Paradox lost: Explaining and modeling seemingly random individual behavior in social dilemmas
| Published date | 01 April 2011 |
| Author | Stephen Wendel,Joe Oppenheimer,Norman Frohlich |
| DOI | 10.1177/0951629811398687 |
| Date | 01 April 2011 |
| Subject Matter | Articles |
Article
Paradox lost: Explaining and
modeling seemingly random
individual behavior in social
dilemmas
Journal of Theoretical Politics
23(2) 165–187
©The Author(s) 2011
Reprints and permission:
sagepub.co.uk/journalsPermissions.nav
DOI:10.1177/0951629811398687
jtp.sagepub.com
Joe Oppenheimer
Department of Government and Politics, University of Maryland, USA
Stephen Wendel
Department of Government and Politics, University of Maryland, USA
Norman Frohlich
Asper School of Business, University of Manitoba, Canada and Department of Social and Preventive Medicine,
University of Montreal, Canada
Abstract
Despite a large body of experimental data demonstrating consistent group outcomes in social
dilemmas, a close look at individual behavior at the micro level reveals a more complicated story.
From round to round, individual behavior appears to be almost random. Using a combination of
formal deduction and agent-based simulations, we argue that any theory of individual choice that
accounts for the observed behavior of real people is likely to require 1) premises of probabilistic
choice, 2) preferences that are a function of others’ previous behavior (i.e., context dependent),
and 3) preferences that are other-regarding rather than simply self-interested. We present a
model that fits the requirements.
Keywords
agent-based modeling; other-regarding preferences; rationality; self-interest context-dependent
preferences; social dilemmas; voluntary contribution mechanism.
Mancur Olson (1965) first formally described the collective action problem in rational
choice terms, and provided an explanation for why groups of rational self-interested1
Corresponding author:
Joe Oppenheimer, Department of Government and Politics, University of Maryland, College Park, MD 20742,
USA
Email: joppenheimer@gvpt.umd.edu
166 Journal of Theoretical Politics 23(2)
individuals often fail to achieve attainable group benefits. The standard interpretative
model of collective action problems, which flows from his insights, presumes that one
can explain group (political) behavior as an n-person prisoner dilemma game (n-PD)
(see Ledyard, 1995). This model uses the workhorse behavioral assumptions of micro-
economics: rational, self-interested, non-probabilistic, separable preferences, hereafter
referred to as standard rational choice theory (or RCT). Problematically, a quarter cen-
tury of experimental data on collective action has produced results that contradict the
theory’s predictions. At the group level, there are well-documented problems, such as
non-conformity to the corner solution of no contributions, and the singular success of
any level of communication in driving cooperation.
However, here our attention is focused on remedying what we consider to be a bigger
failure of the theory: poor prediction at the individual level. The problem is that RCT
predicts an equilibrium behavioral outcome and yet the experimental data on individual
behavior appears almost chaotic.
We build upon alternative lines of research, in experimental economics and psy-
chology, which have developed more nuanced models of individual decision making.
Specifically,we argue, as we have earlier (Frohlich et al., 1984, 1996, 2004) that the stan-
dard interpretation of self-interest, that one only cares for the outcomes as they affect
oneself, needs amendment. In addition, we maintain that a probabilistic rational choice
model needs to replace the standard non-probabilistic descriptors of choice.2This pro-
gram is in line with observations in Frohlich and Oppenheimer (2006 [hereafter F&O])
which conjectured that an explanation for fluctuating individual contributions might be
based on individuals’ having complex context-dependent preferences: preferences that
depend upon the behavior of others in previous rounds. These conjectures were neither
fully developed nor tested there.
Wendel and Oppenheimer (2010 [hereafter W&O]) show that one can use
agent-based simulations to develop plausible specifications for a theory of complex
context-dependent preferences that generate the troubling aspects of the patterns of
contributions observed in the experiments. They suggested that analyses of situations
stemming from such simulations could help test specific models of behavior jointly with
further laboratory settings.
Using a combination of formal deduction and agent-based simulations, we here argue
that any theory of individual choice that accounts for the observed behavior of real
people is likely to require 1) preferences that are dependent upon the payoffs going to
others (i.e., other regarding behavior), 2) preferences that are related to others’ previous
efforts as compared to one’s own (i.e., some sort of taste for ‘fairness’3) and, 3) a specifi-
cation as to how preferences are probabilistically related to choice. We develop a feasible
rational choice model (i.e., within standard utility theory) with these properties to explain
behavior in terms of differentiable, continuous utility functions.
We conclude by noting that probabilistic behavior alters the strategic calculations of
actors, and hence what one can expect from particular institutions. Thus, some reformu-
lation of evaluative and predictive theories of institutions should be undertaken to take
into account these new elements, introduced into the traditional model. In particular, con-
clusions of models of social choice and voting based on the traditional theories may need
to be partially rethought or amended.4This would also extend to models of collective
action situations with little opportunity for learning or deduction about individual level
To continue reading
Request your trial