PLEASE SCROLL DOWN FOR ARTICLE

7KLVDUWLFOHZDVGRZQORDGHGE\>&RNHO\(GZDUG@2Q0DUFK$FFHVVGHWDLOV$FFHVV'HWDLOV>VXEVFULSWLRQQXPEHU@3XEOLVKHU7D\ORU)UDQFLV,QIRUPD/WG5HJLVWHUHGLQ(QJODQGDQG:DOHV5HJLVWHUHG1XPEHU5HJLVWHUHGRIILFH0RUWLPHU+RXVH0RUWLPHU6WUHHW/RQGRQ:7-+8.

(QF\FORSHGLDRI/LEUDU\DQG,QIRUPDWLRQ6FLHQFHV7KLUG(GLWLRQ3XEOLFDWLRQGHWDLOVLQFOXGLQJLQVWUXFWLRQVIRUDXWKRUVDQGVXEVFULSWLRQLQIRUPDWLRQKWWSZZZLQIRUPDZRUOGFRPVPSSWLWOHaFRQWHQW W

,QIRUPDWLRQ8VHIRU'HFLVLRQ0DNLQJ(GZDUG7&RNHO\D/DHO-6FKRROHUD*HUG*LJHUHQ]HUDD&HQWHUIRU$GDSWLYH%HKDYLRUDQG&RJQLWLRQ0D[3ODQFN,QVWLWXWHIRU+XPDQ'HYHORSPHQW%HUOLQ*HUPDQ\

2QOLQHSXEOLFDWLRQGDWH'HFHPEHU

7RFLWHWKLV$UWLFOH&RNHO\(GZDUG76FKRROHU/DHO-DQG*LJHUHQ]HU*HUG,QIRUPDWLRQ8VHIRU'HFLVLRQ0DNLQJ(QF\FORSHGLDRI/LEUDU\DQG,QIRUPDWLRQ6FLHQFHV7KLUG(GLWLRQ

Full terms and conditions of use: http://www.informaworld.com/terms-and-conditions-of-access.pdf

This article may be used for research, teaching and private study purposes. Any substantial orsystematic reproduction, re-distribution, re-selling, loan or sub-licensing, systematic supply ordistribution in any form to anyone is expressly forbidden.

The publisher does not give any warranty express or implied or make any representation that the contentswill be complete or accurate or up to date. The accuracy of any instructions, formulae and drug dosesshould be independently verified with primary sources. The publisher shall not be liable for any loss,actions, claims, proceedings, demand or costs or damages whatsoever or howsoever caused arising directlyor indirectly in connection with or arising out of the use of this material.

Information Use for Decision Making

Edward T. CokelyLael J. SchoolerGerd GigerenzerCenter for Adaptive Behavior and Cognition, Max Planck Institute for Human Development,Berlin, Germany

AbstractHow should we use information to make good decisions? Historically, the view has been that normativelysuperior decision making is the product of complex optimization processes that rationally consider andintegrate all available information. Such optimization processes are well beyond the capabilities of meremortals and in many cases are computationally intractable by any means. Fortunately, optimizationprocesses are not always necessary: Less can be more. Simple decision processesi.e., heuristicsusedin the right environments enable fast, frugal, and adaptive decision making that can be as good as, orbetter than, even the most complex optimization processes. In what follows, we introduce research onecological rationality and the science of adaptive heuristics. Our review includes 1) a brief history of thestudy of decision making; 2) a discussion of simple yet computationally precise heuristics, and how,when, and why they lead to superior performance; and 3) examples of how simple heuristics are starting tobe used in the information sciences, such as in database literature prioritization or in the development ofmore user-friendly technologies. Although it may seem conventionally paradoxical, intelligent and adap-tive information use often requires that information be ignored.

INFORMATION USE FOR DECISION MAKING

In the modern world, one of the leading causes of death isheart disease. As such it is easy to imagine that manypeople will search for information on its causes, symptoms,and treatments. But how will they search? One candidate isa keyword search in a search engine such as Google, whichas of April 2008 returned about 35,500,000 heart diseaseresults in less than 2/10 of a second. This is not only a lot ofinformation, it is too much information. Given the expo-nential rates of information growth in science and technol-ogy this number will likely continue to grow.[1] Can peoplewith limited knowledge, time, and computational capaci-ties effectively search through vast amounts of informationand make good decisions under conditions of high uncer-tainty? Can a scientific understanding of the cognitive pro-cesses involved in human information search and decisionmaking help us improve information technology? In short,we believe the answer to both questions is yes.

To begin to illustrate our perspective, consider a recentstudy conducted by Michael Lee and his colleagues.[2]

Lee et al. compared the performance of a leading researchdatabase (i.e., PsycINFO) with the performance of twovery different systems, each based on a model of humandecision making. The first model was a rational modelthat attempted to combine all information in an optimal(or near-optimal) way. This model was inspired by whathas traditionally been regarded as a normative theoryof decision making. Relevance was determined using a

Bayesian learning algorithm allowing additive weightingand integration of all available information from cuessuch as authors, journals, keywords (title and abstract),language of the publication, and others. When some cuewas associated (or unassociated) with a relevant articlethe probability that another article with that cue would berelevant (or irrelevant) was updated. Eventually, searchwas prioritized by selecting articles with optimal cueconfigurations as those of the highest relevance.

The second decision model developed by Lee et al.[2]

for the literature prioritization task used a rather different,perhaps even paradoxical process. In sharp contrast to arational model, the one-reason decision-making modelignored a considerable amount of the available informationin order to prioritize choices. This model, inspired by re-search on simple, adaptive heuristics[3] used cues for infer-ring and assigning relevance, as did the rational model.However, in this case cues were first ordered by validityfrom most strongly associated and relevant to least rele-vant. Subsequently, a literature search prioritized results byselecting articles based solely on the most relevant cuethat discriminated between options (for a more detaileddiscussion of validity, discrimination, cue ordering, andone-reason decision-making, see take-the-best later in thisentry). Rather than attempting to integrate and use all possi-ble sources of information, the one-reason decision-makingmodel made choices based on a simple heuristic process(i.e., a rule of thumb), namely it prioritized results basedonly on the single most important cue that discriminated.

Encyclopedia of Library and Information Sciences, Third Edition DOI: 10.1081/E-ELIS3-120044539Copyright # 2010 by Taylor & Francis. All rights reserved. 2727

Inform

ationTe

chno

logy

Inform

ationVisu

alization

Down

load

ed B

y: [

Coke

ly,

Edwa

rd]

At:

14:1

1 4

Marc

h 20

10

To assess prioritization performance, Lee et al.[2] com-pared both decision-making models (rational, one-reason)with the performance of PsycINFO. As expected, resultsindicated that all models returned relevant articles atabove chance levels. More interestingly, and to somemore surprisingly, the one-reason decision-making modelconsistently outperformed the rational model, which inturn outperformed PsycINFO. These results add to agrowing body of evidence suggesting that more complexdecision making processes do not necessarily provide su-perior decision-making performance. Instead, superiordecision making can result from simple heuristic processeswhen used in the right environments.

So how is it possible that a simple heuristic was able toperform so well even when compared to a sophisticatedoptimization process? Part of the answer is that in ourfundamentally uncertain world, simplicity often leads torobustness and efficiency. Heuristics improve perfor-mance by ignoring potentially misleading information.Heuristics also confer other benefits as this type of choiceprocess tends be fast and efficient, relying on radicallylimited information search and decision rules. As well, inhuman, animal, and machine cognition, simple decisionprocesses can be fine-tuned and can exploit the fit bet-ween our environmental task constraints and our availablecapacities, such as a humans limited attentional capacityand our adaptive ability to forget.[46] In these ways andothers, simple heuristics tend to produce and enable eff-ective, adaptive judgment and decision making in manysituations.[3,7]

In this entry we will review a research program foc-used on adaptive behavior and cognitionor how simpleheuristics, in the right environments, can make us smarter.Our entry will begin with a discussion of the history ofdecision-making theory and will then turn to issues ofadaptive cognition, ecological rationality, and computa-tional models of heuristics. Next, we will provide someexamples of a few widely used heuristics. Finally, weclose with a discussion of implications for the design ofinformation-rich environments.

Perspectives on Rationality: A Brief History

The emergence of the modern debate on human rational-ity, or how people make decisions and what qualifies asa good decision, can be traced in large part to the Agesof Reason and Enlightenment (i.e., seventeenth andeighteenth centuries, respectively). During these timeslogic and careful, justifiable reasoning became highlyprized by philosophers, empiricists, and political actorsalike. As an example, consider the astronomer and physi-cist Pierre-Simon Laplace. Laplaces legacy includes sem-inal contributions to probability theory; however, moreimportant for our purposes, he also provided a descrip-tion of a fictional omniscient being that captured theZeitgeist of the times. This being, known as Laplaces

superintelligence, was envisioned as one who would knowall the details of past and present and with this knowledgecould readily make good choices and predict the futurewith perfect certainty.[8]

For many people, Laplaces vision of a decision makerwho is omniscient and computationally unbounded mayseem like an elaborate fantasy. Yet this fantasy or someversion of it is fundamental to much of the research andtheory in the modern decision sciences. Some readers willfind this surprising, or ironically unreasonable, but modelsof rational man and homo economicus are among themost central and influential models used in decision sci-ence. According to neoclassical economic theory peoplebehave as if they were unboundedly rational and makeoptimal (but not necessarily perfect) choices as if they hadsolved a complicated decision calculus.[9,10] These deci-sions can be described by optimization processes thatreflect peoples maximization of their own subjectiveexpected utilities (i.e., personal values) via multiattributeintegration calculations. Such theories are at the coreof dozens of models of decision making, including mod-ern theories of motivation, attitudes, and moral judg-ments.[3,7] However, even if this approach has providedinteresting and useful theory, these models often conflictwith empirical evidence. Psychological science hasclearly demonstrated that this is not how real people withlimited resources (i.e., time, attention, memory) use infor-mation to make decisions.[7,1013] Perhaps even more inter-estingly, these complicated and time-consuming processesare not even necessarily required for good, adaptive decisionmaking.[3,7]

In the mid-twentieth century, Herbert Simon[1416]

introduced his notion of bounded rationality. Simon ar-gued that, among other things, people have only limitedtime, knowledge, and cognitive resources, and thus hu-man decision makers cannot carry out the types of opti-mization computations that were (and still are) oftenassumed to underlie rational decision making. In part inreaction to Simons notion of bounded rationality, a moremodern version of a rational decision maker was devel-oped by some decision researchers. The new theory, char-acterized by optimization under constraints, was onceagain concerned with a complex optimization calculusthat could be used to make rational choices, but in thesecases the optimization processes were also subject to anyone of a number of constraints.[17] To illustrate, in somemodels information search was necessary (i.e., no moreomniscience) and so search processes were described andthus carried a cost. However, as a simplifying assumptionthe information search processes were often describedby yet another optimization calculus wherein search wasterminated according to the optimal costbenefit ratio. Asthe number of constraints increases so too do the compu-tational complexities of the search optimization functionsthat must be solved, even in simple decisions. In thisway, optimization under constraints can be even more

2728 Information Use for Decision Making

Information

TechnologyInform

ationVisualization

Down

load

ed B

y: [

Coke

ly,

Edwa

rd]

At:

14:1

1 4

Marc

h 20

10

computationally demanding than what is required of theunboundedly rational agents. In efforts toward greaterpsychological plausibility, the optimization under con-straints models created in some cases an even more psy-chologically unrealistic, and computationally unbounded,superintelligent decision maker.[17]

To be fair, although optimization models are not accurateprocess models of how real people make decisions, they doin some cases predict peoples decisions or tell us about whatoutcomes or decisions should be favored.[1820] For thesereasons, optimization models can be valuable tools. Never-theless, optimization models are as-ifmodels of human deci-sion making: Many decision makers behave as if they useoptimization processes even though they rely on otherprocesses.[2123] Unfortunately, most research focusing onas-if models is essentially uninterested in and ill-equippedto understand the psychological dynamics involved in de-cision making. Hence, the as-if approach to modeling deci-sion making leaves unanswered at least two crucialquestions. First, how do people actually use information tomake decisions? Second, given our known cognitive con-straints (i.e., our bounded rationality), how is it that realdecision processes can approximate and in some cases out-perform optimization processes? To answer these ques-tions, we turn to two different but related researchprograms that study human decision making and heuristics.

In the 1970s, Daniel Kahneman and Amos Tverskydeveloped the heuristics and biases research program incognitive psychology.[11,24,25] This program was aimedat understanding how people actually make decisions.Toward this end researchers worked to reveal cognitiveprocesses (i.e., heuristics) by focusing on judgment errorsand biases. However, in order to identify errors onemust have normative assumptions (i.e., what is the appro-priate standard for an accurate or good judgment), anissue that is not without controversy.[26,27] In the case ofthe heuristics and bias approach, it was assumed thathuman cognition should be compared to a very specificset of rational, normative standards such as the outcomesof optimization processes and logic. Specifically, theheuristics and biases program searches for errors that areevidenced when peoples judgments deviate from anestablished fact. . . [or] an accepted rule of arithmetic,logic, or statistics (p. 493).[28] Indeed, it has demon-strated rather unequivocally that people do not reason inaccord with content-blind logical laws or optimizationprocesses, and that people often use heuristic processesfor judgments and decisions.[10,29] The research and find-ings on heuristics and biases have played key roles inshaping psychological and behavioral decision makingresearch, contributing to the development of new researchfields such as behavioral law and economics.[24,30]

In spite of its many successes, the heuristics and biasesprogram has its limitations. One of our most serious con-cerns is that the program has emphasized ways in whichheuristics are associated with errors, which has led some

to an interpretation that heuristic use is a problem thatneeds to be corrected. In this light heuristics are seen asinferior choice processes designed to be used by computa-tionally disadvantaged individuals. In contrast, other res-earch demonstrates that heuristics (e.g., satisficing) areoften powerful tools.[16,31,32] As illustrated by the literatureprioritization example in the introduction, in real-worldenvironments these simple processes can enable adaptivedecision making, matching or outperforming even the mostsophisticated and time-consuming optimization processes.

Ecological Rationality: Computational Models ofHeuristics

Consider for a moment a Darwinian inspired perspectiveon decision making. On this view, the goals and needs oforganisms, such as finding food, securing mates, or pro-tecting offspring, may or may not benefit from cognitionthat is logically coherent. For these organisms, fitness isbest served when cognition can be tuned to ecologicalconstraints. An organisms success will rely on the extentto which its cognition and behavior can benefit from andexploit features of its internal and external environment,regardless of how well these processes actually adhere tological norms. An organisms ability to survive and repro-duce depends on the fit between: 1) its evolved and devel-oped capacities, 2) its cognitive processes, 3) and thestructure of its natural environment. The analysis of the fitbetween capacities, processes, and environment is knownas the study of ecological rationality.

Understanding the relationship between the mind andits environment serves as a starting point for the study ofadaptive behavior and cognition.[32] This adaptive frame-work has deep roots in psychology. For example, HerbertSimon has argued that Human rational behavior isshaped by a scissors whose two blades are the structureof task environments and the computational capabilitiesof the actor (p. 7).[16] If one wants to understand howpeople make judgments and decisions, and why and whenthese processes work, one cannot examine only processes(as most psychologists do) or only environments (as mosteconomists and sociologists do). Studying how one scis-sor blade cuts, or fails to cut, does not tell us how, why, orwhen the scissors will actually work.

In concert with the study of ecological rationality, asecond key concept for the research program on adaptivebehavior and cognition is the adaptive toolbox. This adap-tive toolbox is conceived of as a collection of preciselydefined cognitive heuristics and other adaptive processesthat can be used both consciously and unconsciously (e.g.,intuitively) to solve problems in the real world. Organ-isms use a number of specific tools that are well-suited tocertain task environments, not entirely unlike specificwrenches and screwdrivers that are designed for specificchores. In this way, there is no general all-purpose heuris-tic (or optimization calculus) that can provide the best

Information Use for Decision Making 2729

Inform

ationTe

chno

logy

Inform

ationVisu

alization

Down

load

ed B

y: [

Coke

ly,

Edwa

rd]

At:

14:1

1 4

Marc

h 20

10

solution for every judgment or decision. In contrast, heuris-tic application is more narrow and constrained, althoughmany heuristics can be used more or less effectively acrossa range of situations (e.g., a screwdriver can also be used toopen a can of paint). Perhaps most critically, by definingheuristics as formal computational models, one can thenidentify (e.g., via simulations) exactly when, how, and whycertain heuristics will succeed or fail. Theoretically, thiscollection of adaptive heuristics provides a precisely def-ined, psychologically plausible alternative to as-if optimi-zation models of human decision making.

So how is it that these simple heuristic processes canmatch and outperform complex optimization processes?The answer has at least two key parts. The first part has todo with computational tractability. Regardless of the ext-raordinary increases in computing power that we haveenjoyed in recent years, for most decisions there simplyis no optimization strategy that can be computed.[3] Opti-mization, except in radically simplified environments, islargely computationally intractable. Even simple pro-blems turn out to be so complex as to be impenetrable tooptimization techniques. This is true of problems that arewell-defined such as in games like Chess or Go, as well asfor the many ill-defined problems we face in the modernworld such as choosing an outfit, a dinner, a career, or aspouse. In all of these cases, heuristic processes are notonly valuable but they are absolutely essential.

The second reason heuristics can perform well is thatthey are robust and reduce the chance of fitting noise inour environment (i.e., overfitting). Because we live in afundamentally uncertain world, information in our envi-ronment (i.e., environmental signals) consists, of both reli-able and unreliable content (i.e., noise), or in other words,information that is diagnostic and information that couldlead us astray. Particularly in situations that involve highuncertainty, the trick for intelligent decision making is toignore information, such as seemingly informative butultimately irrelevant information (e.g., past performanceof stocks for predicting changes in valuation). If we per-formed an optimization calculation for every decision thatincluded every regularity from every previous situation,our ability to accurately predict outcomes in new environ-ments would be crippled as it would be based on manykinds of random, non-diagnostic noise. In contrast, bec-ause heuristics are simple and exploit cognitive capacitiessuch as our adaptive capacity for forgetting[4] we canreduce the risk that we will overfit non-diagnostic envi-ronmental signals. Moreover, because adaptive heuristicsrely on a limited search of the available information thereis a good chance that they will avoid most sources of therandom noise while still benefiting from a focus on themost reliable sources of information.[33]

In addition to the performance benefits noted aboveadaptive heuristics confer at least one more set of eco-logically important advantages: Heuristics can providefast and frugal decision making. Because heuristics can

exploit our capacities (both evolved and developed), suchas our remarkable capacity for highly accurate recogni-tion memory, the time and energy needed for these deci-sion processes are minimized. Heuristics provide essentialcompetitive advantages (e.g., minimizing time, search,effort invested) whether one is making decisions in thewild or in the boardroom. This frugality may benefit usphysiologically as our brain, which is roughly 2% of ouroverall mass, commonly requires between 15% and 25%of our daily energy and oxygen budget. Even simple actsof deliberative, effortful processing can significantly inf-luence and deplete our energy stores.[34] In these waysand others simple heuristics tend to enable adaptive deci-sions in ecological environments.

Heuristic Building Blocks

Advances in the science of adaptive behavior and cogni-tion rely in large part on identifying and modelingexact heuristic processes. These processes are describedby dividing heuristics into different building blocks (i.e.,subprocesses), often including: 1) search processes,2) stopping rules, and 3) decision rules, which are com-posed of either unique processes or nested combinationsof other more basic heuristics. Consider the take-the-bestheuristic.[34] Take-the-best belongs to a family of onereason decision-making heuristics and has proven itselfto be a common and effective decision tool. Take-the-bestcan be used when one is faced with a choice between twodifferent options. In these cases, take-the-best orders andconsiders different cues (reasons) for making its decisionby selecting the best option with the first cue that discri-minates between the options. In simpler language, con-sider a preference situation in which one is trying todecide which of two colleges to attend. One way youcould make a decision would be to focus on only a singlefactor, the one that seems most important. For some peo-ple that might be prestige, and thus when considering bothHarvard and say San Francisco State University (SFSU),one would likely pick Harvard. However, if the mostimportant consideration was which school was affordable,you might then select the public school (SFSU) wheretuition is far less. In either case, the process is roughlythe same and requires only a very minimum amount ofsearch and reasoning. First identify and order the mostimportant cues that you will use to decide and then makeyour choice based on the first discriminating cue (i.e., thefirst cue in which one option is clearly better than theother). Indeed, in investigations of real world choicessuch as the one described above (but also includingmultiple-cue inference tasks where there are known cor-rect answers such as inferring which of two cities islarger) we see that people often make decisions withsimple processes like take-the-best.[36,37]

More formally, take-the-best is made of three preciselydefined building blocks including:

2730 Information Use for Decision Making

Information

TechnologyInform

ationVisualization

Down

load

ed B

y: [

Coke

ly,

Edwa

rd]

At:

14:1

1 4

Marc

h 20

10

1. Search rule: Search through the cues in order of theirvalidity (correlation with criteria). Look up the cuevalues of the cue with the highest validity first.

2. Stopping rule: If one object has a positive cue valueand the other does not (or is unknown), then stopsearch and proceed to Step 3. Otherwise, exclude thecue and return to Step 1. If no more cues are found,guess.

3. Decision rule: Predict that the object with the positivecue value has the higher value on the criterion.

Again, each of these rules precisely defines behavior,leads to a testable prediction, and therefore can be com-putationally modeled or evaluated empirically. Given thisprecision, researchers have also used mathematical analy-sis and simulations to tell us about the ecological rational-ity of heuristics, describing for example the environmentswherein take-the-best will perform better and worse.[38]

To illustrate, one can analytically prove that take-the-bestis associated with superior decision making in environ-ments that have a non-compensatory cue structure. Anytime the sum of the cue weights does not add up to thesum of the best (remaining) cue, take-the-best will per-form very well. We might imagine a task of trying tomodel which libraries have experienced a budget growthof at least 10%. In this example we will assume that theecological validities (i.e., the relative frequency withwhich a binary cue correctly predicts the criterion) of thecues are one, one-half, one-fourth, and one-eighth, wherelending rates of the library are the single most importantfactor (i.e., more lending is most strongly associated withstrong growth), the current size of the library is one-halfas important, and location is half again as important. Inthis case, the structure is non-compensatory because thesum of the cue weights can never add up to (or compen-sate for) the difference between the first cue (lendingrates) and all other cues (Fig. 1).[39] Even if one finds thatanother library is larger and is in the perfect location, etc.,these factors could not compensate if that library had lowlending rates. Findings such as these can tell us both whena simple heuristic will fail and when it will work betterthan or at least as well as more complex decision pro-cesses. These findings also allow for insights into whatdecision (or inference) should be made as well as whatinformation should actually be used and ignored in thedecision-making process.

Two Memory-Based Heuristics

Many heuristics benefit by taking advantage of evolvedand developed capacities. In humans, some capacities aremore fundamental than others, in the sense that they de-velop earlier and persist longer. Recognition memory isone such capacity. In many ways our recognition memoryis much more sensitive and reliable than the vast majority

of other mnemonic processes, something that becomesincreasing apparent as we age. To better appreciate thiscapacity we invite you to try an exercise: Do your best torecall the names of the seven dwarves from the storySnow White and the Seven Dwarves. Although many ofus will have trouble recalling these characters namesnearly everyone, or at least most Americans, will imme-diately recognize them (Can you recognize the seven realdwarves in this list of nine: Bashful, Doc, Sloppy,Dopey, Grumpy, Pudgy, Happy, Sleepy, andSneezy). As noted, recognition not only reflects a fun-damental aspect of memory, it also provides a powerfulcue for adaptive decision making.

Considerable research now demonstrates that whenfacing a choice people often rely on their recognition, orlack thereof, to help inform their inferences. This heuris-tic process has been formalized in the recognition heuris-tic, which is modeled as follows: if one of two objects isrecognized and the other is not, then infer that the recog-nized object has the higher value with respect to thecriterion.[40] The interesting theoretical finding is thatsometimes less is more, such that ignorance will actuallyimprove our decision making in a variety of environ-ments. Imagine for instance that your task is to selectwhich of two stocks to invest in. One company might benamed Coca Cola Company and the other Ameritech. Oneway to decide between the two companies is simply toinvest in the company that is recognized and familiar. Ifyou do this, and thus decide according to the recognitionheuristic, youd likely select Coca Cola. Although it is asimple strategy that relies largely on ones ignorance,research shows that it can match or outperform expertstock portfolios, in certain common environments[41,42]

(for boundary conditions see also Andersson).[43]

Fig. 1 A noncompensatory environment. In this environmentone can see that the first (binary) cue will always make predic-tions that are always as accurate as any linear combination ofall cues.

Information Use for Decision Making 2731

Inform

ationTe

chno

logy

Inform

ationVisu

alization

Down

load

ed B

y: [

Coke

ly,

Edwa

rd]

At:

14:1

1 4

Marc

h 20

10

By way of analysis and computational modelling it ispossible to understand the conditions in which recognitionwill be likely to provide good (and not so good) infer-ences. Research indicates that recognition will lead togood decisions whenever it is correlated with a decisioncriterion. Consider the task of trying to decide which oftwo cities is larger (in population), selecting betweenDetroit and Houghton (MI), for example. In this case,recognition may be diagnostic as cities that have morepeople are also more likely to be mentioned in the mediaand have more major sports teams, companies, airports,events, and so on. Bigger cities are therefore more likelyto be recognized and not recognizing a city becomes avaluable basis for inference. Of course, recognition doesnot guarantee that one city is larger than another (e.g.,Hollywood, CA, vs. Fresno, CA). However, when thereis a strong correlation between recognition and the deci-sion criteria, recognition will tend to provide accurateinferences.

The fluency heuristic[44,45] is a second fundamentalmemory-based inference process (for a recent review seeKelley).[46] Considerable converging evidence revealsthat fluency, which is roughly defined as our subjectiveexperience of the ease of information processing (cf.availability), often dramatically influences our confi-dence, judgment, and decision making.[24,25,4749] If weagain consider the real-world task of selecting stocks wefind that people tend to judge that unrecognized compa-nies with names that are easier to read (pronounce) aremore likely to be profitable. This is true in laboratorystudies and can also be seen in actual stock trading duringinitial public offerings.[50,51] Indeed, there is a long tradi-tion of studying fluency and availability. Here we focuson more recent research done within the adaptive frame-work. Specifically, we have developed the ACT-R fluencyheuristic that formalized: If one of two objects is morefluently processed, then infer that this object has thehigher value with respect to the criterion (p. 612).[4] Inthis case, fluency is precisely defined as the speed withwhich declarative memory chunks can be retrieved inthe computational cognitive architecture ACT-R. As withthe recognition heuristic and take-the-best, the ACT-Rfluency heuristic relies on a simple but powerful onereason decision-making process. The critical difference isnot whether or not one recognizes some bit of informationbut how quickly and easily one recognizes it. The moreeasily one option is recognized the more likely it is toinfluence our decision, a process that is fundamental to avariety of adaptive inferences. Understanding these pro-cesses provides insight into the adaptive nature of ourcognitive architecture. More generally, understanding theprocesses used by one of the most complex and efficientadaptive devices knowni.e., the human mindcan helpenable the design of better innovative engineering andinformation technology solutions.[52]

Heuristics, Usability, and Environmental Design

The work of Lee et al.[2] on search prioritization serves asan excellent example of the potential for using adaptiveheuristics to improve information technology. However,this is only one side of the usability issue. Opportunitiesalso exist to study user decision making to further under-stand the heuristic processes commonly relied upon in data-base and library type environments. Currently there is asmall but growing body of work using process-oriented orotherwise ecologically grounded approaches that may bevaluable for future research efforts.[53,54] For example,some research focuses on variations in human search strate-gies in database environments[55] such as differences in theprocesses of children and adults (Bilal and Kirby, 2002).[61]

In these studies, key results indicate that specific informa-tion search strategies can be identified that tend to system-atically vary with age. Similarly, there is a somewhat moredeveloped tradition of studying human search behavior ingeneral consumer choices that has begun to explore Internetor database-type environments.[5759] These studies havealso been successful in identifying search strategies, withsome going a step further and identifying stopping rulesand the environmental constraints that mediate heuristicselection.[60] For instance, Browne et al. have started toidentify the relationships between environmental factorsand search processes, suggesting that there often exists apositive relationship between task complexity, searchtimes, and stopping rules. Similarly, other recent work hasused our understanding of human decision-making heuris-tics to improve humancomputer interface design. Byeliminating non-diagnostic or highly redundant (correlated)product-attribute information researchers limit the redun-dant search processes of users, which allows consumers tobetter cope with information overload effects including theso-called too much choice effect.[62]

There are also a number of other connections betweenspecific heuristics and library and information searchbehaviors that have yet to be explored. As noted, bothrecognition and fluency tend to influence everydaychoices. In a library environment it is easy to imaginethat these heuristics would influence decisions such aswhich book(s) to order or read (e.g., an author is recog-nized, an author is more quickly recognized); whatsearch queries to use (e.g., do you recognize any resultsfrom these keywords); and when to stop searching (e.g.,when further search yields little in the way of relevantrecognizable information; stop search when informationis no longer easily recognized or fluent). However, thesesimple connections and the review provided here canonly serve as an introduction. There are many otherheuristics and aspects of information use and searchprocesses that may prove to be very influential in thedesign of user-friendly technology (for a more completeoverview see Gigerenzer).[7,32]

2732 Information Use for Decision Making

Information

TechnologyInform

ationVisualization

Down

load

ed B

y: [

Coke

ly,

Edwa

rd]

At:

14:1

1 4

Marc

h 20

10

CONCLUSION

Over the last few decades there has been considerablefocus on the fact that people often use heuristics to makedecisions. Fortunately, in the right environment thesesimple heuristic processes make us smart by simplifyingand speeding decisions, and by helping us ignore irrele-vant information. In our complex and uncertain worldsimple heuristics sometimes represent the very best deci-sion making processes available. Whether designing betterinformation technology, better decision support systems, orbetter decision environments one lesson is clear: Superiordecision making does not necessarily require more com-plex processes. Less can be much more. In these ways andothers, the future holds great opportunities and promise forengineering simplicity.

REFERENCES

1. Egghe, L.; Rao, I.K.R. Classification of growth modelsbased on growth rates and its applications. Scientometrics1992, 25, 546.

2. Lee, M.D.; Loughlin, N.; Lundberg, I.B. Applying onereason decision-making: The prioritisation of literaturesearches. Aust. J. Psychol. 2002, 54 (3), 137143.

3. Gigerenzer, G.; Todd, P.M.; and the ABC Research Group.Simple Heuristics That Make Us Smart; Oxford UniversityPress: Oxford, 1999.

4. Schooler, L.J.; Hertwig, R. How forgetting AIDS heuristicinference. Psychol. Rev. 2005, 112, 610628.

5. Anderson, J.R.; Schooler, L.J. Reflections of the environ-ment in memory. Psychol. Sci. 1991, 2, 396408.

6. Schooler, L.J.; Anderson, J.R. The role of process in therational analysis of memory. Cogn. Psychol. 1997, 32 (3),219250.

7. Gigerenzer, G.; Selten, R., Eds. Bounded Rationality: TheAdaptive Toolbox; MIT Press: Cambridge, MA, 2001.

8. Gigerenzer, G. Bounded and rational. In ContemporaryDebates in Cognitive Science (Contemporary Debates inPhilosophy No. 7); Stainton, R.J., Ed.; Blackwell: Oxford,U.K., 2006.

9. Hastie, R. Problems for judgment and decision making.Annu. Rev. Psychol. 2001, 52, 652683.

10. Shafir, E.; Tversky, A. Decision making. In Thinking;Smith, E.E., Osherson, D.N., Eds.; MIT Press: Cambridge,MA, 1995; Vol. 3.

11. Kahneman, D.; Slovic, P.; Tversky, A. Judgment underUncertainty: Heuristics and Biases; Cambridge Press:New York, 1982.

12. Payne, J.W.; Bettman, J.R.; Johnson, E.J. Behavioral deci-sion researchA constructive processing perspective.Annu. Rev. Psychol. 1992, 43, 87131.

13. Payne, J.W.; Bettman, J.R.; Johnson, E.J. The AdaptiveDecision Maker; Cambridge University Press: Cambridge,U.K., 1993.

14. Simon, H.A. Rational choice and the structure of the envi-ronment. Psychol. Rev. 1956, 63, 129138.

15. Simon, H.A. Models of Man: Social and Rational; Willey:New York, 1957.

16. Simon, H.A. Invariants of human behavior. Annu. Rev.Psychol. 1990, 41, 119.

17. Sargent, T.J. Bounded Rationality in Macroeconomics;Oxford University Press: New York, 1993.

18. Chater, N.; Oaksford, M. Ten years of the rational analysisof cognition. Trends Cogn. Sci. 1999, 3 (2), 5765.

19. Chater, N.; Oaksford, M.; Nakisa, R.; Redington, M. Fast,frugal, and rational: How rational norms explain behavior.Organ. Behav. Hum. Decis. Process. 2003, 90 (1), 6386.

20. Chater, N.; Tenenbaum, J.B.; Yuille, A. Probabilistic mod-els of cognition: Conceptual foundations. Trends Cogn.Sci. 2006, 10, 287291.

21. Brandstatter, E.; Gigerenzer, G.; Hertwig, R. The priorityheuristic: Making choices without trade-offs. Psychol.Rev. 2006, 113 (2), 409432.

22. Weber, E.U.; Johnson, E.J.; Milch, K.F.; Chang, H.;Brodscholl, J.C.; Goldstein, D.G. Asymmetric discountingin intertemporal choiceA query-theory account. Psychol.Sci. 2007, 18, 516523.

23. Cokely, E.T.; Kelley, C.M. Cognitive abilities and superiordecision making under risk: A protocol analysis and pro-cess model evaluation. Judgement and Decision making.2009, 4, 2033.

24. Kahneman, D.; Tversky, A., Eds. Choices, Values andFrames; Cambridge University Press and the Russell SageFoundation: New York, 2000.

25. Tversky, A.; Kahneman, D. Judgment under uncertainty:Heuristics and biases. Science 1974, 185, 11241130.

26. Anderson, J.R. Is human cognition adaptive? Behav. BrainSci. 1991, 14 (3), 471484.

27. Gigerenzer, G. On narrow norms and vague heuristics:Reply. Psychol. Rev. 1996, 103, 592596.

28. Kahneman, D.; Tversky, A. On the study of statisticalintuitions. In Judgment under Uncertainty: Heuristicsand Biases; Kahneman, D., Slovic, P., Tversky, A., Eds.;Cambridge University Press: Cambridge, U.K., 1982.

29. Kahneman, D. A perspective on judgment and choiceMapping bounded rationality. Am. Psychol. 2003, 9, 697720.

30. Gigerenzer, G.; Engel, C., Eds. Heuristics and the Law;MIT Press: Cambridge, MA, 2006.

31. Gigerenzer, G. Fast and frugal heuristics: The tools ofbounded rationality. In Handbook of Judgment and Deci-sion Making; Koehler, D., Harvey, N., Eds.; Blackwell:Oxford, U.K., 2004.

32. Gigerenzer, G.; Brighton, H. Homo heuristicus: why bi-ased minds make better inferences. Topics in CognitiveSciences. 2009, 1, 107143.

33. Brighton, H.; Gigerenzer, G. Bayesian brains and cognitivemechanisms: Harmony or dissonance? In The ProbabilisticMind: Prospects for Rational Models of Cognition; Chater,N., Oaksford, M., Eds.; Oxford University Press: Oxford,U.K., 2007.

34. Gailliot, M.T.; Baumeister, R.F. The physiology of will-power: Linking blood glucose to self-control. Pers. Soc.Psychol. Rev. 2007, 11, 303327.

35. Gigerenzer, G.; Goldstein, D.G. Reasoning the fast andfrugal way: Models of bounded rationality. Psychol. Rev.1996, 103, 650669.

Information Use for Decision Making 2733

Inform

ationTe

chno

logy

Inform

ationVisu

alization

Down

load

ed B

y: [

Coke

ly,

Edwa

rd]

At:

14:1

1 4

Marc

h 20

10

36. Galotti, K.M. Decision structuring in important real-lifechoices. Psychol. Sci. 2007, 18 (4), 320325.

37. Dhami, M.K. Psychological models of professional deci-sion making. Psychol. Sci. 2003, 14, 175180.

38. Katsikopoulos, K.V.; Martignon, L. Naive heuristics forpaired comparisons: Some results on their relative accu-racy. J. Math. Psychol. 2006, 50, 488494.

39. Martignon, L.; Hoffrage, U. Why does one-reason decisionmaking work? A case study in ecological rationality. InSimple Heuristics That Make Us Smart; Gigerenzer, G.,Todd, P.M., the ABC Research Group, Eds.; OxfordUniversity Press: New York, 1999; 119140.

40. Goldstein, D.G.; Gigerenzer, G. Models of ecologicalrationality: The recognition heuristic. Psychol. Rev. 2002,109, 7590.

41. Borges, B.; Goldstein, D.G.; Ortmann, A.; Gigerenzer, G.Can ignorance beat the stock market? In Simple heuristicsthat Make Us Smart; Gigerenzer, G., Todd, P.M., the ABCResearch Group, Eds.; Oxford University Press: NewYork, 1999.

42. Ortmann, A.; Gigerenzer, G.; Borges, B.; Goldstein, D.G.The recognition heuristic: A fast and frugal way to invest-ment choice? In Handbook of Experimental EconomicsResults: Vol. 1 (Handbooks in Economics No. 28); Plott,C.R., Smith, V.L., Eds.; North-Holland: Amsterdam, 2008;9931003.

43. Andersson, P.; Rakow, T. Now you see it now you dont:The effectiveness of the recognition heuristic for selectingstocks. Judgment Decis. Making 2007, 2, 2939.

44. Jacoby, L.L.; Brooks, L.R. Nonanalytic cognitionmem-ory, perception, and concept-learning. Psychol. Learn.Motiv. Adv. Res. Theory 1984, 18, 147.

45. Jacoby, L.L.; Dallas, M. On the relationship between auto-biographical memory and perceptual learning. J. Exp.Psychol. Gen. 1981, 3, 306340.

46. Kelley, C.M.; Rhodes, M.G. Making sense and nonsenseof experience: Attributions in memory and judgment. InPsychology of Learning and Motivation: Advances inTheory and Research; Ross, B.H., Ed.; 2002; Vol. 41,293320.

47. Jacoby, L.L.; Kelley, C.; Brown, J.; Jasechko, J. Becomingfamous overnightLimits on the ability to avoid uncon-scious influences of the past. J. Pers. Soc. Psychol. 1989,56, 326338.

48. Jacoby, L.L.; Woloshyn, V.; Kelley, C. Becoming famouswithout being recognizedUnconscious influences ofmemory produced by dividing attention. J. Exp. Psychol.Gen. 1989, 118, 115125.

49. Tversky, A.; Kahneman, D. Availability: A heuristic forjudging frequency and probability. Cogn. Psychol. 1973, 5,207232.

50. Alter, A.L.; Oppenheimer, D.M. Predicting short-termstock fluctuations by using processing fluency. Proc. Natl.Acad. Sci. 2006, 103, 93699372.

51. Cokely, E.T.; Parpart, P.; Schooler, L.J. On the link betweencognitive control and heuristic processes. In Proceedingsof the 31st Annual Conference of the Cognitive ScienceSociety, Taatgen, N.A., van Rijn, H., Eds.; Austin, TX:Cognitive Science Society, 2009; 29262931.

52. Pirolli, P.; Fu, W.T. SNIF-ACT: A model of informa-tion foraging on the world wide web. In Proceedings ofthe Ninth International Conference on User Modeling,Johnstown, PA, 2003.

53. Marewski, J.N.; Galesic, M.; Gigerenzer, G. Fast and fru-gal media choices. In Media Choice: A Theoretical andEmpirical Overview, Hartmann, T., Ed.; Routledge: NewYork and London, 2009; 107128.

54. Van Maanen, L.; Marewski, J. N. Recommender systemsfor literature selection: A competition of decision makingand memory models. In Proceedings of the 31st AnnualConference of the Cognitive Science Society, Taatgen, N.A.,van Rijn, H., Eds.; Austin, TX: Cognitive Science Society,2009; 29142919.

55. Thatcher, A. Information-seeking behaviours and cogni-tive search strategies in different search tasks on theWWW. Int. J. Indus. Ergon. 2006, 36, 10551068.

56. Dresang, E.T. The information-seeking behavior of youth inthe digital environment. Libr. Trends 2005, 54, 178196.

57. Ozanne, J.L.; Brucks, M.; Grewal, D. A study of informa-tion search behavior during the categorization of new pro-ducts. J. Consum. Res. 1992, 18, 452463.

58. Guo, C. A review on consumer search: Amount and deter-minants. J. Bus. Psychol. 2001, 51, 505519.

59. Zhang, J.; Fang, X.; Sheng, O.R.L. Online consumersearch depth: Theories and new findings. J. Manage.Inform. Syst. 20062007, 23, 7195.

60. Browne, G.J.; Pitts, M.G.; Wetherbe, J.C. Cognitive stop-ping rules for terminating information search in onlinetasks. Mis Quart. 2007, 31, 89104.

61. Bilal, B.; Kirby, J. Differences and similarities in informa-tion seeking: Children and adults as Web users. Inform.Process. Manage. 2002, 38, 649670.

62. Fasolo, Barbara; McClelland, Gray H; Todd, Peter M.Escaping the tyranny of choice: When fewer attributesmake choice easier. Marketing. 2007, 7 (1). 1326.

2734 Information Use for Decision Making

Information

TechnologyInform

ationVisualization

Down

load

ed B

y: [

Coke

ly,

Edwa

rd]

At:

14:1

1 4

Marc

h 20

10