J/DM 1998 schedule

November 21 - 23
Wyndham Anatole Hotel
Dallas, Texas

Saturday, Nov 21

Graduate Student Social Event (9pm in the hotel sports bar)

10:15-11:55

Psychonomic Society: J/DM I (Grand Ballroom B)

10:15-10:25

Preferences for Future Generations
Gretchen B. Chapman (Rutgers U.)
Three scenarios presented life- saving programs that differed in the generation and time (year) in which lives were saved and the age of the recipients. Eighty-three undergraduates indicated the number of lives saved that would make the program just as attractive as one that saved 100 lives now. Almost half the subjects showed zero discounting, valuing present and future lives equally. The remaining subjects showed low discount rates (3%-5% annually) and marked consistency across questions.

10:30-10:40

Failure to Demonstrate the Sunk Cost Effect in a Field Experiment
Carla R. Scanlan & Hal R. Arkes (Ohio U.)
Students participating in a term-long dormitory meal plan were randomly selected to receive either a $200 reduction in the cost of their meal plan or certificates worth $200 at a local mall. We obtained results contrary to the sunk cost effect: Compared with those given gift certificates, those given a price reduction in their meal plan tended to eat a higher proportion of their meals in the dormitory.

10:45-11:00

Modeling Individual Differences in Risk Attitude.
Lola L. Lopes & Gregg C. Oden (U. of Iowa)
SP/A theory is used to model individual differences in risk attitude. The parameters of the model describe two criteria that contribute independently to risky choice. The SP (security-potential) criterion captures differences in the relative attention paid to best-case and worst-case gamble outcomes. The A (aspiration) criterion captures differences in the impact of outcome goals on choice. Fitted parameter values reveal systematic processing differences between subjects selected for high and low security mindedness.

11:05-11:25

The Susceptibility of Young and Old Adults to Positive and Negative Outcomes of Recent Decisions.
Vered Rafely & Itiel E. Dror (Southampton U.) & Jerome R. Busemeyer (Indiana U.)
Young and old participants performed a decision-making task in which the outcomes of recent decisions were manipulated. The young participants turned more conservative in response to negative outcomes, but were not affected by positive outcomes. In contrast, the older participants did not change their decisions in response to either positive or negative outcomes. The data are interpreted in terms of sequential sampling models and deterioration in accessing recent memory with aging.

11:30-11:50

Oppositional Deliberation: Toward Explaining Overconfidence and Its Cross-Cultural Variations
J. Frank Yates (U. of Michigan), Ju-Whei Lee (Chung Yuan U.), Hiromi Shinotsuka (Hokkaido U.), & Winston R. Siech (U. of Michigan)
People are often overconfident in their general knowledge, and cultures differ reliably in this tendency. In the present research, subjects "thought aloud" as they responded to general knowledge questions. Two key results emerged: (1) Overconfidence occurred strongly only when respondents approached their task via the generation of arguments, which typically neglected arguments inconsistent with the answers respondents actually chose. And (2) respondents in different cultures differed substantially in their inclination to generate such "choice-opposing" arguments.

1:30-3:15

Psychonomic Society: J/DM II (Grand Ballroom B)

1:30-1:45

Explaining Attribute Framing Effects: The Half-Full/Half- Empty Phenmenon.
Irwin P. Levin & Gary J. Gaeth (U. of Iowa), Sandra L. Schneider, (U. of S. Florida) & Victoria L. Phillips & Kirsten I. Redalen (U. of Iowa)
The half-full/half-empty cup, the procedure that has a 50% success rate/50% failure rate, and the beef that is 85% lean/15% fat are all examples of attribute framing. We provide support for an associative model of attribute framing effects by showing that subjects are more apt to endorse items such as "hopeful" and "optimistic" in response to scenarios that include positively framed attributes than to scenarios that include negatively framed attributes.

1:50-2:10

Framing Effects in Medical Treatment Choices: Anchoring Not Risk Preference.
Sandra L. Schneider & Andrea L. Washburne (U. of S. Florida)
Previous framing studies have demonstrated conflicting results regarding preferences for medical treatment options involving tradeoffs between risks. Combining treatment features from McNeil et al. (1982) and O'Connor et al. (1985), we manipulated short-term survival outlook, quality of life, and long-term life expectancy to determine how these factors interact in positive and negative frames. We also measured ratings of option quality and perceived riskiness. Risk aversion predominated in both frames, but framing effects also occurred.

2:15-2:35

Effects of Task Sequence, Difficulty, and Type on Probability Judgments.
Thomas S. Wallsten & Hongbin Gu (U. of North Carolina, Chapel Hill)
Following training, respondents estimated probabilities regarding from which of two a priori equally likely equal-variance normal distributions observations were drawn. Task difficulty varied within respondents (hard, d' = 1, vs. easy, d' = 1.75) and task sequence varied between. The hard-easy effect was strongly moderated by sequence, due to carryover of response-criterion settings. The subjective estimates in this and related tasks depended on sample likelihood ratios. In contrast, they depend on distance from a neutral point in general knowledge tasks.

2:40-2:50

Decision Making and Affective Experience in Mathematical Problems.
Michel Cabanac (U. Laval) & Jacqueline Guillaume (U. of Grenoble)
This study measured the affective experience of 12 subjects reading correct and incorrect versions of 50 mathematical short problems (Questionnaire I) including 200 items. This was followed by a multiple-choice mathematical test with the 50 problems (Questionnaire II) using the same 200 items and offering the correct and incorrect answers. Subjects tended to choose correct as well as incorrect responses corresponding to their highest affective rating within each problem. In all cases the subjects' behavior was higher than chance level, and thus, followed the trend to maximize pleasure. This result supports the hypothesis according to which the key to decision making lies in the affective dimension of conscious experience.

2:55-3:10

On-Line Hypothesis Evaluation: Number of Hypotheses, Number of Hypothesis Changes, and Individual Differences in Working Memory.
Eric G. Freedman, Paul Britton, Matthew Woodruff, & Raymond Vernagus (U. of Michigan, Flint)
High and low working memory (WM) individuals evaluated a single or pair of hypothesis by viewing seven pieces of evidence resulting in two, one, or zero hypothesis changes. Next, subjects decided which final hypothesis was correct. As the hypothesis changes increased, low WM individuals exhibited a relatively greater increase in reaction times and a relatively greater decrease in accuracy than did high WM individuals. High WM individuals were faster and more accurate than low WM individuals.

3:30-5:25

Psychonomic Society: J/DM III (Grand Ballroom B)

3:30-3:50

Evaluating a Class of Utility Theories.
Richard A. Chechile & Sue F. Butler (Tufts U.)
A wide class of utility theories can be subsumed under the Miyamoto (1988) generic utility formulation. Chechile and Cooke (1997) tested this utility framework by measuring model-dependent scaling parameters. One of these parameters should be invariant if the generic utility theory is valid. However, that parameter systematically varied. The present study resolves a methodological problem with the Chechile and Cooke experiment, and still results in finding evidence contradictory to the generic utility theory.

3:55-4:15

Confidence in Aggregation of Expert Opinion
David V. Budescu & Adrian K. Rantilla (U. of Illinois,, Champaign)
Two studies examined how different configurations of expert opinions affect decision-maker confidence. In Study 1, confidence was higher when experts saw the same information, when fewer experts gave opinions, and with higher expert opinion values. Study 2 replicates Study 1 findings, and also suggests that confidence increases as the range of expert opinions (amount of expert disagreement) decreases. These results are discussed within the general framework of aggregation of expert opinion.

4:20-4:40

Cue Priorities in Making Risky Choice
X. T. Wang (U. of South Dakota)
Decision makers often use social and verbal cues in choice problems according to predetermined priorities. We investigated how this ordering is related to domain- specific social factors and ecological validity of the cues. In our study of risk preference, cues of kinship, anonymity, group size, group constitution, and verbal framing were used by participants in decreasing priority. This ordering from more to fewer social cues affects aspiration levels of a decision maker and consequent risk preferences.

4:45-5:00

The Nature of Adaptive Decision Making Under Presentation Constraints.
Douglas H. Wedell & Stuart M. Senter (U. of South Carolina)
We examined accuracy, effort, and strategy differences in a task in which participants viewed sets of apartments described along four dimensions. Sets were designed to identify strategies from choice or judgment patterns. Information was presented by alternative or by dimension. Consistent with the adaptive decision-maker hypothesis, dimensionwise participants were more efficient (higher accuracy, reduced looking times), but greater efficiency in choice was not due to different strategy use. Strategy differences were found in judgment.

5:05-5:20

Generalization of PMM Therapy to Polychotomous Cues
David Slegers, Gregory L. Brake, & Michael E. Doherty (Bowling Green State U.)
Gigerenzer et al. have explored their PMM approach to decision making, which posits a noncompensatory decision rule based on only one cue, but PMM applies only to dichotomous cues. We extend it to the case of polychotomous cues, employing an analogy to the limitation on the number of categories people use in making absolute judgments (Miller's 7 +- 2). The 7 +- 2 model correctly predicts nearly as well as a regression model using all of the predictor variables.

Sunday, Nov 22

8:30-9:00

Registration & Continental Breakfast

9:00-10:40

• Symposium: Interactive Judgment and Decision Making Symposium
  Organizer: Derek J. Koehler (U. of Waterloo)
  • Theory and Research on Uncertainty and Behavior in Judge Advisor Systems
    Janet A. Sniezek (U. of Illinois at Urbana-Champaign)
    A Judge Advisor System is a set of persons participating in a decision process in different roles: the Judge is responsible for the final decision and may solicit or receive advice from one or more Advisors. It is argued that most important decisions are made by JASs rather than individuals or groups, and that the decision processes of JASs differ from those of groups or individuals in critical ways. Theory and research on the unique aspects of JASs will be discussed, with an emphasis on information search, cognitive and goal conflict, and confidence assessments.
  • Evaluation Errors in Interactive Decisions: When Evaluators Perceive Others To Be Too Conservative
    Chip Heath (Duke U.), Marc Knez (U. of Chicago), & George Wu (U. of Chicago)
    We study situations where an outside view may cause outsiders to inappropriately perceive insiders as too conservative. In two studies, targets answered open-ended questions and provided cash equivalents (CEs) for their answers; "outside" evaluators also provided CEs for the target's answers. In a low knowledge condition, targets provided lower CEs than evaluators. Target's CEs were more accurate than evaluators' (calibration and resolution were superior), but, in a social perception task, evaluators rated targets as being overly conservative. Outside judges, when they have a less complex view of the world, may incorrectly evaluate inside advisors as being overly risk- averse or conservative.
  • Social Influence Under Uncertainty: Effects of Evaluating Typical and Extreme Advice on Confidence
    Kristine M. Kuhn (Princeton U.), David Spurlock (U. of Illinois at Urbana- Champaign), & Janet A. Sniezek (U. of Illinois at Urbana-Champaign)
    Two experiments on the impact of considering advice for a task with high uncertainty long-range forecasts of global events are presented. In the first, subjects were given anonymous forecasts (advice) that were either typical or very unusual and wrote either supporting or contrary reasons for their advisor's forecasts; the two manipulations had interactive effects on subjects' confidence in their own subsequent forecasts. Results of the second experiment indicate that different degrees of effort engendered by the experimental manipulations best explain their effects on confidence. Social influence as a source of alternative decision hypotheses and the complexity of assessing effects in real-world settings are discussed.
  • Unintended Influence of Exposure to Another's Hypothesis on Confidence and the Generation of Alternative Hypotheses
    Derek J. Koehler (U. of Waterloo)
    Two experiments demonstrate that exposure to somebody else's hypothesis induces an anchoring effect on the generation of alternative hypotheses. Although such an influence can be rationally justified by treating the person's hypothesis as relevant information, results of the present experiments suggest that the anchoring effect may not be entirely deliberate. Instead, people apparently misattribute the resulting similarity between their own hypothesis and that of the person's to which they have been exposed as evidence of accuracy. As a result, generation of alternative hypotheses following exposure to that of another person can have an unintended biasing influence on subsequent confidence assessments.
• Symposium: Decision Analytic Concepts Applied to Prostate Cancer Screening and Treatment Decisions: Progress and Prospects
  Organizer: Robert M. Hamm (U. of Oklahoma Health Sciences Center)
  Speakers:
  • Robert Hamm, PhD. (U. of Oklahoma Health Sciences Center) Introduction.
  • James Mold, MD (U. of Oklahoma Health Sciences Center)
    Although prostate cancer screening is often viewed as a great benefit for all men, in fact it can be a trap for the physician and for the patient. Prostate cancer grows slowly, and it often would be appropriate for men with a short life expectancy (less than a decade) not to treat it (by removing the prostate) but instead to live with it. However, many patients, who would choose not to treat, are nonetheless interested in getting a prostate cancer screening test. Doctors who recognize there is no point to screening for such men often gamble (hoping that the test will be negative) rather than trying to convince the patient that it would be better not to know. Is there a better way to communicate with patients about the risks and benefits of screening?
  • Michael W. Kattan, PhD (Baylor College of Medicine)
    Decision analysis of the decision whether to treat a cancer discovered by screening (and deemed treatable) takes into account both the predictable side effects (postoperative morbidity, incontinence, impotence) and the anticipated increase in life expectancy. Markov analysis lets the analysis take account of quality of life at various points in the future. The decision is sensitive to individual patient's preferences for living with the side effects. The decision analytic model can be individualized for patients, who can visit a web site, complete a utility assessment regarding incontinence and impotence, and be given a recommendation whether they should get treatment. Problems inherent in keeping a decision theoretic model up to date will be discussed.
  • Scott Cantor, PhD (MD Anderson Cancer Center, U. of Texas)
    There have been two decision analyses of the decision whether to screen for prostate cancer, one evaluating a single screening, the other evaluating annual screening. Logically, each includes the decision whether to treat (described by Kattan), and adds considerations of the reassurance provided by a negative screen as well as the anxiety induced by knowing one has a cancer, especially pertinent for those who will opt not to treat. Problems of integrating evaluations of time limited anxiety with evaluations of death and long term quality of life are discussed.
  • Robert J. Volk, PhD (Univ of Texas Medical Branch at Galveston)
    Whose utilities count? The majority of men in whom prostate cancer is discovered are married. The wife would share in the outcomes of any decision, including incontinence and impotence. A study of husbands' and wives' evaluations of these outcomes showed large differences within the couples. Generally, women were more willing than their husbands to risk longterm incontinence and impotence to increase the husband's expected life span. Problems of integrating discordant spousal preferences are discussed.
  • Robert Hamm, PhD. (U. of Oklahoma Health Sciences Center)
    How may we assist men in making rational decisions (decisions consistent witht their values) regarding prostate cancer screening? Tools proposed for presenting decision relevant facts for medical decisions (probabilities of outcomes, evaluations of outcomes) are described. Issues to be discussed include that presentation of all the facts is unbiased but perhaps difficult to comprehend, while presentation of vivid examples is memorable and influential but may be biased.

10:40-11:10

Break

11:10-12:25

• Paper Session
  • The Springs of Action: Affective and Analytical Information Processing in Choice
    Ellen Peters & Paul Slovic (U. of Oregon and Decision Research)
    Affective processes are predicted to play a critical role in the construction of choices. In two studies, we successfully used paper-and-pencil measures of individual differences in affective information processing to predict choices in a card selection task modified from Damasio (1994). The results were particularly strong when the processing of rewards and punishments were considered separately. Individual differences in analytical information processing did not relate to choice as predicted. While society has valued analytical, deliberative thought more than other emotional ways of knowing, affect, in interaction with analytical thought, appears to strongly influence decisions and judgments.
  • Is It True Bronze Medalists Have More Fun?
    A. Peter McGraw & Barbara Mellers (The Ohio State U.)
    Medvec, Madey, and Gilovich (1995) claimed that bronze medal winners in the 1992 Summer Olympics appeared happier than silver medal winners, due to systematic upward and downward counterfactual comparisons. We investigate the effects of prior expectations in a series of laboratory and real-world experiments. We find that people tend to make upward counterfactual comparisons when judging the most likely alternative outcomes. We also find that prior expectations have strong effects on perceived happiness. Finally, we examine the effect of prior expectations on athlete's emotions in the 1996 Summer Olympics and compare our findings to those obtained by Medvec et al.
  • The Impact of Negative and Positive Affect on Managers' Risky Decision Making
    Kimberly Moreno (Virginia Polytechnic Institute and State U.), Thomas Kida (U. of Massachusetts at Amherst), & James F. Smith (U. of Massachusetts at Amherst)
    This study examines the impact of affective reactions on managers' risky behavior in both gain and loss contexts. We propose that an affective reaction's underlying negative or positive valence is the critical factor influencing decision behavior. The findings indicate that affective reactions to data within a decision context can result in risk taking for gains and risk aversion for losses, contrary to the predictions of prospect theory. Since this research was conducted on experienced corporate managers, the study demonstrates the importance of affect in expert decision making. The results also indicate that affect influences experienced managers to choose alternatives with lower economic value, suggesting that managers consider both financial utility and the utility of affect when evaluating risky alternatives.
• Paper Session
  • 'Learning' With No Feedback: A Test of Belief and Choice Reinforcement Models in Games
    Roberto Weber (California Institute of Technology)
    Theories of learning in games map a history of observed outcomes into choices. These theories are inconsistent with the possibility that behavior in games may systematically change as a result of factors other than experienced outcomes. This paper uses the p-Beauty Contest game to test the assumption that learning mainly takes place through reinforcement. The main experimental treatment is a no- feedback condition, in which outcomes are not reported to the players until the end of the experiment. Traditional models of learning predict that there should be no convergence towards the Nash equilibrium. Preliminary results indicate some convergence towards the equilibrium.
  • The Development of Expertise in Decision Making
    Ilan Fischer (Ben Gurion U. of the Negev, Israel)
    This study traces the utilization of decision strategies, as a decision-maker develops expertise in a given domain. The study is driven by a Cognitive Clasification Structure (CCS) model, which distinguishes between three operation modes: Screening, Discrimination and Classification. The amount and structure of the information that is processed and utilized for decision making varies systematically across the three modes. In two experiments, 180 subjects performed a repeated categorical decision task with feedback. Subjects participated in six sessions of 300 decisions each. The findings reveal a dynamic use of all three CCS processes. The most frequently used decision mode is the Discrimination process. The ratio between Classification and Screening gradually changes as the DM gains more experience.
  • Psychological Space of Decision Similarity and Its Influence on Decision Strategies
    Yuri Tada & Elke U. Weber (The Ohio State U.)
    In exploring factors that determine the choice of decision strategies in everyday decision making, knowing how people categorize those decisions would be useful. If people are indeed adaptive decision makers who select strategies in a systematic fashion, there must be in their mind some representation of the similarity between decisions. Using multidimensional scaling, we found two different three-dimensional configurations, depending on whether participants were instructed to sort decisions in terms of unspecified "similarity" or in terms of "similarity in making these decisions." A subsequent study indicated that these dimensions indeed appeared to influence the selection of different decision strategies.

12:25-1:30

Lunch

1:30-2:00

Einhorn Award Address, Einhorn Award Winner announced and introduced by Eric J. Johnson

2:00-3:00

Invited Address by Robert Winkler (Duke U.): Combining Forecasts

3:00-4:30

Poster Session I

4:30-5:45

• Paper Session
  • On Evaluation Goals and Sequential Order Effects: The Relationship between Hedonic and Informational Evaluation
    Gal Zauberman (The Fuqua School of Business, Duke U.) & Dan Ariely (The Sloan School, Massachusetts Institute of Technology)
    This paper examines the effect of the sequential order of information on the final evaluation under two evaluation modes, Hedonic and Informational. The main thesis of this paper is that the Evaluation-Goal impacts the integration rules, and thus the observed sequential order effects. The prediction is a stronger primacy effect under informational evaluation, and stronger recency effect under hedonic evaluation. In two experiments subjects were asked to evaluate identical sequences of information (ascending and descending) under each evaluation goal. The hypotheses were supported in both experiments. That is, there was an interaction between the Evaluation-Mode and the ascending and descending patterns. The exact pattern of results differed between retrospective and predictive evaluations.
  • The Relationship Between Time Constraints and Time Pressure
    Lehman Benson III, Markus Groth, & Lee Roy Beach (U. of Arizona)
    Time constraints and time pressure often are often used interchangeably. We suggest that this is incorrect, and differentiate between the two, conducting four experiments designed to examine the relationship between time constraints and feelings of time pressure. The literature, and common sense, both suggest that three factors influence the perception of time pressure (1) the time available, (2) the time required, and (3) the importance of the task (motivation). The first two experiments involved a laboratory task in which subjects were told both the time required to do a hypothetical task (Rr) and the time available to do it (Ra). Experiments 3 and 4 involved a task in which subjects solved arithmetic problems for pay. Implications for past and future research on decision making under time pressure are discussed.
  • When the Pleasure of Winning Exceeds the Pain of Losing
    Jeff T. Larsen & Barbara A. Mellers (The Ohio State U.)
    Prospect theory holds that equal probability gambles to win or lose some amount of money should be avoided, even when stakes are low. In Experiment 1, however, when presented with gambles to win or lose $0, $1, $3, $6, $12, $25, $50 or $100, participants' preferences peaked at $6, rather than $0. In another experiment, participants presented with the $0, $1, $3, $6, and $12 gambles preferred the $3 gamble; those presented with the $0, $12, $25, $50, and $100 gambles preferred the $25 gamble. These and data from two other experiments suggest a desire for gambles of intermediate value.
• Paper Session
  • Judgment under Sample Space Ignorance
    Michael Smithson (Australian National U.), Thomas Bartos (Australian National U.), & Kazuhisa Takemura (Tsukuba U.)
    Although the incorporation of 'ambiguity' in decision research is now fairly commonplace, little attention has been paid to how people judge probabilities or make decisions under sample space ignorance, i.e., knowing what all possible outcomes are. Walley (1991, 1996) has developed an imprecise probability framework for probability inference that begins with vacuous lower and upper probabilities when the sample space is completely unknown, and then updates those estimates as evidence becomes available. This paper presents findings from six studies that investigate aspects of Walley's framework in combination with related behavioral research and theories of judgment under ambiguity.
  • A Good Chance Is Not a Number: Determinants and Implications of Verbal Probabilistic Terms
    Karl Halvor Teigen (U. of Tromso, Norway)
    Verbal probability phrases are of two distinct kinds: Positive ('possible', 'a good chance') and negative ('not certain', 'doubtful'). Positive phrases focus on the occurrence of a target event, whereas negative phrases focus on its non-occurrence. This makes them different from numerical probabilities, which can be given both positive and negative interpretations, dependent upon context. Thus verbal probabilities are, in one sense, less ambiguous than numbers. The directionality of verbal phrases can be shown to have consequences for predictions, decisions, and probabilistic inferences (e.g. the conjunction fallacy). Verbal probabilities (in contrast with numerical estimates) are strongly affected by closeness of the target outcome.
  • The Interpretation of "Likely" Depends on the Context, But "70%" is 70%--Right?: Associative vs. Rule-Based Processing of Uncertainty Information
    Elke U. Weber (The Ohio State U.) & Paul D. Windschitl (U. of Iowa)
    Past research demonstrates that interpretations of verbal uncertainty forecasts ("likely") depend on the context to which they refer, presumably as the result of rule-based mental processes and the vagueness of verbal expressions. Three experiments demonstrate that precise numeric forecasts ("70%") are also susceptible to context effects. We show that associative rather than rule-based processing of uncertainty information best explains why perceptions of uncertainty vary with context even when precise and credible numeric forecasts are provided. We discuss the implications of the fact that uncertainty information appears to be processed within two semi-independent processing systems, one associative and the other rule-based.

5:45-7:15

Poster Session II and Reception

Monday, Nov 23

8:00-9:00

Business Meeting & Continental Breakfast

9:00-10:40

• Paper Session
  • "That's the Best You Can Do?" When Negative Evidence Increases Confidence
    Craig R. M. McKenzie, Susanna M. Lee, & Karen K. Chen (U. of California at San Diego)
    Using a legal scenario, three experiments examined change in confidence following presentation of each of two sides of a dispute. We argue that change does not depend simply on a case's strength, but on the difference between its strength and its minimum acceptable strength. A case can fall below this level and increase confidence that the competing account is most accurate. Furthermore, the minimum acceptable strength of the second case is influenced by the strength of the first case. Case strength is evaluated relative to a non-zero reference point because it is understood that a case represents a biased sample of available evidence.
  • Priming for Expertise and Confidence in Choice: Evaluating the Global Improves Calibration for the Specific
    Oleksandr S. Chernyshenko & Janet A. Sniezek (U. of Illinois at Urbana-Champaign)
    It is well-known that item-level confidence is higher than more general self evaluations for difficult decision tasks. But can the latter debias the former? Participants provided choices and confidence assessments for items from 20 different knowledge areas. For some items, participants rated their expertise in the area before seeing the item. This priming for expertise improved calibration, with less overconfidence on difficult items as well as less underconfidence on easy items. The direction of the item bias reduction is consistent with a dual-source explanation of the over/underconfidence phenomenon. Theoretical and practical implications of the observed reduction in miscalibration are discussed.
  • When Knowledge is a Bad Thing: Empirical Demonstrations of Less-Is-More Effects
    Daniel G. Goldstein (Harvard U.)
    This talk presents a challenge to the view that more information is always better than less for the quality of inferences. Focusing on two-alternative inferences made from multiple cues, it is shown that the addition of valid, predictive information can decrease inferential accuracy in natural and artificial systems. This phenomenon, deemed the less-is-more effect, was discovered as a consequence of the simple recognition heuristic for inference. Three empirical demonstrations of less-is-more effects are presented. The most challenging of these makes the prediction, contrary to intuition, that knowledge test scores should decrease with repeated testing due to a "false fame" effect.
  • Familiarity Bias in Relative Likelihood Judgment
    Craig R. Fox & Jonathan Levav (Duke U.)
    Probability theory and most descriptive models of judgment under uncertainty imply that belief orderings over events and their complements should mirror each other (e.g., H is more likely than L iff not-L is more likely than not-H). This principle is violated in several surveys in which we asked people to assess the relative likelihood of familiar versus unfamiliar events. In particular, respondents are biased to view familiar events (and their complements) as more likely than unfamiliar events (and their complements). Further studies suggest that the familiarity bias may be less pronounced among subjects who are asked to judge the probability of each event rather than which event is more likely.
• Paper Session
  • Taking One for the Team: Group Utility and Personal Dissatisfaction
    Jonathan Levav (Duke U.) & Dan Ariely (MIT)
    Many individual decisions, such as what dish to choose in a restaurant, take place in a group context wherein group members voice their choices sequentially. In this paper we examine the impact of this dynamic decision process on individuals' choices and satisfaction with their outcomes. We propose that group members alter their decisions contingent upon the choices of other members, yielding a pattern of variety-seeking on the group level. This group contingency sometimes results in choices that undermine personal satisfaction and increase regret. We find support for our hypotheses in one pilot study and two field studies where we tracked consumers' order of dishes and drinks.
  • Initial Evaluation of Integrative Offers in Multi-Issue Negotiations
    Simone Moran & Ilana Ritov (Ben-Gurion U., Israel)
    This research examined evaluation of initial offers by inexperienced negotiators. Our findings suggest that (a) Due to coding of values as gains or losses relative to the "even split" reference point, integrative offers appear less attractive than distributive offers. (b) Although receiving integrative offers may improve understanding of one's own profit schedule, it does not lead to better understanding of mutual interests. (c) The composition of counteroffers depends upon the composition of initial offers. We conclude by proposing that integrative offers may be beneficial to the initiator, not due to any social message they presumably convey, but simply because they establish within-issue advantageous anchors.
  • Theories of Commitment, Altruism and Reciprocity: Evidence from Linear Public Goods Games
    Rachel Croson (U. of Pennsylvania)
    Theories of commitment, altruism and reciprocity have all been invoked to explain and describe observed behavior in public goods and social dilemma situations. In particular, commitment theories have been used to explain behaviors like water conservation and voting. Theories of altruism are applied in explanation of contributions to charities and intergenerational transfers and bequests. Theories of reciprocity have been invoked to explain gift exchange and labor market decisions. This paper describes a set of experiments which distinguish between these competing theories by testing their comparative statics predictions in a linear public goods setting. Results provide strong support for reciprocity theories over either theories of commitment or of altruism.
  • Cross-National Differences in Risk Preference
    Christopher K. Hsee (The U. of Chicago) & Elke U. Weber (The Ohio State U.)
    A series of studies using different methods show a consistent cross-national difference in choices under risk that the Chinese were less risk averse than the Americans. Yet when asked to predict the choices of their counterparts in the other country, both American and Chinese respondents predicted that the Chinese would be more risk averse. The cross-national difference in risk preference is explained by a "cushion" hypothesis which suggests that people in a collectivist society, such as China, are more likely to receive help if they are in need (i.e., they could be "cushioned" if they fell), and consequently, they are less risk-averse.

10:40-11:10

Break

11:10-12:10

Invited Address by Alvin E. Roth (U. of Pittsburgh): Towards a Cognitive Game Theory

12:15-1:45

Luncheon
Student Poster Award Winner announced by Sandra L. Schneider
Presidential Address by Elke U. Weber (The Ohio State U.), "From Shakespeare to Spielberg (with apologies to Paul Slovic): Some Reflections on Modes of Decision Making." Introduced by Irwin P. Levin, President-Elect.

1:45-3:25

• Symposium: On Mice & Men: Effects of Sequences' Characteristics and Duration
  Organizer: Dan Ariely (MIT)
  • Temporal Integration and Duration Neglect in the Evaluation of Rewarding Brain Stimulation by Laboratory Rats
    Peter Shizgal & Bonnie Sonnenschein (Center for Studies in Behavioral Neurobiology, Concordia U.)
    In the evaluation of temporally extended experiences, certain salient features, such as peak value, terminal value, and slope, appear to play a privileged role. To predict the effect of changing the temporal profile of an experience on its remembered value, the weighting of different features must be described and the rules employed in processing them understood. One approach to this problem takes advantage of the tight control over intensity and duration afforded by direct electrical brain stimulation in laboratory animals. When delivered to appropriate brain sites, such stimulation can be intensely rewarding; rats can be trained readily to initiate such stimulation, and their performance is tied closely to its temporal and spatial parameters. We have developed and tested a model that links the subjective intensity of the rewarding effect to stimulation strength and duration. Intensity is modeled as a logistic function of stimulation strength; the input strength required to produce a half-maximal intensity declines hyperbolically with duration. A simple peak rule, which suffices to account for existing data, is employed to translate the real-time profile into a stored record. As predicted by this model, the recorded intensity increases initially with duration but eventually levels off; the higher the stimulation strength, the shorter the time required for the intensity of the rewarding effect to approach asymptote. Beyond this critical time, duration is neglected. The implications of these and related results for the study of intertemporal choice in humans will be discussed.
  • Intensities and Intensity Changes: The Impact of Memory and Distraction on Encoding of Sequences' Characteristics
    Dan Ariely (MIT)
    The emerging literature on the evaluation of sequences has convincingly demonstrated that when people summarize experiences they do not simply combine (integrate) the intensity of their actual experiences. Instead, decision makers extract only a few key aspects of these sequences (gestalt characteristics), and it is these gestalt characteristics that are than used to form overall evaluations of the sequence as a whole. The current work examines the relative role of a few of these gestalt characteristic. Two general classes of characteristic are examined, one that is based on the actual intensities of a sequence and one that is based on the relationships between intensities over time. In addition, we examine the impact of delay and distraction on the relative impact of these two types of characteristic. The results show that the relationships between intensities over time (i. e. slop) plays a large role on both retrospective and prospective evaluations. In fact its role is so large that evaluation sometimes violate dominance relationships between the sequences. Finally, the role of the transient intensities even increases with delay and distraction.
  • Encoding Sequences and Conveying Information in Ratings and Choice: The Case of Sequences' Characteristics and Duration
    George Loewenstein (Center for Advanced Study in the Behavioral Sciences)
    Recent research shows that people's evaluations of temporally extended experiences do not correspond to a simple integration of pleasure and pain over time but are sensitive to a few key features of the experience, such as improvement or deterioration over time, and peak and end levels. In this line of research, a provocative assertion, made first by Fredrickson and Kahneman (1993) is that duration is a feature that is not incorporated into such evaluations, a pattern which was labeled "duration neglect." Duration neglect is of great potential significance since it can, in principle, produce severe choice anomalies such as violations of dominance. The vast majority of studies examining duration neglect have not, however, examined choices between extended experiences, but rather ratings of such experiences. We argue that ratings can sometimes be a biased measure for studying whether duration is neglected in choice, for reasons that relate to norms in decision making (Kahneman & Miller, 1986) and conversational norms (Grice, 1975). We demonstrate the problem with ratings with two parallel studies in which subjects retrospectively evaluate sequences of aversive sounds, one in which subjects rated the sequences, and the other in which they chose between them. Our major finding that duration neglect is more pronounced in the rating condition, but less so in the choice condition raises questions about the conditions under which duration neglect observed in ratings actually leads to duration neglect in choice.
• Symposium: Legal Applications of J/DM Research
  Organizer: Russell Korobkin (U. of Illinois College of Law)
  • The Status Quo Bias and Contract Law
    Russell Korobkin (U. of Illinois College of Law)
    Standard legal analysis of contract "default rules," that is, rules that govern a relationship between contracting parties only if the parties fail to specify their own preferred terms suggests that if the costs of bargaining are low, the content of default rules will matter little. This paper demonstrates that, in certain circumstances, contracting parties will view the default rules as part of the status quo, and will thus be reluctant to contract around these rules even when the costs of bargaining are low. Thus, default rules are "stickier" than previously thought. The paper then addresses the question of how, in light of these results, policy makers should determine the content of default rules.
  • Judging in Hindsight
    Jeff Rachlinski (Cornell Law School)
    The "hindsight bias" affects judgments of liability in the legal system. The bias makes adverse outcomes seem more predictable than they really were, and as a consequence defendants can be held liable for adverse outcomes that they could reasonably believed to have been unlikely. The legal system, however, seems to be aware of the bias and has adapted well to its influence. When possible, courts have developed methods of making judgments that avoid reliance on the hindsight bias. Nevertheless, there are as yet untaken adaptations that the courts could pursue to ameliorate the influence of the bias further.
  • The Scales of Justice: The Effect of Cap Magnitude on the Likelihood of Settlement
    Linda Babcock (Heinz School of Public Policy & Management, Carnegie-Mellon U.)
    A major issue in the current debate over litigation reform is whether to limit the amount of damages plaintiffs can receive. This paper conducts a series of experiments to explore the behavioral mechanisms by which damage caps affect the prospects for settlement. Our results indicate that when the damage cap is relatively low, it increases the likelihood of settlement. However, preliminary evidence suggests that a very high damage cap can actually reduce the prospects for settlement by providing plaintiffs with an upward "focal point". This implies that in a particular state, a cap statue will improve settlement rates for big cases and worsen them for small cases.
  • When Are Jurors Impressed By Probabilistic Evidence?
    Jonathan J. Koehler (U. of Texas at Austin)
    The way in which probabilistic evidence is presented to legal factfinders can have a profound impact on the perceived strength of that evidence. This study shows that presentations of DNA statistics that call attention to the number of others in a reference population who would also match by coincidence tend to weaken the perceived value of the evidence. Presentations that focus on the target suspect's small coincidental match probability without reference to others who would also match by coincidence tend to strengthen the perceived value of the evidence. For example, in one hypothetical murder case that included DNA match evidence, many subjects believed that there was more than a 99% chance that the suspect was the source of the DNA evidence. But when the match statistic was presented in a slightly different (but accurate) way, nearly as many subjects believed that there was less than a 1% chance that the matching suspect was the source when the evidence! A general theory will be offered.

3:25

Adjourn