). As such, a driver may not consider his or her action to be a risky one even though it increases his or her chances of being in a crash or becoming severely injured in a crash. This definition of risky-driving behavior also assumes a baseline from which to assess the increase in risk or crash severity. This baseline is set by societal standards. In the case of speeding, for example, the baseline may be the speed limit, "the speed of traffic flow," or the speed that is safe for the current conditions.

When a driver approaches a situation in which an action may be required, for example a young driver approaching a signalized intersection where the light has changed from green to yellow, the model proposes that an analysis of possible courses of action (COAs) is conducted. If the driver only knows about, or is only able to produce, a single possible action, then that action is performed. This outcome is represented by the arrow that exits the courses-of-action box (only 1 COA) and terminates at the objective driving behavior part of the model (dotted line). If only one action is possible, then no decision is made, the behavior could be either risky or safe, and the driver might or might not perceive the risks in taking the action. In the example of the young driver, he or she may either always brake when the light changes to yellow, or he or she may always continue through the intersection, regardless of all other considerations. Some researchers have suggested that many driving behaviors, in particular those related to risky driving, frequently are based on only one course of action (e.g., Jørgensen, 1988; Wagenaar, 1992).

In driving, however, there is nearly always more than one objective possible course of action, regardless of how unpleasant some of the other actions might be (e.g., the driver could always stop driving). If there is more than one perceived course of action, then the model supposes that the driver uses a decision process to choose a single course of action from the set of possible actions. This set of actions may be exhaustive or may only contain two alternatives. For example, when approaching an intersection where the light has changed from green to yellow, some possible courses of action are to continue at the same speed through the intersection, accelerate through the intersection, or brake and stop before the intersection. The driver may consider all three or may only consider a subset of the courses of action. As proposed by Yates and Stone (1992a), the driver evaluates each course of action by determining a subjective worth for each action. An increase in the subjective worth for a course of action means an increase in the likelihood that that course of action is chosen (Yates & Stone, 1992a). The choice of course of action is based on some decision rule that takes into account the subjective worth for each possible course of action.

The subjective worth is a complex combination of perceived risk and other considerations⁽¹⁾. In the case of driving, the perceived risk of a certain course of action is a combination of the perceived probability of getting in a crash and its perceived severity, and the perceived probability of getting a citation and its perceived severity. The driver may perceive no risk of being in a crash or getting a citation, in which case the perceived risk for that course of action would not be a feature in the determination of that course of action's worth. The other considerations are features of the course of action that either result in perceived costs or in perceived benefits. A list of example considerations, derived from the risk taxonomy of Jacoby and Kaplan (1972), is included in Figure 2. For example, within a certain course of action there may be a time savings benefit, a financial cost, a social benefit, and a small physical cost, in addition to the perceived risk assigned to the action. The considerations that are included in the decision making process, their likelihood of occurring, and their magnitudes are all subjectively determined for that course of action at that time. Thus, for example, a young driver may heavily weight the perceived social benefits of running a red light (e.g., the driver may think it makes him look brave to his peers, thus, gaining perceived social status), and only minimally weight the injury costs, should a crash occur. The fact that nonrisk considerations might be perceived as highly beneficial for a risky-driving behavior means that high risk courses of action could be assigned a high subjective worth.

Once a course of action has been decided upon, it is performed. For the explanatory purposes of the model, the selected course of action has a perceived risk associated with it that is either zero or above zero. If the perceived risk is above zero, that is, the driver thinks that there is at least some chance of a crash or a citation from law enforcement, then the resulting behavior itself can still be either risky or safe, depending upon the socially-defined baseline for that driving situation. Those who engage in a risky-driving behavior, perceive a risk for that behavior, and have more than one perceived course of action, are defined as risk taking. Thus, the young driver who is approaching an intersection in which the light is yellow and decides to accelerate through the intersection, even though he knows he has other options, is engaging in risk taking if he knows he could get in a crash or receive a traffic citation from law enforcement. By the same argument, the driver who accelerates through the intersection because he is heavily weighting a time savings benefit (i.e., not having to wait through another cycle of the lights), is also risk taking if he perceives some chance of getting in a crash or being cited by law enforcement. On the other hand, if the perceived risk for that course of action is zero (i.e., the driver perceives no risk), then the behavior can also be risky or safe. If the action is considered by society to be a risky-driving behavior, the behavior is perceived by the driver as risk free, and there is more than one perceived course of action, then we define that person's behavior in this situation as risk ignorant. For example, the young driver approaching the intersection may wrongly think that other drivers at the intersection will watch out for him (i.e., no crash risk) and that since he sees no patrol cars, he will not get a traffic citation (i.e., no law enforcement risk). Given this lack of perceived risk, the young driver may continue into the intersection simply because he does not want to wait. A driver who has selected a course of action in which there is no perceived risk can also engage in safe driving behaviors (the two arrows terminating in the safe driving behavior box of the model shown in Figure 2); the person may still drive safely simply because he believes that people should follow traffic laws.

The goal of traffic safety researchers should be to get drivers, in particular young drivers, to follow the decision making pathway to safe driving behaviors represented by the thick arrows in Figure 2. Those drivers who engage in safe driving by this pathway in the model are the ones who recognize the risks associated with possible driving behaviors and choose the behavior that is safe. Those drivers arriving at safe driving by the other pathway do not adequately perceive the risks of driving and may end up driving in a risky manner in other situations.

Also shown in the model are points where traffic safety messages and programs (interventions) can be applied. One potential intervention point is when drivers first determine the courses of action available to them. Drivers can be made more aware of the many courses of action available to them when they are driving, or be helped to improve their ability to actually perform other actions (as may be the case in learning to drive). Another potentially fruitful focus for interventions is the basic decision making process. As has been recently tried by NHTSA, it may be possible to train young drivers to make better driving decisions (e.g., NHTSA, 1996). Another point of intervention is at the risk perception level. Messages and programs to change perceptions of traffic violation enforcement risk or of crash risk, might lead to less risky courses of action being chosen by young drivers. Focusing on other considerations evaluated in determining the worth of an action is another avenue for interventions. Programs and messages could attempt to get young drivers to consider information that they do not already use or to more appropriately weight the significance of the information that they do use.

While the model is specific to a single decision at a certain time, it can help us to understand why certain people may be prone to engage frequently in risky-driving behaviors and in other risky or problem behaviors (e.g., Barnes & Welte, 1988; Donovan, 1993; Elliott, 1987; Evan, Wasielewski, & von Buseck, 1982; Jessor, 1987a, 1987b; Jessor & Jessor, 1975, 1977). The model proposes that a risky behavior is the outcome of a decision process, and that persons engage in that behavior because of the benefits (or absence of losses) that they get from the action. It is reasonable to assume that the subjective aspects of the process are similar in other driving situations (i.e., how risk is perceived, weightings for considerations, etc.). If so, then we would predict that in driving situations involving risk, similar outcomes would occur because similar information processing is occurring. As many researchers have shown (e.g., Barnes & Welte, 1988; Donovan, 1993; Elliott, 1987; Evan, Wasielewski, & von Buseck, 1982; Jessor, 1987a, 1987b; Jessor & Jessor, 1975, 1977), risky driving is only one type of risky behavior that a person tends to engage in. Therefore, it is reasonable to think that an individual would apply the same decision making processes (and would have similar influences on this process) to most situations in which high risk behavior is undertaken.

More to the point of the current project, the model can also help us to understand why risky-driving behaviors seem to decline with age. As discussed in this literature review on cognitive development, several cognitive skills and abilities develop with age. Cognitive changes in the speed of information processing, memory capacity, attention, decision making ability, and general knowledge of the world, could all positively influence the decision making process leading to safe driving. It is important to remember that this literature review only covers the internal factors (see Figure 1). A thorough understanding of risky-driving behaviors among young drivers also requires a review of external factors, such as peer influences, family, and school. These factors are related to the decision making process (e.g., as considerations) and must be considered in the creation of effective traffic safety messages and programs (interventions).

The literature review is divided into 12 sections. The review focuses primarily on cognitive development from about 10 to 24 years of age. In addition, if literature reporting research on sex differences⁽²⁾ was found, then it was included where applicable. While the topics are presented as chapters in this review, there is significant overlap among topics. The first section is about memory; that is, the processes that allow a person to retain knowledge over time. This section discusses the various types of memory and their development. Traffic safety messages, or the effects of programs that are not remembered or not recalled when necessary, will have little positive influence on safe driving behaviors. The second section, attention, discusses the factors related to the development of how people focus cognitive resources on perceptual or mental tasks, including selective, divided, and sustained attention. A traffic safety message or program that is not attended to may be misperceived or missed completely, reducing or eliminating any chance for message or program effectiveness. The third section, learning, discusses three processes by which people acquire information: classical conditioning, operant conditioning, and observational learning. Because many traffic safety messages and programs have a learning component, an understanding of how people learn is paramount for development of effective messages and programs. The fourth section, reasoning, discusses several ways in which people draw conclusions from their knowledge of the world and how these processes develop with age. Because many traffic safety messages and programs use logical arguments to influence safe driving behaviors, it is important to understand how reasoning processes operate and develop in humans.

The fifth section of the literature review, motivation, examines the factors that initiate and influence the intensity of behaviors. Because these factors are numerous, the review discusses only two motivations that seem to influence risky-driving behaviors in young drivers: sex and the need for stimulation (sensation seeking). It is important to understand and consider young driver motivation when developing traffic safety messages and programs because risky driving, like other behaviors, is motivated by something and one needs to provide a motivation for following the recommendations in a traffic safety message or program. Also included in the review, in the sixth section, is a discussion of the development of risk perception and factors that contribute to the misperception of risk. As we have already mentioned, risk perception is an integral part of the decision making process that leads to risky-driving behaviors and has the potential to be influenced by traffic safety messages and programs. The seventh section discusses the development of problem solving and decision making in a general way. While influenced by all other cognitive factors and their age-related limitations, general deficits of problem solving and decision making ability are discussed. Again, an understanding of these processes is integral to understanding risky-driving behaviors. Also covered in this review are some of the social factors that influence how people think, known as social cognition. This eighth section covers those topics in the development of social cognition that are likely to have an effect on driving: attribution and social schemata/scripts. The ninth section focuses on attitude formation and change. Because appropriate traffic safety behaviors may be influenced by attitudes towards driving and traffic safety, knowledge about how attitudes develop, endure, and change is necessary for constructing effective massages and programs. The tenth section examines briefly the development of verbal ability, that is, all use of language. Traffic safety messages and programs use verbal means to convey information. If the verbal ability of the recipient is not accounted for, then message or program effectiveness may be diminished. Because many traffic safety messages and programs deal with moral issues in driving, the eleventh section is a review of moral development. This section examines a theory of moral development and many factors that influence moral thinking. Finally, because of the influence of his ideas on the field of cognitive development, we conclude our review with a section on Piaget's theory of cognitive development.

Memory
Table of Contents

Introduction In a very real sense, our memories determine who we are, what we do, and what we think. Whenever we maintain information over time, we are using our memories. Memory is therefore a critical feature of all cognitive processes. Traffic safety messages that cannot be remembered or recalled effectively when necessary will have little or no impact on a person's driving behaviors. It is, therefore, critical for readers interested in constructing traffic safety messages to understand how human memory develops and functions.

While there are many models of human memory, it is useful to conceptualize memory as composed of three stages (e.g., Atkinson & Shiffrin, 1968; Kintsch & Buschke, 1969; Klatzky, 1980): sensory memory⁽³⁾, short-term memory, and long-term memory. This model of memory, known as the Atkinson-Shiffrin model, has been challenged by some researchers who cite experiments that suggest that the distinctions between these memory stages are somewhat unclear (e.g., Baddeley, 1984; Craik & Levy, 1976; Wicklegren, 1973). Despite this lack of agreement, we use the Atkinson-Shiffrin model as an efficient organizational framework for describing the empirical data about memory development and function.

Short-Term Memory

Short-term memory (STM) has been described as working memory (Klatzky, 1980) because it is the type of memory used for ongoing cognitive activities. It has also been characterized as the conscious part of memory where activities such as decision making, reasoning, symbol manipulation, and problem solving take place (Siegler, 1991). Klatzky (1980) provides a useful analogy for STM:

"It may be helpful to think of STM as a workbench in a workroom where a carpenter is building a cabinet. All her materials are neatly organized on shelves around the walls of the room. Those materials that she is immediately working with--tools, boards... and so on--she brings from a shelf and places on the bench, leaving a space on the bench where she can work. When the bench gets too messy, she may stack material in orderly piles, so that more can be fit onto the bench." (Klatzky, 1980, pp. 88.)

In this analogy, the materials are bits of information and the materials stacked neatly on shelves are analogous to information in long-term memory to be discussed next. The analogy is useful because it describes many empirical properties of STM, such as a limited capacity and the ability to use organizational strategies to increase the capacity.

The first study to investigate the capacity of STM was conducted by the German psychologist Hermann Ebbinghaus (1885/1960). Ebbinghaus developed a procedure whereby he learned various length lists of "nonsense syllables," such as LAR or SIF. He learned each list by reading aloud to himself the list and then attempting to recall it in the order it was read. If he made a mistake, he would reread the list and again he would attempt to recall the list. This procedure continued until he could recall the entire list in the correct order. He tallied the number of times he had to reread the list. Ebbinghaus discovered that if the list contained seven nonsense syllables or less, he could recall the list perfectly after only one reading. This finding suggested that the capacity of STM was seven items.

Subsequent research on STM capacity with adults confirmed Ebbinghaus' finding and showed that the capacity was the same for many other nonrelated items, such as digits or letters (e.g., Howard, 1983; Pollack, 1953). However, if the digits formed familiar numbers or the letters formed familiar words, then many more digits or letters than seven could be recalled perfectly after a single reading. For example, suppose that the following list of 12 digits were read to you: 1, 8, 1, 2, 1, 7, 7, 6, 1, 9, 4, 2. Since there are 12 items in the list, you would not be able to recall them in order after a single reading--the number of digits exceeds STM capacity. However, if you were to notice that the digits formed three important years in American history, 1812, 1776, and 1942, you would easily be able to recall the list after one reading. As in the workbench analogy, if you can stack up information in organized piles, then more can fit on the workbench. In a seminal paper, Miller (1956) called this type of information organization "chunking" and showed that the STM capacity was 7 ± 2 chunks; that is, it ranges from five to nine meaningful chunks of information. Thus, provided information can be chunked, STM can hold a large amount of information.

The durability of information in STM has also been measured. Consider what strategy you might employ to remember a telephone number said aloud. Most people would repeat this number continuously until they dial the telephone. This strategy, called rehearsal, allows information to remain in STM for as long as rehearsal continues (Klatzky, 1980). (Rehearsal is also one process that helps to move information into the long-term memory store.) However if rehearsal is prevented, by asking the person to perform an intervening cognitive task like counting backwards, studies with adults show that information is lost completely within about 15-to-20 seconds (e.g., Brown, 1958; Peterson & Peterson, 1959).

Are there age-related differences in STM? Studies on STM capacity have shown that children can recall fewer symbols than adults (e.g., Chi, 1976; Dempster, 1981; Keating & Bobbitt, 1978). This result could indicate that up to about age 13 or 14, STM capacity increases (e.g., Pascual-Leone, 1989a). However, other researchers (e.g., Brainerd, 1983; Siegler, 1991) have suggested two alternative explanations for the results: 1) Adults know more about the world and may be able to use this information to chunk information more efficiently; and 2) Adults are more likely to have learned and know when to use strategies, such as rehearsal, to help maintain information in STM. In any case, the fact that adults can store more material in STM means that they are better equipped than children to understand and think about information.

There are also consistent age-related differences in the speed at which information is processed in STM, with processing speed increasing with age (e.g., Hale, 1990; Kail, 1986; 1988; Keating & Bobbitt, 1978). STM processing speed is typically studied using a paradigm developed by Sternberg (1966; 1969) called memory scanning. In this paradigm, the person is given a set of stimuli, such as the letters R, W, L, B, and S. The set usually contains fewer items than the capacity of STM. The person is then asked if a test stimulus, such as Y, is contained in the set of stimuli. The reaction time (RT) for a correct yes or no answer is measured. RT is taken as the amount of time for information to be processed in STM. Typically, as the number of items in the given set of stimuli increases, RT also increases by an equal ratio (Sternberg, 1966), suggesting that more processing time is required when more information is involved.

Keating and Bobbitt (1978) used the memory scanning procedure to investigate STM processing speed in 9, 13, and 17-year-olds. For stimulus set sizes of one, three, and five, they found average RTs to be the longest for the 9-year-olds, shorter for the 13-year-olds, and the shortest for the 17-year-olds. RT decreased by roughly one-third between each age-group. Thus, 17-year-olds processed information more than twice as fast as the 9-year-olds in this experiment. Similar results using different tasks have been obtained (e.g., Hale, 1990; Kail, 1986; 1988).

In summary, the research on short-term memory shows, among other things, that the speed at which information is processed and the amount of information that can be processed increases significantly with age among young people. Therefore, any cognitive process requiring short-term memory, such as decision making, reasoning, or understanding a traffic safety message will proceed at a slower rate for children and adolescents than for young adults. The amount of information that can be considered at the same time will also be less for children and adolescents. Consider the situation in which a driver is approaching a signalized intersection and the light changes to yellow. The driver is faced with a complex decision that requires rapid processing. The driver will have to consider the speed he or she is going, the distance to the intersection, the amount of time for the yellow cycle, the conditions of the roadway, the presence of other vehicles, and several other dimensions before deciding whether to brake or proceed into the intersection. If the driver is speeding, as young drivers often do, then the decision time can be quite short. The research we have discussed suggests that younger drivers, because of slower processing speed and lack of ability to consider several dimensions of a problem at the same time, are likely to inappropriately enter intersections on a yellow light. Further, the occurrence of this high risk driving behavior should be most frequent in young drivers who are speeding. A direct observation study of driver behavior at yellow lights supports this prediction. The study discovered that younger drivers violated a red light more often than older drivers, especially if they were traveling over the speed limit (Koneni, Ebbesen, & Koneni, 1976).

Long-Term Memory

Long-term memory (LTM) stores our experiences and knowledge. All that we know and have thought about is stored in LTM. It is believed that the capacity of LTM is unlimited for both adults and children--at least no study to date has been able to measure the capacity (e.g., Tulving, 1974). Adults, of course, differ from younger people in how much they have been able to store in LTM; that is, adults tend to be more experienced, especially in driving-related knowledge, than young adults or adolescents. A study by Chi and Koeske (1983), however, showed that expertise is not necessarily age-related. They showed that young children can develop expertise about something (i.e., in this case dinosaurs), that far surpasses the knowledge base for this topic in most adults.

As the name implies, the durability of information in LTM is quite long and can be impressively accurate (Shepard, 1967; Allen & Reber, 1980; Bahrick, Bahrick, & Wittlinger, 1975). In one study, Bahrick, et al. (1975) tested adult recognition of faces from high school year books 35 years later and found that people could accurately identify whether a face was in their year book 90 percent of the time! Since older people can remember events that took place when they were young, it is reasonable to conclude that the LTM durability does not change with age.

Despite such impressive capacity and durability of LTM, we are more likely to be aware of cases in which we cannot remember something. The fact that both adults and children sometimes forget things seems at odds with the large capacity of LTM. It is believed by some researchers that the cause of forgetting in LTM is not from loss of information but rather from an inability to retrieve the information (Tulving & Psotka, 1971; Ratcliff & McKoon, 1989). Good illustrations of this inability are instances in which a person feels like he or she knows a certain fact but cannot quite remember it (e.g., Nelson & Narens, 1980). This experience has been called the tip-of-the-tongue (TOT) phenomenon. Consider a study in which adults were read dictionary definitions of infrequently used words, such as "An instrument used by navigators for measuring the angular distance of a star from the horizon." (Brown & McNeill, 1966). If the persons could not recall the defined word, they were asked to report how many syllables were in the word, what letter it began with, or to name words that it rhymed with. People were quite accurate in reporting the features of the word, even though they could not recall it. In our example, the defined word is "sextant." The TOT phenomenon illustrates the effortful nature of retrieval from LTM and shows that LTM contains information of which we must not be consciously aware.

Experiences such as the TOT phenomenon also help to illustrate the different ways of retrieving information from LTM. If the adult in the example above was given a list of words that contained the word sextant, it is likely that he or she would have been able to easily pick it out of the list. That is, it is easier to recognize something than it is to recall it. Because of this distinction, psychologists typically distinguish between recognition and recall when investigating LTM retrieval (e.g., Anderson & Bower, 1972; Klatzky, 1980). This distinction is important when considering age-related differences in LTM retrieval.

As with adults, several studies have shown that children's recognition memory is quite impressive (Brown & Campione, 1972; Brown & Scott, 1971; Daehler & Bukatko, 1977) . For example, Brown and Scott (1971) showed 4 and 5-year-old girls and boys 32 pictures cut out of a children's book. Some of the pictures were shown to them once and some were shown twice. Recognition of the pictures as long as 28 days later was tested by individually showing the children the 32 pictures mixed in with an additional 12 filler pictures. For each picture, the children reported whether or not they had seen it before. The results showed that the children tested after 7 days were about 94 percent accurate in their recognition of pictures seen twice, and children tested after 28 days were about 75 percent accurate for the same pictures. Pictures seen only once were recognized correctly at a lower rate. No sex differences were reported. This impressive recognition accuracy is comparable to similar studies conducted with adults (e.g., Shepard, 1967). Further, studies that have compared adults with children on recognition memory of realistic pictures, abstract pictures, and abstract forms have shown no reliable recognition differences between adults and children (Nelson, 1971; Nelson & Kosslyn, 1976).

It is clear that even children have impressive recognition memory. However, as discussed by Kail (1986), the typical stimuli used in developmental studies of recognition memory are pictures of single objects. Kail further pointed out that children's recognition often involves much more complexity; that is, scenes that contain multiple objects. Using a paradigm similar to the one utilized by Brown and Scott (1971), several studies have shown that recognition accuracy for scenes containing multiple objects increases with age (e.g., Hock, Romanski, Galie, & Williams, 1978; Mandler & Robinson, 1978; Newcombe, Rogoff, & Kagen, 1977). For example, Newcombe, et al. (1977) tested recognition accuracy of 6-year-olds, 9-year-olds, and adults for several pictures of scenes, each of which contained multiple objects arranged in a natural setting. When tested after 5 days, recognition accuracy for the scenes was about 50 percent for the 6-year-olds, 80 percent for the 9-year-olds, and 90 percent for the adults. Collectively, this literature shows that retrieval of information from LTM is identical for adults and children when the to-be-remembered item is shown to the person (recognition) and is not too complex, such as a single object or simple scene. However, if the item is complex, such as in a real-world scene, recognition accuracy continues to improve up through adulthood.

Several authors have cited inexperience as a contributing factor to the elevated crash rates of young drivers (e.g., Catchpole, Cairney, & Macdonald, 1994; Eby, 1995b; Hodgdon, Bragg, & Finn, 1981; McPherson, McKnight, & Weidman, 1983; Pelz & Schulman, 1971). In terms of LTM processes, inexperience is related to a lack of knowledge about factors such as driving situations, vehicle handling, problem solving, and decision making strategies. Clearly, a major goal of driver education and graduated licensing programs is to allow young drivers to develop their expertise with driving while minimizing their crash risk.

Attention
Table of Contents

Introduction

The word attention is used frequently in everyday language and, depending upon its usage, has several meanings. Cognitive psychologists, however, define attention as a process of concentrating or focusing of limited cognitive resources to facilitate perception or mental activity (e.g., Anderson, 1985; Bernstein, Roy, Srull, & Wickens, 1991; Broadbent, 1958; Kahneman, 1973; Matlin, 1989). Attention is a conscious process; that is, it is usually under voluntary control. For example, you have focused your attention on these words in order to be able to read them and, if your radio was on, you could shift your attention there in order to understand the radio announcer. Thus, attention is a process that is necessary for information processing--the information will get into memory only if it has been attended to.

For a traffic safety message or program to be effective, it must attract the attention of the target persons. The most important factor determining what people focus on is their level of interest in the information (e.g., Miller, 1982; Miller & Shannon, 1984). Thus, it is important to know what will attract the attention of the target audience and adjust the message format and content to utilize these interests.

In order to perceive, interpret, and understand a message, such as a traffic safety message, children or adults must be able to shift their attention to the appropriate message while filtering out other stimuli (called selective attention). Further, they might devote only part of their attention to the message (called divided attention). They must also be able to focus their attention long enough to receive the entire message (called sustained attention). A lack of capacity in any of these three attentional processes could lead to either a misunderstanding of traffic safety messages or a complete failure to receive them.

Selective Attention

Whenever we are awake, we are bombarded constantly with sensations. In order to make sense of this chaos, we have to determine what is most important to us and focus our attention on it. Parasuraman (1986) defined selective attention as, "a process in which the observer attempts to attend selectively to some stimuli, or some aspect of stimuli, or to some task, in preference to other stimuli or tasks." (pp. 43-2). Thus, selective attention is the process that determines which information sources we will consider.

Numerous studies have shown that adults are better able to ignore irrelevant information than children (see Dempster, 1993; Lane & Pearson, 1982; and Miller, 1990 for reviews). One of the most frequently studied phenomena in experimental psychology, the Stroop test, is a test of selective attention (Stroop, 1935). In this test, a literate person is required to name the color of ink used to print an incongruent color-word. For example, the person might be shown the word GREEN printed in red ink. In this example, the task would be to say "red" as quickly as possible, ignoring the meaning of the printed word. The time required to respond is typically compared to the time required to name a nontext color chip, with the difference in response time taken as the amount of attentional interference produced by the text. Response times on the Stroop test show clear age differences.

In one of the most thorough investigations of response times in the Stroop test by age, Comali, et al. (1962) found that performance on the Stroop test was strongly related to age. As shown in Figure 3, Comali, et al. found that the difference between response times in the Stroop task and naming color patches was greatest for 7-year-olds (the youngest age tested) and decreased fairly consistently up to about age 18. The differences remained fairly constant from 18 years of age to the middle age group (35 to 44 years of age) and increased significantly for the people in the 65-to-80-year-old age group. The effect of sex was not studied. Because greater response time differences are interpreted as a person having greater difficulty attending selectively to a relevant stimulus, these results show that the ability to selectively attend to important information develops up to about age 18, where it remains constant for the majority of the life span. Sometime after 44 years of age, selective attention ability declines appreciably.

Figure 3: Response time on the Stroop test by age (Comali, et al., 1962).

Converging evidence for the increasing selective attention ability with age is also found in selective-listening tasks (Cherry, 1953). In this approach, selective attention is investigated by having a person wear stereo earphones in which different messages are played in the left and right ears. The person is typically asked to listen to only one of the messages while ignoring the other. Selective attention ability is measured by determining the accuracy with which the person can report information about the to-be-attended-to message; in other words, how well the person can ignore the irrelevant message.

Many studies have shown that selective-listening ability increases with age (see, e.g., Doyle, 1973; Geffen & Wale, 1979; Geffen & Sexton, 1978; Hiscock & Kinsbourne, 1977, 1978; Maccoby & Konrad, 1966, 1967; Pearson & Lane, 1991; Sexton & Geffen, 1979). For example, Sexton and Geffen (1979, experiment 2) studied 7, 11 and 20-year-olds in a selective-listening task. In this study, the subject wore earphones that played different recorded lists of words to each ear. The subject was told to ignore the list in one ear and to listen for a target-word in the other ear. When the subject heard the target-word in the proper ear, he or she pressed a button. Subjects were scored on the accuracy of identifying the target word, with higher accuracy being attributed to better selective attention ability. The researchers found that accuracy increased monotonically with age with no difference between sexes. Thus, the ability to ignore irrelevant auditory messages and to focus on a relevant message, such as a traffic safety message, improves at least up to about 20 years of age.

In addition to the Stroop test and selective-listening tasks, selective attention ability has been shown to improve with age in the processing of pictures (e.g., Day & Stone, 1980), the classification of three-dimensional objects (e.g., Pick, Christy, & Frankel, 1972) and stick figures (e.g., Smith, Kemler, & Aronfreed, 1975), the ability to sort cards based upon the characteristics of the cards (e.g., Strutt, Anderson, & Well, 1975), and exploratory searching tasks (Miller & Seier, 1994). Collectively, these studies show that the ability to focus attention on relevant and important information, and thereby to be most prepared to receive and process that information, is poor at 7 years of age but develops markedly during the next 10 to 15 years.

Divided Attention

A topic closely related to selective attention is divided attention. In tasks requiring selective attention, the person attempts to ignore irrelevant stimuli while focusing on a relevant stimulus. In a divided attention task, a person attempts to focus attention on more than one stimulus. Bernstein, et al. (1991) have defined divided attention as, "...devoting psychological resources to more than one task or stimulus at a time." (pp. 202). In nearly all driving situations, drivers must divide their attention among several tasks.

Divided attention is studied by having a person attempt to perform two tasks at once or to attend to two stimuli at once. One of the most frequently cited studies on this topic found that people generally are not very good at dividing attention (Neisser & Becklin, 1975). The method typically used to present two stimuli to a person is a variation of the selective-listening task described earlier. In this variation, the person is instructed to listen to the messages in both ears (called a dichotic-listening task) and produce a response when he or she hears a target word or event in either ear. In order for the researcher to determine which ear the subject is responding to, the subject is typically asked to press a button with the right hand if the target word is heard in the right ear, press a button with the left hand if it is heard in the left ear, and press buttons with both hands if the target word is heard simultaneously in both ears.

Using a dichotic-listening task, Sexton and Geffen (1979, experiment 3) investigated divided attention abilities of 7, 11, and 20-year-olds. They found several interesting results. First, divided attention ability was poor, with all people identifying accurately only about 55 percent of the target words. Second, accuracy increased significantly between 7 and 11 years of age. There was no consistent difference in accuracy between 11 and 20 years of age. Finally, there were no consistent effects of sex on dichotic-listening accuracy. Thus divided attention ability seems to peak quickly (by 11 years of age), is generally poor, and does not seem to be affected by the sex of the person. Additional support for these conclusions comes from a variety of studies using several divided attention tasks (e.g., Hiscock & Kinsbourne, 1978; Pearson & Lane, 1991; Schiff & Knopf, 1985).

Sustained Attention

Once the most important stimulus has been selectively attended to, we must also be able to maintain our focus of attention on the stimulus in order to effectively process it. Parasuraman (1986) defined sustained attention as, "a process of maintaining attention to a critical stimulus or aspect of a stimulus for a sustained period of time." (pp. 43-3). The duration of sustained attention is sometimes called the attention span (e.g., Bjorklund, 1995). However, this phrase is misleading because it suggests that a time value (e.g., 5 minutes) can be assigned to the duration of sustained attention. This is not the case. The typical result in the literature is that performance on a task requiring sustained attention declines gradually over time (e.g., Parasuraman, 1986); that is, the person makes more errors, takes longer to respond, or, by self-report, has more difficulty with the task as duration increases. The rate of decline in performance is closely related to the characteristics of the task, with interesting tasks showing smaller performance decrement rates over time (see Parasuraman, 1986, for an extensive review).

The fact that both selective and divided attention abilities develop with age, suggests that sustained attention would follow a similar developmental trend. Several studies have shown that the ability for sustained attention does indeed increase with age Crow & Crow, 1963; Gutteridge, 1935; Murphy-Berman, Rosell, & Wright, 1986; Mussen, Conger, & Kagan, 1974; Tyler, Foy, & Hutt, 1979; van Alstyne, 1932). In particular, Murphy-Berman, et al. (1986) investigated sustained attention in children 7-to-16 years of age. Using a microcomputer to display simple drawings, participants were required to watch for a target picture that appeared in a certain part of the display screen. When the target picture appeared, they pressed a button as quickly as they could. Because the presentation rate of the pictures was adjusted for each child, a testing session lasted from 20-to-30 minutes. The results showed that sustained attention abilities generally improved between 7 and about 12 years of age, leveled off until 14 or 15 years of age, and then improved significantly for the 16-year-old subjects. There was a general sex trend, with females showing slightly better sustained attention ability across all ages, but it was not statistically significant. Thus, it appears that sustained attention ability does follow developmental trends that are similar to selective and divided attention trends.

Learning
table of contents

Introduction

Learning has been defined as any relatively permanent change in behavior or thinking that results from past experiences (e.g., Bernstein, Roy, Srull, & Wickens, 1991). This rather broad definition accounts for the simplest learning such as habituating to a stimulus, as well as complex learning such as mastering a musical instrument or driving a car. The ability to learn, of course, is present at birth. Studies have shown that neonates can imitate the tongue protrusion of an adult and possibly other adult gestures (e.g., Anisfeld, 1991; Meltzoff & Moore, 1989). Other studies show that neonates habituate to visual and auditory stimuli (e.g., Bridger, 1961; Cole & Cole, 1989; Kellman & Spelke, 1983) and can recognize their mother's voice (e.g., DeCasper & Fifer, 1980).

Despite the innate presence of learning processes, learning is an integral component of cognitive development. Further, an understanding of learning processes is important for those interested in developing messages and programs that attempt to improve safe driving practices. Most traffic safety messages are designed to either change how people think about a traffic safety issue or to change people's safety behaviors. In other words, they are designed to educate people. Therefore, a comprehensive understanding of learning processes is central in the development of effective traffic safety messages and programs. Three learning processes are particularly relevant: classical conditioning, operant conditioning, and observational learning (Leahey & Harris, 1997).

Classical Conditioning

One basic learning process is the ability to identify relationships or make associations between events; for example, blue skies result in dry weather, or driving immediately before 8:00 am leads to sitting in rush hour traffic. The simplest kind of associative learning, classical conditioning, involves the association between reflexive responses (such as many emotional responses) and a stimulus (such as food or a person).

The scientific investigation of classical conditioning began with the now famous studies of the Russian physiologist Ivan Pavlov (1927). In his Nobel Prize winning studies of canine digestive physiology, Pavlov noticed that the dogs he was studying would salivate initially only when food was presented. However, over time, the dogs would also salivate at the sight of the previously neutral experimenter who fed them. Pavlov recognized that a reflexive response (salivating in the presence of food) could be associated with nonfood stimuli, such as the sight of an experimenter or the sound of a bell.

Around the same time as Pavlov's work, a pair of American psychologists demonstrated classical conditioning in humans (Watson & Rayner, 1920). Watson and Rayner were interested in the conditioning of emotional responses. They chose to study how fear toward a previously neutral object might develop. They selected as their subject a 9-month-old male called "Albert B." or as he later became known, Little Albert. Watson and Rayner first determined a set of objects toward which Little Albert showed no fear, including a rat and several other small furry objects. They then discovered a stimulus that caused a fear reaction in their young subject--striking a metal bar with a hammer. This stimulus startled Little Albert and then produced crying. During conditioning, Albert was given a tame rat toward which he had previously shown no fear. When he first touched the rat, Watson and Rayner struck the metal bar with the hammer. This process was repeated for 7 days. Afterward, Albert was presented with the rat and no metal bar was struck. Albert showed a fear response towards the rat. The previously neutral object now produced fear for Little Albert. Further, Watson and Rayner showed that this fear also generalized to other furry objects such as a rabbit, a fur coat, a dog, a Santa Claus mask, and cotton balls. Thus, after only 7 days of fear conditioning, Little Albert now showed fear toward a wide range of objects. When Albert was tested 1 month later, his fear responses had been lessened but were still present.

Since the work of Pavlov (1927) and Watson and Rayner (1920), the properties of classical conditioning have been firmly established. Before discussing these properties, four classical conditioning terms need to be defined (Leahey & Harris, 1997). The stimulus that produces a response without any learning required (e.g., the sound of a hammer striking a metal bar) is called the unconditioned stimulus (US), while the reflexive response (e.g., fear) is called the unconditioned response (UR). The stimulus that, through learning (conditioning), leads to a response that is like the UR is called the conditioned stimulus (CS; e.g., the rat) and the response is called a conditioned response (CR). Three factors can affect the probability of a conditioned association being acquired:

1) The probability of a CS producing a CR increases with the number of pairings of the CS and US (Pavlov, 1927; Watson & Rayner, 1920). In other words, the more times the loud noise was paired with the rat, the more likely it was that Little Albert would associate a fear response with the rat.

2) The probability of a CS producing a CR increases with decreases in the temporal interval between presentation of CS and the US (e.g., Pavlov, 1927; Ross & Ross, 1971). In other words, the conditioned fear response toward the rat (the CS) was more likely to occur if the interval between presentation of the rat to Little Albert and hitting of the metal bar with the hammer was short. Studies have shown that the optimal interval between the CS and US is .5-to-1 second (Ross & Ross, 1971).

3) The probability of a CS producing a CR increases with the intensity of the UR. That is, a highly intense unconditioned response, such as the pain from an injection, can lead to a CS-CR association even after a single pairing.

Fortunately for Little Albert, once the CS-CR association has been established, it will not remain forever. With repeated exposure to the CS without the US, the CS-CR association will eventually disappear or become extinct. In other words, if Little Albert were to be continually presented with the rat but no loud noise, he would eventually lose his fear of the rat. The likelihood of this extinction increases with the number of exposures to the rat without the loud noise. Little Albert's fear response, however, was not experimentally extinguished (Harris, 1979).

Operant (Instrumental) Conditioning

Classical conditioning suggests a way in which two stimuli that occur together can become associated. But, clearly, people can also learn by doing something and seeing what happens. For example, children learn to say "please" in order to get something that they want. Because in this situation the individuals are operating on their environment and are instrumental in producing some outcome, this type of learning has been called both instrumental (Thorndike, 1898) and operant (Skinner, 1938) conditioning. It differs from classical conditioning in that people have control over events.

In operant conditioning, an action occurs that is followed by some outcome. If the outcome is positive, then the action is likely to be repeated. Psychologists call this type of outcome a reinforcer. An outcome can be reinforcing if it is pleasant (positive reinforcer) or if it removes something that is unpleasant (negative reinforcer). Outcomes can also be punishing. If the outcome is punishing, the action that led to it will become less likely to be repeated. Thus, through both reinforcement and punishment, new behaviors are learned and others are extinguished. Several factors affect whether and how a behavior is operantly conditioned.

1) The effectiveness of the reinforcer or punishment to change behavior is decreased as the amount of time between the behavior and outcome increases (e.g., Kalish, 1981). For example, many behaviors that people typically engage in have punishing consequences, such as drinking too much alcohol or not wearing a safety belt. One reason these behaviors continue is that the negative consequences are often delayed too much to be effective in changing the behavior. There is some evidence that the incidence of drunk driving may be reduced by decreasing the time between a drunk driving arrest and adjudication of the case (Streff & Eby, 1994).

2) The effectiveness of the reinforcer or punisher to change behavior increases with the magnitude of the reinforcer or punisher (e.g., Holmes & Robbins, 1987). For example, a crash in which a person was seriously injured because of a lack of safety belt use will be more effective in producing later belt use than a crash in which the person merely receives a cut or bruise.

3) A behavior does not have to be reinforced every time it occurs in order for that behavior to be conditioned (e.g., Skinner, 1961). The reinforcer may be presented on a fixed schedule, say, for example, after every third time a child raises his or her hand before speaking in class. The reinforcer may also be presented on a variable schedule, say after the fourth, tenth, and nineteenth time the child raises his or her hand before speaking in class. Both schedules of reinforcement have been shown to produce operant behaviors. At least one study has shown the effectiveness of a variable reinforcement schedule in increasing safety belt use. In this study, police randomly pulled over drivers and rewarded them with prizes for wearing safety belts (Mortimer, Goldsteen, Armstrong, & Macrina, 1988).

As in classical conditioning, learned behaviors are not necessarily permanent. In operant conditioning, a conditioned behavior will stop if it continues without reinforcement. The time required for this extinction is dependent upon the reinforcement schedule used to produce the behavior. If the behavior was reinforced every time it occurred (the quickest way to condition the behavior), then it will stop quickly. If the behavior was conditioned on a fixed schedule, then it will stop less rapidly. If, however, the behavior was conditioned on a variable schedule, it will stop the least rapidly.

Numerous traffic safety programs have utilized operant conditioning paradigms to increase positive traffic safety behaviors or decrease negative ones with varying levels of success (e.g., Marchetti, Hall, Hunter, & Stewart, 1992; Mortimer, Goldsteen, Armstrong, & Macrina, 1988; Roberts & Fanurik, 1986; Wilde, 1985; Wodarski, 1987). For example, Roberts and Fanurik (1986) showed impressive increases in elementary school children's belt use following a program in which belted children received tokens they could redeem for toys. In this evaluation, belt use increased over 50 percentage points immediately after the program and then gradually declined over the following 2 months. The use rate, however, stabilized at a level that was higher than the preprogram level. These findings are in agreement with the principles of operant conditioning. Had the token been given out on a variable schedule (e.g., 2 randomly selected days each week) rather than the fixed schedule that was used, then the increased belt use would have, most likely, lasted for a much longer period of time.

Traffic law enforcement programs work on the principle of using punishment, or the threat of punishment, to reduce the likelihood of unsafe driving practices. The effectiveness of such programs has been shown in studies of drinking and driving (e.g., Kinkade & Leone, 1992; Streff & Eby, 1994), safety belt use (e.g., Eby & Christoff, 1996; Jonah, Dawson, & Smith, 1982; Ulmer, Preusser, & Preusser, 1994); bicycle helmet use (e.g., Cameron, Vulcan, Finch, & Newstead, 1994; Coté, et al., 1992; Healy & Maisey, 1992; Macknin & Medendorp, 1994), and motorcycle helmet use (e.g., Chinier & Evans, 1987; Lund, Williams, & Womack, 1991; Streff, Eby, Molnar, Joksch, & Wallace, 1993).

Observational Learning

Much of human learning conforms to the principles of both classical and operant conditioning, which require direct experience of either a stimulus pairing or an action and its consequences. However, it is clear that some learning does not follow conditioning principles or require direct experience. All of us do not need to receive head injuries or be thrown in jail to know that we should wear a helmet when cycling, and should not drive while drunk. Humans can benefit from the experiences of others in order to learn behaviors and their consequences. Such learning is called observational learning or vicarious conditioning (Bandura, 1965).

In a series of experiments, Bandura (1965) convincingly demonstrated observational learning in nursery school children. Using an equal number of boys and girls, Bandura showed children a film of an adult interacting aggressively with an inflatable, adult-sized punching bag doll called a "Bobo" doll. All children saw the adult punch, kick, and throw objects at the doll and then strike it with a hammer. One-third of the children then saw the adult get punished for his or her behavior, one-third saw the adult get rewarded, and the remaining one-third saw no consequences. Individually, children were then placed alone in the room with the Bobo doll and their behavior was observed. Bandura found that those children who watched the adult get rewarded for aggressiveness initiated more imitative acts on the Bobo doll than those who saw the adult receive a punishment or no consequences at all. Overall, boys imitated more aggressive acts than girls. Bandura then rewarded all children for imitating the adult and found no differences between groups, except that boys generally imitated more aggressive acts than girls.

These results show two important findings. First, observing the consequences of someone else's behaviors can affect the behaviors a child chooses at a later date. Behaviors that were observed being punished were less likely to be imitated by children. Second, even though the behaviors were not imitated, when prompted and rewarded, all children showed that they had learned the behavior, even if they saw it being punished.

In several studies, Bandura (1977, 1986, 1989) has identified four factors that influence observational learning:

1) Attention. Observational learning will not occur unless the person is paying reasonably close attention to the person or people performing the behavior.

2) Memory. Observational learning will not occur unless the person can remember the actions and consequences at a later time.

3) Ability to Reproduce the Action. Observational learning will not occur if the person cannot reproduce the action (the reason why we all cannot play violin after watching Itzhak Perlman perform).

4) Motivation. Observational learning will not occur unless the person has some reason for performing the behavior. (See the chapter on motivation for further information on this topic.)

Undoubtedly, many driving behaviors, both good ones and bad ones, are learned through observation. DiBlasio (1986), for example, has reported that driving while drunk and riding with a drunk driver are partly acquired through a person's association with his or her peer group. He also mentions that observation of family models (children's parents, stepparents, and guardians) engaging in drunk driving behaviors is positively correlated with similar behaviors in children. Carlson and Klein (1970) compared the traffic conviction histories of men and their sons and found a positive correlation. The authors concluded that a greater proportion of driving behavior is learned through familial contact than through institutional contact (e.g., driver training). The principles of observational learning should be considered when developing traffic safety programs and messages.

Reasoning
table of contents

Introduction

A prominent feature of human cognition is our ability to draw conclusions based upon the things that we have learned. This ability is so commonplace in everyday cognition that people frequently do not even notice when they are engaging in it. For example, if a friend were to introduce you to a young woman and an old man as his sister and father, respectively, you might conclude that the old man is also the woman's father. You would not have been told this, but you might reason that siblings have the same father, so the woman's father must be the old man. In most cases you would be correct, unless your friend and the woman are step-siblings. The process by which people draw conclusions from their knowledge of the world is called reasoning (e.g., Evans, Newstead, & Byrne, 1993; Garnham & Oakhill, 1994; Halpern, 1989; Johnson-Laird & Byrne, 1991). When people reason, they are generating a belief (i.e., conclusion) that has been inferred from what they know. Thus, reasoning is intimately related to most other cognitive abilities including learning, moral development, verbal ability, memory, attitude formation, and problem solving.

Many traffic safety messages and programs use logic to both teach and convince drivers to drive safely. Further, much of what people learn about the driving task and driving situations is based upon reasoning processes. Therefore, a thorough understanding of how reasoning develops and the problems people have with reasoning is necessary for the development of effective traffic safety messages.

The study of childhood and adult reasoning usually makes a distinction between two types of reasoning processes: deductive and inductive reasoning. In deductive reasoning, a person begins with information that is known to be true (called premises) and draws a conclusion based upon this information. If the information is true and the person follows the rules of logic, then the conclusion is both truthful and valid (e.g., Evans, Newstead, & Byrne, 1993; Halpern, 1989; Johnson-Laird & Byrne, 1991). Thus, in deductive reasoning, the conclusion must necessarily follow from the premises. In inductive reasoning, on the other hand, a person starts with information that may or may not be correct and then draws a conclusion (such as an hypothesis). The person then collects information to support or refute that conclusion (e.g., Halpern, 1989; Holland, Holyoak, Nisbett, & Thagard, 1986). For example, a young driver might conclude that safety belt use causes injuries rather than prevents them. This young driver might then talk with friends, read the newspaper, or watch the television to learn about safety belt use and injury outcomes in crashes. If the rules of logic are followed by the young driver, a single instance in which a safety belt prevents injury during a crash should cause him or her to modify or abandon the falsified belief.

Deductive Reasoning

Research has shown that children, even those as young as preschool age, can draw deductively valid conclusions (e.g., Dias & Harris, 1988, 1990; Hawkins, Pea, Glick, & Scribner, 1984). However, other studies have found that both adults (e.g., Evans, Newstead, & Byrne, 1993; Johnson-Laird & Byrne, 1991; Newstead & Evans, 1995; Wason & Johnson-Laird, 1972; Woodworth & Sells, 1935) and children (e.g., Galotti & Komatsu, 1989; Kuhn, 1989; Markovits, 1993; Markovits, Schle