Heuristics: Tools for an Uncertain World
Media
Part of Heuristics: Tools for an Uncertain World
- Title
- Heuristics: Tools for an Uncertain World
- extracted text
-
Heuristics:
Tools for an Uncertain World
HANSJÖRG NETH and GERD GIGERENZER
Abstract
We distinguish between situations of risk, where all options, consequences, and
probabilities are known, and situations of uncertainty, where they are not. Probability theory and statistics are the best tools for deciding under risk but not under
uncertainty, which characterizes most relevant problems that humans have to solve.
Uncertainty requires simple heuristics that are robust rather than optimal. We
propose to think of the mind as an adaptive toolbox and introduce the descriptive
study of heuristics, their building blocks, and the core capacities they exploit. The
question of which heuristic to select for which class of problems is the topic of the
normative study of ecological rationality. We discuss earlier views on the nature of
heuristics that maintained that heuristics are always less accurate because they
ignore information and demand less effort. Contrary to this accuracy–effort trade-off
view, heuristics can lead to more accurate inferences—under uncertainty—than
strategies that use more information and computation. The study of heuristics opens
up a new perspective on the nature of both cognition and rationality. In a world of
uncertainty, Homo sapiens might well be called Homo heuristicus.
INTRODUCTION
I suppose it is tempting, if the only tool you have is a hammer, to treat everything as if
it were a nail.
Abraham H. Maslow (1966, p. 15f.)
Heuristics are the answer to Simon’s question:
How do human beings reason when the conditions for rationality postulated
by the model of neoclassical economics are not met?
(Simon, 1989, p. 377)
Both Simon’s question and our answer to it seem deceptively simple. Simon
argues against the backdrop of a historic tradition that views the maximization of some notion of utility, the prescriptions of probability theory, or logical
Emerging Trends in the Social and Behavioral Sciences. Edited by Robert Scott and Stephen Kosslyn.
© 2015 John Wiley & Sons, Inc. ISBN 978-1-118-90077-2.
1
2
EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES
consistency as the hallmarks of human rationality. Ever since Fermat and Pascal’s development of probability theory, Bayes’ prescription of how to integrate novel and known information, and Laplace’s proposal that probability
theory was “basically just common sense reduced to calculus” (1814/1951,
p. 196), probability theory has been seen by many as the exclusive criterion
for rational behavior.
An alternative notion of rationality takes into consideration the ecological
nature of human cognition, and argues that instead of one mighty hammer,
we have many adaptive tools at our disposal and routinely rely on these
heuristics. However, heuristics have often been equated with systematic
biases. To unlock the positive potential of heuristics, it is necessary to
acknowledge that it can be misguided to capture or combat real-world
complexity with similarly complicated methods. Instead, the irreducible
uncertainty of everyday environments often calls for simple solutions.
In this essay, we assert that people routinely rely on heuristics—not because
they are irrational, but because they have to make decisions under uncertainty, where risk is not calculable. In these situations, they indeed should rely
on heuristics, provided they do so in an ecologically rational way. To make this
case, we begin by exploring different realms of rationality and illuminating
Simon’s notions of bounded rationality and satisficing in response to his question (section titled “Three Realms of Rationality”). To provide the necessary
background to our answer, we then explicate the nature of heuristics (section
titled “The Nature of Heuristics: Trade-offs, Biases, or Adaptive Tools?”). In
the subsequent section, we introduce the descriptive study of heuristics, the
study of the so-called adaptive toolbox (section titled “The Scientific Study
of Heuristics”). Following that, we extend Simon’s descriptive question to
a normative one: Can we describe the structure of environments in which
a given heuristic works or fails, in comparison with some other strategies?
This analysis of the match between heuristics and environments is called
the study of a heuristic’s ecological rationality (section titled “The Normative
Study of Ecological Rationality”). Finally, we end with the study of intuitive
design, that is, designing heuristic tools for improving expert decisions “in
the wild” (section titled “Intuitive Design: Decision Making in the Wild”).
THREE REALMS OF RATIONALITY
In Hans Christian Andersen’s tale The Emperor’s New Clothes, the emperor
falls prey to two tricksters who pretend to weave the most exquisite fabrics.
Because these are universally believed to be invisible to anyone excessively
stupid or unfit for their position, neither the emperor nor his ministers dare
to disclose that they cannot see anything. Only when the emperor publicly
Heuristics: Tools for an Uncertain World
3
Table 1
Three Realms of Rationality: Certainty, Risk, and Uncertainty
Realm
Type of
Problem
Certainty
All options and
Deductive
consequences are known
inference
for certain (known knowns)
All options and
Inductive
consequences are known,
inference
and their probabilities can
be reliably estimated
(known unknowns)
Ill-posed or ill-defined
Heuristic
problems (unknown
inference
unknowns)
Risk
Uncertainty
Type of
Inference
Appropriate People
Tool
Are …
Logic
Intuitive
logicians
Probability
theory,
statistics
Intuitive
statisticians
Heuristics,
Homo
ecological
heuristicus
rationality
parades his nonexistent clothes does a small child call the hoax by crying out:
“But the emperor has nothing on!”
When considering the question of how rational decisions should be made,
part of the intellectual history of Western thought has succumbed to an
analogous collective illusion—the illusion that formal logic and probability
theory are sufficient for solving all relevant problems. However, applications of probability theory to unstable situations with high uncertainty have
repeatedly failed. Examples are value-at-risk computations by large banks,
which have missed every crisis and prevented none, and the end-of-year
predictions of the euro-dollar exchange rates, which are notoriously inaccurate (Gigerenzer, 2014). The emerging science of heuristics assumes the role
of the small child in Andersen’s tale and aims to promote simple rules of
thumb as an alternative route to rational behavior.
To understand this point, we need to carve up the landscape of rationality
into three realms that are characterized by the type of information available
to the organism whose rationality is being examined (see Table 1). When
all necessary information to solve a given problem, that is, all options and
all consequences, are known for certain, the problem is of the type “known
knowns,” and belongs to the realm of certainty. In a certain world, theories of
logic model how to draw conclusions from premises. Although it is tempting
to view logic as providing the “laws of thought,” we should keep in mind
that deductive inference is utterly conservative, and that certainty is rare.
By explicating only information that is already contained in the premises,
it never allows inferences to be generalized or drawn beyond the realm of
known instances.
4
EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES
Drawing inductive inferences involves risk, a second realm of rationality
that is characterized by probability theory and various theories of statistical
inference. In a risky world, it is assumed that all available options and consequences are known and that the probability distribution is also known or
can be reliably estimated. These problems are known as “known unknowns,”
and the proper tool for solving them is statistical inference, to be made on the
normative basis of probability theory.
Beyond the realms of certainty and risk exists a third realm in which not
all options, consequences, or probabilities are/can be known: the realm
of uncertainty. These are the kinds of problems to which Simon’s question
refers. In this world of “unknown unknowns,” finding the best solutions
by means of statistical optimization techniques is, by definition, no longer
feasible. Instead, heuristics can be helpful to find good solutions. In his
classic distinction between risk and uncertainty, Frank Knight emphasized
that “a measurable uncertainty, or ‘risk’ proper ( … ) is so far different from an
unmeasurable one that it is not in effect an uncertainty at all.” (Knight, 1921,
p. 20). When laying out the Foundations of Statistics, Savage (1954) made it
very clear that probability theory pertained only to predictable scenarios
without surprises—even planning a picnic is out of this realm, according to
Savage, because unexpected things may happen.
The distinction between the three realms illustrates why the theories of
logic and probability theory can rule supremely within their respective
realms but are overextended as universal standards of rationality. Similarly,
Simon’s question does not challenge the model of neoclassical economics
where it is applicable, but instead asks how human beings behave beyond its
reach, such as when probability distributions cannot be estimated because
some options or consequences are unknown. In his search for an answer,
Simon coined the term bounded rationality and proposed that humans use
heuristics such as satisficing (Simon, 1955): Rather than attempting to find
an optimal solution by searching through all alternatives and incorporating
all available information, individuals can set themselves an aspiration level,
sequentially inspect options, and choose the first one that meets or exceeds
this level. Satisficing is an example of a heuristic decision strategy that works
well “without ever making probability calculations” (Simon, 1955, p. 118).
But because satisficing provides no guarantee for finding the best solution,
it seemed to many as if Simon merely replaced the ideal of optimization on
the basis of extensive data collection and deliberation with the more modest
goal of finding acceptable solutions with reasonable amounts of time and
effort. This juxtaposition of decision quality and effort contributed to the
view that heuristics inevitably involve trade-offs (e.g., between accuracy and
speed), which we will discuss in the next section. Note, however, that Simon
Heuristics: Tools for an Uncertain World
5
did not embrace heuristics as inferior alternatives to optimization but as our
only viable option in a fundamentally uncertain world.
THE NATURE OF HEURISTICS: TRADE-OFFS, BIASES,
OR ADAPTIVE TOOLS?
STRATEGIES FOR DISCOVERY, SEARCH, AND EFFORT REDUCTION
The notion of heuristics has undergone dramatic conceptual changes.
Based on the Greek term for “serving to find out or discover,” heuristics
feature prominently in a variety of sciences and have played the roles
of both villain and hero in accounts of human rationality (see Groner,
Groner, & Bischof, 1983). In physics, Albert Einstein presented his Nobel
prize-winning ideas on the emission of light in 1905 as “a heuristic point of
view” to indicate that his proposal was valuable but incomplete (Holton,
1988, pp. 360–361). Gestalt psychologists presented heuristic principles such
as “looking around” as useful solution strategies (Wertheimer, 1923/1938),
and mathematicians identified heuristics as essential methods for discovering mathematical proofs (Pólya, 1945). This inspired early researchers on
human problem-solving to apply the notion of heuristic search to systematic
methods of guiding and constraining mental navigation through large
problem spaces (Newell & Simon, 1972). According to the portrayal of
people as adaptive decision-makers (Payne, Bettman, & Johnson, 1993),
people constantly negotiate trade-offs with their decision environments and
can choose to sacrifice accuracy to gain efficiency or reduce effort.
FROM VAGUE LABELS TO FORMAL MODELS
This mix of positive and negative connotations surrounding the use of
heuristics has led to two distinct interpretations in psychology. Within the
framework of heuristics and biases (Tversky & Kahneman, 1974), heuristics
are interpreted as suboptimal shortcuts that inevitably lead to systematic
errors and cognitive illusions. The problem with the “representativeness
heuristic” (Kahneman & Tversky, 1972) and similar vague notions was that
these were not formally defined and thus could be used to explain almost
everything after the fact, even A and non-A, that is, a behavior as well
as the opposite behavior (Ayton & Fisher, 2004). Consider the gambler’s
fallacy: After a series of n reds on the roulette table, the expectation of
another red decreases. This fallacy was attributed to people’s reliance on the
representativeness heuristic because “the occurrence of black will result in
a more representative sequence than the occurrence of an additional red”
(Tversky & Kahneman, 1974, p. 1125). Next, consider the hot-hand fallacy,
which is the opposite belief: After a basketball player scores a series of n hits,
6
EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES
the expectation of another hit increases. This belief was equally attributed
to representativeness because “even short random sequences are thought to
be highly representative of their generating process” (Gilovich, Vallone, &
Tversky, 1985, p. 295). No scientific model should be able to explain both a
phenomenon and its contrary. But a vague label can do this by changing its
meaning.
Despite general acknowledgments that heuristics are “quite useful”
(Tversky & Kahneman, 1974, p. 1124) and “highly economical and usually
effective” (p. 1131), those benefits are typically qualified by emphasizing
that heuristics “lead to severe and systematic errors” (p. 1124). Kahneman
(2003, p. 1449) made it perfectly clear that he views heuristics as sources of
systematic deviations from the normative ideal of rationality:
“Our research attempted to obtain a map of bounded rationality, by exploring
the systematic biases that separate the beliefs that people have and the choices
they make from the optimal beliefs and choices assumed in rational-agent
models.”
By contrast, we advocate a brighter view of heuristics and pursue a
different research agenda based on formal models of heuristics. Within the
framework of the fast-and-frugal heuristics program (Gigerenzer, Todd, &
the ABC Research Group, 1999; Gigerenzer & Gaissmaier, 2011; Gigerenzer,
Hertwig, & Pachur, 2011), heuristics are seen as adaptive tools that allow
people to make accurate, efficient, and robust decisions under uncertainty.
Rather than focusing on psychological follies and fallacies or compiling
catalogs of human deviations from rational norms, we assume that heuristics
are routinely relied on and often for good reasons, and aim to discover,
analyze, and develop their positive potential. In studying when and why
people use heuristics, we adopt the following definition:
Definition: Heuristics are adaptive tools that ignore information to make
fast and frugal decisions that are accurate and robust under conditions
of uncertainty.
Several aspects of this definition are noteworthy.1 First, describing heuristics as tools emphasizes their methodological nature and status as means
to make decisions, rather than as ends in themselves. Yet, in some cases,
understanding the heuristic can help to clarify what exactly the end is
1. Two aspects frequently mentioned in other accounts (e.g., Evans, 2008, Kahneman 2003) are notably
absent from our definition, namely, that heuristics (i) are used “unconsciously” and (ii) reflect the workings of some “System 1.” As we believe that any heuristic can be applied both consciously or unconsciously and prefer testable process models of heuristics to opaque descriptions that tend to exhaust
themselves in long lists of dichotomies, we prefer to abstain from such characterizations (see Kruglanski
& Gigerenzer, 2011 for a critical review).
Heuristics: Tools for an Uncertain World
7
(Gigerenzer & Selten, 2001). In psychological studies of human cognition,
making decisions typically involves mental strategies. But our notion
of heuristics also includes algorithmic procedures for expert systems in
applied domains (e.g., in financial, legal, or medical contexts; see section
titled “Intuitive Design: Decision Making in the Wild”). In either case, we
consider it essential that heuristics are formulated as precise process models
that yield testable predictions. Second, the qualifier adaptive implies that
a successful use of heuristics depends on experience and the degree of
fit between decision strategies and environmental structures (see below).
Third, our definition agrees with previous accounts that a primary motivation for using heuristics is making fast and frugal decisions but differs
in emphasizing that this increase in efficiency does not necessarily entail
a decrease in effectiveness or decision quality. Whether the benefits of
heuristics come at a cost is not an a priori issue or matter of opinion, but an
empirical question. In fact, one of the surprising discoveries is that heuristics
can avoid the trade-off between speed and accuracy and often yield more
accurate and robust results than more effortful methods. Finally, the natural
habitat of heuristics is the realm of uncertainty. As we have argued in the
previous section, this is hardly much of a constraint, as this realm comprises
the vast majority of situations under which “the conditions for rationality
postulated by the model of neoclassical economics are not met” (Simon,
1989, p. 377).
In short, existing perspectives on heuristics paint a deeply ambivalent
picture of heuristics by portraying them as valuable strategies involving
trade-offs, as vices leading to systematic biases, or as adaptive tools allowing
for accurate and robust decisions in a fast and frugal manner. To disentangle
this conceptual confusion, the scientific study of heuristics needs to examine
both descriptive and normative research questions. We will ask and answer
these questions in the following sections.
THE EMERGING SCIENCE OF HEURISTICS
The descriptive question when studying heuristics is:
Which heuristics do people use?
The result of this investigation yields an arsenal of heuristics that are organized in an adaptive toolbox. Unable to discuss all the heuristics that it contains (see Table 2 for an illustrative but incomplete list), we provide some
examples before discussing their general properties in more detail.
8
Source
Goldstein and
Gigerenzer (2002)
Schooler and
Hertwig (2005)
Gigerenzer and
Goldstein (1996)
Martignon et al.
(2003)
Dawes (1979)
DeMiguel et al.
(2009), Hertwig
et al. (2002),
Messick (1993)
Heuristic
Recognition heuristic
Fluency heuristic
Take-the-best
Fast-and-frugal trees
Tallying (unit-weight
linear model)
Equality heuristic; 1/N
Allocate resources equally to each of N alternatives
To estimate a criterion, do not estimate their importance,
but instead count the number of favoring cues
To classify an object, go sequentially through a number of
cues and stop search as soon as a cue allows doing so
(similar to take-the-best but for classification tasks)
Infer which of two alternatives has the higher value: (i)
search through cues in order of their validity, (ii) stop
search as soon as a cue discriminates, and (iii) choose
the alternative favored by this cue
If one alternative is recognized faster than another, infer that
it has the higher criterion value
If one of two alternatives is recognized, infer that it has the
higher value on the criterion
Definition
Table 2
Heuristics in the Adaptive Toolbox
Equality judgments
Counting
Ordering, estimation
Recognition memory,
estimation (of
recognition validity)
Recognition fluency,
estimation (of fluency
validity)
Ordering, estimation (of
the order of cue
validities)
Capacity exploited
9
Simon (1955)
Wübben and von
Wangenheim
(2008)
Johnson and
Goldstein (2003)
Axelrod (1984)
Boyd and Richerson
(2005)
Satisficing
Hiatus heuristic
Default heuristic
Tit-for-tat
Social imitation
Adapted from Todd et al. (2012, p. 9f).
McBeath, Shafer,
and Kaiser (1995),
Gigerenzer (2007)
Gaze heuristic
Imitate the behavior of the majority of people in your peer
group or of your most successful peer
Cooperate first, then imitate your partner’s most recent
behavior
If a default option exists, follow it
To identify active customers, select those who have made a
purchase within a temporal threshold
To choose an option, search alternatives sequentially and
select the first that exceeds your aspiration level
To catch a flying object, fixate your eye on it and move to
maintain a constant viewing angle
Social imitation
Social imitation and
retribution
Detecting defaults
Setting and detecting
temporal thresholds
Setting aspiration
levels, comparison
Tracking, locomotion
10
EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES
RECOGNITION HEURISTIC
How do people infer which of two alternatives is bigger or better according
to some criterion? For instance, which of the following two cities has more
inhabitants:
Detroit or Milwaukee?
About 60% of a sample of people living in the United States correctly
answered this question with “Detroit.” Surprisingly, the proportion of
British or German citizens correctly answering the same question was
80–90% (Gigerenzer & Goldstein, 2011). Does this mean that Europeans
know more about US cities than US citizens do? No—in fact, the Europeans
benefit from knowing less. Because many of them recognize the name of
Detroit but have not heard of Milwaukee, they can apply the recognition
heuristic (RH; Goldstein & Gigerenzer, 2002):
If only one of two alternatives is recognized, infer that it has the higher value
with respect to the criterion.
The RH makes the bold prediction of a less-is-more effect: If the recognition
validity (defined as the proportion of correct inferences when only one alternative is recognized) exceeds the knowledge validity (defined as the proportion
of correct inferences when both alternatives are recognized), an increase in
the number of recognized alternatives beyond some level causes a decrease
in accuracy. And, although the RH is so simple that it may seem naïve, it has
been successfully used to predict consumer choices, election results, the winners of the Wimbledon tennis matches, and the performance of investment
portfolios (for overviews, see Marewski, Pohl, & Vitouch, 2010, 2011).
1/N HEURISTIC
How should we allocate available resources to assets, for example, when
investing money? According to traditional finance theory, maximizing
profit implies a trade-off between risk and return. The Nobel-prize winning
mean-variance model of Markowitz (1952) solved this problem by maximizing profit for a given level of variance (risk). An alternative strategy is the
simple 1/N heuristic, which appears as early as in the Talmud:
Allocate all resources equally to each of N alternative assets.
Benartzi and Thaler (2001) observed that investors tend to use this “naïve
diversification” method in their contributions to retirement savings plans
and concluded that this may incur substantial costs. But, when DeMiguel,
Heuristics: Tools for an Uncertain World
11
Garlappi, and Uppal (2009) compared the performance of the Markowitz
model and its modern variants with the 1/N heuristic, the mathematically
sophisticated strategies failed to outperform the seemingly naïve heuristic.2
THE SCIENTIFIC STUDY OF HEURISTICS
The systematic study of heuristics requires precise models of the cognitive
processes and their building blocks, and an analysis of the cognitive or
perceptuo-motor core capacities that heuristics exploit. Heuristics typically
recruit and rely on evolved and learned core capacities (such as recognition
memory, which is known to precede conscious recollection; Ratcliff &
McKoon, 1989), and exploit regularities in the external environment (e.g.,
that the recognition of city names tends to be positively correlated with
population size). Although no complete taxonomy exists, heuristics can be
sorted into categories such as recognition-based rules (RH, fluency heuristic), heuristics relying on one good reason (take-the-best, gaze heuristic,
hiatus heuristic), and trade-off heuristics (tallying, 1/N; see Gigerenzer
& Gaissmaier, 2011, for additional examples). Note that heuristics are not
isolated from each other, but share knowledge structures and frequently
contain three common building blocks:
Search rules specify what information is considered and in which order or
direction information search proceeds.
Stopping rules specify when the search for information is terminated.
Decision rules specify how the final decision is reached.
By combining building blocks, heuristics can be adopted and adjusted to
form new heuristics. For instance, the 1/N heuristic for financial allocation
decisions is a specialized instance of a more general equality rule (Messick,
1993) that also appears to govern parental investments (Hertwig, Davis,
& Sulloway, 2002). Similarly, the so-called hiatus heuristic (Wübben & von
Wangenheim, 2008), which companies use to predict which customers will
purchase in the future, is related to optimal stopping rules considered in
animal foraging theory (Green, 1984) and research on humans when switching between multiple tasks (Payne, Duggan, & Neth, 2007). Additionally,
combining building blocks or heuristics can yield new heuristics, such
as when merging satisficing with tallying to create a new heuristic that
selects an alternative as soon as a certain number of criteria are met. These
transformations show that the adaptive toolbox does not consist of a fixed
2. Markowitz himself used 1/N instead of his own model of portfolio optimization (Benartzi & Thaler,
2001, p. 80).
12
EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES
set of heuristics but is a flexible and fluid collection that can be shaped by
the demands of adaptive agents, the current task, and task environments.
Crucially, our stipulation that heuristics must be specified as formal process
models (or explicit algorithms) turns the study of the adaptive toolbox into a
productive research program that not only yields new questions—including
“How do we select heuristics?” (Rieskamp & Otto, 2006) and “For which
cognitive and environmental niches are they most suited?” (Marewski &
Schooler, 2011)—but also makes their performance measurable by yielding
testable predictions. This enables us to address the question of heuristic
performance as an empirical issue rather than accepting their alleged
inferiority as a predetermined conclusion. For a given heuristic, this task
involves specifying the conditions under which it works or fails as measured
by some criterion of performance, such as accuracy.
THE NORMATIVE STUDY OF ECOLOGICAL RATIONALITY
With the help of precise models of heuristics, Simon’s descriptive question
can be extended to a normative one:
When should people rely on heuristics rather than on more complex strategies?
WHY HEURISTICS WORK
To answer the normative question of heuristics, we crucially need to
understand why heuristics work (e.g., Gigerenzer, 2008). The answers can be
expressed in both ecological and statistical terms. Ecologically, identifying
the conditions under which heuristics work well must consider the interplay
between a particular strategy and the structure of its task environment.
Heuristics generally tend to be successful if the conditions of ecological
rationality are met, that is, if (i) a high degree of match exists between the
heuristic and the structure of the task environment and (ii) the organism has
the core capacities necessary for applying the heuristic (Todd, Gigerenzer,
& the ABC Research Group, 2012). Only when agents have particular
capacities is using heuristics simple or effortless per se. For instance, aiming
to construct a robot that applies the gaze heuristic (see Table 2) to intercept
a flying object would demand specific core capacities that robots do not yet
have, such as tracking moving objects against a noisy background.
The statistical answer to the normative question is based on the difference
between fitting models to data and predicting future events on the basis
of past observations. A more flexible model (e.g., with additional free
parameters) will always provide an equal or better fit to existing data than
Heuristics: Tools for an Uncertain World
13
a simpler model will (Gigerenzer & Brighton, 2009, for examples). But, the
main purpose of scientific models is not to explain post hoc, but to predict
similar events in the future.3 Mathematically, the relevant characteristics
of different types of models can be analyzed in terms of the bias–variance
dilemma (Geman, Bienenstock, & Doursat, 1992), which decomposes a
model’s expected prediction error into three parts:
prediction error = (bias)2 + variance + noise
In this sum, the bias term describes the average accuracy of an algorithm’s
predictions, the variance term describes the variation in a model’s predictions
given different samples, and the noise term contains unsystematic deviations,
such as measurement error. Crucially, a more flexible model generally tends
to exhibit both lower bias and higher variance. A model with high bias will
tend to underfit the data (i.e., miss existing patterns), whereas a model with
high variance will typically overfit them (i.e., “explain” even random noise).
Consequentially, a successful model must balance bias and variance in order
to achieve high predictive accuracy.
How can the bias–variance dilemma explain the success of heuristics such
as 1/N? Markowitz’s (1952) model assigns a weight to every available asset
and needs to estimate all means and variances, as well as the covariance
matrix. What makes the mean–variance model prone to error are these
estimates, not the complexity of its calculations. By contrast, the 1/N
heuristic completely ignores historic information and assigns a fixed weight
of 1/N to every asset. Thus, the Markowitz model is highly flexible and
reduces bias at the expense of increased variance, whereas 1/N is likely to
exhibit a high bias but no variance. If a large amount of relevant data were
available and the world of investment were stable, the Markowitz model
would outperform 1/N. But the key is that we are dealing with uncertainty rather than with risk: Even with an abundance of financial data, the
dynamics of financial markets are such that only a small proportion of the
available data may actually be relevant to the current economic conditions.
And, when relevant data are sparse, the variance term tends to dominate
the bias term and thus increase the overall prediction error of complex
models.
Taken together, these considerations allow us not only to provide an
answer to the normative question (When should we rely on heuristics?), but
also to disentangle the conceptual confusion about heuristics by explaining
why older, nonadaptive notions of heuristics were unable to discover and
comprehend their benefits. Heuristics work well when they are adapted to
3. Otherwise, the model “it was due to fate” would perfectly (and parsimoniously) “explain” everything.
14
EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES
solving problems in uncertain environments. Situations with high levels of
uncertainty, high degrees of complexity (e.g., a large number of options),
and small amounts of relevant data tend to favor simple models over more
flexible ones. When the primary concerns are predictive accuracy and an
understanding of generative mechanisms—that is, success in uncertain environments, or designing scientific models—simple heuristics can outperform
more flexible ones by yielding more robust results. In a world of risk, it is
sensible to collect all data and integrate it according to the laws of probability
theory. But in situations of uncertainty, where an optimal solution can no
longer be calculated, heuristics are adaptive tools that make uncertainty
manageable.
INTUITIVE DESIGN: DECISION-MAKING IN THE WILD
The descriptive study of the adaptive toolbox of individuals and institutions
and the normative study of the conditions of ecological rationality have
important consequences for the design of psychological theories. First, the
study of the adaptive toolbox provides an alternative to the widespread
understanding of human behavior in terms of internal traits, preferences, or
attitudes: Behavior is a function of cognition and environment in tandem.
This view has been applied to better understand moral behavior (Fleischhut & Gigerenzer, 2013), the nature of intuition (Gigerenzer, 2007), legal
decision-making (Dhami, 2003), and social behavior in general (Hertwig,
Hoffrage, & the ABC Research Group, 2013).
The answers to the descriptive and normative question have been used
to improve experts’ decision-making and decisions “in the wild” (Gigerenzer et al., 2011). In this program of intuitive design, the resulting decision
systems are designed to fit the intuitive strategies people use. For instance,
fast-and-frugal trees (Martignon, Vitouch, Takezawa, & Forster, 2003,
Table 2) have been designed for coronary care unit allocation (Green &
Mehr, 1997), for diagnosing depression (Jenny, Pachur, Lloyd Williams,
Becker, & Margraf, 2013), for reducing civilian casualties at military checkpoints (Keller, Czienskowski, & Feufel, 2014), and for detecting vulnerable
banks (Aikman et al., 2014; Neth, Meder, Kothiyal, & Gigerenzer, 2014). In
all this applied work, the design of heuristics can provide efficient tools in
an uncertain world.
As mentioned in the beginning, optimization is not possible under uncertainty, but good decisions are. Heuristics are tools toward this end, which, as
far as we know, humans have always relied on to solve adaptive problems.
Moreover, given that humans, past and present, spend most of their time in
the twilight of uncertainty, the nature of Homo sapiens might well be found in
Homo heuristicus.
Heuristics: Tools for an Uncertain World
15
REFERENCES
Aikman, D., Galesic, M., Gigerenzer, G., Kapadia, S., Katsikopoulos, K. V., Kothiyal,
A., Murphy, E., & Neumann, T. (2014). Taking uncertainty seriously: Simplicity
versus complexity in financial regulation. Bank of England Financial Stability
Paper, 28.
Axelrod, R. M. (1984). The evolution of cooperation. New York, NY: Basic Books.
Ayton, P., & Fisher, I. (2004). The hot hand fallacy and the gambler’s fallacy: Two
faces of subjective randomness? Memory and Cognition, 32, 1369–1378.
Benartzi, S., & Thaler, R. H. (2001). Naive diversification strategies in defined contribution saving plans. American Economic Review, 91(1), 79–98.
Boyd, R., & Richerson, P. J. (2005). The origin and evolution of cultures. New York, NY:
Oxford University Press.
Dawes, R. M. (1979). The robust beauty of improper linear models in decision making. American Psychologist, 34(7), 571–582.
DeMiguel, V., Garlappi, L., & Uppal, R. (2009). Optimal versus naive diversification:
How inefficient is the 1/N portfolio strategy? Review of Financial Studies, 22(5),
1915–1953.
Dhami, M. K. (2003). Psychological models of professional decision making. Psychological Science, 14(2), 175–180.
Evans, J. S. B. T. (2008). Dual-processing accounts of reasoning, judgment, and social
cognition. Annual Review of Psychology, 59, 255–278.
Fleischhut, N., & Gigerenzer, G. (2013). Can simple heuristics explain moral inconsistencies?. In R. Hertwig, U. Hoffrage & the ABC Research Group, Simple heuristics
in a social world (pp. 459–486). New York, NY: Oxford University Press.
Geman, S., Bienenstock, E., & Doursat, R. (1992). Neural networks and the bias/
variance dilemma. Neural Computation, 4(1), 1–58.
Gigerenzer, G. (2007). Gut feelings: The intelligence of the unconscious. New York, NY:
Penguin.
Gigerenzer, G. (2008). Rationality for mortals: Risk and rules of thumb. New York, NY:
Oxford University Press.
Gigerenzer, G. (2014). Risk savvy: How to make good decisions. New York, NY: Allan
Lane.
Gigerenzer, G., & Brighton, H. (2009). Homo heuristicus: Why biased minds make
better inferences. Topics in Cognitive Science, 1(1), 107–143.
Gigerenzer, G., & Gaissmaier, W. (2011). Heuristic decision making. Annual Review
of Psychology, 62(1), 451–482.
Gigerenzer, G., & Goldstein, D. G. (1996). Reasoning the fast and frugal way: Models
of bounded rationality. Psychological Review, 103, 650–669.
Gigerenzer, G., & Goldstein, D. G. (2011). The recognition heuristic: A decade of
research. Judgment and Decision Making, 6(1), 100–121.
Gigerenzer, G., Hertwig, R., & Pachur, T. (Eds.) (2011). Heuristics: The foundations of
adaptive behavior. New York, NY: Oxford University Press.
Gigerenzer, G., & Selten, R. (2001). Rethinking rationality. In G. Gigerenzer &
R. Selten (Eds.), Bounded rationality: The adaptive toolbox (pp. 1–12). Cambridge, MA:
MIT Press.
16
EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES
Gigerenzer, G., Todd, P. M., & the ABC Research Group (1999). Simple heuristics that
make us smart. New York, NY: Oxford University Press.
Gilovich, T., Vallone, R., & Tversky, A. (1985). The hot hand in basketball: On the
misperception of random sequences. Cognitive Psychology, 17, 295–314.
Goldstein, D. G., & Gigerenzer, G. (2002). Models of ecological rationality: The recognition heuristic. Psychological Review, 109(1), 75–90.
Green, R. F. (1984). Stopping rules for optimal foragers. The American Naturalist,
123(1), 30.
Green, L., & Mehr, D. R. (1997). What alters physicians’ decisions to admit to the
coronary care unit? Journal of Family Practice, 45(3), 219–226.
Groner, R., Groner, M., & Bischof, W. (1983). Approaches to heuristics: A historical
review. In R. Groner, M. Groner & W. Bischof (Eds.), Methods of heuristics (pp. 1–17).
Hillsdale, NJ: Lawrence Erlbaum Associates.
Hertwig, R., Davis, J. N., & Sulloway, F. J. (2002). Parental investment: How an equity
motive can produce inequality. Psychological Bulletin, 128(5), 728–745.
Hertwig, R., Hoffrage, U., & the ABC Research Group (2013). Simple heuristics in a
social world. New York, NY: Oxford University Press.
Holton, G. J. (1988). Thematic origins of scientific thought: Kepler to Einstein (2nd ed.).
Cambridge, MA: Harvard University Press.
Jenny, M. A., Pachur, T., Lloyd Williams, S., Becker, E., & Margraf, J. (2013). Simple
rules for detecting depression. Journal of Applied Research in Memory and Cognition,
2(3), 149–157.
Johnson, E. J., & Goldstein, D. (2003). Do defaults save lives? Science, 302(5649),
1338–1339.
Kahneman, D. (2003). Maps of bounded rationality: Psychology for behavioral economics. American Economic Review, 93, 1499–1475.
Kahneman, D., & Tversky, A. (1972). Subjective probability: A judgment of representativeness. Cognitive Psychology, 3(3), 430–454.
Keller, N., Czienskowski, U., & Feufel, M. (2014). Tying up loose ends: A method for
constructing and evaluating decision aids that meet blunt and sharp-end goals.
Ergonomics, 1–13PMID: 24852324.
Knight, F. H. (1921). Risk, uncertainty and profit. Chicago, IL: University of Chicago
Press.
Kruglanski, A. W., & Gigerenzer, G. (2011). Intuitive and deliberate judgments are
based on common principles. Psychological Review, 118(1), 97–109.
Marewski, J. N., Pohl, R. F., & Vitouch, O. (2010). Recognition-based judgments and
decisions: Introduction to the special issue (Vol. 1). Judgment and Decision Making,
5(4), 207–215.
Marewski, J. N., Pohl, R. F., & Vitouch, O. (2011). Recognition-based judgments and
decisions: Introduction to the special issue (Vol. 2). Judgment and Decision Making,
6(1), 1–6.
Marewski, J. N., & Schooler, L. J. (2011). Cognitive niches: An ecological model of
strategy selection. Psychological Review, 118(3), 393–437.
Markowitz, H. M. (1952). Portfolio selection. The Journal of Finance, 7, 77–91.
Heuristics: Tools for an Uncertain World
17
Martignon, L., Vitouch, O., Takezawa, M., & Forster, M. R. (2003). Naïve and yet
enlightened: From natural frequencies to fast and frugal decision trees. In D. Hardman & L. Macchi (Eds.), Thinking: Psychological perspectives on reasoning, judgment
and decision making (pp. 189–211). Chichester, UK: Wiley.
Maslow, A. H. (1966). The psychology of science. New York, NY: Harper & Row.
McBeath, M. K., Shaffer, D. M., & Kaiser, M. K. (1995). How baseball outfielders determine where to run to catch fly balls. Science, 268(5210), 569–573.
Messick, D. M. (1993). Equality as a decision heuristic. In B. A. Mellers & J. Baron
(Eds.), Psychological perspectives on justice: Theory and applications (pp. 11–31).
Cambridge, MA: Cambridge University Press.
Neth, H., Meder, B., Kothiyal, A., & Gigerenzer, G. (2014). Homo heuristicus in the
financial world: From risk management to managing uncertainty. Journal of Risk
Management in Financial Institutions, 7(2), 134–144.
Newell, A., & Simon, H. A. (1972). Human problem solving. Englewood Cliffs, NJ:
Prentice Hall.
Payne, J. W., Bettman, J. R., & Johnson, E. J. (1993). The adaptive decision maker.
Cambridge, UK: Cambridge University Press.
Payne, S., Duggan, G., & Neth, H. (2007). Discretionary task interleaving: Heuristics
for time allocation in cognitive foraging. Journal of Experimental Psychology: General,
136(3), 370–380.
Pólya, G. (1945). How to solve it: A new aspect of mathematical method. Princeton, NJ:
Princeton University Press.
Ratcliff, R., & McKoon, G. (1989). Similarity information versus relational information: Differences in the time course of retrieval. Cognitive Psychology, 21(2), 139–155.
Rieskamp, J., & Otto, P. E. (2006). SSL: A theory of how people learn to select strategies. Journal of experimental psychology: General, 135(2), 207–236.
Savage, L. J. (1954). The foundations of statistics (2nd ed.). New York, NY: Dover.
Schooler, L. J., & Hertwig, R. (2005). How forgetting aids heuristic inference. Psychological Review, 112(3), 610–628.
Simon, H. A. (1955). A behavioral model of rational choice. The Quarterly Journal of
Economics, 69(1), 99–118.
Simon, H. A. (1989). The scientist as problem solver. In K. Klahr & D. Kotovsky
(Eds.), Complex information processing: The impact of Herbert A. Simon (pp. 377–398).
Hillsdale, NJ: Lawrence Erlbaum.
Todd, P. M., Gigerenzer, G., & the ABC Research Group (2012). Ecological rationality:
Intelligence in the world. New York, NY: Oxford University Press.
Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and
biases. Science, 185(4157), 1124–1131.
Wertheimer, M. (1923/1938). Untersuchungen zur Lehre von der Gestalt. (English
excerpts: Laws of organization in perceptual forms). In W. Ellis (Ed.), A source book
of Gestalt psychology (pp. 71–88). New York, NY: Hartcourt, Brace and Co.
Wübben, M., & von Wangenheim, F. (2008). Instant customer base analysis: Managerial heuristics often ‘get it right’. Journal of Marketing, 72(3), 82–93.
18
EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES
HANSJÖRG NETH SHORT BIOGRAPHY
Hansjörg Neth is researcher in Social Psychology and Decision Sciences at the
University of Konstanz, Germany, and adjunct researcher at the Center for
Adaptive Behavior and Cognition (ABC) at the Max Planck Institute for Human
Development, Berlin. His research focuses on the theoretical analysis of
adaptive behavior, interactive cognition, and ecological rationality, as well as
applied aspects of choice and heuristic decision-making under uncertainty
(e.g., in financial contexts). He has served as acting chair of Cognition, Emotion, and Communication at the University of Freiburg, taught Cognitive and
Decision Sciences at the University of Göttingen, and was research assistant
professor in Cognitive Science at the Rensselaer Polytechnic Institute. He
holds a PhD in psychology from Cardiff University, UK.
GERD GIGERENZER SHORT BIOGRAPHY
Gerd Gigerenzer is Director at the Max Planck Institute for Human Development and Director of the Harding Center for Risk Literacy in Berlin. He
is former Professor of Psychology at the University of Chicago and John M.
Olin Distinguished Visiting Professor, School of Law at the University of
Virginia. He is also Member of the Berlin-Brandenburg Academy of Sciences
and the German Academy of Sciences, and Batten Fellow at the Darden
Business School, University of Virginia. Awards for his work include the
AAAS Prize for the best article in the behavioral sciences, the Association
of American Publishers Prize for the best book in the social and behavioral
sciences, and the Communicator Award of the German Research Foundation. His award-winning popular books Calculated Risks, Gut Feelings, and
Risk Savvy have been translated into 21 languages.
-
Heuristics:
Tools for an Uncertain World
HANSJÖRG NETH and GERD GIGERENZER
Abstract
We distinguish between situations of risk, where all options, consequences, and
probabilities are known, and situations of uncertainty, where they are not. Probability theory and statistics are the best tools for deciding under risk but not under
uncertainty, which characterizes most relevant problems that humans have to solve.
Uncertainty requires simple heuristics that are robust rather than optimal. We
propose to think of the mind as an adaptive toolbox and introduce the descriptive
study of heuristics, their building blocks, and the core capacities they exploit. The
question of which heuristic to select for which class of problems is the topic of the
normative study of ecological rationality. We discuss earlier views on the nature of
heuristics that maintained that heuristics are always less accurate because they
ignore information and demand less effort. Contrary to this accuracy–effort trade-off
view, heuristics can lead to more accurate inferences—under uncertainty—than
strategies that use more information and computation. The study of heuristics opens
up a new perspective on the nature of both cognition and rationality. In a world of
uncertainty, Homo sapiens might well be called Homo heuristicus.
INTRODUCTION
I suppose it is tempting, if the only tool you have is a hammer, to treat everything as if
it were a nail.
Abraham H. Maslow (1966, p. 15f.)
Heuristics are the answer to Simon’s question:
How do human beings reason when the conditions for rationality postulated
by the model of neoclassical economics are not met?
(Simon, 1989, p. 377)
Both Simon’s question and our answer to it seem deceptively simple. Simon
argues against the backdrop of a historic tradition that views the maximization of some notion of utility, the prescriptions of probability theory, or logical
Emerging Trends in the Social and Behavioral Sciences. Edited by Robert Scott and Stephen Kosslyn.
© 2015 John Wiley & Sons, Inc. ISBN 978-1-118-90077-2.
1
2
EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES
consistency as the hallmarks of human rationality. Ever since Fermat and Pascal’s development of probability theory, Bayes’ prescription of how to integrate novel and known information, and Laplace’s proposal that probability
theory was “basically just common sense reduced to calculus” (1814/1951,
p. 196), probability theory has been seen by many as the exclusive criterion
for rational behavior.
An alternative notion of rationality takes into consideration the ecological
nature of human cognition, and argues that instead of one mighty hammer,
we have many adaptive tools at our disposal and routinely rely on these
heuristics. However, heuristics have often been equated with systematic
biases. To unlock the positive potential of heuristics, it is necessary to
acknowledge that it can be misguided to capture or combat real-world
complexity with similarly complicated methods. Instead, the irreducible
uncertainty of everyday environments often calls for simple solutions.
In this essay, we assert that people routinely rely on heuristics—not because
they are irrational, but because they have to make decisions under uncertainty, where risk is not calculable. In these situations, they indeed should rely
on heuristics, provided they do so in an ecologically rational way. To make this
case, we begin by exploring different realms of rationality and illuminating
Simon’s notions of bounded rationality and satisficing in response to his question (section titled “Three Realms of Rationality”). To provide the necessary
background to our answer, we then explicate the nature of heuristics (section
titled “The Nature of Heuristics: Trade-offs, Biases, or Adaptive Tools?”). In
the subsequent section, we introduce the descriptive study of heuristics, the
study of the so-called adaptive toolbox (section titled “The Scientific Study
of Heuristics”). Following that, we extend Simon’s descriptive question to
a normative one: Can we describe the structure of environments in which
a given heuristic works or fails, in comparison with some other strategies?
This analysis of the match between heuristics and environments is called
the study of a heuristic’s ecological rationality (section titled “The Normative
Study of Ecological Rationality”). Finally, we end with the study of intuitive
design, that is, designing heuristic tools for improving expert decisions “in
the wild” (section titled “Intuitive Design: Decision Making in the Wild”).
THREE REALMS OF RATIONALITY
In Hans Christian Andersen’s tale The Emperor’s New Clothes, the emperor
falls prey to two tricksters who pretend to weave the most exquisite fabrics.
Because these are universally believed to be invisible to anyone excessively
stupid or unfit for their position, neither the emperor nor his ministers dare
to disclose that they cannot see anything. Only when the emperor publicly
Heuristics: Tools for an Uncertain World
3
Table 1
Three Realms of Rationality: Certainty, Risk, and Uncertainty
Realm
Type of
Problem
Certainty
All options and
Deductive
consequences are known
inference
for certain (known knowns)
All options and
Inductive
consequences are known,
inference
and their probabilities can
be reliably estimated
(known unknowns)
Ill-posed or ill-defined
Heuristic
problems (unknown
inference
unknowns)
Risk
Uncertainty
Type of
Inference
Appropriate People
Tool
Are …
Logic
Intuitive
logicians
Probability
theory,
statistics
Intuitive
statisticians
Heuristics,
Homo
ecological
heuristicus
rationality
parades his nonexistent clothes does a small child call the hoax by crying out:
“But the emperor has nothing on!”
When considering the question of how rational decisions should be made,
part of the intellectual history of Western thought has succumbed to an
analogous collective illusion—the illusion that formal logic and probability
theory are sufficient for solving all relevant problems. However, applications of probability theory to unstable situations with high uncertainty have
repeatedly failed. Examples are value-at-risk computations by large banks,
which have missed every crisis and prevented none, and the end-of-year
predictions of the euro-dollar exchange rates, which are notoriously inaccurate (Gigerenzer, 2014). The emerging science of heuristics assumes the role
of the small child in Andersen’s tale and aims to promote simple rules of
thumb as an alternative route to rational behavior.
To understand this point, we need to carve up the landscape of rationality
into three realms that are characterized by the type of information available
to the organism whose rationality is being examined (see Table 1). When
all necessary information to solve a given problem, that is, all options and
all consequences, are known for certain, the problem is of the type “known
knowns,” and belongs to the realm of certainty. In a certain world, theories of
logic model how to draw conclusions from premises. Although it is tempting
to view logic as providing the “laws of thought,” we should keep in mind
that deductive inference is utterly conservative, and that certainty is rare.
By explicating only information that is already contained in the premises,
it never allows inferences to be generalized or drawn beyond the realm of
known instances.
4
EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES
Drawing inductive inferences involves risk, a second realm of rationality
that is characterized by probability theory and various theories of statistical
inference. In a risky world, it is assumed that all available options and consequences are known and that the probability distribution is also known or
can be reliably estimated. These problems are known as “known unknowns,”
and the proper tool for solving them is statistical inference, to be made on the
normative basis of probability theory.
Beyond the realms of certainty and risk exists a third realm in which not
all options, consequences, or probabilities are/can be known: the realm
of uncertainty. These are the kinds of problems to which Simon’s question
refers. In this world of “unknown unknowns,” finding the best solutions
by means of statistical optimization techniques is, by definition, no longer
feasible. Instead, heuristics can be helpful to find good solutions. In his
classic distinction between risk and uncertainty, Frank Knight emphasized
that “a measurable uncertainty, or ‘risk’ proper ( … ) is so far different from an
unmeasurable one that it is not in effect an uncertainty at all.” (Knight, 1921,
p. 20). When laying out the Foundations of Statistics, Savage (1954) made it
very clear that probability theory pertained only to predictable scenarios
without surprises—even planning a picnic is out of this realm, according to
Savage, because unexpected things may happen.
The distinction between the three realms illustrates why the theories of
logic and probability theory can rule supremely within their respective
realms but are overextended as universal standards of rationality. Similarly,
Simon’s question does not challenge the model of neoclassical economics
where it is applicable, but instead asks how human beings behave beyond its
reach, such as when probability distributions cannot be estimated because
some options or consequences are unknown. In his search for an answer,
Simon coined the term bounded rationality and proposed that humans use
heuristics such as satisficing (Simon, 1955): Rather than attempting to find
an optimal solution by searching through all alternatives and incorporating
all available information, individuals can set themselves an aspiration level,
sequentially inspect options, and choose the first one that meets or exceeds
this level. Satisficing is an example of a heuristic decision strategy that works
well “without ever making probability calculations” (Simon, 1955, p. 118).
But because satisficing provides no guarantee for finding the best solution,
it seemed to many as if Simon merely replaced the ideal of optimization on
the basis of extensive data collection and deliberation with the more modest
goal of finding acceptable solutions with reasonable amounts of time and
effort. This juxtaposition of decision quality and effort contributed to the
view that heuristics inevitably involve trade-offs (e.g., between accuracy and
speed), which we will discuss in the next section. Note, however, that Simon
Heuristics: Tools for an Uncertain World
5
did not embrace heuristics as inferior alternatives to optimization but as our
only viable option in a fundamentally uncertain world.
THE NATURE OF HEURISTICS: TRADE-OFFS, BIASES,
OR ADAPTIVE TOOLS?
STRATEGIES FOR DISCOVERY, SEARCH, AND EFFORT REDUCTION
The notion of heuristics has undergone dramatic conceptual changes.
Based on the Greek term for “serving to find out or discover,” heuristics
feature prominently in a variety of sciences and have played the roles
of both villain and hero in accounts of human rationality (see Groner,
Groner, & Bischof, 1983). In physics, Albert Einstein presented his Nobel
prize-winning ideas on the emission of light in 1905 as “a heuristic point of
view” to indicate that his proposal was valuable but incomplete (Holton,
1988, pp. 360–361). Gestalt psychologists presented heuristic principles such
as “looking around” as useful solution strategies (Wertheimer, 1923/1938),
and mathematicians identified heuristics as essential methods for discovering mathematical proofs (Pólya, 1945). This inspired early researchers on
human problem-solving to apply the notion of heuristic search to systematic
methods of guiding and constraining mental navigation through large
problem spaces (Newell & Simon, 1972). According to the portrayal of
people as adaptive decision-makers (Payne, Bettman, & Johnson, 1993),
people constantly negotiate trade-offs with their decision environments and
can choose to sacrifice accuracy to gain efficiency or reduce effort.
FROM VAGUE LABELS TO FORMAL MODELS
This mix of positive and negative connotations surrounding the use of
heuristics has led to two distinct interpretations in psychology. Within the
framework of heuristics and biases (Tversky & Kahneman, 1974), heuristics
are interpreted as suboptimal shortcuts that inevitably lead to systematic
errors and cognitive illusions. The problem with the “representativeness
heuristic” (Kahneman & Tversky, 1972) and similar vague notions was that
these were not formally defined and thus could be used to explain almost
everything after the fact, even A and non-A, that is, a behavior as well
as the opposite behavior (Ayton & Fisher, 2004). Consider the gambler’s
fallacy: After a series of n reds on the roulette table, the expectation of
another red decreases. This fallacy was attributed to people’s reliance on the
representativeness heuristic because “the occurrence of black will result in
a more representative sequence than the occurrence of an additional red”
(Tversky & Kahneman, 1974, p. 1125). Next, consider the hot-hand fallacy,
which is the opposite belief: After a basketball player scores a series of n hits,
6
EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES
the expectation of another hit increases. This belief was equally attributed
to representativeness because “even short random sequences are thought to
be highly representative of their generating process” (Gilovich, Vallone, &
Tversky, 1985, p. 295). No scientific model should be able to explain both a
phenomenon and its contrary. But a vague label can do this by changing its
meaning.
Despite general acknowledgments that heuristics are “quite useful”
(Tversky & Kahneman, 1974, p. 1124) and “highly economical and usually
effective” (p. 1131), those benefits are typically qualified by emphasizing
that heuristics “lead to severe and systematic errors” (p. 1124). Kahneman
(2003, p. 1449) made it perfectly clear that he views heuristics as sources of
systematic deviations from the normative ideal of rationality:
“Our research attempted to obtain a map of bounded rationality, by exploring
the systematic biases that separate the beliefs that people have and the choices
they make from the optimal beliefs and choices assumed in rational-agent
models.”
By contrast, we advocate a brighter view of heuristics and pursue a
different research agenda based on formal models of heuristics. Within the
framework of the fast-and-frugal heuristics program (Gigerenzer, Todd, &
the ABC Research Group, 1999; Gigerenzer & Gaissmaier, 2011; Gigerenzer,
Hertwig, & Pachur, 2011), heuristics are seen as adaptive tools that allow
people to make accurate, efficient, and robust decisions under uncertainty.
Rather than focusing on psychological follies and fallacies or compiling
catalogs of human deviations from rational norms, we assume that heuristics
are routinely relied on and often for good reasons, and aim to discover,
analyze, and develop their positive potential. In studying when and why
people use heuristics, we adopt the following definition:
Definition: Heuristics are adaptive tools that ignore information to make
fast and frugal decisions that are accurate and robust under conditions
of uncertainty.
Several aspects of this definition are noteworthy.1 First, describing heuristics as tools emphasizes their methodological nature and status as means
to make decisions, rather than as ends in themselves. Yet, in some cases,
understanding the heuristic can help to clarify what exactly the end is
1. Two aspects frequently mentioned in other accounts (e.g., Evans, 2008, Kahneman 2003) are notably
absent from our definition, namely, that heuristics (i) are used “unconsciously” and (ii) reflect the workings of some “System 1.” As we believe that any heuristic can be applied both consciously or unconsciously and prefer testable process models of heuristics to opaque descriptions that tend to exhaust
themselves in long lists of dichotomies, we prefer to abstain from such characterizations (see Kruglanski
& Gigerenzer, 2011 for a critical review).
Heuristics: Tools for an Uncertain World
7
(Gigerenzer & Selten, 2001). In psychological studies of human cognition,
making decisions typically involves mental strategies. But our notion
of heuristics also includes algorithmic procedures for expert systems in
applied domains (e.g., in financial, legal, or medical contexts; see section
titled “Intuitive Design: Decision Making in the Wild”). In either case, we
consider it essential that heuristics are formulated as precise process models
that yield testable predictions. Second, the qualifier adaptive implies that
a successful use of heuristics depends on experience and the degree of
fit between decision strategies and environmental structures (see below).
Third, our definition agrees with previous accounts that a primary motivation for using heuristics is making fast and frugal decisions but differs
in emphasizing that this increase in efficiency does not necessarily entail
a decrease in effectiveness or decision quality. Whether the benefits of
heuristics come at a cost is not an a priori issue or matter of opinion, but an
empirical question. In fact, one of the surprising discoveries is that heuristics
can avoid the trade-off between speed and accuracy and often yield more
accurate and robust results than more effortful methods. Finally, the natural
habitat of heuristics is the realm of uncertainty. As we have argued in the
previous section, this is hardly much of a constraint, as this realm comprises
the vast majority of situations under which “the conditions for rationality
postulated by the model of neoclassical economics are not met” (Simon,
1989, p. 377).
In short, existing perspectives on heuristics paint a deeply ambivalent
picture of heuristics by portraying them as valuable strategies involving
trade-offs, as vices leading to systematic biases, or as adaptive tools allowing
for accurate and robust decisions in a fast and frugal manner. To disentangle
this conceptual confusion, the scientific study of heuristics needs to examine
both descriptive and normative research questions. We will ask and answer
these questions in the following sections.
THE EMERGING SCIENCE OF HEURISTICS
The descriptive question when studying heuristics is:
Which heuristics do people use?
The result of this investigation yields an arsenal of heuristics that are organized in an adaptive toolbox. Unable to discuss all the heuristics that it contains (see Table 2 for an illustrative but incomplete list), we provide some
examples before discussing their general properties in more detail.
8
Source
Goldstein and
Gigerenzer (2002)
Schooler and
Hertwig (2005)
Gigerenzer and
Goldstein (1996)
Martignon et al.
(2003)
Dawes (1979)
DeMiguel et al.
(2009), Hertwig
et al. (2002),
Messick (1993)
Heuristic
Recognition heuristic
Fluency heuristic
Take-the-best
Fast-and-frugal trees
Tallying (unit-weight
linear model)
Equality heuristic; 1/N
Allocate resources equally to each of N alternatives
To estimate a criterion, do not estimate their importance,
but instead count the number of favoring cues
To classify an object, go sequentially through a number of
cues and stop search as soon as a cue allows doing so
(similar to take-the-best but for classification tasks)
Infer which of two alternatives has the higher value: (i)
search through cues in order of their validity, (ii) stop
search as soon as a cue discriminates, and (iii) choose
the alternative favored by this cue
If one alternative is recognized faster than another, infer that
it has the higher criterion value
If one of two alternatives is recognized, infer that it has the
higher value on the criterion
Definition
Table 2
Heuristics in the Adaptive Toolbox
Equality judgments
Counting
Ordering, estimation
Recognition memory,
estimation (of
recognition validity)
Recognition fluency,
estimation (of fluency
validity)
Ordering, estimation (of
the order of cue
validities)
Capacity exploited
9
Simon (1955)
Wübben and von
Wangenheim
(2008)
Johnson and
Goldstein (2003)
Axelrod (1984)
Boyd and Richerson
(2005)
Satisficing
Hiatus heuristic
Default heuristic
Tit-for-tat
Social imitation
Adapted from Todd et al. (2012, p. 9f).
McBeath, Shafer,
and Kaiser (1995),
Gigerenzer (2007)
Gaze heuristic
Imitate the behavior of the majority of people in your peer
group or of your most successful peer
Cooperate first, then imitate your partner’s most recent
behavior
If a default option exists, follow it
To identify active customers, select those who have made a
purchase within a temporal threshold
To choose an option, search alternatives sequentially and
select the first that exceeds your aspiration level
To catch a flying object, fixate your eye on it and move to
maintain a constant viewing angle
Social imitation
Social imitation and
retribution
Detecting defaults
Setting and detecting
temporal thresholds
Setting aspiration
levels, comparison
Tracking, locomotion
10
EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES
RECOGNITION HEURISTIC
How do people infer which of two alternatives is bigger or better according
to some criterion? For instance, which of the following two cities has more
inhabitants:
Detroit or Milwaukee?
About 60% of a sample of people living in the United States correctly
answered this question with “Detroit.” Surprisingly, the proportion of
British or German citizens correctly answering the same question was
80–90% (Gigerenzer & Goldstein, 2011). Does this mean that Europeans
know more about US cities than US citizens do? No—in fact, the Europeans
benefit from knowing less. Because many of them recognize the name of
Detroit but have not heard of Milwaukee, they can apply the recognition
heuristic (RH; Goldstein & Gigerenzer, 2002):
If only one of two alternatives is recognized, infer that it has the higher value
with respect to the criterion.
The RH makes the bold prediction of a less-is-more effect: If the recognition
validity (defined as the proportion of correct inferences when only one alternative is recognized) exceeds the knowledge validity (defined as the proportion
of correct inferences when both alternatives are recognized), an increase in
the number of recognized alternatives beyond some level causes a decrease
in accuracy. And, although the RH is so simple that it may seem naïve, it has
been successfully used to predict consumer choices, election results, the winners of the Wimbledon tennis matches, and the performance of investment
portfolios (for overviews, see Marewski, Pohl, & Vitouch, 2010, 2011).
1/N HEURISTIC
How should we allocate available resources to assets, for example, when
investing money? According to traditional finance theory, maximizing
profit implies a trade-off between risk and return. The Nobel-prize winning
mean-variance model of Markowitz (1952) solved this problem by maximizing profit for a given level of variance (risk). An alternative strategy is the
simple 1/N heuristic, which appears as early as in the Talmud:
Allocate all resources equally to each of N alternative assets.
Benartzi and Thaler (2001) observed that investors tend to use this “naïve
diversification” method in their contributions to retirement savings plans
and concluded that this may incur substantial costs. But, when DeMiguel,
Heuristics: Tools for an Uncertain World
11
Garlappi, and Uppal (2009) compared the performance of the Markowitz
model and its modern variants with the 1/N heuristic, the mathematically
sophisticated strategies failed to outperform the seemingly naïve heuristic.2
THE SCIENTIFIC STUDY OF HEURISTICS
The systematic study of heuristics requires precise models of the cognitive
processes and their building blocks, and an analysis of the cognitive or
perceptuo-motor core capacities that heuristics exploit. Heuristics typically
recruit and rely on evolved and learned core capacities (such as recognition
memory, which is known to precede conscious recollection; Ratcliff &
McKoon, 1989), and exploit regularities in the external environment (e.g.,
that the recognition of city names tends to be positively correlated with
population size). Although no complete taxonomy exists, heuristics can be
sorted into categories such as recognition-based rules (RH, fluency heuristic), heuristics relying on one good reason (take-the-best, gaze heuristic,
hiatus heuristic), and trade-off heuristics (tallying, 1/N; see Gigerenzer
& Gaissmaier, 2011, for additional examples). Note that heuristics are not
isolated from each other, but share knowledge structures and frequently
contain three common building blocks:
Search rules specify what information is considered and in which order or
direction information search proceeds.
Stopping rules specify when the search for information is terminated.
Decision rules specify how the final decision is reached.
By combining building blocks, heuristics can be adopted and adjusted to
form new heuristics. For instance, the 1/N heuristic for financial allocation
decisions is a specialized instance of a more general equality rule (Messick,
1993) that also appears to govern parental investments (Hertwig, Davis,
& Sulloway, 2002). Similarly, the so-called hiatus heuristic (Wübben & von
Wangenheim, 2008), which companies use to predict which customers will
purchase in the future, is related to optimal stopping rules considered in
animal foraging theory (Green, 1984) and research on humans when switching between multiple tasks (Payne, Duggan, & Neth, 2007). Additionally,
combining building blocks or heuristics can yield new heuristics, such
as when merging satisficing with tallying to create a new heuristic that
selects an alternative as soon as a certain number of criteria are met. These
transformations show that the adaptive toolbox does not consist of a fixed
2. Markowitz himself used 1/N instead of his own model of portfolio optimization (Benartzi & Thaler,
2001, p. 80).
12
EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES
set of heuristics but is a flexible and fluid collection that can be shaped by
the demands of adaptive agents, the current task, and task environments.
Crucially, our stipulation that heuristics must be specified as formal process
models (or explicit algorithms) turns the study of the adaptive toolbox into a
productive research program that not only yields new questions—including
“How do we select heuristics?” (Rieskamp & Otto, 2006) and “For which
cognitive and environmental niches are they most suited?” (Marewski &
Schooler, 2011)—but also makes their performance measurable by yielding
testable predictions. This enables us to address the question of heuristic
performance as an empirical issue rather than accepting their alleged
inferiority as a predetermined conclusion. For a given heuristic, this task
involves specifying the conditions under which it works or fails as measured
by some criterion of performance, such as accuracy.
THE NORMATIVE STUDY OF ECOLOGICAL RATIONALITY
With the help of precise models of heuristics, Simon’s descriptive question
can be extended to a normative one:
When should people rely on heuristics rather than on more complex strategies?
WHY HEURISTICS WORK
To answer the normative question of heuristics, we crucially need to
understand why heuristics work (e.g., Gigerenzer, 2008). The answers can be
expressed in both ecological and statistical terms. Ecologically, identifying
the conditions under which heuristics work well must consider the interplay
between a particular strategy and the structure of its task environment.
Heuristics generally tend to be successful if the conditions of ecological
rationality are met, that is, if (i) a high degree of match exists between the
heuristic and the structure of the task environment and (ii) the organism has
the core capacities necessary for applying the heuristic (Todd, Gigerenzer,
& the ABC Research Group, 2012). Only when agents have particular
capacities is using heuristics simple or effortless per se. For instance, aiming
to construct a robot that applies the gaze heuristic (see Table 2) to intercept
a flying object would demand specific core capacities that robots do not yet
have, such as tracking moving objects against a noisy background.
The statistical answer to the normative question is based on the difference
between fitting models to data and predicting future events on the basis
of past observations. A more flexible model (e.g., with additional free
parameters) will always provide an equal or better fit to existing data than
Heuristics: Tools for an Uncertain World
13
a simpler model will (Gigerenzer & Brighton, 2009, for examples). But, the
main purpose of scientific models is not to explain post hoc, but to predict
similar events in the future.3 Mathematically, the relevant characteristics
of different types of models can be analyzed in terms of the bias–variance
dilemma (Geman, Bienenstock, & Doursat, 1992), which decomposes a
model’s expected prediction error into three parts:
prediction error = (bias)2 + variance + noise
In this sum, the bias term describes the average accuracy of an algorithm’s
predictions, the variance term describes the variation in a model’s predictions
given different samples, and the noise term contains unsystematic deviations,
such as measurement error. Crucially, a more flexible model generally tends
to exhibit both lower bias and higher variance. A model with high bias will
tend to underfit the data (i.e., miss existing patterns), whereas a model with
high variance will typically overfit them (i.e., “explain” even random noise).
Consequentially, a successful model must balance bias and variance in order
to achieve high predictive accuracy.
How can the bias–variance dilemma explain the success of heuristics such
as 1/N? Markowitz’s (1952) model assigns a weight to every available asset
and needs to estimate all means and variances, as well as the covariance
matrix. What makes the mean–variance model prone to error are these
estimates, not the complexity of its calculations. By contrast, the 1/N
heuristic completely ignores historic information and assigns a fixed weight
of 1/N to every asset. Thus, the Markowitz model is highly flexible and
reduces bias at the expense of increased variance, whereas 1/N is likely to
exhibit a high bias but no variance. If a large amount of relevant data were
available and the world of investment were stable, the Markowitz model
would outperform 1/N. But the key is that we are dealing with uncertainty rather than with risk: Even with an abundance of financial data, the
dynamics of financial markets are such that only a small proportion of the
available data may actually be relevant to the current economic conditions.
And, when relevant data are sparse, the variance term tends to dominate
the bias term and thus increase the overall prediction error of complex
models.
Taken together, these considerations allow us not only to provide an
answer to the normative question (When should we rely on heuristics?), but
also to disentangle the conceptual confusion about heuristics by explaining
why older, nonadaptive notions of heuristics were unable to discover and
comprehend their benefits. Heuristics work well when they are adapted to
3. Otherwise, the model “it was due to fate” would perfectly (and parsimoniously) “explain” everything.
14
EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES
solving problems in uncertain environments. Situations with high levels of
uncertainty, high degrees of complexity (e.g., a large number of options),
and small amounts of relevant data tend to favor simple models over more
flexible ones. When the primary concerns are predictive accuracy and an
understanding of generative mechanisms—that is, success in uncertain environments, or designing scientific models—simple heuristics can outperform
more flexible ones by yielding more robust results. In a world of risk, it is
sensible to collect all data and integrate it according to the laws of probability
theory. But in situations of uncertainty, where an optimal solution can no
longer be calculated, heuristics are adaptive tools that make uncertainty
manageable.
INTUITIVE DESIGN: DECISION-MAKING IN THE WILD
The descriptive study of the adaptive toolbox of individuals and institutions
and the normative study of the conditions of ecological rationality have
important consequences for the design of psychological theories. First, the
study of the adaptive toolbox provides an alternative to the widespread
understanding of human behavior in terms of internal traits, preferences, or
attitudes: Behavior is a function of cognition and environment in tandem.
This view has been applied to better understand moral behavior (Fleischhut & Gigerenzer, 2013), the nature of intuition (Gigerenzer, 2007), legal
decision-making (Dhami, 2003), and social behavior in general (Hertwig,
Hoffrage, & the ABC Research Group, 2013).
The answers to the descriptive and normative question have been used
to improve experts’ decision-making and decisions “in the wild” (Gigerenzer et al., 2011). In this program of intuitive design, the resulting decision
systems are designed to fit the intuitive strategies people use. For instance,
fast-and-frugal trees (Martignon, Vitouch, Takezawa, & Forster, 2003,
Table 2) have been designed for coronary care unit allocation (Green &
Mehr, 1997), for diagnosing depression (Jenny, Pachur, Lloyd Williams,
Becker, & Margraf, 2013), for reducing civilian casualties at military checkpoints (Keller, Czienskowski, & Feufel, 2014), and for detecting vulnerable
banks (Aikman et al., 2014; Neth, Meder, Kothiyal, & Gigerenzer, 2014). In
all this applied work, the design of heuristics can provide efficient tools in
an uncertain world.
As mentioned in the beginning, optimization is not possible under uncertainty, but good decisions are. Heuristics are tools toward this end, which, as
far as we know, humans have always relied on to solve adaptive problems.
Moreover, given that humans, past and present, spend most of their time in
the twilight of uncertainty, the nature of Homo sapiens might well be found in
Homo heuristicus.
Heuristics: Tools for an Uncertain World
15
REFERENCES
Aikman, D., Galesic, M., Gigerenzer, G., Kapadia, S., Katsikopoulos, K. V., Kothiyal,
A., Murphy, E., & Neumann, T. (2014). Taking uncertainty seriously: Simplicity
versus complexity in financial regulation. Bank of England Financial Stability
Paper, 28.
Axelrod, R. M. (1984). The evolution of cooperation. New York, NY: Basic Books.
Ayton, P., & Fisher, I. (2004). The hot hand fallacy and the gambler’s fallacy: Two
faces of subjective randomness? Memory and Cognition, 32, 1369–1378.
Benartzi, S., & Thaler, R. H. (2001). Naive diversification strategies in defined contribution saving plans. American Economic Review, 91(1), 79–98.
Boyd, R., & Richerson, P. J. (2005). The origin and evolution of cultures. New York, NY:
Oxford University Press.
Dawes, R. M. (1979). The robust beauty of improper linear models in decision making. American Psychologist, 34(7), 571–582.
DeMiguel, V., Garlappi, L., & Uppal, R. (2009). Optimal versus naive diversification:
How inefficient is the 1/N portfolio strategy? Review of Financial Studies, 22(5),
1915–1953.
Dhami, M. K. (2003). Psychological models of professional decision making. Psychological Science, 14(2), 175–180.
Evans, J. S. B. T. (2008). Dual-processing accounts of reasoning, judgment, and social
cognition. Annual Review of Psychology, 59, 255–278.
Fleischhut, N., & Gigerenzer, G. (2013). Can simple heuristics explain moral inconsistencies?. In R. Hertwig, U. Hoffrage & the ABC Research Group, Simple heuristics
in a social world (pp. 459–486). New York, NY: Oxford University Press.
Geman, S., Bienenstock, E., & Doursat, R. (1992). Neural networks and the bias/
variance dilemma. Neural Computation, 4(1), 1–58.
Gigerenzer, G. (2007). Gut feelings: The intelligence of the unconscious. New York, NY:
Penguin.
Gigerenzer, G. (2008). Rationality for mortals: Risk and rules of thumb. New York, NY:
Oxford University Press.
Gigerenzer, G. (2014). Risk savvy: How to make good decisions. New York, NY: Allan
Lane.
Gigerenzer, G., & Brighton, H. (2009). Homo heuristicus: Why biased minds make
better inferences. Topics in Cognitive Science, 1(1), 107–143.
Gigerenzer, G., & Gaissmaier, W. (2011). Heuristic decision making. Annual Review
of Psychology, 62(1), 451–482.
Gigerenzer, G., & Goldstein, D. G. (1996). Reasoning the fast and frugal way: Models
of bounded rationality. Psychological Review, 103, 650–669.
Gigerenzer, G., & Goldstein, D. G. (2011). The recognition heuristic: A decade of
research. Judgment and Decision Making, 6(1), 100–121.
Gigerenzer, G., Hertwig, R., & Pachur, T. (Eds.) (2011). Heuristics: The foundations of
adaptive behavior. New York, NY: Oxford University Press.
Gigerenzer, G., & Selten, R. (2001). Rethinking rationality. In G. Gigerenzer &
R. Selten (Eds.), Bounded rationality: The adaptive toolbox (pp. 1–12). Cambridge, MA:
MIT Press.
16
EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES
Gigerenzer, G., Todd, P. M., & the ABC Research Group (1999). Simple heuristics that
make us smart. New York, NY: Oxford University Press.
Gilovich, T., Vallone, R., & Tversky, A. (1985). The hot hand in basketball: On the
misperception of random sequences. Cognitive Psychology, 17, 295–314.
Goldstein, D. G., & Gigerenzer, G. (2002). Models of ecological rationality: The recognition heuristic. Psychological Review, 109(1), 75–90.
Green, R. F. (1984). Stopping rules for optimal foragers. The American Naturalist,
123(1), 30.
Green, L., & Mehr, D. R. (1997). What alters physicians’ decisions to admit to the
coronary care unit? Journal of Family Practice, 45(3), 219–226.
Groner, R., Groner, M., & Bischof, W. (1983). Approaches to heuristics: A historical
review. In R. Groner, M. Groner & W. Bischof (Eds.), Methods of heuristics (pp. 1–17).
Hillsdale, NJ: Lawrence Erlbaum Associates.
Hertwig, R., Davis, J. N., & Sulloway, F. J. (2002). Parental investment: How an equity
motive can produce inequality. Psychological Bulletin, 128(5), 728–745.
Hertwig, R., Hoffrage, U., & the ABC Research Group (2013). Simple heuristics in a
social world. New York, NY: Oxford University Press.
Holton, G. J. (1988). Thematic origins of scientific thought: Kepler to Einstein (2nd ed.).
Cambridge, MA: Harvard University Press.
Jenny, M. A., Pachur, T., Lloyd Williams, S., Becker, E., & Margraf, J. (2013). Simple
rules for detecting depression. Journal of Applied Research in Memory and Cognition,
2(3), 149–157.
Johnson, E. J., & Goldstein, D. (2003). Do defaults save lives? Science, 302(5649),
1338–1339.
Kahneman, D. (2003). Maps of bounded rationality: Psychology for behavioral economics. American Economic Review, 93, 1499–1475.
Kahneman, D., & Tversky, A. (1972). Subjective probability: A judgment of representativeness. Cognitive Psychology, 3(3), 430–454.
Keller, N., Czienskowski, U., & Feufel, M. (2014). Tying up loose ends: A method for
constructing and evaluating decision aids that meet blunt and sharp-end goals.
Ergonomics, 1–13PMID: 24852324.
Knight, F. H. (1921). Risk, uncertainty and profit. Chicago, IL: University of Chicago
Press.
Kruglanski, A. W., & Gigerenzer, G. (2011). Intuitive and deliberate judgments are
based on common principles. Psychological Review, 118(1), 97–109.
Marewski, J. N., Pohl, R. F., & Vitouch, O. (2010). Recognition-based judgments and
decisions: Introduction to the special issue (Vol. 1). Judgment and Decision Making,
5(4), 207–215.
Marewski, J. N., Pohl, R. F., & Vitouch, O. (2011). Recognition-based judgments and
decisions: Introduction to the special issue (Vol. 2). Judgment and Decision Making,
6(1), 1–6.
Marewski, J. N., & Schooler, L. J. (2011). Cognitive niches: An ecological model of
strategy selection. Psychological Review, 118(3), 393–437.
Markowitz, H. M. (1952). Portfolio selection. The Journal of Finance, 7, 77–91.
Heuristics: Tools for an Uncertain World
17
Martignon, L., Vitouch, O., Takezawa, M., & Forster, M. R. (2003). Naïve and yet
enlightened: From natural frequencies to fast and frugal decision trees. In D. Hardman & L. Macchi (Eds.), Thinking: Psychological perspectives on reasoning, judgment
and decision making (pp. 189–211). Chichester, UK: Wiley.
Maslow, A. H. (1966). The psychology of science. New York, NY: Harper & Row.
McBeath, M. K., Shaffer, D. M., & Kaiser, M. K. (1995). How baseball outfielders determine where to run to catch fly balls. Science, 268(5210), 569–573.
Messick, D. M. (1993). Equality as a decision heuristic. In B. A. Mellers & J. Baron
(Eds.), Psychological perspectives on justice: Theory and applications (pp. 11–31).
Cambridge, MA: Cambridge University Press.
Neth, H., Meder, B., Kothiyal, A., & Gigerenzer, G. (2014). Homo heuristicus in the
financial world: From risk management to managing uncertainty. Journal of Risk
Management in Financial Institutions, 7(2), 134–144.
Newell, A., & Simon, H. A. (1972). Human problem solving. Englewood Cliffs, NJ:
Prentice Hall.
Payne, J. W., Bettman, J. R., & Johnson, E. J. (1993). The adaptive decision maker.
Cambridge, UK: Cambridge University Press.
Payne, S., Duggan, G., & Neth, H. (2007). Discretionary task interleaving: Heuristics
for time allocation in cognitive foraging. Journal of Experimental Psychology: General,
136(3), 370–380.
Pólya, G. (1945). How to solve it: A new aspect of mathematical method. Princeton, NJ:
Princeton University Press.
Ratcliff, R., & McKoon, G. (1989). Similarity information versus relational information: Differences in the time course of retrieval. Cognitive Psychology, 21(2), 139–155.
Rieskamp, J., & Otto, P. E. (2006). SSL: A theory of how people learn to select strategies. Journal of experimental psychology: General, 135(2), 207–236.
Savage, L. J. (1954). The foundations of statistics (2nd ed.). New York, NY: Dover.
Schooler, L. J., & Hertwig, R. (2005). How forgetting aids heuristic inference. Psychological Review, 112(3), 610–628.
Simon, H. A. (1955). A behavioral model of rational choice. The Quarterly Journal of
Economics, 69(1), 99–118.
Simon, H. A. (1989). The scientist as problem solver. In K. Klahr & D. Kotovsky
(Eds.), Complex information processing: The impact of Herbert A. Simon (pp. 377–398).
Hillsdale, NJ: Lawrence Erlbaum.
Todd, P. M., Gigerenzer, G., & the ABC Research Group (2012). Ecological rationality:
Intelligence in the world. New York, NY: Oxford University Press.
Tversky, A., & Kahneman, D. (1974). Judgment under uncertainty: Heuristics and
biases. Science, 185(4157), 1124–1131.
Wertheimer, M. (1923/1938). Untersuchungen zur Lehre von der Gestalt. (English
excerpts: Laws of organization in perceptual forms). In W. Ellis (Ed.), A source book
of Gestalt psychology (pp. 71–88). New York, NY: Hartcourt, Brace and Co.
Wübben, M., & von Wangenheim, F. (2008). Instant customer base analysis: Managerial heuristics often ‘get it right’. Journal of Marketing, 72(3), 82–93.
18
EMERGING TRENDS IN THE SOCIAL AND BEHAVIORAL SCIENCES
HANSJÖRG NETH SHORT BIOGRAPHY
Hansjörg Neth is researcher in Social Psychology and Decision Sciences at the
University of Konstanz, Germany, and adjunct researcher at the Center for
Adaptive Behavior and Cognition (ABC) at the Max Planck Institute for Human
Development, Berlin. His research focuses on the theoretical analysis of
adaptive behavior, interactive cognition, and ecological rationality, as well as
applied aspects of choice and heuristic decision-making under uncertainty
(e.g., in financial contexts). He has served as acting chair of Cognition, Emotion, and Communication at the University of Freiburg, taught Cognitive and
Decision Sciences at the University of Göttingen, and was research assistant
professor in Cognitive Science at the Rensselaer Polytechnic Institute. He
holds a PhD in psychology from Cardiff University, UK.
GERD GIGERENZER SHORT BIOGRAPHY
Gerd Gigerenzer is Director at the Max Planck Institute for Human Development and Director of the Harding Center for Risk Literacy in Berlin. He
is former Professor of Psychology at the University of Chicago and John M.
Olin Distinguished Visiting Professor, School of Law at the University of
Virginia. He is also Member of the Berlin-Brandenburg Academy of Sciences
and the German Academy of Sciences, and Batten Fellow at the Darden
Business School, University of Virginia. Awards for his work include the
AAAS Prize for the best article in the behavioral sciences, the Association
of American Publishers Prize for the best book in the social and behavioral
sciences, and the Communicator Award of the German Research Foundation. His award-winning popular books Calculated Risks, Gut Feelings, and
Risk Savvy have been translated into 21 languages.