Decision processes brown bag, 1999

12-1:30 PM, Mondays, Steinberg-Dietrich 204

Sept 27George Wu Large slice bias in bargaining
Oct 4Nathan Novemsky Does the Order of Hedonic Experiences Matter? It Depends on When You Ask.
Oct 11Barbara Malt Similarity and the linguistic categorization of common objects.
Oct 18Stijn Van Osselaer The Value of Brands: Exploring Causes and Moderators of Cue Interaction in Consumer Learning
Oct 25Eldar Shafir On the Evolution of Cognition and Rationality
Nov 1Rachel Croson International Trust Games: The Effect of Communication and Social Distance
Nov 8INFORMS(no brown bag)
Nov 15Leaf Van Boven Egocentric empathy gaps
Dec 6Jon Baron Deontological vs. utilitarian values
Jan 17Gretchen Chapman Value for the Future and Preventive Health Behavior
Jan 24Drazen Prelec A solution to the wine tasting problem
Jan 31no talk scheduled
Feb 7Claudia Keser Principals' Principles When Agents' Actions Are Hidden
Feb 14George Loewenstein Arbitrary Coherence: Duration-sensitive pricing of hedonic stimuli around an arbitrary anchor
Feb 21Jay Koehler Betrayal aversion
Feb 28Baruch Fischhoff What's worth knowing?
Mar 6Doug Medin Universal and Cultural Features of Folktaxonomy, Folkecology, and Agro-forestry Practice
Mar 13no talk, spring break
Mar 20Maureen Cropper What is a statistical life worth?
Mar 27William Schulze Automobile Safety and the Value of Statistical Life in the Family: Valuing Reduced Risk for Children, Adults and the Elderly
Apr 3Reid Hastie The Role of Explanations in Judgments and Decisions
Apr 10Richard Thaler
(4:30, 350 SH-DH)
Doing Economics (and Finance) without Homo Economicus
(2nd annual Decision Processes Distinguished Lecture)
Apr 17Elke Weber Perception is relative: Implications of psychophysics for models of risky decision making
Apr 24James Hammitt Willingness to pay to reduce mortality risks
May 8Jon Elster Belief-motivation interactions ROOM 217 SH-DH

Talks given in the spring of 1999

George Wu, University of Chicago, "LARGE SLICE BIAS IN BARGAINING."

We present three studies that demonstrate a "large slice bias": negotiators overestimate their share of the surplus in distributive bargaining. Our studies show that the effect persists even when expectations about the size of the bargaining zone are manipulated. However, the bias becomes smaller in a small bargaining zone. We offer several explanations for the bias.

Nathan Novemsky, Princeton: Does the Order of Hedonic Experiences Matter? It Depends on When You Ask.

This paper focuses on hedonic contrast effects. Hedonic contrast occurs when a later experience in a sequence is enjoyed more following a worse experience then following a better experience. Although recent studies have found no evidence that hedonic contrast exists in real-time measures, consumer choices in several paradigms (e.g., Loewenstein and Prelec 1993; Ratner, Kahn, and Kahneman 1999) suggest that consumers may expect contrast effects in their experiences. The present research examines whether individuals hold an implicit theory that they will experience contrast effects in sequential experiences. We present the results of four experiments that suggest that 1) people expect hedonic contrast effects even when they do not experience such effects, 2) people remember having experienced hedonic contrast effects when in fact they did not, and 3) people's expectations about contrast effects are related to the choices they make. It seems that individuals' belief in contrast rather than actual real-time contrast is driving choice behavior.

Barbara Malt (Lehigh University)
Similarity and the linguistic categorization of common objects.

How do people choose a name for an object when they want to refer to it? Many central psychological models and views of categorization assume that similarity is the basis for recognizing objects as belonging to kinds, and that, further, identifying an object's kind leads in a straightforward way to choice of a name for the object. I suggest that the relation between similarity and use of a name may not be as straightforward as has generally been assumed because naming is part of a communication process, whereas recognition is not. The name selected for an object may reflect requirements for successful communication that are not relevant to the recognition process. In the work presented here, I address this possibility through a cross-linguistic comparison of naming and similarity judgments of common objects.

Consider the observation that the boundaries for linguistic categories (that is, groups of objects called by the same name) may differ from language to language. For instance, in English, a large stuffed seat for one person is given the same name as a wooden chair, but in Chinese, it is given the same name as a large stuffed seat for two or more people (things that in English would be called "sofa). The linguistic boundary between "chair" and "sofa" therefore is not the same in Chinese as in English. Would speakers of Chinese perceive the large stuffed seats for one person and the large stuffed seat for multiple people as more similar to each other than speakers of English, who call them by different names? In the study to be presented, we asked whether speakers of different languages that have divergent linguistic category boundaries for a set of objects show differences in their perception of the similarity among the objects, and whether those differences parallel the divergences in how they name the objects.

We collected data from native speakers of English, Chinese, and Spanish to address these questions. Participants sorted 60 common containers into piles according to their similarity, and then they named each of the objects. We found that speakers of the three languages showed substantially different patterns of naming for the set of containers. Speakers of Chinese, for instance, named most of the objects using only four different names, whereas speakers of Spanish used 15 different names. In contrast, however, speakers of all the languages saw the similarities among the objects in much the same way, as reflected in high correspondences among the derived similarity matrices and other measures. The observed patterns of naming therefore cannot arise only from the similarities speakers of the three languages see among the objects.

I will suggest three possible ways in which complexity in naming may arise. These mechanisms may lead to differences in patterns of naming across language without affecting perceived similarity among the objects. I will also present preliminary data looking at how non-native speakers of English name the same set of objects in English. Places where the linguistic category boundaries applied by non-native speakers using English diverge from those applied by native speakers may reflect lack of exposure to the language-specific output of the mechanisms creating complexity in naming.

Stijn van Osselaer (University of Chicago)
The Value of Brands: Exploring Causes and Moderators of Cue Interaction in Consumer Learning

When consumers make purchasing decisions, they have to evaluate the consumption outcomes or benefits afforded by the products they are considering. Because consumption outcomes usually cannot be assessed directly in the store, consumers learn to use directly observable product cues (e.g., brand names, ingredient info) as predictors of consumption outcomes. In this talk, I will briefly discuss an experiment showing that the subjective predictive value of brand names depends on the presence and outcome-relationship of other cues (e.g., sub-brand names, ingredients). After this introduction, I will focus on recent experiments that try to distinguish between retrospective, causal reasoning explanations of cue interaction effects and prospective, associative explanations. Finally, I plan to discuss some new experiments investigating cue interaction in multiple-outcome scenarios--in which consumers learn to predict multiple benefits instead of one.

Eldar Shafir (Princeton University)
On the Evolution of Cognition and Rationality

Some common assumptions about evolutionary psychology are considered and their implications for the study of human rationality are discussed. Familiar behavioral violations of the rational theory of choice and of simple logic are reviewed, and recent work in evolutionary psychology that reinterprets those violations is critically evaluated. Whereas evolutionary psychology hopes to provide a novel perspective, it is argued that this perspective is of limited relevance to the experimental study of rationality. Instead, it is suggested that the lessons derived from the rationality debate that has engaged psychologists, economists, and decision theorists, can help shed light on the nature of cognition that emerges from selection and adaptation. Examples from animal behavior are considered and their implications are discussed.

Leaf Van Boven (Cornell University)
Egocentric Empathy Gaps

We examined people's perceptions of the endowment effect, or the tendency to think a commodity is worth more simply because one owns it. Five studies documented egocentric empathy gaps between owners and buyers: Each side overestimated the similarity between their own valuation of a commodity and the valuation of people on the other side, thereby underestimating the endowment effect. This underestimation led participants to make decisions that cost them money. When informed of the true difference between owners' and buyers' valuations, participants tended to rate greed as a more likely explanation of the other role's valuation than the endowment effect. We discuss the robustness and implications of these findings.

Jonathan Baron (Penn)
Deontological vs. utilitarian values

I will review recent work on omission bias, protected values and (possibly) the bias toward indirect harm. The general idea is that several biases are linked together because of principles that concern characteristics of action other than its consequences. Relevant papers are in my web pag:
How serious are expressions of protected values? (with S. Leshner)
The bias toward indirect harm (with E. Royzman)
Protected values and omission bias (with I. Ritov)
Protected values (with M. Spranca)
I will not discuss all these papers, and some of what I will discuss is not written yet.

Gretchen Chapman (Rutgers)
Value for the Future and Preventive Health Behavior

Many everyday decisions appear to exemplify inter-temporal choices, or trade-offs between an immediate and a delayed benefit. Although a sizeable body of research has explored preference for delayed outcomes, little research has examined whether such time preference is actually related to real-world behavior. Three studies examined the relationship between time preference and preventive health behaviors which involve an up-front cost for a long-term benefit. In the first study, influenza vaccination showed only a small relationship to one of four time preference measures. In the second and third studies, time preference showed no relationship to multiple measures of adherence to a medication regimen to control high blood pressure or high cholesterol. Thus, surprisingly, it appears that time preference, as measured with questionnaire scenarios, bears little if any relationship to actual behaviors that exemplify inter-temporal trade-offs.

Claudia Keser
University of Montreal
"Principals' Principles When Agents' Actions Are Hidden"

We examine the behavior of subjects in a simple principal-agent game with hidden action. While subjects in the role of agents tend to choose the actions which maximize their expected profits, subjects in the role of principals offer contracts which differ from the theoretical predictions. We identify three principles of contract design: (1) The agent's remuneration for the better outcome is at least as high as the remuneration for the worse outcome. (2) The agent must not risk making a loss. (3) The net profit of the agent should not be higher than the net profit of the principal.

George Loewenstein
Carnegie-Mellon University
Arbitrary Coherence: Duration-sensitive pricing of hedonic stimuli around an arbitrary anchor.

In three experiments, subjects stated their wilingness to accept pain - from listening to annoying sounds - in exchange for payment (WTA). Subjects were presented with annoying sounds of different durations, indicated their WTA, and received the sounds and payment that resulted from their prices. At the onset of each experiment subject were asked to listen to the sound. Since the sound was very simple, from that point subjects had full information about the hedonic experience. After the initial exposure subjects were asked to state whether, hypothetically, they would be willing to listen to the noise for 30 seconds for either a large or small payment. Subsequently, their actual WTA was elicited to listen to the noise for different intervals (10, 30 and 60 seconds in the first experiment). WTA values exhibited a pattern that we label "arbitrary coherence." Suggestive of arbitrariness, prices were powerfully influenced by the arbitrary high/low anchor accompanying the hypothetical question. Suggestive of coherence, prices were systematically related to noise duration. The first study documented the effect at the individual level, the second in experimental markets, and the third examined more deeply the effect of the initial anchor.

Jay Koehler (University of Texas at Austin)
"Betrayal Aversion"

An egregious form of betrayal occurs when agents cause the very harm that they were entrusted to guard against. Examples include the military leader who commits treason and the automobile airbag that causes death upon deployment. We conducted three experimental studies on betrayal and found that people's aversion to it is reflected in their responses to criminal betrayals, safety device betrayals, and the risk of future betrayals. We conclude that people are willing to incur greater risks of the very harm they wish to avoid (death) in order to avoid the mere possibility of betrayal. Study 1 finds that people wish to punish betrayers more than non-betrayers for identical crimes. Study 2 finds that people are more likely to seek punitive damagesfor harm caused by a malfunctioning safety product (betrayal) than for identical harm caused by a malfunctioning non-safety product (non-betrayal). Study 3 finds that people are willing to double their risk of death from automobile crashes, fires, and diseases to avoid a small possibility of death by safety device betrayal. Specifically, more people select a safety device that is associated with a 2% death riskthan one that is associated with a 1% death risk when there is an additional .01% chance that the low risk (1%) safety device may cause death (betrayal). But when additional .01% risk is described as a non-betrayal risk, most people select the safety device that reduces their overall risk of death. Theoretical support for betrayal aversion is found in classical sociological theory on trust and in recent literature on the role of affect, emotion, moral beliefs and other non-rationalist influences on risky decision making. Alternative explanations for betrayal aversion are considered and promising avenues for further research are identified.

Baruch Fischoff, Carnegie Mellon University
What's Worth Knowing?

Often, we have a very small window of opportunity for communicating a large set of potentially relevant facts. The size of that window may be set by properties of the channel (e.g., a warning label, an on-screen disclaimer), the message (e.g., technical concepts each requiring extensive exposition), or the recipient (e.g., short attention span, limited confidence in the communicator). The talk will describe a general approach to prioritizing information. It combines an integrated assessment of the relevant science with semi-structured examination of recipient beliefs. Examples will focus on health, safety, and environmental risks (e.g., medical informed consent, voluntary regulation of consumer products, infectious disease, workplace safety). The approach suggests some general strategies for the integration of basic and applied research.

Doug Medin, Northwestern University
Universal and Cultural Features of Folktaxonomy, Folkecology, and Agro-forestry Practice.

There is considerable evidence for universal principles of folk-taxonomic systems, at least in traditional societies. Nonetheless, cultures may differ dramatically with respect to relational or folkecological conceptions of nature. This talk describes research with three populations that live in more or less the same area (Peten, Guatemala) and engage in similar activities (hunting, agro-forestry) but have distinct mental models of the forest. Furthermore, these differing models are linked to equally distinctive differences with respect to sustainable and nonsustainable forestry practice. Social network data suggest that some but not aspects of these mental models may be socially transmitted. This work has implications for research on environmental decision making.

Maureen Cropper (University of Maryland and the World Bank)
"What Is A Statistical Life Worth?"

I will report on the results of a survey that I am conducting with Alan Krupnick, Anna Alberini, Bernie O'Brien (and others) that asks persons 40-75 what they would pay for a product that would reduce their risk of dying over the next 10 years by a stated amount. The survey is computerized (with a voice-over) and begins with a significant segment on risk communication. We've just completed a sample of 1,000 persons in Hamilton, Ontario. The good news is that they pass external and internal scope tests. The mean Value of a Statistical Life is much lower than figures used by the USEPA--about $600,000. The survey will be administered at a larger sample in the U.S. (but not before March 20).

William Schulze and Timothy Mount
Automobile Safety and the Value of Statistical Life in the Family: Valuing Reduced Risk for Children, Adults and the Elderly

Little work has been done theoretically or empirically to obtain the value of a statistical life (VSL) for children or the elderly. This paper addresses both of these issues by first presenting a theoretical model of how families value risk and then examining family automobile purchases. Automobile safety is shown to be a family public good, where the marginal cost of purchasing and operating a safer automobile is set equal to the usage- weighted sum of the values of statistical life of family members. We use data on automobile purchases to estimate how much single car families of different composition (in terms of children, adults and the retired) spend on safety to impute the VSL of each age group. We find that children are valued much more highly and the elderly less than most existing studies suggest. This result comes, in great part, from an analysis of the fatal accident data that shows that fragility--the susceptibility to death in an accident of fixed severity--increases with age. Also, we show that an important factor for survival in two-vehicle accidents is the relative weight of the vehicles involved. The models of survival in fatal accidents are used to estimate standardized risks of mortality in different types of vehicles. These standardized risks are then used in hedonic models of the purchase price and fuel efficiency of a specified vehicle to determine the capital costs and the operating cost of reducing the risk of mortality.

Reid Hastie, University of Colorado
The Role of Explanations in Judgments and Decisions

Explanations pervade our conscious thought about the everyday world. They are the 'glue' that hold together our conceptual categories, attitudes, mentyal models, and the scenarios that guide our decisions. This talk will survey some recent research on the functions of explanations in category-based inferences and decisions, and speculate about their role in socially consequential situations involving strategic decisions, jury verdicts, and consumer choice.

Elke Weber (Columbia University)
"Perception is relative: Implications of psychophysics for models of risky decision making"

In this talk I will show the relevance of psychophysical regularities such as Weber's law and signal-detection theory for models of decision making under risk and uncertainty. Using a generalization of the risk--return conceptualization of risky choice used in finance, I will show that the perception of the riskiness of choice alternatives by humans (and lower animals) is affected by outcome magnitude and context in ways that are predictable by well-known psychophysical regularities. I will also present evidence suggesting that people's choices under uncertainty are different when they learn about the likelihood of outcomes experientially than when this information is provided symbolically, a result that raises questions about the generalizability of insights and models from studies that use one or the other method of providing probability information.

Jim Hammitt (Harvard University)
Willingness to Pay to Reduce Mortality Risk

1) Background Risks and the Value of a Statistical Life.
We examine the effects of background mortality and financial risks on the rate of substitution between wealth and mortality risk (the value of a statistical life or VSL). The effects of background risk depend on the individual's risk aversion and prudence in the states where he survives and dies. The presence of a competing mortality risk decreases VSL. An unfair background financial risk reduces VSL if the individual is risk averse and prudent in both states; if he is risk averse but imprudent, the effect is ambiguous. A positive correlation between financial and mortality risks is improves welfare and increases VSL under plausible assumptions.

2) Evaluating the Effect of Visual Aids on Willingness to Pay for Reductions in Mortality Risk
We investigate how estimates of willingness to pay (WTP) for mortality risk reduction, elicited using contingent valuation (CV), are influenced by using alternative visual aids to communicate the magnitude of risk change. We find that WTP depends on the magnitude of risk reduction for three groups of respondents presented with alternative visual aids, but not for the subsample that did not receive any visual aid. Estimated WTP is consistent with the theoretically predicted proportionality to risk change for the subgroups presented with a logarithmic scale or an array of 25,000 dots, but not for the subgroups receiving a linear scale or no visual aid. These results suggest that CV can provide valid estimates of WTP for mortality risk reduction if appropriate methods are used to communicate the risk change to respondents.