Homo moralis goes to the voting booth: Coordination and information aggregation
| Published date | 01 April 2022 |
| Date | 01 April 2022 |
| Author | Jean-François Laslier,Ingela Alger |
| DOI | 10.1177/09516298221081811 |
| Subject Matter | Articles |
Homo moralis goes to the
voting booth: Coordination
and information aggregation
Ingela Alger
Toulouse School of Economics, CNRS, University of Toulouse Capitole,
and Institute for Advanced Study in Toulouse, France
Jean-François Laslier
Paris School of Economics (CNRS)
Abstract
This paper revisits two classical problems in the theory of voting—viz. the divided majority prob-
lem and the strategic revelation of information—in the light of evolutionarily founded partial
Kantian morality. It is shown that, compared to electorates consisting of purely self-interested
voters, such Kantian morality helps voters solve coordination problems and improves the informa-
tion aggregation properties of equilibria, even for modest levels of morality.
Keywords
Voting, Homo moralis, kantian morality, social dilemmas
1. Introduction
The question of individual cooperation is a puzzle for social theories because cooperation
should be sustained when efficient for the group but might be in contradiction with effi-
ciency at the individual level. This puzzle appears under various disguises in different
disciplines: Evolutionary Biology (Nowak and Sigmund, 2005), Ethology (deWaal,
1996), Economics (Moulin, 1995), Political theory (Ostrom, 1998) or Social
Philosophy (Binmore, 1994). In the light of recent results from the literature on the evo-
lutionary foundations of human motivation, we use formal game theory to revisit two
Corresponding author:
Ingela Alger,Toulouse School of Economics, CNRS, University of Toulouse Capitole, and Institutefor Advanced
Study in Toulouse, France.
Email: ingela.alger@tse-fr.eu
Article
Journal of Theoretical Politics
2022, Vol. 34(2) 280–312
© The Author(s) 2022
Article reuse guidelines:
sagepub.com/journals-permissions
DOI: 10.1177/09516298221081811
journals.sagepub.com/home/jtp
classic cooperation dilemmas faced by voters whose ability to communicate with each
other is limited: the divided majority problem and the Condorcet jury theorem.
The divided majority problem. In elections with at least three candidates, it is some-
times in the interest of supporters of two of the candidates to coordinate their votes on one
of them in order to block the other candidates (Cox, 1997). Instrumentally motivated
voters can be induced to vote strategically if their vote stands a chance of being
pivotal. In large enough electorates, however, pivotality becomes irrelevant. Hence it
may be necessary to resort to other explanations for why and how voters achieve such
coordination.
The Condorcet jury theorem. In situations where individuals receive private and
informative signals about the true state of nature, and where the preferences of the
voters are aligned—e.g., a jury that wishes to convict a person only if she is guilty—effi-
cient information aggregation is typically achieved if voters vote blindly according to the
signal they receive (Condorcet, 1785). However, as is known since Austen-Smith and
Banks (1996), a sophisticated voter would realize that, given that the others vote accord-
ing to the signal they receive, she should condition her vote on the event that she is
pivotal. In some settings this may lead her to conclude that it is better to vote against
the signal she observes. This surprising result casts doubt on the efficiency of majority
(and super-majority) rules as a procedure to aggregate information.
We contribute to both literatures by modelling voting behaviours based on preferences
that entail a form of universalization. Specifically, we adopt the view that voters have
Homo moralis preferences, which have been shown to be favoured by evolution by
natural selection (Alger and Weibull, 2013). Homo moralis reasons as follows: when
contemplating a course of action, she evaluates what her material payoff would be if—
hypothetically—each other individual of the population she belongs to were to follow
the same course of act ion with probability κ∈[0,1]. Although one might dispute
whether this behaviour captures the whole significance of Immanuel Kant’smorality,it
incorporates a key ingredient of this construct (and, arguably, of most moral theories):
the “universalization”principle (Kant, 1785; Roemer, 2019). One obtains the standard
materialistic Homo oeoconomicus for κ=0 and the Kantian model of Laffont (1975)
for κ=1. Values of κbetween 0 and 1 trigger partial Kantian universalization, and the
parameter κcan be interpreted as a level of morality. Hence, Homo moralis preferences
reconcile the theory of ethical voting with that of purely instrumental voters: it encom-
passes the purely instrumental motive as a special case, and spans a continuum of
degrees of partial universalization, up to and including full universalization. So the
model also stands on its own feetas a model of ethical behaviour,independently of its evo-
lutionary foundations. In practice, moraljudgments and actions are often guidedby the uni-
versalization principle (Levine et al., 2020) and Homo moralis can be read as modelling a
particular “ethical”behaviour based on partial universalization. Moreover, there is exten-
sive evidence that moral motivations are important for most voters (Blais, 2000).
Theories with ethically motivated voters are not new. In the literature the most
common formalization of an ethical voter comes in the form of the rule utilitarian
(Harsanyi, 1980, 1992) who selects a voting strategy which, if chosen by all other
rule-utilitarian voters, would maximize their aggregate material payoff. The partial
Kantian morality captured by Homo moralis preferences is a less demanding ethical
Alger and Laslier 281
concept than rule utilitarianism in two respects. First, the Homo moralis player does not
take into account the payoffs of the other players, only her own. Second, Homo moralispre-
ferences induces the voter to ponder what the outcome would be, if —hypothetically—
some fraction (not all) of the other voters selected the same strategy as her. It will be
seen that even small values of the universalization parameter κsometimes lead to results
that differ significantly from the ones obtained with instrumentally motivated voters.
Our formalization of ethical voters should not be confused with group-based voting
models (Coate and Conlin, 2004; Feddersen and Sandroni, 2006) where, by assumption,
strategic decisions are made at the collective level: in these models an ethical voter
chooses a strategy based on the anticipation that other ethical voters will effectively
also choose the same strategy. Along this line, two recent contributions have examined
one of the problems studied in this paper (the divided majority problem). Li and Pique
(2020) adopt the view that some voters are rule utilitarians, who select a voting strategy
which, when chosen by all voters in the divided majority, maximizes their utility. Bouton
and Ogden (2021) assume that voting strategies are taken at the level of the group, so that
each supporter of a particular candidate acts in the interest of the group of like-minded
supporters (it is as if they applied were rule utilitarians on behalf of the group). By con-
trast, in our model decisions are individually decided.
Since Rousseau (1755), many scholars have expressed the idea that political psych-
ology should not be cut from its possible biological roots: see for instance Schubert
(1982), Petersen (2015), Sidanus and Kurzban (2013) or Bergner and Hatemi (2017).
Within this stream of research, Homo moralis is a theoretical model that does not link
to empirical genetics but to the pure theory of evolution and stability of interactive behav-
iour. As in Economics (Lesourne et al., 2006) the evolutionary approach complements
the now-standard but still criticized theory of rational choice (Downs, 1957; Green and
Shapiro, 1994; Stephenson et al., 2018), by refining the concept of Nash equilibrium
(which does not contain by itself notions of stability or convergence
1
). In this work,
we use a distinction between (material) payoff and (subjective) utility, which is justified
by evolutionary considerations (Alger and Weibull, 2019). We therefore alter the object-
ive function. Still, we do compute the (Nash) equilibria of the games played with the
altered objective. We will here not attempt a full evolutionary study of these games,
but we will make the essential distinction between strict equilibria, where best responses
are well defined and which are known to be robust and stable under most dynamics, and
flat equilibria, in which all strategies are indifferent and that lack stability and robustness.
The paper is organized as follows. In section 2. we formally describe the Homo
moralis model. Then we consider two classical problems in the theory of voting: the
divided majority problem under plurality rule in section 3. and the question of strategic
revelation of information in the Condorcet jury setting in section 4.. For each of these
questions we study the implications of the hypothesis of evolutionary Kantian morality.
The last section is a short conclusion and proofs are in the Appendix.
2. Who is Homo moralis?
The evolutionary argument that implies precisely the behaviour termed “Homo moralis”
rests on the ideas that at least part of the fitness an individual achieves depends on the
282 Journal of Theoretical Politics 34(2)
To continue reading
Request your trial