Environmental Cognition Research Paper

At the present time, environmental protection enjoys broad support. It is expressed in public opinion polls (e.g., Dunlap et al. 2000), in the popularity of nature programming on television, and in the growth of ‘socially responsible’ mutual funds, which control 13 percent of the US market and typically screen firms for their environmental behavior. However, the fate of the environment depends on the actual support (or opposition or indifference) emerging in specific decisions. These choices are enormously varied, in their options, consequences, and decision-makers—not to mention their national, social, cultural, institutional, and economic settings.

The diversity and complexity of these decisions poses a challenge to researchers. One response is performing case studies, fully respecting the particularities of each situation. Environmental studies scholars have produced many such accounts (Slovic and Dixon 1992). Their observations shape and constrain the theories of environmental scientists, seeking general, testable theories about environmental behavior. These scientists have typically adopted one of the two classic compromises for producing general accounts of heterogeneous phenomena. (a) Holistic approaches look for recurrent behavioral archetypes, groups of people who respond in predictable, but different ways (e.g., Cotgrove 1979, Douglas and Wildavsky 1982). For example, one might posit ‘ecofreaks’ who resolutely oppose any intrusion into pristine areas or ‘technocrats’ who automatically dismiss such concerns. (b) Reductionist approaches assume that all individuals respond to common factors, but interpret and weight them differently— sometimes leading to different actions (Stern 1992). In them, eco-freaks and technocrats (and everyone else) would attend to similar outcomes and processes. However, they might view them very differently, in terms of their attitudes and beliefs. Both kinds of account coexist in dual-process theories of attitude and cognition. In them, holistic first impressions create an orientation toward a stimulus (e.g., an environ mental cause or insult). A parallel process involves analytical thinking, elaborating that impression. The result might be interpreting it in terms of more detailed (holistic) archetypes or (reductionist) cross-cutting concerns. Such analysis can overturn first impressions. However, it can also be constrained by them. Many common cognitive processes encourage the perseveration of beliefs. For example, people tend to seek confirmatory, rather than contradictory evidence; to resolve ambiguity in favor of current beliefs; to exaggerate the definitiveness of small samples of evidence (Gilovich et al. in press). How hard people work to understand an issue should depend on how important it seems, what returns they expect from their efforts to learn, and how great is their general need for cognition. Those factors, in turn, reflect and shape the opportunities that they have to affect their environment.

1. Holistic Accounts

A general account of environmental behavior is only useful if researchers can unambiguously translate complex real-world situations into its necessarily abstract terms. For holistic theories, that translation process may be holistic itself, attempting to identify specific cases with theoretical archetypes. Holistic interpretation allows astute observers to use the full set of available data. Moreover, if those observers write well, the coherence of their accounts should help readers to retain and integrate the information that is presented. Archetypes are a natural way to organize evidence, even if that sometimes leads to neglecting discordant details.

Unfortunately, holistic interpretation also allows motivated observers to exploit the richness of the evidence, in order to fashion desired accounts. For example, they might exploit the privileged position of hindsight, so as to craft interpretations that affirm their theories. Or, they might strategically identify other people with positive or negative archetypes, in order to advance a political agenda. The very act of categorizing others is an exercise of power. It can reduce those individuals to stereotypes and present the observer as possessing superior insight (and perhaps even morals). It can feed prejudices and benefit from the natural tendency to create simplistic views of other people.

These possibilities are not lost on the partisans in environmental conflicts, who value the opportunity to demean their opponents and flatter their allies. As a result, holistic characterizations are often offered and evaluated by individuals without scientific credentials. Conversely, scientists can have a political following because of the conclusions they reach, independent of the quality of their work. Pundits can drape themselves in the trappings of academia, seeking its rights, but not responsibilities, in order to achieve political ends. The narrative character of holistic accounts means that both professional and amateur ones have the same outward form. Commentators may have insights to offer. However, they are in a different business, without the scholarly constraints of sharing sources and reflecting previous research. Thus, readers need to check the scientific pedigree of environmental narratives. What training did the authors have? What peer-review screens did their work pass?

With ongoing environmental issues, more systematic measurement of holistic archetypes might be possible. If the archetypes are clearly defined, it should be possible to operationalize them in an individual difference measure (e.g., Dake 1992). If these measures pass conventional psychometric criteria (e.g., reliability, construct validity), then they allow objective tests of their theory (e.g., ‘ ‘‘corporate greens’’ have the deepest commitment to consumption-related issues’; ‘environmental pessimists have the greatest scientific knowledge and institutional distrust’). Of course, this strategy also requires similarly systematic characterization of the other elements of the prediction: what the environmental issues are, how those individuals perceive them, what their observable behavioral responses are, what meaning they attach to those actions, and so on.

If archetypes cannot be measured authoritatively, then individuals may be characterized by more observable categories, such as gender, race, income, or nationality. The price paid for the precision of such characterizations is losing the theoretical force of a well-crafted archetype. For example, surveys in the US have found particularly high environmental concern among African–Americans (Vaughan 1993). Speculatively, that concern could reflect greater direct exposure to environmental problems (so that they have more reasons for concern), greater awareness of their immediate surroundings, greater physiological sensitivity (e.g., because of multiple exposures or weakened immune systems), greater suspicion of societal institutions, fewer resources for self-protection, and so on. An analogous array of hypotheses follows from Slovic’s (1999) finding of similar attitudes toward the environment across most US populations. The one exception was white males who, on average, were more favorable toward technology and less concerned about the environment. Characterizing the subset of white males responsible for this overall group difference is a starting point for a theory of archetypes, sorting out the experiences or circumstances creating these distinctive views.

1.1 Reductionist Approaches

Comparable measurement issues confront reductionist approaches, looking at how people respond to a common set of concerns, as expressed in different environmental issues. In these models, individuals evaluate an action in terms of its expected effects on the probability of occurrence for each of a set of consequences. Those consequences might be experienced personally (e.g., changes in taxes, wilderness opportunities, social standing), socially (e.g., changes in health, justice, property rights), or environmentally (e.g., changes in biodiversity, the introduction of novel life forms). An action is attractive to the extent that it increases the chances of good things happening and decreases the chances of bad ones. These models belong to the general class of rational-actor models. They preserve the basic logic of orderly decision-making found in neoclassical economics, but vary in how strongly they endorse its normative and optimality assumptions.

The theoretical challenge for researchers is identifying a concise set of consequences, with important impacts in multiple situations. If that set exists, then general statements are possible, regarding ‘what matters to people, when it comes to the environment.’ Although an appealing aspiration, such sweeping summaries are limited by the actual relevance of the issues. For example, social norms may loom very large for curbside recycling, where all one’s neighbors can see what one does. Those norms may have little effect on roadside littering when no one is looking—unless the norms have been internalized as a personal value or the action (not littering) has become a matter of habit.

Thus, differences in the importance of issues can reflect (earnestly considered) differences in situations, and not just (hypocritical) situational ethics. As a result, some of the most successful attempts to predict behavior from beliefs and attitudes have focused on specific domains, such as home energy conservation (Stern 1992). These studies find that conservation increases when people see greater benefits and fewer barriers. These patterns can be sought in everyday circumstances or in the context of interventions intended to encourage such behaviors (e.g., recycling or political campaigns). Good communication should improve citizens’ understanding of the issues, making their actions more consistent with their values (and more predictable).

Like holistic approaches, these models face the challenge of assessing beliefs and values, for the people whose behavior they hope to predict. Even a valid model can fail empirical tests, if its terms are measured poorly. One form of protection for reductionist models is provided by their additive structure: they sum the impacts of favorable and unfavorable factors, allowing good and bad expectations to cancel one another. Such models are relatively insensitive to errors in measurement and specification (Dawes 1988). As a result, they will have some predictive validity as long as they include rough approximations of the relevant factors—or factors correlated with them. Unfortunately, such robustness inhibits theoretical development. It is hard to interpret the weights estimated for the factors in the model or to know whether they are actual drivers of behavior or just surrogates. An important exception to this rule arises when one consideration overrides all others, as something that must or must not be done. Such a decision model is noncompensatory, in the sense that there is no way to overcome the stigma associated with an action—or to diminish the moral high ground that it confers (Flynn et al. 2001).

2. Perceived Risks And Benefits Of Environmental Changes

Assessing beliefs and values is a central activity for cognitive and social psychology. Assessing them for environmental issues has posed significant challenges to the state of those sciences. Environmental changes can involve novel, unintuitive processes, surrounded by uncertainty and couched in unfamiliar terminology. They may extend over time, individuals, and species, in ways that are hard to grasp, much less evaluate. They pose choices as esoteric and fateful as those arising from modern medicine (e.g., advanced directives regarding life support, genetic screening, rationing scarce resources among competing needs). The environment suffers directly when people make poor choices. It suffers indirectly when investigators mismeasure what people want regarding the environment—hence send inaccurate signals to policy makers and citizens. These practical concerns have encouraged environmental social scientists to break new ground in these traditional assessment activities.

The difficulties that face investigators, in posing these questions, echo the difficulties that face individuals, when life poses them. As a result, the research agenda has combined descriptive and prescriptive goals, trying to understand current behavior in order to facilitate better informed behavior in the future. Those improvements are sought both in beliefs and values. People need to understand how the environment works and how changes can affect their lives (health, emotions, children, etc.). They need to understand themselves well enough to know what values to assign to these impacts.

As a result of these applied concerns, studies of beliefs often have an evaluative framing. They ask, not just what do people know, but whether people know enough to make effective decisions. When people face specific decisions, the adequacy of their knowledge can be evaluated relatively precisely—especially compared to general tests of environmental knowledge, which generally have no analytical foundation for selecting items. A decision-making perspective allows specifying what people need to know, and not just what it would be nice to know.

In the context of values, the evaluative framing asks whether people have succeeded in constructing stable opinions about specific environmental choices, consistent their relevant basic values. It recognizes that, even when a topic is familiar (e.g., declining fish populations, invasive species, suburban sprawl, vector-borne disease), the specific choice may not be. As a result, individuals cannot just ‘read off’ a response, from a universal utility or value function. In effect, they may not know what they want. In such cases, researchers need to decide whether to take people as they are, with fragmentary positions, or help them to derive more reflective ones. The former strategy risks a sin of omission if respondents are swept along by the incomplete set of issues evoked by a particular question. The latter strategy risks a sin of commission, if interacting with respondents distorts, rather than deepens, their understanding.

These questions of competence can be asked regarding both citizens and technical specialists. The answers have implications for the roles that these individuals should play in environmental decision making. When citizens lack the desired competence, a case can be made for rule by experts (or for intensive education). When even the experts don’t know very much, a case can be made for moving slowly (unless that allows even more poorly understood processes to go on unchecked). The current debate over genetically modified organisms revolves on whether anyone understands them well enough to make sound choices. Those who doubt that even the experts know enough to proceed often invoke the ‘precautionary principle.’ At times, it is just a (holistic) slogan of distrust. At times, it is a request (or demand) for incorporating deeper forms of uncertainty into cost, risk, and benefit analyses. These concerns pose significant challenges to the measurement and characterization of expert knowledge (Funtowicz and Ravetz 1990, Kammen and Hassendzahl 1999).

The emerging science of belief and value assessment draws on two converging research traditions. One leads from psychophysics to attitude research. The second leads from decision theory to behavioral decision theory.

2.1 Psychophysical Measurement

Emerging from the natural sciences, many early psychologists focused on what came to be called psychophysics, determining the psychological equivalent of physical stimuli. The conceptualization underlying these studies holds that a combination of physiological and psychological mechanisms translates those stimuli into an internal state of arousal. If asked, individuals can report on that state, either verbally (by word or number) or by some other action (e.g., pressing on a handgrip, adjusting a rheostat to show the light intensity equivalent to the loudness of a tone).

The most straightforward environmental applications of the psychophysical approach ask people how they feel about a specific environment (e.g., how natural, attractive, rich … is it?). They might be asked to use direct ratings (Daniel and Vining 1983) or to make similarity judgments, which are then subjected to multi-dimensional scaling (Garling 1976). Alternatively, they might be asked to sort pictures that vary in the roles played by natural and built objects (Cantor 1977). Using pictures frees respondents from having to encode verbally what they see and like. As a result, it may capture attitudes that people have difficulty expressing, because they lack the words or hesitate to use them (e.g., because of social unacceptability or uncertainty about terminology). One set of properties, identified in such studies, is coherence, complexity, legibility, and mystery (Kaplan and Kaplan 1989).

Psychophysical studies create a significant interpretative burden for investigators, who must decode the meaning for respondents of both stimulus and response (Fischhoff and Furby 1988). Just what features did respondents notice? What did they overlook? What did respondents believe about the services that the focal environment provides and the peril it faces? Did they detect hints (intended or inadvertent), regarding what to say? Would they respond similarly in less hypothetical circumstances? It takes a suite of converging studies, with suitable manipulation checks, to establish what people think they are evaluating and how deep their understanding is. Such interpretative issues have been at the center of the continuing controversy regarding contingent valuation, a method advanced by some resource economists for monetizing goods not traded in any conventional marketplace (e.g., the survival of an endangered species, the improvement of atmospheric visibility) (Mitchell and Carson 1989). In the attitude measurement tradition, interviewers ask respondents how much they are willing to pay in order to prevent an adverse environmental change or to accept as compensation for it. If valid, the responses to these questions could be used in cost-benefit analyses. In that role, they would represent environmental changes that would otherwise be neglected in policy making dominated by economic analyses. However, these are intellectually demanding questions. By definition, they require explicit, quantitative evaluations of complex, novel stimuli with many potentially relevant details.

Attitudes have both strengths and weaknesses as anchors for evaluating environmental goods. The ease with which they are evoked means that individuals typically have some feelings to draw upon, whatever question they face. On the other hand, such initial feelings can be hard to overcome when more analytical responses are required (e.g., willingness to pay for a specific good) (Kahneman et al. 1999). Strong feelings can blur the details of serious environmental questions. Debates over the validity of contingent valuation have focused on whether respondents are properly sensitive to the amount (or ‘scope’) of the environmental change being evaluated (Arrow et al. 1993). The death of migratory birds saddens many people. Envisioning many dead birds creates more sadness than the death of few. The challenge is to translate degrees of sadness into appropriate monetary equivalents. Doing so may require intensive, interactive processes, which depart from the nonreactive norm of attitude research.

Even if such elicitation processes are successful, questions can be asked about the appropriateness of their goal. At the extreme, monetization assumes that everything of value has an economic equivalent. A more modest aspiration is to assign economic values wherever that is possible, but without assuming that they are everything. However, one might still worry that the focus on economics ‘anaesthetizes moral feeling’ (Tribe 1972). Furthermore, it would take a very disciplined process not to let monetized effects swamp diverse other ones.

Despite the residual ambiguity of their results, these studies place important constraints on studies using other procedures. Methodologically, features that demonstrate a robust influence in attitude studies should be explicitly represented in the stimuli used with other methods. Otherwise, respondents need to guess about the features that are missing, based on those that are present. Substantively, the basic preferences reflected in attitude studies should be observed in studies using in other methods. If not, then doubts must be raised about the robustness of the preferences (or the methods).

In the attitude research tradition, evaluation focuses on a single option. Formally speaking, those evaluations could reflect the alternatives as well, in terms of the opportunity costs of forgoing those other options. However, in cognitive terms, the nature and consequences of those alternative actions will inevitably be less clear than those of the focal one. With many options, the alternatives may not even be fully enumerated. However, even when there are but two complementary options (e.g., develop don’t develop), thinking about one need not call the other to mind. As a result, each option might evoke different concerns. Asking about it alone does not allow individuals to express their full suite of relevant values.

2.2 Decision Theoretic Measurement

The early days of the modern environmental movement presented many technology managers with opposition that they could not, or would not, understand. One natural response was that their opponents were misinformed. The initial research response was to determine how well people understood the magnitude of different risks, conceptualized as annual fatality rates. The research found (a) a strong correlation between lay and statistical estimates; (b) systematic deviations from the best-fit curve, partially attributable to the relative availability (in memory) of examples of the different causes of death; (c) considerable ordinal (and even ratio–scale) consistency in judgments across response mode (e.g., fatalities vs. relative death rates); and (d) considerable inconsistency in absolute estimates, across response modes. Thus, respondents seemed to have a fairly robust intuitive feeling for relative risk levels, which emerged however these (unusual) questions were asked (Lichtenstein et al. 1978).

However, rather different judgments emerged when respondents judged ‘risk,’ rather ‘average year fatalities.’ Moreover, the risk judgments were better predictors of their attitudes toward technologies (Slovic et al. 1979). Many studies have sought the determinants of subjective definitions of ‘risk.’ One early candidate was ‘catastrophic potential,’ in the sense that people see greater risk in technologies capable of claiming many lives at a single time. Anecdotal support for this hypothesis can be found in the attention drawn to plane crashes and public concern over nuclear power (despite citizens’ recognition that, in an average year, the death rate from nuclear power is negligible). Subsequent research suggested, however, that catastrophic potential was not as significant as the uncertainty associated with technologies that can produce large accidents (Slovic et al. 1984). Avoiding uncertain technologies reflects risk aversion, rather than an attitude toward how deaths are distributed over time. These are different ethical stances, with different policy implications. Concern over catastrophic potential means providing incentives for locating technologies remotely and disincentives for creating technologies that could, conceivably, affect large numbers of people. Concern over uncertainty means providing incentives for research that sharpens risk estimates and disincentives for highly innovative technologies (which can’t be known that well).

The search for defining attributes of ‘risk’ can be traced to Starr’s seminal (1969) claim that, for a given level of benefit, the public tolerates higher levels of risk when they are incurred voluntarily (e.g., skiing), rather than imposed involuntarily. As evidence, he presented aggregate estimates of annual deaths and economic benefits from eight sources. In his plot, he sketched in two parallel ‘acceptable risk’ lines, an order of magnitude apart, for the voluntary and involuntary risks. This sort of revealed preference analysis assumes that society manages to achieve acceptable risk-benefit tradeoffs from technologies. If so, then that tradeoff should be seen in lay perceptions of risks and benefits (which drive political processes).

A study testing this hypothesis asked citizens to judge 30 technologies in terms of current risks and benefits, acceptable risks (given the benefits), voluntariness, and eight other features (e.g., catastrophic potential, dread, controllability, known to science, known to the public) (Fischhoff et al. 1978). It found:

(a) discrepancies between lay and statistical estimates of risk and benefit;

(b) a weak correlation between estimates of current risks and benefits (so society was not seen as exacting greater benefit from riskier technologies);

(c) no increase in the risk-benefit correlation, when mean judgments of the risk attributes were partialed out (indicating that respondents did not see society as having set a double standard for, say, voluntary and involuntary technologies);

(d) a belief that the acceptable level of risk for most (but not all) technologies was lower than respondents’ current estimate (rejecting the hypothesis of effective societal risk management);

(e) a significant correlation between judgments of current benefits and acceptable risk levels (indicating a willingness to have more risk with greater benefit); and

(f) a large increase in predicting acceptable risk levels from perceived benefits, when many of the risk attributes were partialed out (suggesting the desire for a double standard, accepting greater risk when, say, it is incurred voluntarily).

3. Prescriptive Applications Of Cognitive Research

3.1 Setting Public Policy

If robust, such results provide guidance to public policies affecting the environment. At the simplest level, they show a willingness to accept risk-benefit tradeoffs (possibly surprising some critics of lay rationality). They also show attributes of risk that might be considered when designing technologies or regulatory mechanisms. Different definitions of risk can lead to different social priorities. That is true even for different ways of measuring a single attribute. For example, mortality risk can be summarized as probability of premature death or as lost days of life expectancy. The latter measure puts a special premium on deaths of young people (where many days are lost with each life). With the former measure, a death is a death (Crouch and Wilson 1981, Fischhoff et al. 1981). Either could consider the distribution of risk across the population (e.g., among those who do and do not benefit from a technology), or leave equity considerations to others. Technical specialists often follow professional conventions when computing their estimates. Nonetheless, those definitions still incorporate ethical considerations. Whether citizens (or their representatives) are consulted is a reflection on accepted political processes.

In cognitive and administrative terms, dealing with a large set of attributes is unwieldy (and, at some point, intractable). As a result, many studies have looked for redundancy among risk attributes. Using various data-reduction techniques, typically they have found that much of the variance in ratings of the attributes can be accounted for by two factors. One seems to reflect how well risks were known, by science and by those exposed to it. The second seems to reflect how much risk is dreaded, inequitable, involuntary, and personally threatening.

If such issues are on citizens’ minds, democratic societies need ways to incorporate them in public policies, lest trust be eroded. Institutionalizing analytical procedures that incorporate public values (e.g., definitions of risk) is one possibility. Processes that directly involve citizens have increasingly been advocated, with environmental issues both driving and reflecting societal pressure for more consultative government and industry (e.g., Canadian Standards Association 1997, UK Royal Commission on Environmental Pollution 1998). The US Environmental Protection Agency (1993) fostered some 50 state and regional efforts to set priorities among risks. These panels’ ability to structure their work as they saw fit, and to receive clear explanation of technical matters, seemed critical to their success (Davies 1996). Nonetheless, it was unclear what features any particular panel had considered, nor what weight it gave to each feature. Without such transparency, the credibility of its conclusions could lack credibility, for individuals who were not party to the deliberations. Furthermore, skeptics might wonder whether participants really understood the environmental risks that they were judging.

3.2 Ensuring Citizen Competence

Thus, cognitive approaches to preference elicitation draw upon two of the major streams in psychology, as it has developed over the past century-plus. The psychophysical stream envisions values as emerging from direct perceptual processes; these visceral responses, then, require intellectual effort, in order to translate them into terms that address specific policy questions. The decision theory stream envisions specific values as being derived intellectually from basic values; these, then, must be connected with individuals’ life experiences, lest they be but transient artifacts of the elicitation procedure.

Both streams recognize the following steps that individuals must traverse, if they are to formulate meaningful positions—with either researchers or policy makers as their audience:

(a) Encode the task—in order to understand the choices being offered.

(b) Access relevant personal values—in order to structure the evaluation process.

(c) Integrate the relevant values, as interpreted in the specific context—in order to construct a preference.

(d) Translate that preference into terms recognizable by the investigator or policy maker.

The first step to ensuring such shared understanding is creating a full task specification, characterizing all features that matter to investigators or respondents (e.g., Fischhoff and Furby 1988). How hard investigators have to work to convey these task definitions depends on how far they share a common frame of reference with respondents. For example, as the vote approaches on a widely debated referendum, many citizens know what commitment it implies and can respond reliably to a poll eliciting their values. When that occurs, many details ‘go without saying.’ Omitting them both saves time for less obvious details and respects conversational norms.

Communicating the remaining details has been approached from both a bottom-up and a top-down perspective. The former involves creating a supply curve for facts, focusing presentations on the facts that respondents most need to know, but currently do not. If successful, it allows using recipients’ attention most wisely, as well as evaluating the adequacy of the understanding that they achieve. Sometimes, only a few things really matter. Sometimes, many do. Top-down communication takes advantage of people’s well-known ability to absorb information more quickly, when they can organize it into chunks, clusters of facts cognitively processed as units. Mnemonists have mastered integrating diverse facts into highly flexible templates. Ordinary people tend to organize information spontaneously into less coherent, domain-specific mental models (Johnson-Laird 1983). Invoking one takes advantage of existing beliefs, allowing some task features to go without saying, while making others easier to absorb. However, mental models can also encourage unintended inferences, while hindering the processing of unexpected features. Like other metaphors, they require some interpretation, and may evoke varying images. Thus, here as elsewhere, empirical research is needed to design the best communications, then to evaluate their success (Fischhoff 1999).

4. Conclusion

Environmental decisions are enormously diverse. Individuals can have a hard time grasping the issues involved and deciding how they feel about them. These difficulties are part of the task of researchers, hoping to characterize or facilitate environmental decision making. Driven by the importance of these problems, environmental social scientists have cast a wide net, for potentially relevant methods and theories—extending and integrating them along the way. They make it possible to go beyond speculation and anecdote, regarding what people believe and want about specific environmental issues. Most people report caring about the environment. Its fate and theirs depend on how well they understand what is happening with it and convey their desires for its future.

Bibliography:

1. Arrow K, Solow R, Portney P, Leamer E, Radner R, Schuman H 1993 Report of the NOAA panel on contingent valuation. Federal Register 58: 4601–14
2. Canadian Standards Association 1997 Risk Management (CSA-850). Canadian Standards Association, Ottawa
3. Cantor D V 1977 The Psychology of Place. Architectural Press, London
4. Cotgrove S F 1979 Catastrophe or Cornucopia. Wiley, New York
5. Crouch E A C, Wilson R 1982 Risk Benefit Analysis. Ballinger, Cambridge, MA
6. Dake K 1992 Myths of nature—culture and the social construction of risk. Journal of Social Issues 48(4): 21–37
7. Daniel T C, Vining J 1983 Methodological issues in the assessment of landscape quality. In: Altman I, Wohlwill J F (eds.) Human Behavior and Environment. Plenum Press, New York, Vol. 16, pp. 39–84
8. Davies J C (ed.) 1996 Comparing Environmental Risks. Resources for the Future, Washington, DC
9. Dawes R M 1988 Rational Choice in an Uncertain World. Harcourt Brace Jovanovich, San Diego, CA
10. Douglas M, Wildavsky A 1982 Risk and Culture. University of California Press, Berkeley, CA
11. Dunlap R, VanLiere K D, Mentig A G, Emmet Jones R 2000 Measuring endorsement of the new ecological paradigm. Journal of Social Issues. 56(3): 425–42
12. Fischhoff B 1999 Why (cancer) risk communication can be hard. Journal of the National Cancer Institute Monographs 25: 7–13
13. Fischhoff B, Furby L 1988 Measuring values: A conceptual framework for interpreting transactions. Journal of Risk and Uncertainty 1: 147–84
14. Fischhoff B, Lichtenstein S, Slovic P, Derby S L, Keeney R L 1981 Acceptable Risk. Cambridge University Press, New York
15. Fischhoff B, Slovic P, Lichtenstein S, Read S, Combs B 1978 How safe is safe enough? A psychometric study of attitudes towards technological risks and benefits. Policy Sciences 8: 127–52
16. Flynn J, Kunreuther H, Slovic P (eds.) 2001 Risk, Media and Stigma. Earthscan, London
17. Funtowicz S O, Ravetz J 1990 Uncertainty and Quality in Science for Policy. Kluwer, London
18. Garling T 1976 Structural analysis of environmental perception and cognition—multidimensional scaling approach. Environment and Behavior 8: 385–415
19. Garner G T, Stern P C 1996 Environmental Problems and Human Behavior. Allyn and Bacon, Boston
20. Gilovich T, Griffin D, Kahneman D (eds.) in press The Psychology of Judgment: Heuristics and Biases. Cambridge University Press, New York
21. Johnson-Laird P N 1983 Mental Models. Cambridge University Press, New York
22. Kahneman D, Ritov I, Schkade D 1999 Economic preferences or attitude expression? Journal of Risk and Uncertainty 19: 203–35
23. Kammen D, Hassendzahl D 1999 Should We Risk It? Princeton University Press, Princeton, NJ
24. Kaplan R, Kaplan S 1989 The Experience of Nature: A Psychological Perspective. Cambridge University Press, Cambridge, UK
25. Lichtenstein S, Slovic P, Fischhoff B, Layman M, Combs B 1978 Judged frequency of lethal events. Journal of Experimental Psychology: Human Learning and Memory 4: 551–78
26. Mitchell R C, Carson R T 1989 Using Surveys to Value Public Goods. Resources for the Future, Washington, DC
27. Royal Commission on Environmental Protection 1998 Setting Environmental Standards. Stationery Office Ltd., London
28. Slovic P 1999 Trust, emotion, sex, politics, and science. Risk Analysis 19: 698–701
29. Slovic S, Dixon T 1992 Being in the World: An Environmental Reader for Writers. Macmillan, New York
30. Slovic P, Fischhoff B, Lichtenstein S 1979 Rating the risks. Environment 21(4): 14–20, 36–9
31. Slovic P, Lichtenstein S, Fischhoff B 1984 Modeling the societal impact of fatal accidents. Management Science 30: 464–74
32. Starr C 1969 Societal benefit versus technological risk. Science 165: 1232–8
33. Stern P C 1992 Psychological dimensions of global environmental change. Annual Review of Psychology 43: 269–302
34. Tribe L H 1972 Policy science: Analysis or ideology? Philosophy and Public Affairs 2: 66–110
35. US Environmental Protection Agency 1993 A Guidebook to Comparing Risks and Setting Environmental Priorities. US Environmental Protection Agency, Washington, DC
36. Vaughan E 1993 Individual and cultural differences in adaptation to environmental risks. American Psychologist 48: 673–80