Tuesday, April 21, 2015
Porter Hall 223-D
Carnegie Mellon University
Pittsburgh, PA
Workshop organized by:
Cleotilde Gonzalez
coty@cmu.edu
Dynamic Decision
Making Laboratory
Social
and Decision Sciences Department
Carnegie Mellon
University
A theoretical distinction has emerged in the past decade regarding how decisions are made from description (explicit definition of risks, outcomes, probabilities) or experience (implicit collection of past outcomes and probabilities). Explanations of choice under descriptive information often rely on Prospect Theory, while experiential choice has been plagued by highly task-specific models that often predict choice in particular tasks but fail to explain behavior even in closely related tasks. Furthermore, in social interactions the information about others (preferences, beliefs, degree of interdependence) may also influence their interactions and choice, but narrow self-interest and complete information is a common assumption in empirical game theory paradigms, limiting our understanding of the types of uncertainty that people face in real-world social interactions, especially when cooperation is welfare enhancing.
The goal of this workshop is to bring recent research that crosses the borders of traditional descriptive or experiential approaches and attempts to address decision making where many levels of information are available.
This event is free and open to the academic community. No registration is required.
Save this event to your calendar!
Google
Calendar iCalendar
& Outlook
Time Note: presenters must leave 10 minutes
for discussion/questions |
Event | Presenter |
8:30-9:00 AM | Gathering & Breakfast | |
9:00-9:15 AM | Information and decisions: Decisions from description
and experience come together in individual and social interactions I will briefly summarize some of the
current research on decisions from description and experience
in individual interactions with an environment and in
interactions with other individuals and agents, in order
to frame the motivation to this workshop. |
Coty Gonzalez |
9:15-10:00 AM | The history of decision from description and Prospect Theory, resulting from interactions between economists and psychologists This lecture describes the current state of the art in modeling risk attitudes as the result of interactions between empirically oriented psychologists and theoretically oriented economists. At several stages in history, the next step forward could be made only by empirical inputs and intuitions from psychologists. At several other stages, the next step forward could be made only by theoretical inputs from economists with advanced technical skills. Modern views on the measurement of utility, beliefs, risk, and ambiguity attitudes could only arise from the merger of ideas from all the fields mentioned. New empirical impulses have recently come from psychological studies on decision from experience, criticizing prospect theory for decisions under risk. Theories on decision under ambiguity, rather than risk, may shed new light on decision from experience. |
Peter Wakker |
10:00-10:45 AM | Decision making and rationalizing cooperation: Preferences, beliefs, and mechanisms There is a large body of evidence showing that a substantial proportion of people cooperate in social dilemmas, even if the interaction is one-shot and completely anonymous. We consider two major endogenous factors that are known to affect cooperative decisions, and in so doing replicate and extend previous empirical research on public goods problems in several important ways. First, we measure social preferences using a relatively new and well validated method that yields results that are both more highly resolved and reliable than with other measures. Concurrently, we elicit beliefs on the individual level using multiple methods, and repeatedly during the experiment. With this rich set of individual level variables, we can make predictions of people's choices in both one-shot and repeated social dilemma interactions. We show that when heterogeneity in people’s tastes and beliefs is taken into consideration, more than 50% of the variance in individual choices can be accounted for by using a simple statistical model. This approach extends rational choice modeling by accounting for behavioral variation in tastes and expectations, and builds towards understanding under what conditions people are willing to cooperate. Applications to environmental decision making are highlighted and particular mechanisms that can promote cooperation are discussed. |
Ryan O. Murphy |
10:45-11:00 AM | Morning break | |
11:00-11:30 AM | Allais from Experience: Choice consistency, rare
events, and common consequences in repeated decisions The Allais Paradox is a well-known
bias in which people’s preferences result in contradictory
choices between two normatively identical gamble pairs.
Studies have shown that these preference reversals depend
on how information is described and presented. In an experiment,
we investigate the Allais gambles in several formats including
an experiential paradigm, where participants make selections
from two blank buttons and get an outcome as a result
of a draw from distributions of outcomes in the selected
gamble. Results indicate that a large proportion of Allais
reversals are found in the traditional descriptive format,
they are reduced when gambles are presented in a descriptive
table format, and they disappear when choices are made
from experience. Although a majority of participants made
consistent choices from experience, the proportion of
individual reversals is similar to that of descriptive
choices. Detailed analyses of experiential choice suggest
interesting behavioral differences between participants
classified as consistent and those classified as reversals:
consistent participants explore and maximize more than
reversal participants. Furthermore, consistent participants
demonstrate a different switching behavior after experiencing
a rare outcome than do reversal participants. In contrast
to the view that people generally underweight rare outcomes
in experiential choice, overweighting or underweighting
may occur within the course of an individual’s experience
depending on the timing of those experiences, the magnitude
of the outcomes observed, and the general accumulated
value of the gambles. |
Jason Harman |
11:30 AM-12:00 PM | Causal knowledge & decision-making Causal knowledge plays multiple key
roles in our decision-making. In this talk, I will briefly
discuss two: (1) our choices are correctly sensitive to
causal direction, as acting on causes can influence effects
but not vice versa; and (2) our causal knowledge can help
solve the problem of action construction/selection. Moreover,
as I will attempt to show in the talk, both of these roles
can be usefully explained—both theoretically and
descriptively—in terms of causal graphical models. |
David Danks |
12:00-1:30 PM | Lunch | |
1:30-2:00 PM | Information gaps for risk and ambiguity We apply a model of preferences for
information to the domain of decision making under uncertainty.
An uncertain prospect exposes an individual to an information
gap. Gambling makes the missing information more important,
attracting more attention to the information gap. To the
extent that the uncertainty (or other circumstances) makes
the information gap unpleasant to think about, an individual
tends to be averse to risk and ambiguity. Yet when an
information gap happens to be pleasant, an individual
may seek gambles providing exposure to it. The model provides
explanations for source preference regarding uncertainty,
the comparative ignorance effect under conditions of ambiguity,
aversion to compound risk, and more. |
Russell Golman |
2:00-2:30 PM | Social value orientations and cooperative behavior
in the prisoner's dilemma We
investigate the relationship between individual differences
in Social Value Orientation (SVO) and cooperative behavior
in iterated Prisoner’s Dilemma (PD) games. SVO is an individual’s
preference about how to allocate resources between the
self and the other. In an experiment, we tested pairs
in four different PD games, some of which being expected
to produce similar cooperative behavior according to the
well-known Rapoport & Chammah's index of cooperation.
A first finding is that this index of cooperation fails
to explain our behavioral data across those games. This
study also reveals that socially oriented participants,
as defined by the SVO, tend to cooperate significantly
more often than individualistic participants in some specific
PD games, but such social characteristics have no effect
in other types of PD games. Furthermore, the SVO as an
individual difference measure is robust and not easily
affected by the repeated experiences of interaction with
another individual. Finally, we find that the more social
is the less socially oriented individual within a pair,
the more mutual cooperation and the less mutual defection
is found within that pair. |
Frederic Moisan |
2:30-3:00 PM | Decision-by-sampling: A utility-free theory of
decision making In this
talk, I will present Decision by Sampling (DbS), a utility-free
theory of decision making. DbS is a model of magnitude
evaluation based on memory sampling and relative judgment.
This model does not rely on stable, underlying value representations
to explain valuation and choice, or on choice behavior
to derive value functions. Instead, preferences concerning
outcomes emerge spontaneously from the distributions of
sampled events and the relative nature of the evaluation
process. I will illustrate the power and relevance of
the model to 3 domains: evaluations of death tolls (e.g.,
following deadly events), evaluations of life-years (e.g.,
how much value ascribe to extending their life expectancy),
and inter temporal choice. |
Chris Olivola |
3:00-3:15 PM | Afternoon break | |
3:15-3:45 PM | Parameter recovery for decision modeling using
choice data We introduce
a general framework to predict how decision sets used
in decision-making experiments impact the quality of parameter
estimates. We applied our framework to cumulative prospect
theory (CPT) to investigate the expected parameter discrimination
achieved by current research practices. Our analysis is
the first to accurately predict the relative estimation
precision of each parameter of CPT. We additionally applied
our framework to analyze the decision sets that were used
to produce the empirical evidence for the description–experience
gap. Specifically, we found that choices based on few
experienced draws from a gamble provided little information
for estimating decision weights when compared to equivalent
description based choices. Therefore, choices between
experienced gambles can be explained by a wider range
of decision weights than choices between equivalent described
gambles, providing an alternative explanation for the
empirical evidence surrounding the description– experience
gap. We conclude with implications for future experiments
designed to estimate parameters from choice data. |
Stephen Broomell |
3:45-4:15 PM | Wait, wait... Don't tell me: Repeated choices
With clustered feedback When
decision makers repeatedly choose between a safe and a
risky asset, the frequency of feedback can influence maximization.
Myopic prospect theory proposes that agents evaluate each
outcome on its own rather than considering the series
of choices as one large game. In combination with loss
aversion, this can lower performance in the large game.
Revealing outcomes less frequently may discourage such
narrow bracketing and improve expected value maximizing
behavior when losses are likely to be observed. We explore
this prediction in decisions from experience, where participants
have to learn about the options from feedback. Our results
show that providing feedback about individual outcomes
less frequently increases the proportion of choices favoring
the risky option when the high outcome occurs only rarely
or when the low and high outcome are equiprobable. When
the high payoff is common, the effect disappears. We also
find that delayed feedback helps overcome recency bias,
where the probability of choosing the risky option depends
on the last observed outcome. |
David Hagmann |
4:15-4:45 PM | An experimental test of theories of behavior in
Allais-type tasks This
paper tests the predictions of expected utility theory
and several of its most prominent alternatives and assesses
their ability to explain behavior in Allais-type tasks.
We investigate whether a transparent presentation reduces
violations of expected utility theory and find that it
does so dramatically. We also investigate the incremental
explanatory power of disappointment aversion, the fanning
out hypothesis, and rank-dependent utility theories, including
cumulative prospect theory with different probability
weighting functions. All these theories can justify the
predictions of expected utility as a special case, yet
all of their incremental explanatory power is attributable
to what we call a zero effect—a change in behavior
when a lottery may provide a zero outcome. |
Elif Incekara Hafalir |
4:45-5:00 PM | Conclusion | Coty Gonzalez |