You are on page 1of 21

Acta Psychologica 104 (2000) 317337

www.elsevier.com/locate/actpsy

Action tendencies and characteristics of environmental risks


Gisela B
ohm
a

a,*

, Hans-R
udiger Pster

b,1

Department of Educational Psychology and Sociology, PH Ludwigsburg, P.O. Box 220, D-71602
Ludwigsburg, Germany
b
Department of Psychology, Technical University of Berlin, Berlin, Germany

Abstract
It is assumed that the mental representation of the causal structure of environmental risks,
i.e., the type of cause and the type of potential consequence, determines which sort of action
tendencies are formed. We propose a model of risk evaluation that includes consequentialist
and deontological judgments as well as specic emotions as mediators of action tendencies.
Four hundred participants took part in an experiment which presented scenario information
about environmental risks. The scenarios diered with respect to (a) causation (human vs.
natural cause; single vs. aggregate causation), (b) consequence (harm to self vs. harm to other
people vs. harm to nature), and (c) geographical distance (proximate vs. distant). Participants
indicated how much they preferred each of 31 prospective behaviors. Factor analyses yielded
ve types of action tendencies: help, aggression, escape, political action, and self-focus. The
causal structure of the risks was systematically related to action tendencies, e.g., environmental risks that are caused by humans, and in particular those caused by a single human
agent, elicit aggressive action tendencies. The ndings conrm that the perceived causal
structure of a specic risk determines whether the focus is upon consequentialist or deontological judgments, which, in turn, elicit specic types of action tendency, mediated by
emotions. 2000 Elsevier Science B.V. All rights reserved.
PsycINFO classication: 3040; 4070
Keywords: Risk perception; Environmental psychology; Emotional responses

Corresponding author. Visiting address: Reuteallee 46, D-71634, Ludwigsburg, Germany. Tel.: +497141-140 685; fax: +49-7141-140 434.
E-mail address: boehm_gisela@ph-ludwigsburg.de (G. Bohm).
1
Present address: German National Research Center for Information Technology, GMD, Darmstadt,
Germany.
0001-6918/00/$ - see front matter 2000 Elsevier Science B.V. All rights reserved.
PII: S 0 0 0 1 - 6 9 1 8 ( 0 0 ) 0 0 0 3 5 - 4

318

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

Environmental risks, especially global environmental risks such as ozone layer


depletion or global warming, are among the most urgent problems to be addressed
by political and societal decision makers. Most of these global risks originate from
human activities. However, the connection between individual choice and action, on
the one hand, and long-term environmental consequences, on the other, is rather
indirect. In this paper, we analyze action tendencies that are elicited by past,
present, or anticipated eects of environmental risks. We argue that action tendencies are determined by the way in which the environmental risk is mentally
construed, i.e., by the cognitive representation of the risk's causal structure. Furthermore, we assume that specic emotions function as mediating factors of action
tendencies in the face of environmental risks.
In the next section, we consider the notion of environmental risk and describe its
typical characteristics. Then, we propose a model of the mental representation of
environmental risks, how risk evaluation is mediated by emotions, and how action
tendencies are formed. Third, an experiment is reported that tests some of the
models main assumptions.
1. The structure of environmental risks
We use the term environmental risk in a rather broad sense, including any large
scale environmental process with potentially negative consequences, i.e., any environmental hazard that is publicly perceived and discussed as a ``risk'' (Bayerische
R
uck, 1993). A large number of environmental risks are anthropogenic, i.e., originate from human activities. On the other hand, since the natural environment
provides important resources for humans, environmental damage can have perilous
consequences for human health and/or living conditions. Hence, humans are both,
perpetrators and victims of environmental risks (Kruse, 1995). Environmental risks
can be seen from two perspectives: As risks for the natural environment, or as risks
from the natural environment (B
ohm, Rost & Spada, 1998). In the former case,
human activities cause environmental damage; e.g., clearing of forests. In the latter
case, environmental changes cause negative eects for humans, e.g., the increased
probability of skin cancer resulting from stratospheric ozone depletion.
A causal chain going from human actions via environmental changes to negative
consequences for mankind is typical for environmental risks. Four basic types of risks
can be distinguished according to the causal structure of human and environmental
factors. The four types result from a 2  2 categorization of anthropogenic vs. natural
causation and consequences for humans vs. those for the natural environment: (1)
ME-risks (manenvironment) occur when humans endanger the natural environment
per se, without necessarily causing negative eects for themselves, e.g., deposing radioactive waste on the moon. (2) EM-risks (environmentman) are environmental
changes beyond human control which jeopardize humans, e.g., volcano eruptions. (3)
MEM-risks (manenvironmentman) are anthropogenic environmental changes
which in turn put humans at risk, e.g., ozone layer depletion (MM-risks, i.e., those
without involvement of the natural environment, of course also exist, but in this paper

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

319

we will only be concerned with environmental risks). Finally, one can think of (4) EErisks (environmentenvironment), i.e., environmental changes that jeopardize the
environment (e.g., forest res in remote areas). The fourth type is not relevant for the
present study. Since no humans are involved in this case, it will not be further discussed in the present paper. Most public debate is probably about MEM-risks such as
pollution. Our main assumption is that these risk types imply specic causal representations that dier with respect to which types of action they most likely produce.
A classication that is similar to our distinction between EM- and MEM-risks has
been proposed by Baum, Fleming and Davidson (1983). These authors distinguish
between natural disasters and technological catastrophes. They argue that many
characteristics of these two types of events are dierent, most notably, natural disasters reect a lack of control whereas technological catastrophes reect a loss of
control over a system that once was under control. Furthermore, technological catastrophes dier from natural disasters in that there are agencies or people to blame
which may result in more focused anger than will natural disasters. Baum et al.
(1983) also show that the psychological eects of these two types of events dier.
Technological catastrophes produce more chronic stress, and the eects are not
limited to the victims of the disaster, which is usually the case for natural disasters.
Rather, loss of condence and credibility may engender eects in people that are not
directly victimized. Thus, Baum et al. (1983) distinguish two event types that resemble two of our risk types. Their argument, that the two types dier with respect to
the psychological eects they produce, is also similar to ours. The dierence between
the two approaches is that we distinguish four types of risk, not only two, and that
our notion of MEM-risks includes, but is not limited to, technological catastrophes.
Apart from their causal structure, environmental risks typically have several of
the following characteristics (B
ohm et al., 1998):
(a) Complex causal processes. Particularly global environmental changes e.g.,
climate change follow an intricate and complex pattern of causal interdependencies that are often nonlinear (Pawlik, 1991).
(b) Delayed and geographically far-reaching consequences. Again, global changes
in particular may have negative consequences that are often extremely delayed
and geographically far-reaching (Frankenberg, 1990; Pawlik, 1991).
(c) Aggregated causation. Normally, e.g., with air pollution, the contribution of a
single individual is neglectably small. Risks arise through the aggregated actions
of many (millions) individuals.
(d) Social dilemma structure. The positive consequences of environmentally
harmful behavior options, e.g., the conveniency of driving a car, are usually immediate and experienced personally by the decision maker, but the negative consequences are delayed and collective, they may even not aict those who cause
them. Vlek (1996), among others, has pointed out that environmental risks constitute a kind of social dilemma situation: The option which is individually rational in the short run turns out to be collectively disastrous in the long run.
(e) Low controllability. Due to the aggregated causation of many environmental
risks, there is little opportunity for an individual to control or prevent potential
damage.

320

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

(f) Relevance of ethical considerations. Due to the social dilemma structure of environmental risks, solutions of the dilemma often hinge on ethical or moral considerations. Morally, the best choice is not necessarily the one with the best
consequences, but one which satises some deontological rule. Such rules often
touch upon ethical considerations such as justice and equity.
2. Components of the evaluative process
In this study, we focus on action tendencies caused by the perception of an environmental risk; i.e., action tendencies as reactions to an event with potential
negative consequences. For instance, how do people react to an instance of pollution
about which they learn that harmful consequences may result? We do not study the
decisions and actions that lead to the risky event in the rst place, such as actions
that lead to pollution in the example above, as is done in studies on pro-environmental or environmentally harmful behavior (Gardner & Stern, 1996). We assume
that actions with respect to risks are at the end of a process that starts with a mental
representation of the perceived risk, which then is cognitively evaluated, and this
evaluation in turn causes specic emotions, which eventually elicit specic action
tendencies.
2.1. Cognitive evaluation: consequentialist vs. deontological
We assume that two evaluative components are relevant when evaluating environmental risks: Evaluation of consequences and evaluation with respect to deontological principles. Consequentialist evaluations are about potential losses, i.e., they
refer to the anticipation of uncertain negative consequences, and to an assessment of
the seriousness of these consequences (Yates & Stone, 1992a). Deontological evaluation refers to the judgment whether ethical values or principles concerning what one
ought to do are violated, no matter what the consequences are. Deontological
evaluations focus upon the actions themselves (i.e., the actions causing the risk), not
upon the consequences and their harmfulness, i.e., they are non-consequentialist
(Baron, 1994). Philosophers distinguish between teleological, i.e., consequence-based
(e.g., utilitarism), and deontological moral theories. Deontological moral theories
deal with the morality of actions, irrespective of their consequences. We interpret the
notion of moral values in the latter sense and use the term deontological evaluations.
Human actions are usually perceived as intentional, which implies that the agent
is seen as responsible and that the consequences are considered avoidable. However,
in the context of environmental behavior, the issue of responsibility is often more
complex, particularly because of the aggregated causation and the social dilemma
structure of many environmental risks. We therefore assume that it is not necessary
that a human agent who causes detrimental eects for the environment, explicitly
intended to cause that damage in order to elicit deontological evaluations. The
damage might be a side-eect or be altogether unwanted; this can hold for aggregated as well as for individual causation. Consider, for instance, a truck driver, who

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

321

has a heart attack resulting in an accident with hazardous chemicals. We assume that
it is primarily the involvement of a human agent, not intentionality, that triggers
deontological considerations and judgments of moral blameworthiness (e.g., inferences about the truck drivers boss as being irresponsible). The human agent need
not be a specic individual, but can also be an organization, acting more or less
intentionally.
We propose that it is the perceived causal structure of risks which determines the
relative salience of consequentialist and deontological evaluations. If a person
thinks that the risk is caused by natural forces, which are normally construed as an
inevitable fact of life, he or she will focus on the risks consequences. Hence, lossbased evaluations of consequences will dominate. If, in contrast, a person thinks
that the risk is caused by human agents, then he or she will focus on the actions
which caused the risk; here, deontological evaluations of the agents actions will
dominate. Deontological evaluations, we assume, are even more pronounced if the
causal agent is clearly identiable, e.g., a specic person, or a specic organization.
On the other hand, if the agent is not identiable as a single unit, because the
aggregate actions of a larger number of individuals are the cause of the risk,
deontological evaluations will be attenuated (though still more intense than in the
case of natural causal forces). Of course, consequentialist evaluations are always
more or less relevant, since the very concept of risk implies some type of potential
negative outcome. Thus, the basic assumption here is that the relative salience of
the two evaluative components is largely determined by the perceived causal
structure of the risk.
2.2. Loss-based and ethical emotions
It has been shown that ecological risks can lead to very intense emotional reactions, such as anger, fear, and outrage (McDaniels, Axelrod & Slovic, 1995). We
assume that both evaluative aspects consequentialist and deontological elicit
emotional reactions, but that dierent specic emotions result from these two aspects. We distinguish two types of emotions, which correspond to the evaluative
components: loss-based and ethical (or moral) emotions. Loss-based emotions result
from risk perception in the narrow sense, i.e., from the subjective experience of loss,
harm, and danger. More specically, prospective loss-based emotions are based upon
the anticipation of future negative events, e.g., fear, worry, or apprehension; retrospective loss-based emotions are aective reactions to events that have already taken
place, e.g., sadness, pity, or frustration. Ethical emotions, on the other hand, may
result from violations of deontological principles, examples are anger, outrage, and
guilt (Lazarus, 1991; Ortony, Clore & Collins, 1988; Smith & Ellsworth, 1985).
Which of these emotions actually occurs depends, among other factors, on the seriousness of the loss, or the seriousness of the violation of a deontological principle;
e.g., small anticipated losses might lead to worry, large losses to fear. Also, the two
evaluative paths are not entirely independent. For instance, the size of a loss that
results from a blameworthy action may increase anger and fury.

322

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

2.3. Action tendencies


Emotions motivate behavior, and dierent specic emotions initiate dierent
types of action tendencies (Frijda, Kuipers & ter Schure, 1989; Roseman, Wiest &
Swartz, 1994). Recent neuropsychological studies yield even stronger evidence that
without the ability to experience emotions, no tendencies, no decisions, hence no
intentional actions can be formed, even if cognitive functions work properly
(Damasio, 1994). Hence, we assume that emotions are closely associated with the
formation of action tendencies. Prospective emotions such as fear or worry motivate
to prevent potential damage and may thus initiate pro-environmental behavior
(Kannapin, Pawlik & Zinn, 1998; Martens & Rost, 1998). Retrospective emotions
such as anger and sadness may give rise to intentions to remedy, rehabilitation, or
punishment (Nerb, Spada & Wahl, 1998).
Confronted with an environmental risk, a large variety of action tendencies can
arise. For example, one might try to act towards the agent who caused the risk, e.g.,
to prevent him or her from further detrimental actions, or to attack or punish him or
her. This class of aggressive actions is directed towards the causal agent of the environmental risk, if such a causal agent can be identied. A second class of actions,
i.e., helping behavior, is directed towards the victims that suer from the negative
consequences of environmental risks. A person might want to help directly, or might
want to partake in actions that indirectly prevent further harm or damage, e.g.,
donations. A third class of actions consists of social or political actions, these may be
directed towards the prevention of the risk in the future, e.g., voting for a particular
party or participating in a demonstration, or towards punishing the responsible
agent, e.g., boycotting. Finally, there is the class of ``do-nothing'' actions, such as
avoiding to deal with the risk and its consequences, denial of the severity of the
problem, or escape from the situation.
All of these action classes are associated with emotion types that are typical for
them: fury or anger might lead to aggressive behavior, pity to helping behavior, fear
and worry to preventive political actions, and frustration and sorrow to avoidance
and escape. However, a unique correspondence between one specic emotion and
one specic action will hardly be observed; under realistic conditions, mixtures or
blends of emotions as well as of action tendencies will arise. Again, we assume that
the relative salience of action tendencies is a function of the salient emotions, and,
hence, originate in the mental representation of the risks causal structure.
2.4. Hypotheses
In sum, we expect that the relative salience of the two evaluative components,
consequentialist and deontological, depends on the risks perceived causal structure,
i.e., whether the risk is of type ME, EM, or MEM. These three types dier with
respect to their causation in that they are either anthropogenic or of natural origin.
The consequences of the three types dier in that they aect either the natural environment or humans. Deontological evaluation is expected to be dominant when the
risk is anthropogenic as compared to caused by natural processes. Correspondingly,

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

323

ethical emotions such as anger and, hence, aggressive action tendencies should be
stronger when the evaluative focus is deontological rather than consequentialist. On
the other hand, loss-based emotions such as fear, and, hence, preventive political
action tendencies should be more pronounced when the focus is on potential losses
rather than on deontological considerations. In addition, we generally assume that
aective reactions and action tendencies are more intense if the consequences of the
risk are geographically near, as this implies greater personal relevance. In the following experiment, characteristics of the environmental risks (causation, consequences, geographical distance) are varied by providing specic scenario
descriptions; evaluative focus, consequentialist and deontological evaluations,
emotions, and action tendencies are assessed as dependent variables.
3. Method
3.1. Participants
Four hundred citizens of the city of Bremen, Germany, volunteered to participate,
59% were female and 41% male, their age ranges from 16 to 77 yr (mean age: 33.61).
They were recruited by advertisements in a local newspaper and were paid for
participation. We did not aim at a sample that is fully representative of the general
population, since this is not as relevant in this type of experimental work as it is in
survey research. As one may expect from the recruitment procedure, there are some
deviations from representativeness with respect to age, sex, and education: females,
young persons (age 1524), and persons with higher education (high school diploma)
are overrepresented.
3.2. Design
Four independent variables were manipulated by providing dierent scenario
information:
(a) Causation with three levels: natural causation, anthropogenic-aggregate causation, and anthropogenic-single (identiable) agent causation.
(b) Consequences with four levels: no negative consequences, negative consequences for the natural environment, negative consequences for other humans,
negative consequences for oneself.
(c) Geographic distance (as a proxy for personal relevance) with two levels: close
vs. far from the evaluating person.
(d) Scenario with four levels: Four semantically dierent but formally equivalent
scenarios (vulcano, gas, spores, and mosquitoes).
3.3. Materials
Four ctitious scenarios about environmental risks were constructed. We used
four scenarios rather than only one in order to increase reliability of eects by

324

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

comparing scenarios and to increase generalizability across dierent risk domains.


The core events in these four scenarios are:
(a) a vulcano starts erupting gases and ashes;
(b) a new gas called ``Polypiperin'' arises in the atmosphere;
(c) spores from new water plants accumulate in a river;
(d) mosquitoes spread out in a new area.
For each scenario 19 versions were constructed that correspond to the possible
combinations of causation, consequences, and geographic distance with three, four,
and two levels, respectively. Complete orthogonal manipulation would yield 24
versions, but the following combinations were omitted:
(a) all three combinations of negative consequences for self with large geographic
distance, and
(b) the two combinations of causation by nature with negative consequences only for nature (this combination corresponds to EE risks, i.e., risks without human involvement).
An abbreviated description of the versions of the vulcano scenario follows as an
example (for descriptions of the other scenarios see Appendix A).
A vulcano, after having been inactive for a long period, becomes active and emits
gases and ashes. The vulcano is either close to where the participant lives or far away
in the West of USA (geographic distance). The eruption is caused either by minor
earthquakes (natural causation), or by drillings for mineral resources that are undertaken either by the mining company ``Metal Industries'' (anthropogenic causation
single agent), or by many individuals (anthropogenic causation aggregate). Experts
agree either that no negative consequences for humans or nature are to be expected
(no negative consequences), or that the gases and ashes are harmless to people but
may damage the metabolism of plants (negative consequences for nature), or that the
gases and ashes may lead to serious damages to the health of children (negative
consequences for others), or that gases are detrimental to the health of people in
general, including the self (negative consequences for self ).
3.4. Dependent variables
Five types of dependent variables were measured that correspond to the components of the evaluative process.
3.4.1. Evaluative focus
The evaluative focus, i.e., whether the participant focuses upon potential consequences or deontological principles, was measured by providing six consequentialist
and six deontological arguments why something should be done about the situation
described in the scenario. Participants ranked the 12 arguments according to their
persuasiveness if they were to convince an acquaintance to sign a list. Examples of
deontological arguments are ``. . . because the rights of other humans should not be
violated'', ``. . . because future generations should not be deprived of their basis of
subsistence'', ``. . . because the life of animals and plants should not be disregarded'';
examples of consequentialist arguments are ``. . . because the potential harmful

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

325

consequences may be very serious'', ``. . . because high nancial costs could result'',
``. . . because of the possible impact on health''. The deontological arguments were
chosen in such a way that they expressed principles that are directed towards valuing
either other human beings or the natural environment. The consequentialist arguments emphasize either a specic loss category (nancial costs, health damage, environmental damage), or one of Yates and Stones (1992b) risk components
(uncertainty, probability of loss, signicance of loss). Each participants mean rank
of the deontological arguments was used as the dependent measure; lower ranks
indicate higher importance of deontological arguments compared to consequentialist
arguments.
3.4.2. Moral blameworthiness
Blameworthiness serves as a measure of the result of deontological evaluation,
i.e., whether the participant thinks that deontological principles are violated. Participants indicated on a seven-point rating scale ranging from 1 (not at all) to 7 (very
strongly) the extent to which they considered the situation described in the scenario
as morally blameworthy.
3.4.3. Perceived riskiness
Perceived riskiness serves as a measure of the result of a consequentialist evaluation, i.e., whether the participant fears serious consequences. Perceived riskiness
was measured by four ratings, obtained on seven-point scales: (a) How high would
you rate the overall risk of the situation? (no risk at all very high risk), (b) How
threatening do you nd the situation? (not at all threatening very threatening), (c)
How dangerous do you nd the situation? (not at all dangerous very dangerous), (d)
How probable is it that the situation will lead to harmful consequences? (very unlikely very likely). The mean of these ratings was used as an index for perceived
riskiness.
3.4.4. Emotions
Participants rated the intensity of each of 18 specic emotions: anger, pride, indignation, relief, fear, trouble, regret, envy, worry, sadness, contempt, guilt, fury,
outrage, hope, helplessness, admiration, sympathy. The question was: ``When you
think of this situation, how intensely do you feel ... ?'', the rating scale ranged from 1
(not at all) to 7 (very strongly). The emotion terms were selected from Ortony et al.,
(1988) and from pilot studies in which we obtained emotions in free-response format.
The emotions were chosen in such a way that about as many ethical as loss-based
emotions were included and that the emotion was plausible vis a vis the scenario.
3.4.5. Action tendencies
A sample of 31 behaviors was presented to participants who rated for each behavior how much they felt inclined to perform it. A complete list of the behaviors is
included in Table 2. The behaviors were selected so that (a) they covered all sorts of
behaviors that are relevant in the context of environmental risks, (b) for each
emotion some behaviors were included that matched that emotion, and (c) the action

326

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

types discussed in the section about evaluative components were covered: aggression,
help, political action, and escape/avoidance.
3.5. Procedure
The study was run in groups of approximately eight, but not more than 10 participants. Each participant completed a questionnaire in which he or she performed
all tasks four times, i.e., for each of the four scenarios, but each scenario was presented in a dierent version. Versions of scenarios (i.e., experimental conditions)
were randomized, but with the restrictions that for each participant each scenario
was presented in a dierent version, and that each experimental condition eventually
included the same number of participants. Order of scenarios was also randomized.
For each of the four scenarios evaluative focus, perceived riskiness, moral blameworthiness, aective evaluation, and action tendencies were assessed. Questions
within each group of dependent measures were presented in random order.
4. Results
First, we report factor analyses that provide derived measures for analyses of
variance. Then, eects of the independent variables on evaluations, emotions, and
action tendencies are analyzed by multifactorial analyses of variance.
4.1. Factor analyses
4.1.1. Specic emotions
A factor analysis (based on the complete correlation matrix with correlations
computed across all participant scenario combinations as cases, N 1619) yielded
four factors with eigenvalue greater than 1 (see Table 1; percentages of explained
variance: 22.9%, 13.1%, 12.6%, and 12.6%). The rst factor is interpreted as referring
to Ethical Emotions, reecting outrage, indignation, fury, anger, and contempt. The
second factor is interpreted as a Retrospective Loss-Based Emotions factor, with
high loadings for regret, sympathy, trouble and sadness. Emotions with high loadings on Factor 3, Positive Valence, are relief, pride, and admiration, but also envy.
The fourth factor is interpreted as a Prospective Loss-based Emotions factor, reecting fear, worry, and helplessness. The results of this factor analysis support our
proposed theoretical distinction between ethical, prospective loss-based, and retrospective loss-based emotions.
An index for each of the three emotion types (positive emotions are not further
considered in this paper) was computed by taking the mean of those emotions with
high loadings on the respective factor. Thus, the ethical emotion index was computed
as the mean intensity rating for outrage, indignation, fury, anger, and contempt, the
prospective loss-based emotion index as the mean of fear, worry, and helplessness, and
the retrospective loss-based emotion index as the mean of regret, sympathy, trouble,
and sadness.

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

327

Table 1

Rotated (VARIMAX) factor loadings of specic emotionsa


Emotion

Factor 1
(ethical)

Factor 2
(loss-based retrospective)

Factor 3
(positive)

Factor 4
(loss-based prospective)

Outrage
Indignation
Fury
Anger
Contempt
Regret
Sympathy
Trouble
Sadness
Guilt
Relief
Pride
Admiration
Envy
Hope
Fear
Worry
Helplessness

0.86
0.84
0.84
0.84
0.73
0.12
0.11
0.33
0.38
0.39
)0.01
0.02
)0.18
0.05
0.01
0.39
0.35
)0.02

0.14
0.09
0.17
0.08
0.24
0.77
0.74
0.59
0.58
0.43
)0.01
)0.03
0.02
0.08
0.36
0.13
0.20
0.13

)0.08
)0.04
)0.04
)0.07
0.06
0.00
0.06
)0.02
0.04
0.14
0.78
0.77
0.68
0.61
0.42
)0.04
)0.11
0.01

0.16
0.19
0.20
0.22
0.02
0.26
0.09
0.47
0.41
)0.04
)0.02
0.02
)0.05
)0.04
0.00
0.76
0.74
0.72

a
Italiced loadings indicate that the emotion item was used to compute the index of the corresponding
emotion factor.

4.1.2. Action tendencies


The factor analysis of the 31 behaviors resulted in ve factors with eigenvalues
greater than 1, the rotated loadings are depicted in Table 2. The ve factors account
for 20.5%, 13.1%, 10.1%, 9.9%, and 5.7% of the variance, respectively. The factors
can be interpreted as Tendency to Help/Prevention (Factor 1), Tendency to Aggression/Retaliation (Factor 2), Tendency to Escape/Avoidance (Factor 3), Social/
Political Actions (Factor 4), and Self-focused Behaviors (Factor 5). An index for
each action type (i.e., help, agression, escape, political action, self-focus) was computed by taking the mean rating of all behaviors corresponding to the respective
factor (Table 2).
4.2. Analyses of variance
The dependent variables evaluative focus, moral blameworthiness, perceived
riskiness, emotions, and action tendencies were each analyzed with a 3  4  2  4
(Causation Consequences Geographic Distance Scenario) multifactorial analysis of variance. Signicant eects were further analyzed by Bonferroni post-hoc
comparisons.
4.2.1. Evaluative focus
Only causation was found to be signicant, F 2; 1532 4:46; p 0:012:
Deontological arguments are more important for anthropogenic causation both

328

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

Table 2

Rotated (VARIMAX) factor loadings of action tendenciesa


Action tendency
Help/prevention
. . . I would try to help to reduce
or limit damage
. . . I would do something to
improve the situation
. . . I would personally help
those aicted
. . . I would do something so
that potential harmful
consequences may be prevented
or at least mitigated
. . . I would try to employ
preventive measures
. . . I would personally initiate
actions to improve the situation
. . . I would do anything to stop
what is going on
. . . I would try to comfort those
aicted.
. . . I would personally
participate in actions to
improve the situation
. . . I would donate to an
environmental or relief
organization that takes action
against the situationb
Aggression/retaliation
. . . I feel like hitting the one
who is responsible
. . . I feel ready to spit the one
who is responsible into the face
. . . I feel like yelling at the one
who is responsible
. . . I feel like taking vengeance
. . . I would like to sue the one
who is responsible
. . . I would try to coerce the one
responsible into rehabilitating
for what happened
Escape/avoidance
. . . I feel like running away
. . . I would like not to know
anything about it
. . . I would like to forget
everything as soon as possible
. . . I feel like lamentating

Factor 1
(help)

Factor 2
(aggr.)

Factor 3
(escape)

Factor 4
(political)

Factor 5
(self)

0.83

0.08

)0.03

0.17

0.07

0.81

0.09

)0.02

0.26

)0.03

0.76

0.12

0.03

0.05

0.23

0.75

0.09

)0.06

0.32

0.02

0.75

0.07

0.12

0.07

)0.13

0.70

0.16

)0.16

0.32

0.14

0.69

0.25

)0.05

0.35

0.02

0.69

0.12

0.22

)0.08

0.21

0.67

0.15

)0.11

0.37

0.18

0.48

0.01

0.09

0.47

0.18

0.10

0.87

0.10

0.09

0.09

0.13

0.83

0.07

0.18

0.15

0.10

0.82

0.13

0.18

0.16

0.09
0.26

0.75
0.61

0.16
0.03

0.05
0.49

0.21
0.05

0.33

0.55

)0.02

0.47

)0.03

0.16
)0.15

0.05
)0.04

0.76
0.64

)0.04
0.12

0.09
0.23

)0.10

0.10

0.63

)0.03

0.04

0.09

0.10

0.61

0.15

0.37

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

329

Table 2 (continued)

Action tendency
. . . I would try to get as far
away as possible
. . . I would not know what to
do
. . . I wish that I could undo
everything that happened
Social/political action
. . . I would like to make it a big
issue in the media
. . . I would participate in a
demonstration so that the
situation or potential damage
gets prevented
. . . I would boycott the one
who caused the situation
. . . I would write a letter to the
newspaper
Self-focus
. . . I feel like patting myself on
my shoulder
. . . I feel like slapping myself in
the face
. . . I feel like crying
. . . I would like to pour out my
heart to a friend

Factor 1
(help)

Factor 2
(aggr.)

Factor 3
(escape)

Factor 4
(political)

Factor 5
(self)

0.28

0.18

0.61

)0.22

)0.20

)0.25

0.03

0.59

)0.01

0.01

0.33

0.22

0.48

0.26

)0.11

0.35

0.21

0.10

0.63

)0.01

0.44

0.16

)0.03

0.62

0.10

0.10

0.43

0.07

0.60

)0.14

0.34

0.25

)0.06

0.55

0.18

0.05

0.11

)0.01

)0.11

0.62

0.06

0.25

0.16

0.10

0.57

0.28
0.37

0.20
0.03

0.34
0.30

0.20
0.27

0.48
0.44

a
Italicized loadings indicate that the behavior item was used to compute the index of the corresponding
action tendency factor. Ratings were obtained on scales ranging from 1 (do not agree at all) to 7 (agree
totally). Participants responded to the following question: ``To what extent do you agree with the following statements?''
b
This item was not used when computing action indices, since it does not have a high loading only on one
single factor.

for the single causation (mean rating M 6:57 and aggregate causation condition
M 6:58 than for the natural causation condition M 7:02. Thus, the relative
salience of the evaluative focus on deontological or on consequentialist considerations only depends on the type of causation of the risk.
4.2.2. Moral blameworthiness
The analysis of moral blameworthiness only yields a signicant main eect for
causation, F 2; 1517 11:27; p < 0:001. Ratings of moral blameworthiness are
highest in the anthropogenic-single condition M 4:96, second highest if the risk is
caused by aggregate anthropogenic causation M 4:24, and natural causation is
the least morally blameworthy situation M 2:67.

330

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

4.2.3. Perceived riskiness


The four ratings used to measure perceived riskiness proved to be suciently
internally consistent to be averaged as a risk index (alpha 0.94). Signicant main
eects
for
consequences,
F 3; 1536 75:96; p < 0:001,
for
scenario,
F 3; 1536 4:32; p 0:005, and a signicant interaction between consequences
and scenario were found, F 9; 1536 6:41; p < 0:001. The vulcano scenario is
rated as the most risky M 5:83, and the spores scenario is the least risky
M 4:91. Consequences that aict humans self M 5:93 or others
M 5:87 are most risky, no negative consequences are least risky M 4:78,
and negative consequences for nature are in between M 5:44.
4.2.4. Emotions
For ethical emotions such as anger two signicant main eects are found, one for
causation, F 2; 1537 10:31; p < 0:001, and one for consequences, F 3; 1537
23:25; p < 0:001. Ethical emotions are rated as more intense if negative consequences are expected (mean index is 4.20, 4.12, and 4.18 for the three levels nature,
others, and self, respectively, of the consequences factor) than if no negative consequences are expected M 3:33. With respect to the causation eect, a single
human agent leads to the most intense ethical emotions M 4:60, aggregate anthropogenic causation is second M 4:01, and natural causes M 2:73 arouse
the least ethical emotions. This pattern parallels that of moral blameworthiness.
For prospective loss-based emotions, e.g., fear, a signicant main eect of geographic distance is found, F 1; 1538 4:65; p 0:031. Ratings are more intense
when geographic distance is close M 4:99 than when it is far (M 4.69). Also, a
signicant main eect of consequences is found, F 3; 1538 23:91; p < 0:001.
Potential negative consequences for humans self M 5:37 or others M 5:09
evoke more intense prospective loss-based emotions than negative consequences for
nature (M 4.77), which in turn evoke higher ratings than if no negative consequences occur (M 4.45). This consequence eect parallels the eect found for
perceived riskiness.
For retrospective loss-based emotions such as sadness only the main eect of type
of consequence proved to be signicant, F 3; 1537 20:36; p < 0:001. Retrospective emotions are more intense if negative consequences are possible (mean ratings
are 3.46, 3.53, and 3.34 for self, others, and nature, respectively) than if there are no
potential negative consequences M 2:83.
4.2.5. Action tendencies
An analysis of the derived dependent variable Help/Prevention yields signicant
main eects for geographic distance, F 1; 1533 8:12; p 0:004, and for type of
consequences, F 3; 1533 14:40; p < 0:001. Tendency to help and prevent is
stronger if the risk is geographicly distant M 4:74 than if it is close M 4:37.
Concerning the main eect for consequences, tendency to help increases from no
negative consequences M 4:27 via consequences for nature M 4:52 to consequences for humans self M 4:58 or others M 4:77.

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

331

For Aggression/Retaliation, the main eects of causation, F 2; 1531 5:03;


p 0:007, and consequences, F 3; 1531 7:08; p < 0:001, proved to be signicant.
With respect to consequences, there is more aggression if negative consequences
occur (mean index is 3.25, 3.24, and 3.22 for self, others, and nature, respectively)
than if no negative consequences are expected M 2:80. There is a clear dierentiation concerning causation: Aggression is strongest if the situation is caused by a
single human agent M 3:52, it is less strong for aggregate anthropogenic causation M 3:14, and weakest if the situation is caused by nature M 2:44. This
pattern parallels those of moral blameworthiness and ethical emotions. Thus, anger
and aggression covary with moral blameworthiness, but there must be some potential loss involved in order to evoke those responses.
The results of the tendency to Social/Political Action closely parallel those of
Aggression/Retaliation. Social/Political Actions are signicantly aected by causation, F 2; 1541 5:78; p 0:003, and by type of consequences, F 3; 1541
3:41; p 0:017. Tendency to Social/Political Action increases from natural causation M 3:72 via anthropogenic aggregate causation M 4:40 to anthropogenic
causation with a single agent M 4:98.
No signicant eect was found for the tendency to Escape, even though the entire
model is signicant, F 75; 1533 1:77; p < 0:001.
For self-focus, only a signicant main eect for type of consequences was found,
F 3; 1534 9:73; p 0:007. Self-focus is generally low; stronger for consequences
that aect humans M 1:84 and M 1:94 for others and self, respectively) than for
consequences for nature M 1:71 or no consequences M 1:66.
4.3. Mediational analyses
The reported analyses of variance show eects of the risks perceived causal
structure on each of the dependent variables evaluative focus, perceived risk, moral
blameworthiness, emotions, and action tendencies. In our model, however, we assume specic directional causal relationships. The causal relations are assumed to go
from evaluative focus via cognitive judgments (consequentialist or deontological) to
emotions (loss-based or ethical), and, nally, to action tendencies. Specically,
emotions are assumed to mediate the impact of cognitive judgments on actions. As a
preliminary test, we conducted a mediational analysis as proposed by Baron and
Kenny (1986). These authors suggest that a series of three regression models be
computed in order to establish that the relationship between an independent and a
dependent variable is mediated by a third variable, the mediator (Baron & Kenny,
1986, p. 1177):
(a) regressing the mediator on the independent variable;
(b) regressing the dependent variable on the independent variable, and
(c) regressing the dependent variable on both the mediator and the independent
variable.
Mediation can be inferred if the following three conditions hold:
(a) the independent variable signicantly predicts the mediator in the rst
regression;

332

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

(b) the independent variable also signicantly predicts the dependent variable in
the second regression, and
(c) the mediator signicantly predicts the dependent variable in the third
regression.
If all three conditions are met, the eect of the independent variable should be
weaker in the third equation than in the second; mediation would be perfect if
the eect of the independent variable vanishes after controlling for the
mediator.
We carried out a preliminary analysis to nd some evidence for the mediational
assumptions; it would be premature and the data do not allow us to test a full edged
model of all causal relationships. We considered two of the action tendencies: Help/
Prevention and Aggression/Retaliation, which are the ones that can be assigned most
clearly to consequentialist and deontological evaluation, respectively. Mediational
analyses were conducted for three chains of mediation:
(a) perceived riskiness aects help/prevention tendencies, mediated by prospective loss-based emotions;
(b) perceived riskiness aects help/prevention tendencies, mediated by retrospective loss-based emotions, and
(c) moral blameworthiness aects aggression/retaliation tendencies, mediated by
ethical emotions.
Results are shown in Table 3. The three conditions described above are met for all
three causal chains. Thus, for all three chains, the independent variable aects the
mediator in regression (a), the independent variable aects the dependent variable in
regression (b), and in regression (c) the mediator aects the dependent variable, and
the regression weight of the independent variable is smaller than in regression (b).
Hence, the results of these analyses provide preliminary support for the assumption
that the causal relations go from cognitive evaluations via emotional reactions to

Table 3

Mediational analyses: regression modelsa


Regression (a):
IV ! MV

R2

0.20

0.35

(2) Perceived riskiness ! retrospective loss-based emotions ! help/prevention


0.48
0.19
0.58
0.33
0.46
0.24

0.38

(3) Moral blameworthiness ! ethical emotions ! aggression/retaliation


0.52
0.37
0.39
0.25
0.15

0.41

b (IV)

R2

Regression (c): IV MV ! DV
b (MV)

b (IV)

R2

Regression (b):
IV ! DV

b (IV)

(1) Perceived riskiness ! prospective loss-based emotions ! help/prevention


0.64
0.37
0.58
0.33
0.45

0.48

a
IV: Independent variable; DV: Dependent variable; MV: Mediator variable; b: Unstandardized
regression coecient. All regression coecients and all R2 's are signicant at p < 0:0001.

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

333

action tendencies. This causal chain has been documented for the consequentialist as
well as for the deontological evaluative branch of our model.
5. Discussion
We will rst briey summarize the results. Next, results on action tendencies are
discussed in more detail. Finally, some problems and future research issues are
discussed.
In general, there are remarkably few eects of the four dierent semantic scenarios. Only perceived risk is aected by the type of scenario, especially, the volcano
scenario was perceived as very risky. Thus, some situations are obviously conceived
per se as more risky than others. However, the absence of any eect of the scenario
on the other dependent variables indicates that the model components and their
relations are largely independent of the semantic underpinning. Deontological
evaluations are more salient when causation is anthropogenic than when it is natural. Moral blameworthiness is highest for a single human agent, aggregate anthropogenic causation is less blameworthy, and natural causation is least morally
blameworthy. These basic ndings are all in accordance with our model. Consequentialist evaluations such as perceived riskiness depend on the type of potential
consequences. As expected, negative consequences for humans are perceived as
riskier than negative consequences for nature. Also, the theoretical distinction between prospective loss-based, retrospective loss-based and ethical emotions was
supported by factor analyses. Moral blameworthiness, ethical emotions, tendency to
aggression/retaliation, and tendency to social/political action all show a parallel
eect pattern. They are all inuenced by the risks perceived causation: Risks caused
by a single, identiable human agent are perceived as most blameworthy, give rise to
the most intense ethical emotions and most strongly elicit tendencies to aggression/
retaliation as well as to social/political action. The second group of variables that
exhibits a parallel eect pattern is perceived risk, loss-based emotions particularly
prospective and tendency to help/prevention. These variables are aected by the
type of potential consequences: Perceived risk, loss-based emotions, and tendency to
help/prevention are strongest if consequences are involved that may be harmful to
humans.
In sum, as expected we found that evaluative focus depends on the perceived
causal structure of the risk that is evaluated. Furthermore, we found that deontological considerations (moral blameworthiness) goes together with ethical emotions
and aggressive behavior tendencies. Consequentialist considerations (perceived risk),
on the other hand, correspond to loss-based emotions and a tendency to help. These
results are in accordance with our theoretical model and were corroborated by
mediational analyses.
In this study, the main focus was on the action component, i.e., on the question
which actions or action tendencies are elicited by perceived instances of risky events.
A factor analysis identied ve classes of action tendencies: helping behavior,
aggressive behavior, political action, escape, and self-focus. However, only help,

334

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

aggression and political action are clearly related to the perceived structure of the
risk. As predicted, a preference to help the victims of a risk is caused by geographic
distance. Aggression, in contrast, is determined by the perceived cause of the risk: if a
single human agent can be identied, he or she will be the target of aggressive action
tendencies. Surprisingly, political action is also related to the cause of the risk. We
had expected political actions to result from loss-based emotions, but they covary
with ethical emotions. Interestingly, all three action classes are also inuenced by the
risks consequences, i.e., if consequences are negative, the tendency to act in a specic
way is more intense. Negativity of consequences seems to be the most general factor
that elicits action tendencies; without the potential for negative consequences, the
situation might not be classied as a risk in the rst place, and, hence, no action
tendencies might result. On the other hand, type of cause and geographical distance
are more specic to dierent action classes such as help or aggression. These ndings
indicate a possible modication of the model: if the risks mental representation
implies negative consequences, a general tendency to act is activated; if the representation allows further inferences about causes and distance, specic action tendencies are emphasized and others are inhibited.
All in all, a systematic pattern exists which relates characteristics of the mental
representation of a risk to specic action tendencies. Of course, the mental representation might not be a valid representation of the ``true'' structure of the risk. In
the experiment, we assume that participants take the risks description as valid and
construe a corresponding mental representation. It remains an open question if and
how seriously people verify their subjective mental model before any action is taken.
The degree of uncertainty about ones knowledge of the situation might be an important mediator: if one is not highly condent about ones knowledge and inferences, one might refrain from real action.
Note that what in the experiment was assessed are only action tendencies, not real
actions. It has long been known in attitude research that the correlation between
attitudes, i.e., action tendencies, and behavior, i.e., real action, is far from perfect
(Eagly & Chaiken, 1993). As plausible as the general relationship between the representation of a risk and action tendencies may seem, it does not imply that the
relationship with real actions follows the same pattern.
What we found in this experiment is an eect of the risks perceived causal
structure on a group of dependent variables evaluative focus, moral blameworthiness, perceived risk, emotions, and action tendencies. In a preliminary mediational analysis, we also found some support for the hypothesized directional
relationships among those variables, i.e., that the causal relations go from cognitive
judgments to emotions to action tendencies. Note that this implies a clear stance
towards one of the basic problems of emotion research, i.e., if emotions are caused
by cognitive judgments, or if, vice versa, emotions come rst and change the way
cognitive processes are executed, or are independent of cognitive processing altogether (Zajonc, 1980). This issue remains to be investigated in future studies. A rst
step could be to apply structural equation modeling, a second step would be to
experimentally manipulate emotions and to test their eects on cognition and
behavior.

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

335

Acknowledgements
This research was supported by Grant He 1449/2-2 from the German Research
Association (DFG) as part of the program Human Dimensions of Global Change.
The authors wish to thank Marcus Ladineo, Dorothee K
orner, and Hans J
org
Henning for their help in conducting the experiments and for many constructive
discussions.
Appendix A. Scenarios
A.1. Versions of the gas scenario
In the gas scenario, chemical reactions in the atmosphere produce a new gas called
``Polypiperin''. This takes place either in Germany, which is where our participants
live, or in Japan (geographical distance). The chemical reaction is due either to a
change in the sun`s activity (natural causation), or to a new extremely ecient
cordless transmission procedure that was either developed by the TV station ``Ultra
TV'' that can now transmit many more channels (anthropogenic causation single
agent) or is used for a new kind of mobile phone that has become very wide-spread
and is used by many individuals (anthropogenic causation aggregate). The potential
negative consequences expressed in the expert statements were the same as in the
vulcano scenario.
A.2. Versions of the spores scenario
The core event in the spores scenario is that spores from new water plants are
found in a river. This river was either close to where the participant lives or in the
Mid West of the USA (geographic distance). The new spores stem either from the
aquaria of a fun park called ``Oceanworld'' (anthropogenic causation single agent),
from aquaria of individual households (anthropogenic causation aggregate), or are
carried to the river by water birds (natural causation). According to experts, there
were either no negative consequences to be expected (no negative consequences), the
ecological balance may be seriously disturbed (negative consequences for nature), or
because of contamination of drinking water serious health eects either for children (negative consequences for others) or for humans in general, including the self
(negative consequences for self), could result.
A.3. Versions of the mosquito scenario
In the mosquito scenario, a new kind of mosquito spreads out either in Germany
or in the Northern part of Africa (geographical distance). The mosquitoes are imported from the jungle either by migrant birds (natural causation), by the freight ship

336

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

company ``Afro Line'' (anthropogenic causation single agent), or by many tourists


(anthropogenic causation aggregate). Experts agree that either no negative consequences for humans or nature are likely to occur (no negative consequences), or due
to a pathogenic agent carried by the mosquitoes that epidemics among red deer
may spread out so that some species could even become extinct (negative consequences for nature), or that an infection may seriously damage the health of children
(negative consequences for others), or the health of humans in general, including the
self (negative consequences for self ).

References
Baron, J. (1994). Nonconsequentialist decisions. Behavioral and Brain Sciences, 17, 142.
Baron, R. M., & Kenny, D. A. (1986). The moderatormediator variable distinction in social
psychological research: Conceptual, strategic, and statistical considerations. Journal of Personality
and Social Psychology, 51, 11731182.
Baum, A., Fleming, R., & Davidson, L. M. (1983). Natural disaster and technological catastrophe.
Environment and Behavior, 15, 333354.
Bayerische R
uck, (1993). Risk is a construct. Perceptions of risk perception. M
unchen: Knesebeck.
B
ohm, G., Rost, J., & Spada, H. (1998). Editorial: Psychologische Aspekte von Umweltrisiken. Zeitschrift
f ur Experimentelle Psychologie, 45, 270285.
Damasio, R. D. (1994). Decartes' error, emotion, reason, and the human brain. New York: Putnam's Sun.
Eagly, A. H., & Chaiken, S. (1993). The psychology of attitudes. Fort Worth: Hartcourt Brace Jovanovich.
Frankenberg, P. (1990). Risikokommunikation: Anthropogen induzierte Klimaver
anderungen. In H.
Jungermann, B. Rohrmann, & P. Wiedemann, Risiko-Konzepte, Risiko-Konikte, Risiko-Kommunikation (pp. 281334). J
ulich: Forschungszentrum J
ulich.
Frijda, N. H., Kuipers, P., & ter Schure, E. (1989). Relations among emotions, appraisal, and emotional
action readiness. Journal of Personality and Social Psychology, 57, 212228.
Gardner, G. T., & Stern, P. C. (1996). Environmental problems and human behavior. Boston: Allyn and
Bacon.
Kannapin, O., Pawlik, K., & Zinn, F. (1998). Pr
adiktormuster selbstberichteten Umweltverhaltens.
Zeitschrift f ur Experimentelle Psychologie, 45, 365377.
Kruse, L. (1995). Globale umweltver
anderungen: eine herausforderung f
ur die psychologie. Psychologische
Rundschau, 46, 8192.
Lazarus, R. S. (1991). Emotion and adaptation. New York: Oxford University Press.
Martens, T., & Rost, J. (1998). Der Zusammenhang von wahrgenommener Bedrohung durch
Umweltgefahren und der Ausbildung von Handlungsintentionen. Zeitschrift f 
ur Experimentelle
Psychologie, 45, 345364.
McDaniels, T., Axelrod, L. J., & Slovic, P. (1995). Characterizing perception of ecological risk. Risk
Analysis, 15, 575609.
Nerb, J., Spada, H., & Wahl, S. (1998). Kognition und Emotion bei der Bewertung von Umweltschadensfallen: Modellierung und Empirie. Zeitschrift f 
ur Experimentelle Psychologie, 45, 251269.
Ortony, A., Clore, G. L., & Collins, A. (1988). The cognitive structure of emotions. New York: Cambridge
University Press.
Pawlik, K. (1991). The psychology of global environmental change: Some basic data and an agenda for
cooperative international research. International Journal of Psychology, 26, 547563.
Roseman, I. J., Wiest, C., & Swartz, T. S. (1994). Phenomenology, behaviors, and goals dierentiate
discrete emotions. Journal of Personality and Social Psychology, 67, 206221.
Smith, C. A., & Ellsworth, P. C. (1985). Patterns of cognitive appraisal in emotion. Journal of Personality
and Social Psychology, 48, 813838.

G. B
ohm, H.-R. Pster / Acta Psychologica 104 (2000) 317337

337

Vlek, C. A. J. (1996). Collective risk generation and risk management: The unexploited potential of the
social dilemmas paradigm. In W. B. G. Liebrand & D. M. Messick, Frontiers in social dilemmas
research (pp. 1138). Berlin: Springer.
Yates, J. F., & Stone, E. R. (1992a). Risk appraisal. In J. F. Yates, Risk-taking behavior (pp. 4985).
New York: Wiley.
Yates, J.F., & Stone, E.R. (1992b). The risk construct. In J. F. Yates, Risk-taking behavior (pp. 125).
New York: Wiley.
Zajonc, R. B. (1980). Feeling and thinking: Preferences need no inferences. American Psychologist, 35,
151175.

You might also like