Choice Processing in Emotionally difficult decisions
Choice conflicts may cause negative emotions.
Individuals use a wide variety of processing strategies (from normative to heuristic).
Effort-accuracy theoretical framework
 Trade-off between the accuracy of various decision strategies and the cognitive effort required.
Negative emotions play an important role:
A task may be negatively emotion-laden and characterized as very difficult, even if the
relevant information is easy to comprehend.
The effort-accuracy theoretical framework is a powerful resource for explaining decision-processing
behavior, however people have other goals that may interact with effort and accuracy goals in
decision making.
 We use the framework to make predictions regarding how negative, specifically task-related
emotion influences decision processing.
Decision tasks and decision strategies
Decision tasks often involve selection from a set of alternatives, each defined by some set of
potential future consequences.
 Vary in terms of likelihood of occurrence and their value or desirability.
The information processing carried out during a decision varies depending on such properties as its
complexity and its level of negative task emotion.
3 primary measures of information acquisition behavior:
1) How much information is processed?
2) Whether information is processed
a. selectively across either attributes or alternatives
= different amounts for each attribute or alternative
b. or consistently.
= same amount
3) The pattern of processing:
a. By alternative = multiple attributes of a single alternative are considered before
information about another alternative is processed.
b. By attribute = value several alternatives on a single attribute.
Many different strategies for solving decision problems exist.
1) Weighted additive strategy
All attributes are of differing importance and have a subjective value.
2) Simpler lexicographic strategy
Choosing the alternative with the best value on the most important attribute.
3) Elimination-by-aspects strategy
Involves the elimination of options that do not meet a minimum cutoff value for the most
important attribute, followed by consideration of the second most important attribute etc.
until a single attribute remains.
One important distinction among theories is the degree to which they are compensatory.
 Compensatory decision strategy
A good value on one attribute can compensate for a poor value on another.
 Noncompensatory strategies
A good value in one attribute cannot make up for a poor value on another.
Decision strategies differ in terms of the degree to which they are normatively accurate.
 Can be defined in terms of the maximizing utility.
 Can be defined in terms of the decision process.
=> Weighted additive decision strategy = classic model because it both identifies the above things.
Decision accuracy is important. People use it in decision processes.
 Accuracy goal  Effort goal
An effort-accuracy framework for strategy selection
Cost-benefit perspective = choose between more normative and more heuristic strategies.
 Trade-off between the benefits offered by and the costs extracted by each available strategy.
Bounded rationality
 The decision makers exploit environmental structure to attain reasonable decision accuracy
subject
to de constraints of limited cognitive resources.
 The decision makers are assumed to trade off the increased potential accuracy offered by
more normative decision strategies with the increased effort savings offered by more
heuristic strategies.
The influence of negative emotions?
1) Avoiding the relevant decision
2) Choosing alternatives that minimize likely regret an guilt, protect self-esteem, maximize
security
 How emotion might affect either accuracy or effort?
1) Negative emotion might act as a signal of decision importance or as an incentive
2) Negative emotion might act as a source of effort-taxing decision complexity
Effort-accuracy hypotheses regarding task-induced emotion
Emotions as incentives.
- By signaling the importance of making an accurate decision.
- Tend to signal situations requiring coping efforts to forestall or minimize some threat or
harm.
- Negative emotion = an incentive to perform the decision task well, increasing a decision
maker’s relative emphasis on accuracy over effort and causing a shift to more normative
processing patterns.
 This predicts that increased negative, task-related emotion will lead to more extensive, less
selective, and more alternative-based processing patterns.
Emotions as complexity.
- Decisions involving important goals or difficult conflicts often seem more taxing.
- Emotional arousal (opwinding) related to negative mood limits the attentional capacity
available for decision making.
- As tasks become more complex, people typically shift toward more simplified, heuristic
decision strategies.
 Negative task-related emotion will encourage less extensive, more selective, and more
attribute-based processing.
These are two conflicting prediction!
Coping hypotheses regarding task-induced emotion
Hierboven: Emotion as a factor influencing either the decision maker’s desire for accuracy or the
decision maker’s desire for effort conservation.
Nu: Believe that the motivation to directly reduce the experience of negative emotion may interact
with effort and accuracy goals, and therefore one must consider this additional goal to make
clear
predictions.
Decision behavior under negative emotion may incorporate elements of both increased accuracy
maximization and increased simplicity as directed by emotion minimization concerns.
2 general behaviors for dealing with emotion:
1) Problem-focused coping or direct actions intended to improve the person-environment
relationship eliciting emotion.
2) Emotion-focused coping or indirect actions intended to minimize experienced emotion
through changes in (only) the amount or content of thought about the emotion-eliciting
situation.
 Involve very different types of behavior.
 The two forms of coping are typically used simultaneously.
Problem-focused coping.
= Involves direct attempts to solve a problem.
In decision domains, problem-focused coping should involve processing efforts directed at
identifying the most accurate alternatives.
The motivation to perform accurately should be particularly closely associated with extensive
processing. People find it important to consider all relevant information.
(Slides: De oorzaak van negatieve emoties proberen weg te werken.)
Emotion-focused coping.
= Involve both avoidance and changing the subjective meaning of a situation.
Avoidance can involve simply not making a decision or use of simplified heuristic strategies.
 Emotion-focused coping may operate by motivating the decision maker to avoid
particularly emotion-laden decision processing operations, even though the overall extent
of processing may increase as negative, task-related emotion increases.
Emotion-focused coping motivations may result in more attribute-based processing.
 Attribute trade-offs.
 Noncompensatory decision rules: allow to avoid the negative feelings by explicitly confronting the
potential losses associated with trade-offs between attributes.
A third aspect is selectivity (next to the extent of processing and the degree to which it is attributebased).
 Not clear, contradictory
 Increased emotion will lead to less compensatory processing and therefore increased selectivity
or
 Increased emotion may lead to more extensive processing and therefore decreased selectivity
(Slides: minder negatieve emotie proberen te voelen)
Hypothesis:
Task-related emotion will encourage decision makers to process both more extensively and
more by attribute.
 Argue that the decision maker may use motivation to cope with negative emotion as a guide in
choosing which specific accuracy maximization operations to increase and which particularly
taxing operations to decrease.
-
Emotion may serve as a signal regarding problem importance (hyp 1)
at the same time that people try to minimize the experience of negative emotion through
avoidance (hyp 2)
 We kunnen de hypotheses niet generaliseren.
Overview of methodology
Manipulations of emotion in decision domains
Manipulating task-related emotion.
Necessary to manipulate task-related emotion within a controlled decision-making environment.
 A decision should be more negatively emotion laden to the degree that its unwanted potential
consequences are either more severe or more likely.
 Decision conflict must me present to be decision-related.
Emotion-manipulation check measures.
Checken of mensen die meer negatieve gevoelens zouden moeten hebben die ook werkelijk hebben.
Paper p. 389-390
The mouselab computer program
Bestaat uit een matrix met alle beschikbare informatie. Alle informatie met betrekking tot de keuzes
is verstopt in deze gesloten vakjes. De kandidaat kan de vakjes openen door er met de muis van de
computer op de klikken. Op die manier houdt men de volgorde, gespendeerde tijd en de uiteindelijke
keuze in het oog.
Dependent variables assessing processing
- Time = time spent on each choice trial
- Acquisitions = the number of times information boxes were opened
o The next acquisition could involve the same alternative but a different attribute
o An acquisition could involve the same attribute but a different alternative
- Pattern (tss -1 en 1) =
#transities in product - #transities tss producten
_____________________________________
Som van alle transities
 A pattern score indicating more attribute-based processing (= a lower value) is interpreted as
indicating greater avoidance of trade-offs.
Experiment 1
De deelnemers werden gevraagd zich in te beelden dat ze van een weldadigheidsorganisatie zijn. Ze
moeten 1 kind kiezen dat ze kunnen redden uit een reeks van 5 beschrijvingen.
Er werd een hoge-emotiegroep en een lage-emotiegroep gemaakt. De mensen in de hogeemotiegroep werden voorzien van meer specifieke informative. Ook kregen ze korte beelden van de
kinderen te zien. Deze meer levendige informatie werd verondersteld meer negatieve emoties op te
wekken.
Verder werd ook nog verteld dat ze zich moesten inbeelden dat ze hiervoor de 5 kinderen hielpen,
maar dat ze dit nu gingen terugschroeven tot 1 en dat de overige 4 kinderen geen hulp meer zouden
krijgen.
In de lage emotiegroep deden ze deze acties helemaal niet. Er werd hen verteld dat ze vantevoren
geen enkel kind hielpen en dat ze nu 1 kind kiezen dat ze beginnen te helpen. Er werd gezegd dat de
4 overblijvende kinderen ergens anders hulp zouden krijgen.
Resultaten in table 4 p. 391 (slide 61) stellen dat er een statistisch significant verschil is tussen de
hoge en lage emotiegroepen. Ook al was de totale emotie niet zo groot, toch blijkt er een duidelijk
verschil te zijn. De reacties zullen verschillend zijn.
 Manupulation was successful.
 High-emotion: more acquisitions, more time, more by attribute.
Experiment 2
Manipulatie van emoties is verschillend van exp 1.
 Niet tegelijk de levensechte informatie en het verhaal van de andere 4 kinderen, maar apart.
 Single response mode (zodat het verschil niet kan te maken hebben met de eliminatie van
bepaalde
informatie in de hoge-emotiegroep)
Method
Manipulation.
1) Aan de ene groep (low-base rate) dat de kinderen die niet gekozen werden 10% kans hadden op
andere hulp. Aan de andere groep (high-base rate) dat de kinderen die niet gekozen werden 90%
kans hadden op andere hulp.
 Lower-base rate was expected to increase negative emotion.
2) Idem experiment 1: hoge-emotiegroep kreeg meer, duidelijkere, en levensechte informative.
Manipulation check.
Additional scale measures.
Three additional scale measures assessing how participants appraised their decision tasks were
included to more fully evaluate the alternative hypothesis that negative emotion functions simply as
an incentive.
 Importance
 Threat
 Challenge
Results
Low-base rate condition:
- Feeling more negatively
- Processed more extensively
- More attribute-based processing
- Higher threat
Discussion
As a decision task becomes more inherently emotion laden, participants simultaneously shift toward
more extensive and less alternative based processing patterns.
Negative emotion cannot be treated as only an incentive within the accuracy-effort framework
(there was no increased perception of challenge).
Experiment 3
Verschil met andere twee: manipulatie!
 Examine negative emotion that is related more directly to aspects of decision attributes and
attribute values.
 Altered the pattern of attribute values (high vs. low conflict) to increase the prevalence of threats
to
the decision maker’s goals, and we altered attribute identities (high vs. low trade-off difficulty) to
manipulate the importance of these goals.
Verschil met andere twee: different decision task
 Job choice.
Method
- Choice of 5 jobs, described by 4 attributes.
- High degree of realism.
- Participants in the low-conflict condition could choose an alternative with average to good
values on all of the relevant attributes, whereas participants in the high-conflict condition
had to accept below-average values on at least two attributes.
- Higher trade-off difficulty and higher conflict were both expected to increase negative
emotion.
Session1 Method and Results
- Some attributes are more distressing to trade than others.
- The notion of loss aversion varies by attribute.
- We argue that attributes associated with higher levels of loss aversion are more difficult to
trade off because such trade-offs require that a loss be accepted.
- Importance and loss aversion are related, but not identical, concepts.
Session 1 measures. (zie p. 395)
3 different measures of importance (15 attributes)
3 different measures of loss aversion
Manipulation development.
Spreadsheet
Each participant: high trade-off difficulty, a low trade-off difficulty.
Session 2 results
Expectations:
More emotion-laden decision tasks would be associated with both more extensive and more
attribute-based processing.
- Higher in trade-off difficulty and conflict = more emotion-laden
Manipulation check:
- More negative emotion under high conflict
Processing:
- High-conflict group processed more extensively, making more acquisitions, and taking more
time.
-
Discussion
Hypotheses zijn bewezen.
Sterk bewijst omdat
- verschillende studies
- verschil tussen high- and low-conflict groups
- different decision domains
- different manipulations
Strong evidence that it is necessary to consider the interaction of coping goals with effort and
accuracy goals when attempting to understand decision processing in environments characterized by
task-induced negative emotion.
Mediation analysis
To examine the direct link between the processing patterns and negative emotion.
To demonstrate that mediation, three relationships must hold:
1) The emotion factor significantly affects NEGAVG
2) NEGAVG is significantly related to processing
3) The effect of the emotion factor on processing is weakened or eliminated of NEGAVG is used
as a covariate.
The time course of decision processing
Overview
The findings are novel in that participant seem to be shifting neither toward more normative nor
toward more heuristic decision rules.
Generally, we analyze the time course of decision processing under negative emotion.
 To better isolate the way negative emotion generates the unique overall processing patterns we
observed.
We also consider whether and how selectivity, or the relative amount of effort, is altered over
participants decision-processing episodes.
This analysis addresses the pattern of processing and the relative effort devoted to several decision
aspects over the time course of processing.
Defining processing stages
In the following sections, we analyze several processing indices in terms of the emotion (high vs low)
and place (beginning vs end) factors calculated for our pooled data.
Results
The time course of processing pattern.
Participants engaged in more attribute-based processing under the high (more emotional) level of
the composite emotion factor.
There is much more attribute-based decision processing under high emotion in the beginning stage
of decision processing. By the end of processing, participants processed in an alternative-based
manner regardless of the level of emotion.
We can more directly address the relative amount of effort devoted to attribute- versus alternativebased processing operations over the decision episode by creating variables ATT-total and ALTTOTAL, reflecting the number of attribute- and alternative-based transitions.
 Number of attribute-based transitions increases under high emotion.
The effects of negative emotion on ALT-total are generally the opposite of the effects on ATT-total.
 The mean number of alternative-based transitions decreases under high emotion
 The mean number of alternative-based transitions increasing over the course of processing.
The length of participants’ attribute- and alternative-based processing sequences to address where
in the decision process participants seem most motivated to avoid trade-offs.
 ATTLENGTH = the average length of within-attribute processing sequences.
- Longer under high emotion
 Decision makers will cope with negative emotion in part by shifting toward more
noncompensatory processing strategies.
 ALTLENGTH = the average length of within-alternative processing sequences.
- If decision makers are avoiding trade-offs under negative emotion, we would predict shorter
alternative-based sequences under high emotion.
- A tendency for emotion to encourage shorter alternative-based processing sequences in the
initial stage of processing
- Mostly the inverse of ATTLENGTH
 Decision makers respond to increased task-related negative emotion primarily by engaging
in more attribute-based processing with linger attribute-based sequences in the initial
phase of the decision. Processing became more alternative-based at the end of the
decision, but this result did not vary with the level of emotion.
The time course of selectivity in processing.
VARATT = the variances in the proportions of time spent on each attribute
VARALT = the variances in the proportions of time spent on each alternative
 More selectivity over attributes under higher emotions and at the beginning of processing.
 Emotion encourages processing that is initially more selective over attributes, that is, that
initially concentrates on more extensive examination of fewer attributes.
 More selective over alternatives under low emotion and in the end of stage of processing.
The time course of focus on the chosen alternative.
Participants in both emotion groups processed in a similar, alternative-based manner during the final
stage of processing.
TCHOSEN = the proportion of time spent considering information about one’s ultimately chosen
alternative.
 Higher at the end of processing.
 It appears that participants focused extensively on their to-be-chosen alternative in the final stage
of
decision processing, and they did so regardless of the degree to which their decision task was
emotion laden.
Discussion of the time course analysis
Participants initially approached more emotion-laden decisions with long, attribute-based processing
sequences that were quite selective over attributes.
= The use of noncompensatory strategies.
 Strongest effects during the initial phase of processing.
By the last several acquisitions, all participants tend to display relatively long alternative-based
processing sequences that are selective over alternatives, but not over attributes.
Half of the participants’ time in this phase is spent considering their ultimately chosen alternative.
General discussion
Summary of results
Basic findings and implications.
Our emotion manipulation resulted in more extensive processing, with participants taking a longer
time to choose and opening more information boxes in more threatening decision environments.
This results also in more attribute-based processing.
Results: inconsistent
- It seems implausible that negative emotion functions only as an incentive for increased
accuracy
- It also seems implausible that negative emotion functions as a factor increasing task
complexity
A framework considering accuracy and effort goals explains contingent decision behavior very well
for many decision tasks. This implies that one must sometimes augment this framework to include
other goals that may interact with accuracy and effort concerns.
Analyses of the time course of processing.
It appears that participants in the conditions associated with higher emotion approached their
decisions by initially searching for an attribute-based way to decide.
More emotion-laden situations
 shifting towards more alternative-based rules.
 Noncompensatory, attribute-based decision rule.
…
Because identical patterns of information acquisition can result from differing underlying strategies,
there are competing explanations for our observations of processing.
Emotion and decision performance
Stress can influence the quality of choices in both ways.
In our studies we did not include any measures of decision quality.
The finding that the participants shifted to more attribute-based processing in more emotion-laden
environments indicates that they may have been avoiding explicit trade-offs, and their performance
could be expected to suffer as a result of such actions.
An increase in negative emotion lead to increased choice of the status quo option in a setting where
participants are asked to imagine choosing between a tentatively chosen car (= status quo option) or
other alternatives.
= decision bias
Task-related negative emotions versus negative moods
Negative moods will cause more effortful, complex decision processing.
The use of normative decision strategies should increase with negative mood.
Emotion in the laboratory
Het is niet altijd hetzelfde. Vaak zijn de emoties gemanipuleerd in een laboratorium minder
sterk/anders dan in de realiteit.
Conclusion
Increased negative, task-related emotion associated with a decision problem influences both how
and how much people process information when making a choice. Specifically, the participants shift
to more attribute-based and more extensive processing in more negatively emotional environments.