Individualism: the classical debate

Lecture 19/9

We start with the question about the origin of society. From here we can slope down the
question in the question on the possibility of explaining social behaviours rationally;
obviously fundamental for this second question is to define what is rationality.
The importance of rationality is well-known in western philosophy. Starting from
Aristotele passing from Cartesio and the enlightenment. We are going to analyse the notion
of rationality formalized and systematize in the rational choice theory (RCT). The process of
formalization and systematization starts in the first part of the twentieth century and is
strongly linked to the born of the social sciences. Is very useful starting our analyse at this
moment not because there weren’t these ideas before 1, but because this formalization creates
a benchmark. A necessary point of comparison for all the other’s reflection on this type of
rationality in this contest.
As mentioned before one of the first attempts to define rationality was made by Aristotele
and is well represented by the phrase «man is a political animal». So for the preceptor of
Alessandro Magno rationality was the faculty of reasoning that is a unique characteristic of
human beings, useful for distinguish humans from other animals. Using this definition our
last primary question changes from «what is rationality? » in «What is good reasoning? ».
Now through the use of Aristotele definition we can distinguish two different type of
rationality: the faculty of reasoning and the good use of the faculty of reasoning.
We will see as in the twentieth century the RCT will overcome also the fact that rationality
is a faculty. To be rational will means to act, and behave, in a rational way.
The fundamental distinction that we must look at is the one between practical reasoning
that has an outcome action and theoretical reasoning that has outcome beliefs. Using this
distinction as an assumption doesn’t signific that beliefs don’t belong to both types of
reasoning. Is important not to forget that beliefs contribute also to the formation of the
practical reasoning outcome: action.
Keeping our eyes on Aristotele: you are rational if you have reasons. The faculty of
reasoning isn’t the faculty of creating reasons? Starting from here we can observe how facts
could be reasons and reasons take reality in facts. But we can observe also that behind a
reason there are beliefs, values, and judgments. So, if in the twentieth century to be rational
means to act in a rational way these two observations show us that rational action has many
pillars. We can go further: we can explain action with facts. Always reminding that behind
facts there are beliefs, values, and judgements.
A way in which is possible to explain behaviours is: to make a hypothesis on values and
beliefs. Preliminary to this method is taking as the assumption that: the agent acts rationally
and also that through rationality is possible to make a prediction.

1
Cfr. Adam Smith
 From here we can classify the issue as normative. The new question-related to the
normative classification of the issue is: what shall I do to be a rational being? Therefore,
which action shell I do starting from precise values and beliefs.

Lecture 20/09

We will focus a part of our argumentation on institutions. The fundamental reason for this
choice is that our first question was the one on the origin of society. We can observe that
societies are made of institution, therefore, discovering what institutions are is a fundamental
part of the solution to our first question. Moreover, there are many other reasons that justify
our investigation on institutions.
Institutions are important because they influence humans’ behaviours. This claim is
supported by the fact that they give sense to many things. A good example could be the
values of a university of course if it wasn’t given by the institution of university.
Another reason for pays attention to institutions will be because we can observe a relation
between quality of life and quality of institution. Natural recourses are not a sufficient
condition to be a rich nation. From the other hand it seems that is necessary to have good
institution to be a rich nation. A good example could be the comparison between Swiss and
Bolivia.
Punt another time our eyes on the try of Aristotele to find something that distinguish
humans being from other animals we can identify this thing in institution (cfr. last lectures).
Institution are only humans. There are many other social animals, but none use the mediation
of institution in its form of socialization.
So now my personal question could be to find preliminary definition of institution. For go
further in our research we are going to make a try of analysing and resuming the history of
the question about institution, so trough the history of social sciences.
Preparatory for our analysis is to clarify the approach we want to use. The first
qualification that we can give our approach is pragmatic. This qualification is justified by
the fact that we use it because is useful for producing good results. The assumption that we
use for starting is the individualistic intuition. This intuition in justified by the fact that
without humans wouldn’t exist the human society. So, after these two observations we can
describe our method. The behaviour of a part is corelated to the behaviour of the whole object
of study (principle like the reductionism in physics). Therefore, the relation between the
parts is the starting point for discover how the whole works. Human societies are made by
humans and something that keeps them together 2. Looking at the assumption so we can
summarise the approach: «start from the relation between humans for explain the society».
The definition of this approach like individualistic intuition obviously has all the problems
related to the word individualistic and in consequence to the concept of individualism 3.

2
Cfr. Durkheim mechanical solidarity and organic solidarity.
3
Cfr. History of the European individualism. Is possible that the deep penetration of the concept of
individualism in western philosophy has influenced the method of the social science. But this doesn’t weaken
our approach because in origin it is justified by a pragmatic argument.
 Returning at our historical approach we can start looking at the 19 th century because this is
the period in which were born the social science. They were born by the disjunction from
philosophy. The 19th century is on one hand the century of individualism that take reality in
the romanticism4, and on the other hand the century of the growth of the natural science in the
form of mechanism5. So, social science can’t forget both of these trends.
The discussion on the method of social science takes a good formalisation in the
“methodenstreit”. The discussion between two positions, mainly represented by Carl
Menger and Gustav von Schmoller. The first represent the trend that argues the existence of
general laws in the social science and that those could be find using mathematics as an
instrument. The second argues that economy and social sciences in general are an historical
science. Therefore, that do not exist general laws. This secondo position could be named the
historical school.
A key point in this debate is represented by Max Weber. The scholar proposes a
compromise between the two positions. The possibility of this compromise is given by the
theoretical instrument of the methodological individualism6.
We can find the bases of the metalogical individualism in the first chapter of Economy and
society. This concept appears in a clear way in the for characteristic of the social sciences’
knowledge7. Knowledge in social sciences is interpretative, is causal, is individualistic and is
an ideal-type.
The first characteristic is that the knowledge of the social sciences is interpretative. It is
well explained by Weber himself « Sociology (in the sense in which this highly ambiguous
word is used here) is a science which attempts the interpretive understanding of social action
in order thereby to arrive at a causal explanation of its course and effects ». The second
characteristic is that understanding is causal so that is fundamental to discover the cause of
an action not only the motives. This attribute reminds us that the motives aren’t necessarily
the causes of an action and that the social sciences, in particular in this case sociology have to
discover the real causes of an action. The connection between motive and action is not
necessary. What is necessary is to find the real causes of an action. The third characteristic is
that the understanding is individualistic in the sense that are people that make action not
entities, not institution (example could be that is not Russia that have attacked Ukraine but
the people that constitute the govern of Russia). So, in necessary to look at individual humans
being. A fundamental argument in this sense is that institution and entities couldn’t have
motives for action in the sense of mental states. So, the attempt of the methodological
individualism is to explain the relation between institution and individuals (a mutual
influence relation). The fourth and the last characteristic is that rationality is an ideal type. So
that the model is a representation of a human being, it is an idealisation of a human being. For

4
Cfr. Literature and paintings.
5
Line of thought that starts in the ancient philosophy for exempla in the atomism of Democrito passing through
Cartesio and Hobbes in the 17th century and that becomes nearly total in the 19th century (Loeb Jaques).
6
Cfr. Durkheim that proposes an opposite position. The scholar starts from society not prom individuals. The
Weber’s approach is surely the one that won among the economists.
7
Understanding and knowledge will be used as synonymous in the next few lines.
Weber is useful start from the fact that the actions are acted by a rational agent 8. He is
perfectly conscious of the existence of no rational action (empathetic action) but he prefers to
start with the hypothesis of rationality always reminding the empathetic action exist and
could have their motives and reasons.
These four determinations of knowledge in social sciences are the demonstration that
Weber does not want to abandon the idea of the importance of the context and the idea that
the psychological motives are important. For these reasons this argues are the synthesis of the
two-position described before.

Weber, M. (1921) “The Interpretive Understanding of Social Action”, in Readings in

the Philosophy of the Social Sciences, edited by M. Brodbeck. Macmillan, pp. 19-33.

Weber make the distinction between an action generally conceived and a social action.
An action in general is a deliberate and positive intervention in a situation, or the deliberate
no intervention or also the passive acceptance of a situation. A social action is when the
action takes in consideration other subjects. Our aim is to reach the comprehension of actions.
So, fundamental is the definition of the term “meaning”. The basis for understanding could be
rational or empathic. For the scientific proposals of the program the most problematic part
is the emotional one. The solution is to treat emotional motives as a deviation rational once’s
(the pure/ideal type). The use of this methodological device perfectly shows the sense in
which the knowledge is social sciences is rationalistic.
In general understanding may be of two types direct observational understanding or
explanatory understanding. Both moods of understanding could be used on both rational
and irrational phenomena, both on natural and social phenomena. Every interpretation
attempts for clarity, but it is not always possible. The causal understandings of phenomena
are not always easy. The necessary comparison between the subjective part and the real
things is not always possible with a good accuracy. Many times is necessary to use the
dangerous practice of the thought experiment. The possibility of determining the probability
is one of the pillars of ad adequate causal explanation. Subjective understanding is the
specific characteristic of sociology knowledge.
Meaning actions are always subjective and individual. An action exists only as the
behaviour of an individual. Those are subjectively understandable orientation of behaviours.
It is evident that the fundamental ground of Weber’s view of sociology is the individual. The
individualistic premises leads to the understanding of social actions, the aim of sociology.
This methodology pulls in opposite direction in respect of the organic school of sociology
that attempts to study society starting from the whole within all individual acts.

Durkheim, E. (1892) “What is a social fact?”, in The Rules of Sociological Method.

Palgrave, 50-59.
8
Cfr. the notion of a rational action as the action that is the best for reach the proposes of the action. This is the
notion of rationality that weber use. This notion in called instrumental rationality.
 E. Durkheim in this text defines what is a social fact. This could be an interesting line of
comparison with Weber. On one hand sociology study social action on the other social facts
are the objects of sociology. The difference in the object underlines very well the
methodological difference between the two approaches. Durkheim here is interested in
defining what is a social fact because in his prospective this is the necessary premise for
searching the right method in sociology.
Social facts are distinct from natural facts, which are objects of natural sciences. Social
facts exist outside the consciousness of the individual (cfr. Weber and Serle). Social facts
consist of manners of acting, thinking and feeling external to the individual, which are
invested with a coercive power by virtue of which they exercise control over him. These are
neither organic phenomena nor natural phenomena so they must be called social phenomena.
(e.g., legal, and moral rules, religious dogma, financial systems…) Particular significant for
showing the difference with Weber is the fact that social facts have as substratum the society
not individuals: «it is clear that, not having the individual as their substratum, they can have
none other than society». There are also social facts called current social facts. This group
of fact differ from simply social facts because it is not so easy to identify the pressure, they
are exercising on individuals, from the point of view of an individual.
A paradigmatic example that shows reasons in favour of this definition of social facts is
the growth of children. Education is a continuous effort to impose ways of seeing. Important
is that education is oriented to create a social being.
We do not have to confuse social facts with individual incarnation (social action for
Weber). The fact that individual tendency is sheared not create social facts. What constitute
social facts are beliefs, tendencies…of the group as a collective set. The process that
guarantees for the definition of social fact is the possibility of dissociation from individual
effects of those facts. The facts are general, and the consequences are individual, but this do
not influence the collective nature of social facts.
There are two addition criteria for determine which are social facts and to identifies them:
1) by the coercive power that could be discovered because of the existence of pre-determined
sanctions or the force that the facts opposes to any individual action that threaten it 2) by
accreting how much it is spread in the group, this second criteria at the condition that the
facts exist independently from the group. The second criterion is simply a redefinition of the
first one: if a fact exist in a group and it is independent from the group it exercise pressure
over the individuals.
At the end of the chapter Durkheim give us an explicit definition of what is a social fact:
«A social fact is any way of acting, whether fixed or not, capable of exerting over the
individual an external constraint; or: which is general over the whole of a given society whilst
having an existence of its own, independent of its individual manifestation. »

Watkins, J. W. (1957) “Historical Explanation in the Social Sciences”, British Journal

for the Philosophy of Science 8: 104-117.
 Watkins (K. Popper student) opens his analysis by creating a parallel between mechanism
and methodological individualism, he declared to be a Defensor of these two principles.
Every social situation can be reduced to individuals, it is a particular configuration of
physical, psychological disposition (believes, environment…). This methodological
presupposition is opposite to the social holism or organicism: social systems are governed by
macro laws that are essentially sociological. Laws that are not reducible to individual
behaviour.
The historical declination of this distinction is the fact that for methodological
individualism individual are the only moving agents in history and for social organicism there
are others factor working in the construction of history.
The central claim of methodological individualism, in the way Popper describes it, is that
no social tendency is imposed on human beings, that humans being can in principle always
change all the tendency that seams to act on them.
There are two spheres where methodological individualism does not work: 1) social
regularities, those situation where unpredictable irregularities in human behaviour have a
predictable and regular result (this happens because are situation where variable humans
factors will usually have no influence have here an important influence) e.g. the number of
car accidents do non oscillate very much. 2) In situation where there is an automatic body
responds.
There are some misunderstandings in methodological individualism: 1) It encourages
innocent explanations. It has been objected that with methodological individualism the
psychological state of a person is God-given, in the sense that methodological individualism
negates external constraint on psychological state of individual. Watkins responds that this is
not true because methodological individualism only requires that those psychological
disposition and individual and not collective. 2) methodological individualism has been
confused with psychologism, psychological reductionism. All large-scale social
characteristics are reflection of individual characteristics. Psychologism is a narrow form of
methodological individualism; it used the method but the method do not commit to a
psychologist position.
In the last two section of the paper is discussed about how individualistic presupposition
can lead to important and interesting social discoveries. And how explanation should be
framed in an individualistic frame. The two discussions are reciprocal useful. The holist
methodologies fail because societies are really made by individuals, this sems to be Watkins
explanation. On the other hand, the frame is the correct frame for economics explanation.
Models do not fit on every historical situation. This is the correct frame of a sociological
explanation made starting from the assumption of methodological individualism.

Zahle, J., & Kincaid, H. (2019) "Why Be a Methodological Individualist?", Synthese,

196: 655-675.

Zahle papers analyse the main traditional reasons in favour of Methodological

individualism and conclude that those are not very convincing. In this paper methodological
individualism is conceived as a claim about explanation, individualist explanation. In general,
the debate between methodological individualism and holism is understood as a debate
about the method of explanation. The aim of both positions is to create scientific
explanations.
Starting the analysis are define two different ways of methodological individualism, two
different formulations. 1) Individualist explanation: individualist explanation alone should
be put forward within the social sciences; they are indispensable. Holist explanations may,
and should, be dispensed with (possibility of reinforcing the claim eliminating the holist
possibility). In individual explanation, the individuals are the subjects of explanation, e.g., the
absence of cash in the banks is caused by the fact that all the customers want their money
back. In holist explanation other subjects are the subjects of explanation, e.g. the rise in
unemployment led to a higher crime rate. 2) Microfoundationist account: purely holist
explanations may never stand on their own; they should always be supplemented by accounts
of the underlying individual-level: micro-foundations. Individualism is not necessarily
reductionism; it is not so easy to grasp but it seems possible. Individualists claim to explain
social phenomena case by case without constructing a social type phenomenon. The article
assumes that reductionist positions are probably wrong. Zahle considers now some classical
and important arguments in favour of methodological individualism.
Humans are considered the only causal factor in history, and for this presupposition
methodological individualism is the best way to explain social phenomena. Causation seams
to be necessary only at micro-level, macro-lever is supposed to act in virtue of his parts.
The individual explanations are considered to be deeper that the holistic ones. In this
analysis we are assuming that reductionism fails, so the depth could not be understood as the
facts that holistic explanation are simply an aggregate of individuals one. Individualist
argument in favour of this is game theory in general; it is supposed to explain individuals
behaviours and holist mechanism. The general idea that make us doubt about this argument is
that fact that depth explanation are always preferable?
Human actions are the cause of human agency, and agency requires individual
explanation, not holistic explanation. For Weber fundamental aim of sociology is to
understand subjective understanding of agents, so of individual action and since agency is the
cause of action also individual and subjective agency. Also, the use of quantitative methods
has been useful to development of the attention for individual agency as a fundamental part
of interpretive understanding of social sciences. There is a fundamental argument that support
this view in philosophy; this is the interest in motives. Explain is to finds the motives and
reasons of an action. Reason and intentions are only individual. So, the methodological
individualist analysis is the most appropriate. The holist reply that collective entities could
also have reasons and intentions. Other reasons for denying these individualistic arguments
are to the metaphysical conception of human agency; is wrong to conceive it as the necessary
and sufficient reasons for acting.
Social explanation could not be derived by normativity, and this happens. The key idea
that informs normativity, classical liberalism political philosophical theory, is that individual
actions are a matter of choice. Individual autonomy recognized by the normative dimension
could not be fulfilled by holistic explanation. For this reason individual explanation sems to
be the more adapts.
 The conclusion of the paper is that none of the four lines of argument sems impelled us to
use methodological individualism as a base assumption for our analysis is social sciences. On
the other hand this not imply holistic explanation too.

Methodological Individualism, in Stanford Encyclopaedia of Philosophy

Methodological individualism pretends to be a scientific doctrine. Weber himself says

that is a mistake to think that methodological individualism also implies a system of
individualistic values, this has largely been objected (cfr. L. Goldstein and S. Lukes). But the
commitment to methodological individualism commits at an interpretative pattern of
explanation. The term “methodological individualism” is first used by J. Schumpeter, but he
uses it to refer to the Weberian thought.
Weberian prospective differs from the tradition of atomism, the first generation of
Austrian school such as C. Menger. On one hand atomism commit to the possibility of the
perfect definition of mental stats that leads to the action, on the other hand methodological
individualism no not commit to any particular claim about the content of the mental stats that
lead to the action, the possibility of an irreducible social dimension of the action is still valid.
The most important consequence of Weber theory is that is puts RCT at the hart of the
scientific inquiry about society. Homo economicus is the clearest model of the man of RCT.
The member of the second generation of Austrian School, such as F. von Hayek,
identifies them self in the Weberian theory. Hayek commitment to Weberian prospective
justified the importance of individual prospective, many times unconscious of the general
consequences of its actions e.g., the process that leads to the formation of a path in the
woods. Methodological individualism helps us to show the limitative prospective of each
individual, and using it we cannot lose this fundamental aspect of reality. Useful terminology
used by Watkins for distinguish the type of research of methodological individualist form the
research of holists is the that between “unfinished or half-way explanation” and “rock-
bottom explanation”.
One of the first motivation to commit to methodological individualism is that is a way to
avoid social “grand theory” e.g., Marx, Hegel and Comte. The diffusion in the 80’ of game
theory among social scientist gives a very important pull to methodological individualism.
The reason is that it allows to represent mathematically complex social interaction starting
from individual payoffs e.g., prisoner dilemma game. The contribution of J. Elster has to be
understood in this frame. His critique of Marxism has having been constructed upon
functionalist analysis and having insight an intrinsic teleology is to be understood like the
Watkins difference between “unfinished or half-way explanation” and “rock-bottom
explanation”. For Elster the commitments for methodological individualism and the
commitment for RCT are reciprocal. This connection is helped by the traditional conception
of rationality as instrumental rationality: action is chosen as the most adequate means for
futures ends. Probably Elester fails in making this distinction because in this way he forgets
the Weberian distinction between understand action at an individual level and a specific
model of rationality.
The most identified difficulties in methodological individualist analysis stays in: 1)
statistical analysis, they sems good analysis but do not take in consideration the agents mental
state (maybe because it is impossible). Adding the methodological individualism
characteristics at these analyses sems not to add to much. 2) Sub-intentional explanations, it
seems that deeper explanation is always possible. When we have to stop? When we reach the
biological dimension? 3) microrealization-robusteness 3) fallacy such as the fact that this
prospective helps too much to go in depth without allowing the possibility to consider
collective action problems. Methodological individualism sems to distance large-scale
analysis. Another common fallacy is to identify the “self-interest” as the biological “fitness”
and to assume the existence of some social selection mechanism.

Individualism in practice: rational choice theory (RCT)

Lecture 21/09

After the use of Weber as an introduction to the method the social sciences use, we can
start analyse what is rationality in the RCT. This question is fundamental for discover how
society works; the first question of these lectures.
The rationality in the RCT is instrumental rationality 910: an action is rational if it is
appropriate to the purpose. Therefore, an agent chooses an action to satisfy his preferences.
The preferences are a fundamental instrument for RCT in fact the rational agent is the one
that decide between possibilities using his preferences. An important objection that we will
discuss is that an agent, in an uncertain situation, so in more or less every decision that makes
in real life, can also does not know the best action for reach his proposes. So, for this simple
definition of rationality will act in an un-rational way.
We do not speak about what preference contain 11. RCT want to be over parts. This theory
does not say what we must prefer but wat we must prefer in we have some preferences. Very
useful is the comparison with logic. Logic is an instrument that tell us if our argue are correct
or incorrect starting form specific presupposition. In the same way RCT show us what we
will prefer starting from specific preferences.

9
Cfr. Utilitarisem.
10
Cfr. D. Hume. Rationality is not wat people want. We cannot include people the preferences contain. The
preferences, the desire cannot be included in rationality. Act in a rational way is act looking at preferences and
desire. This is one of the pillar of the liberal theorisation.
11
Cfr. nota 10
 We can find ad important formalisation of RCT in microeconomics 12. The formalisation
starts from tree axioms or principles 13 and if an agent act without breaking these principles he
acts rationally. The first one is consumerism, the second one is completeness and the third
one is transitivity. The consumerism is basically the fact that people prefer to consume more
than less. The completeness is the fact that preferences are complete so that is necessary to
take a decision, in necessary to have preferences, also with the possibility to be indifferent.
Also be indifferent is a preference14. Transitivity tells us that if I prefer A to B and B to C
so I must prefer A to C. The third axiom is fundamental because it create an order list of
elements15. There are many arguments that argues in favour of the truth of the axiom of
transitivity; one of the more important is the money-pomp argument 16. Necessary assumption
for the validity of the third axiom is the certainty.

Lecture 27/09

Now has come the time to analyse the relation between RCT utilitarianism. The
utilitarianism movement is an interdisciplinary field of research between science and
philosophy that born in the first part of the 19 th century. Some of its important exponents are
Jeremy Bentham and looking at Italy Cesare Beccaria. In a very schematic and simple
description we can assert that the utilitarianists tries to explain humans’ behaviours in terms
of utility. Utility is a mental state, something like pleasure or happiness, so the humans being
act attempting to maximise utility.
Economists, in particular Wilfredo Pareto, use some of the ideas of the utilitarianism for
build the RCT. In their systematisation economists reinterpret the concept of utility in
satisfaction. This change happens because the utility concepts of utilitarianists appeared
impossible to quantify17. Creating a link between preference and satisfaction we can say that I
receive more satisfaction if I choose something that I prefer. For make useful consideration
starting from this connection fundamental is the axiom of transitivity. Now if, in virtue of
transitivity, our preferences are order we can create an internal, relative, measure of

12
Cfr. Krugman e Zhok.
13
Cfr. There are many formalisations of the principle of microeconomics.
14
The idea of completeness is strongly linked to the idea of quantification.
15
Cfr. Logical operation on lists.
16
If my preferences are intransitive I prefer B to A and C to B and A to C (if there were transitive I wil prefere
B to A and A to C and B to C), so I create a circle of preference (the opposite of the order list that we can create
starting from transitive preferences). Using the assumption that every time I buy something that a prefer I
increase my satisfaction first I already have A I will buy B for A plus x 1 money, than I will buy C for B and x 2
money at the and I will buy A for C plus x3 money because I prefer A to C. Now I have A, the one that I already
have when I start buying, and a lost x1 plus x2 plus x3.
The example use in the lecture was the one about money and tree markers.
17
Now we have more tools, in particular neuroscience that help us to quantify also happiness. Obviously
reducing happiness at things like the mensuration of some substances.
satisfaction and if satisfaction is the utility of the utilitarianists we can create an ordinal rank
that give us the possibility of measuring the reasons because people act. Pareto is the most
influent economist in this reflection. He creates a mathematic formalisation of the RCT built
starting from the ordinal scale of preferences.

x>y ↔ u(x)>u(y)18

Is very important remind two things. The first is that the scale that we have just created is
ordinal and not cardinal (this s going to be the next passage), the second is that the
economists are conscious that this theory works in conditions of certainty (in the real-life
information is always partial19) and that the condition of uncertainty introduces the concept of
risk.
The first important attempts of risk’s analyses are made in the first part of 19 th century20.
The introduction of probability gives us the possibility to create a new definition of rational
choice: a decision is rational if is the one with the best probability. In others word is
necessary to weight the outcome of an action with its probability (calculate this probability is
another thing).
The model that was utilised was the one that introduce the expected monetary values
(EMV). Specifically, could be interesting and useful to clarify the concept the example of a
lottery:

The decision we must take is to buy or not to buy the ticket. For this theory a rational choice the agent must
choose the possibility with the highest EMV.

Ticket: 1$
Potential win: 1.000.000 $
Winning tickets: 1.100.000 $
Winning buying a ticket is named the event W
P(W) = 1/1.100.000
P(¬W) = 1.099.999/1.100.000

EMV1= 1$ if choose not to buy the ticket

EMV2= (0) P(¬W) + (1.000.000) P(W) = 0,91 if choose to buy the ticket

A situation like this doesn’t take in consideration many factors. The key point is that in
many cases 1 euro of difference is not much but one million is a lot. It seems that with one
euro we are buying the possibility of a million. Saying that the decision of buy the ticket is
irrational is first saying that all the people that buy a lottery ticket are irrational and second is
forgot that for each one the value of money is different. This second consequence shows a
paternalistic approach that is totally different from the idea of the RCT (the theory is an
instrument that with a scale of ordinal preferences tells us what we should prefer).
Continuing in our analyse towards a more precise definition of the RCT is useful to
mention Daniel Bernoulli that remind us that in the operation does before is necessary to use
not the absolute value of money but the relative one. So that is necessary to build a cardinal
18
I prefer x to y if and only if the value of the utility function of x is major of the values of the utility function of
y.
19
Cfr. Zhok normativity of perfect market
20
Cfr. Laplace and his studies on probability.
scale (returns the problem of quantification and of creating a cardinal scale). The solution of
this problem, the problem of cardinal scale (that means quantification), comes in the 20 th
century.
There are many scholars that contributed solve this problem, the most important are F.
Ramsey, Von Neumann, Morgenstern, and L. Savage. Ramsey poses some basis; Von
Neumann and Morgenstern elaborate the theory and L. Savage gives the actual formalization
and systematisation of the RCT. The idea that stays under all this new version of the RCT is
that a rational agent chooses maximizing his expected utility (EU) which is different from
the precedent version in which the aim of the ration agent was to choose the option that
maximise his EMV.

EU=Σ P(xi) u(xi)

x is the outcome of the action and u is the utility function calculated on a cardinal scale.

Von Neumann and Morgenstern discovered a way in which is possible to calculate utility
on a cardinal scale. This is the substantial change between this formulation and the precedent.
This theory has new axioms: completeness, transitivity, continuity, and independence (cfr.
Anand’s definition). The first and the secondo axioms are the same that in the precedent
version on the theory. The third has a mathematic function: it allows us to say that the values
of utility on the cardinal scale belongs to the real numbers. The fourth one is the more
interesting for our proposes.

x>y ↔ [px+(1-p)z]>[py+(1-p)z]

x>y ↔ EU(x)>EU(y) formalisation of the 1942’s theory

Is always useful to think that the social scientist knows perfectly that people acting doesn’t
make this calculation. This reasoning is an ideal type in the sense of Weber. The necessary
premise to all this reflection is that the preferences are consistent.

Anand, P. (1995). Foundations of Rational Choice Under Risk. Oxford University

Press, Ch. 1.

Decision theory is about choosing the act that is the best for our beliefs and desire, ends.
This is the concept of instrumental rationality. Methodological tool for this aim in the
subjective expected utility SEU. SEU define what is rational, so it is also a theory of
irrationality.
Decision theory do not say what people must desire; it respect the values of modern
democracy. The concept of expected value (Expected monetary Values) helps us to weight
payoffs on probability but for P. Arnad is different from the concept of expected utility.
Expected value leads at some problems, the main identified by D. Bernoulli called St
Petersburg paradox21. The solution of the paradox is a function that maps wealth on utility,
21
At what price Peter should participate to the game? For expected value he must pay any price because the
expected value of the lottery is infinite, but the game may ends on the first round so no rational agent would pay
as wealth increase utility decrees, this function is logarithmic so finite. So, there is a price
that would be right to pay for entering the lottery.
There are many views about what probability is. 1) For the classical views, Laplace, it
measures the uncertainty on deterministic events. 2) for the logical or a priori views
probabilities are to be found in logical relation between propositions. 3) another view suggest
that probability is the relative frequency of an event in a set. 4) there is a view that solve
some of the problem of the relative-frequentist approach that probability measures beliefs.
This last approach is the one of F. P. Ramsey. This approach allows us to measure beliefs,
seen as subjective probabilities, through the bet that an agent would made on those beliefs.
The mathematical tools are construction of a lottery of “if…then” and assigning it at an event,
which has a probability, and its negation. Starting from the preferences of the agent is
possible to build an ordinal scale of preferences. The utility of each lottery is calculated
weighting utility of the consequences on probability of antecedent. We reach the conclusion
that the subjective probability, the beliefs that E will occur, can be thought as the division of
two utility differences. Interesting point is the starting choices are arbitrary depends on the
agent. So, those have to respect some conditions that are the axiom of SEU theory.
SEU solve the problem of EV and of the ordinal scale of preferences. Thos are
fundamental steps in the growth of RCT.
Axioms are considered assumption does not postulate; assumption that no rational agent
would refuse. There are may axiomatics formulations of EU; von Neumann-Morgenstern
takes Ramsey approach to beliefs as subjective probabilities. The chapter analyse the Savage
approach to axioms. This formalisation is interesting because Savage do not expect real
decision-makers should not necessarily modify their actions (comparison with Plato ideas).
Savage defines the concepts of world, state, state of the world and consequences these are the
concepts useful to understand the axioms. Completeness, transitivity, The independence
axiom, Resolution independence, Expected Wealth independence, minimal strict preference,
continuity in probability, partial resolution independence.
SEU has been a fundamental step in the development of decision theory, the maximisation
of SEU to select actions. SEU provides the means of measurement for using game theory. In
economics is used in the theory of the consumer or the theory of general equilibrium.
Empirical violation of SEU provides reasons for its normative interpretation.

Hargreaves Heap, S. et al. (1992) The Theory of Choice, Blackwell, Ch.1, especially
pp. 3-14.

The aim is to introduce the different senses of individual rationality. The most common
sense is the instrumental one. Other conceptions of rationality are more open regarding the
conception of individual and conceived action in many ways. A first distinction is between
procedural and expressive.

any price.
 Instrumental rationality idea is first conceived in the Humian conception. The individual
must be capable of comparing the possible satisfactions. Fundamental in the theoretical
construction of modern instrumental rationality are the utilitarianism ideas and the concept of
ordered preferences. The decision in always between bundles of goods that must satisfy the
following conditions: reflexivity, completeness, transitivity, and continuity (given two goods
in a bundle is always possible to construct another indifferent bundle). If an individuum act
according to his preferences he maximises utility, he chooses the bundle that gives him more
utility. This construction of the theory work only in situation (inexistent in real life) where
there is no uncertainty. In situation of uncertainty the individual maximises expected utility.
This is the result of the von Neumann-Morgenstern approach (Savage provides a different
method). There is no interpersonal possibility of comparison, the unit of measurement is still
individual.
The concept of instrumental rationality has limits. In general, first doubts arrive because
may individual and many individual behaviours do not fit with this model e.g., “being
spontaneous” and “going sleep” for an insomniac. There are philosophical difficulties but
also practical ones. Difficulties of instrumental rationality leads to the theoretical construction
of new concepts of rationality.
One of these alternatives’ ideas of rationality in the procedural rationality (bounded
rationality and norms): individua use rules of thumb, procedures, to guid their actions. This
became much more convincing if we think that procedures are fundamental in economize on
scare resources, helps the brain limited computational capacity. Individuals are still
maximizing their expected utility but their rationality in bounded because there in no full
information. Individual action so must be understood in a homo sociologicus framework,
different from the homo economicus framework typical of instrumental rationality. The
interdependency of behaviours excludes from the equation the computational capacity.
Sheard believes in a society are fundamental in coordination. Informal norms help us to be
much more efficient. Individual choices could not be understood only in terms of individuals
attempt to reach satisfaction. Sheared rules are irreducible. Pre-existing rules govern our
behaviour, this is the framework in which instrumental rationality could work. Instrumental
rationality could work in the pre-existing structure of society. This structure allows us to
understood instrumental rationality decision. Historical dimension is irreducible. Procedural
rationality is the limit of instrumental rationality.
Another concept that born from the limits of instrumental rationality in the concept of
expressive rationality (making sense of the self through creative choice). Through many
philosophical reflections (Marx, Kant…) seams that rationality concerns with establishing the
value of ends and purpose. So, actions are more an expression of beliefs on value and not the
attempt to satisfy givens objects. This idea is the idea of expressive rationality. Important part
of this conception in the desire for self-respect that could not belongs to instrumental
conception of rationality. Expressive rationality could make reason of some structure of
preference that could not be possible to justify in an instrumental framework. This in not
allowing whichever structure of preferences.
The three concepts of rationality described in this chapter must be understood not as
mutually exclusive mut mutually complementor. There could be situation of well-defined
objectives where instrumental rationality works perfectly and situation in which other types
of rationality work better.

Lecture 28/09

The notion of rationality is assumed in the theory of rational choice. Two fundamental
things:
1) instrumental rationality: we assume that individuals choose the actions that achieve
most effective to achieve some preferences that we assume people have
2) this notion is neutral about the content of the preference and is only concerned with
some very simple constraints about the shape these preferences must have
The requirement of rationality are structures as in logic, “if… than…”, syntactic relations
on propositions.
The development of these theory went of a century. Economists were the first one to ask
how can we build a theory of human choice that is precise enough to provide rigorous
explanation of human behavior. They started to transform some concept in precise
mathematical theory.
One of the first important theories is the theory of supply & demand: it represents the
relationship between the price and two factors, that are called supply and demand curves.
These curves represent two mathematical functions. The supply curve (on the right) tells us
that the quantity of commodities that is offered (sold) in the market increases with the price:
if the price is higher, we should aspect that more commodities will be sold.
The demand curve is the opposite: the demand that consumers ask for in the market, the
quantity of that good, decline if the price increases: as the price goes up, few quantitative will
be consumed. It is an intuitive motive that does not require justification. The demand curve is
based on individual behavior. What kind of behavior we have to assume in order to have
demand curves that are shaped in this particular shape? These functions are just summaries of
the behavior of millions of consumers and producers, so if we are individualist, we want to
see how these curves are generate by several people.
theory of consumers behavior
One economist took seriously this question and came to an answer that provided the
foundation of the theory of rational choice: we need to assume some simple principles on the
preferences of consumers. 3 principles of the theory of consumer behavior that do not
concern the content of their preferences but are worried about the satisfaction of
consistency/coherence:
1) consumerism: people prefer to consume more rather than less (has limits).
2) completeness: (can’t say I don’t know, have no preferences) they can rank any set of
options (bundles) they are faced with (for any x, y, either x≻y, or y≻x, or x∼y).
3) Transitivity: for any x, y, z, if x≻y and y≻z, then x≻z. ≻: symbol of preference; ∼:
indifference.
But only 2) and 3) are really rational, they are not perfect, but they let a lot of freedom.
 (2) means that if you give someone any pair of options you must express an evaluation,
i.e., a preference; whatever options i give you, you should be able to express a preference,
you should be able to say why you prefer x to y, or y to x, or if you are indifferent.
(3) means that for any three options, if you prefer x to y and you prefer y to z, then you
must prefer x to z; it says that you should be able to rank your options and this ranking must
be an ordered shape (one option cannot be in two positions of the ranking).
These axioms are interpreted as an idealization. It is sufficient for the purpose of analysis
that certain kind of behavior can be described in aggregate.
Exist a function of utility which is represent these principles, U(x) exists in a scale of
different x in which we can assigns numbers to each possible bundle x, such that for every
pair of bundle, xi and xj, when the first is preferred to the second then the number associated
with the first is greater than the number associated with the second, this means that U(x i) >
U(xj); and when you are indifferent between the first and the second you’ll have that the
numbers assigned for the two functions are the same, that is U(xi) = U(xj).
It is the mapping of this latter points that yields to the ‘indifferent curve’, depiction of
these utility functions. We have to remember that unlike traditional utilitarianism, it does not
measure some absolute quantity of sensation, pleasure or desire which can be compared
across individuals. Instead it is only a device which expresses an ordering, and there are any
number of utility functions which could represent that ordering. Is not the absolute value of
the utility assigned by the utility function to each option that matters here. What matters is the
assignment of utility in relation to what other options receive, so that the ordering is
preserved under each function.
The choice under which we have this representation has being developed in a special
condition, we don’t have an uncertainty between actions and outcomes.
2.2 From Expected Value
Decision theory has two fundamental aspects:
1) instrumental rationality from Hume tradition
2) Expected value came from betting games and help to evaluate uncertaintyprospects.
It requires that the worth of action be estimated as the sum of pay-offs (subjective worth a
person attaches to an outcome) weighted by their probabilities.
A preference is the expression of an evaluation that is supposed to take anything into a
count. It corresponds to the output of the process of weighing pros and cons. This kind of
model assumes that a lot of reasoning has been done. A desire is something more primitive, is
just a disposition to choose something, without any sort of evaluation. You can describe
preference of an individual using numbers: if we have a set of preference that are complete,
we can describe the options (x, y, z) as an ordering over ranking.
PARETO UTILITY’S SCALE (which is a scale of preferences): we need a scale in which
we can put the options from the most preferred to the least preferred. Then we can attach
some numbers to the option and the only constraint is that the higher numbers are associated
to the most preferred option. The order of preferences can be expressed with set of numbers.
Which numbers we use is arbitrary. Ordinal scale is a scale that gives information about the
order. We can use this scale because it gives information about the order (from best to worst).
But it does not give us any information about the difference between the preferences in these
options (we do not know how much more he prefers an option than the others). In order to
know this, we need a cardinal scale, that gives an idea about the distance between the options.
With the ordinal scale we are measuring preferences of individuals, i.e., we are measuring
utility. Utility corresponds to the number each consumer associate with a particular option (so
utility corresponds to preference). So we have that:
x≻y ⇔ u(x) > u(y) (from ordinal scale to cardinal scale) where “u” is the function of
utility
Rationality à Utility
The probably most important Italian economist of all time, Pareto (1848-1923) realized
that we can build a theory of utility adopting a theory of rationality. This intuition allowed
economy to bridge an important gap with utilitarianism, a philosophical position that was
used in social science.
All of this has a fundamental problem, it did not know how to measure utility?
In order to do consumers theory, utility is assumed to be consider on an ordinal scale (but
other parts of choice theory require a cardinal scale of utility).
à This theory is a theory of choice under certainty. (It means that people have options,
but they know what the outcome of their decision is going to be)
But there are other kinds of decision where we do not know what the outcome is going to
be. Therefore, this theory has big limitations: it does not consider what happens when we
introduce uncertainty in the picture?
And yet, for a long time, social scientists were not able to extend the theory of rational
choice from the domain of certainty to domain of uncertainty, so a huge part of behavior
could not be explained with this model.
The justification of this attitude was that risky decisions are not driven by rationality.
Uncertainty = Irrational?
There was the idea that risky behavior (behavior under uncertainty) was like gambling, so
it was considered irrational, something that could not be explained using a theory of rational
choice. Why cannot this model be applied in situations with uncertainty? Because to
understand behavior in these situations we have to consider two things: preferences and risk.
Therefore, in order to explain risky behavior, we have to consider people’s beliefs, so we
need a theory that combine preferences and beliefs. From the XVII century intellectuals
started to think about risk and most of them were interested in gambling, theology and
insurances.
E. g. The lottery. You buy a ticket spending few moneys. The lottery has a very large
prize. Typically, the organization that sells the tickets is going to sell a lot of ticket. So, it
must sell tickets in such a quantity that it will recover the prize and have some margin on top
of it. Let’s suppose:
price of ticket = 1 euro
prize = 1 million euro
number of tickets sold = 1100000 euro
Is it rational to buy a ticket?
It might be irrational because the expected monetary value of the ticket of a lottery is
lower than the value of not buying the ticket.
Expected (monetary) value (EMV) of a lottery?
 We have to consider the possible outcomes of the lottery: you win the prize or you do not.
These outcomes are associated to a monetary value: if you do not win, the value is 0, if you
win, the value is 1 million. We have to weigh prize with probability and the probability of not
winning is much larger than the probability of winning.
EMV (ticket) = (0) (0,99999909) + (1000000) (0,00000091) = 0,91
EMV (ticket) = (the prize if you do not win) x (the probability of not winning) + (the prize
if you win) x (the probability of winning) = 0,91
One way to make a decision is to compare this EMV to the EMV of not participating the
lottery. This is much simpler because if you do not participate there is only one outcome: you
are sure that you are not going to win 1 million, but you are also sure that you are not going
to lose your 1 euro.
EMV (no ticket) = 1
Important intellectuals, such as Laplace, came up with the idea that when you are
gambling, or, in general, when you are dealing with a situation of uncertainty, you should
maximize the EMV of an action, so basically you should do a calculation of this kind. It
means that you must compare the EMV of buying the lottery ticket and the EMV of not
buying it, in order to figure out which one is larger (in this case, the second one).
But it might be rational because we should focus not on the monetary value but to the
importance that the amount of money has for each of us and that is a subjective thing
(subjectively, we may value 1 million euros worth the risk).
This consideration led to realize the idea that a rational idea should maximize the expected
monetary value as to simplistic (there are also paradoxes that can be generated), so we
should replaceit with the more realistic idea that a rational person should maximize
the expected utility (a subjective evaluation), so instead of focusing on money prices, we
should focus on the utility that people derive from winning or losing certain amount of
money, then we can apply a rule: you should maximize expected utility.
The concept of a scale of preferences uses the SUBJECTIVE PROBABILITY (De Finetti
e Ramsey), probabilities, cardinal scales, measure beliefs. The weakness is the measurement
process, how can we measure contents of other minds? But Ramsey suggested a solution. He
says that you have not to ask people direct questions of probability but how the decision-
makers would bet to the outcome discussed. There are some axioms that made the decision to
follow some patterns (in particular in Subjective Expected utility theory).
De Finetti's treatise on the theory of probability begins with the provocative statement
“probability does not exist”, meaning that probability does not exist in an objective sense.
Rather, probability exists only subjectively within the minds of individuals. De Finetti
defined subjective probabilities in terms of the rates at which individuals are willing to bet
money on events, even though, in principle, such betting rates could depend on state-
dependent marginal utility for money as well as on beliefs. Most later authors, from Savage
onward, have attempted to disentangle beliefs from values by introducing hypothetical bets
whose payoffs are abstract consequences that are assumed to have state-independent utility.
In this paper, I argue that de Finetti was right all along: probability, considered as a numerical
measure of pure belief uncontaminated by attitudes toward money, does not exist. Rather,
what exist are de Finetti's ‘previsions’, or betting rates for money, otherwise known in the
literature as ‘risk neutral probabilities. But the fact that previsions are not measures of pure
belief turns out not to be problematic for statistical inference, decision analysis, or economic
modeling.
Von Neumann-Morgenstern had another EUT utility as sensitivity to risk. On the other
way Savage made the axioms for the SEU which are 8.
The original approach developed by Von Neumann Morgensten and Savage provided an
alternative method based on:
Uncertainty surrounding decision-making which is captured by a probability distribution
over the states of the world (Savage), which determine the relation between actions and
outcomes. What we mean when we are talking about uncertaintyis that it is a situation in
which there are no good reasons for believing that one probability distribution is more likely
than the others.
2.3. to Expected Utility Theory (EUT)
èUNDER UNCERTANTY
It is the theory that nowadays is used in all social sciences to explain the behavior of
individuals when they are facing risk. The theory says that a rational decision maker
maximizes her expected utility, where EU is defined in this way:
EU(X)= Σ pi(xi) u(xi)
X, Y = risky prospects (or “lotteries”) = an option that has not a sure outcome
x, y … = outcomes (or consequences)
p = probabilities (or beliefs)
u = cardinal utilities (or preferences)
The expected utility of a lottery X is given by the sum of the utility that we can derive
from the various possible outcomes (x, y…) weighted with the probability of that outcome.
The rule that you use is: you must multiply probabilities and utilities and sum the result with
all the possible outcomes. In the case of the lottery:
EU (ticket) = u (0) (0,99999909) + u (1000000) (0,00000091)
The utility of 0 multiply with the probability (which is very high) and the utility of 1
million with the probability.
EU (no ticket) = u (1)
So, gambling may be rational, depending on one’s utility function (how we value losses
and gains), so it is subjective (someone might value 1 million a lot).

Coordination and convention

Lecture 04/10

Competition VS Coordination

Nash Equilibria: a situation in which there is no incentive for changing if no one change.
All cooperative games have at least one Nash equilibrium.
 Matching pennies is a game of pure competition, in this course, we are not very interested
in these games. Here there are no Nash equilibria, the rational strategy in these cases is to
play perfectly casual. With this strategy, the other player could not have any element to
discover the strategy that precisely for this reason is perfectly casual. The battle of sexes is a
coordination game but with asymmetric payoffs. Is a bargaining game, and the benefits of
cooperation are allocated in different ways. Hi-low is a game of coordination but with
symmetric payoffs. It has two Nash equilibria. One of the two is clearly better. In this game is
fundamental the idea of the other beliefs. Stag hunt is a coordination game with multiple
Nash equilibria, there is a safe choice. The evaluation of risk influences a lot the final result
of the game. Prisoner dilemma there is only one Nash equilibrium, but this is not the best
outcome of the game. If the agent acts rationally, according to the rationality definition that
implies maximization of utility, the game ends in an outcome that is not the best one. The
outcome CC is not an equilibrium22 What are important in defining different games are the
relationship between the payoffs not the payoffs number.
In the research of the Nesh equilibria is important to remember the Best Response
Analysis. This is a mechanic method to find Nash equilibria. If we cannot fine Nash
equilibria the game is not a coordination game, and it means that there is no NE in mixed
strategy.
Beliefs are a key point for equilibria selection. In particular the belief about what the
other is going to choose. This point opens the problem of how beliefs come to be, how
behaviours appeared. There are al least two main solutions: 1) searching the solution outside
the game, this seams to be the most realistic interpretation but it does not reach, since now,
satisfying results. Is we search outside the game we cannot fine an explanation and we
abandon partially our scientific and methodological rigid purposes. 2) there is the necessity of
finding a story, players must be in the same outside situation. In this second solution stays a
large part of the abstraction that models like game theoretical ones make.
The solution of this problem, the equilibrium selection for coordination game where we
cannot apply the BRA method stays in conventions. The two main scholars that have
developed this prospective are T. Shelling23 and D. Lewis24. The solution stays in the
expectation that we make of the possible decisions of the other players. The possibility of
communication in guaranteed by the fact that we spook the same language, the problem of
language is very similar to the problem of money. Money convention is guaranteed from the
possibility that the central power has to make the citizens pay taxes.

22

23
(1960) The Strategy of Conflict.
24
(1969) Convention: A Philosophical Analysis
 In practice the problem of equilibria selection is solved through the concepts of salient
and of focal point, elaborated by T. Shelling. Irrelevant factors make a particular strategy
salient (it “stands out”). There are not general and theoretical reasons to determine a focal
point. On the other hand, there are practical reasons, coordination works. The salient
equilibria are the focal point, points that attracts attention. In this way coordination is possible
out of the BRA. We have seen many examples of cognitive silence e.g., paragliders that have
a map and have to coordinate on where to meet without having coordinate their self before
the launch. These are situation where there is not a strong sens of rationality but people still
able to coordination.

Lecture 05/10

The fact that there is not well-defined general reason to understand why a focal point is a
focal point is our cognitive system and our perception becomes very important. There are
many possibilities for saliency e.g., a point could be culturally salient.
D. Lewis apply T. Shelling’s ideas to the problem of social conventions. History, general
and individual, becomes fundamental; precedents become an important case of saliency. For
Lewis:

A regularity R in the behaviour of members of a population P when they are agents in a recurrent situation S
is a convention if and only if, in any instance of S among members of P,
(1) everyone conforms to R;
(2) everyone expects everyone else to conform to R;
(3) everyone prefers to conform to R on condition that the others do, since S is a coordination problem and
uniform conformity to R is a proper coordination equilibrium in S.

The concept of silence stays in the regularity that becomes the coordination point.
There are also other solutions to the coordination problems, the most common solution is
language, that is a convention itself. The concept of salient has been elaborated properly for
understand coordination without communication. Despite its origin is possible to interpret the
concept of focal point also working in language. Language creates salient on a particular
equilibrium and therefore a focal point. As all others convention language works if there is
regularity so Lewis’s theory helps us also in understanding the growth of all those institutions
that control language; those are in ultimate analysis tools of coordination. It is important to
remark that language (in a very general conception sights) do not change payoffs of the game,
it simply allows us to be more efficient in finding salient equlibria and so focal points.
In beliefs for Lewis there is not consistency, so we can search for it at a superior level:
different order of think e.g., I think that you think that I think that you are going to choose D.
There are potentially infinite orders of thinking, but humans usually do not overcome the
third. The solution for finding consistency in the infinite possibility of think orders is
common knowledge e.g., there is a computer on the desk. There are situation where there is
not common knowledge and so orders of thinking could be useful for to reach focal points.
This situation where there is not common knowledge are expressed with conditionals e.g.,
1order: I think I will choose D if I think that you will choose C 2 order: I think I will
choose D if I think that you will choose C if I think that you think that will choose D… We
use focal point to eliminate this conditional, language is a way in which is possible to
eliminate conditionals. For reaching coordination 1 order expectation are enough:

How can we explain coordination by salience? The subjects might all tend to pick the salient as a last resort,
when they have no stronger ground for choice. Or they might expect each other to have that tendency, and
act accordingly; or they might expect each other to expect each other to have that tendency and act
accordingly, and act accordingly; and so on. Or—more likely—there might be a mixture of these. Their first-
and higher-order expectations of a tendency to pick the salient as a last resort would be a system of
concordant expectations capable of producing coordination at the salient equilibrium.

In this chain of cross-references is sufficient that only one of the thinkers thinks that the other
is irrational to make coordination impossible:

We may achieve coordination by acting on our concordant expectations about each other’s actions. And we
may acquire those expectations, or correct or corroborate whatever expectations we already have, by putting
ourselves in the other fellow’s shoes, to the best of our ability. If I know what you believe about the matters
of fact that determine the likely effects of your alternative actions, and if I know your preferences among
possible outcomes and I know that you possess a modicum of practical rationality, then I can replicate your
practical reasoning to figure out what you will probably do, so that I can act appropriately.

Social conventions are the simplest form of social institution.

Guala, F. (2016) Understanding Institutions, Princeton University Press, Ch. 2

(“Games”)

The equilibria approach in studying institution is based on game theory. A game in game
theory is a situation of interactive decision making. Game theory is useful in modelling social
interaction because it allows us to simplify reality. The payoff matrix allows us to reason on
many important things, outcomes, actions, incentives, in a simple way. We use matrix for
representing game with imperfect information, uncertainty. These games are called
simultaneous games. Payoffs allow us to represent preference rankings and therefore utility.
It is also possible to interpret payoffs as real feature of the world, real outcomes of actions.

Nash equilibrium (def.): a profile of strategies (or outcome) such that no player has an incentive to change
her strategy (or action) unilaterally. Each player has no reason or motive to change her strategy provided that
the other players do not change their strategies either. It is a stable state.

Using Nash equilibrium, we can represent the situation where institution semses to be
needed. Situation where institution allow the individuals to reach situation that lonely would
be unreachable. We are going to analyse only equilibria un pure strategies, strategies that do
not imply randomized behaviours. When games have multiple equilibria rise the problem of
equilibrium selection, and institution helps us in this process.
Many games are useful in the study of institution hi-low, stag-hunt…but the most
discussed in the literature is the prisoner dilemma. The prisoner dilemma in not proper a
coordination game because there is only one Nash equilibria, is more appropriate to call in a
cooperation problem not a coordination problem as the other games we have seen (cfr
lecture). The problem arises because many people think that C is the right choice despite D is
the rational one in a game theoretical sense.
The real-life rules, has been demonstrated through many experiments, is conditional
cooperation. This work well when the game is indefinity repeated. In the indefinitely repeated
version of prisoner dilemma conditional cooperation is an equilibrium. Repeated encounters
facilitate cooperation but rise the problem of coordination.
On Shelling view game theory is better at mapping the territory that at explaining
behaviours. This observation resends us to the problem of equilibrium selection. An
equilibrium selection theory is useful because identification of equilibria helps the
formulation of functional explanation of behaviours. Functional explanation are very
common in social sciences. The equilibrium model will support a functional explanation
would support a functional explanation in the form: «the rules exist because they help people
solve a coordination problem». The rules persist because they are equilibria of strategic
games.

Lewis, D. (1969) Coordination and convention and common knowledge, in

Convention, Blackwell, selected paragraphs from Ch. 1 and 2.

In coordination problem each chose depends on the action of the other player. Some
combination of actions are equilibria i.e., a situation in which no agent has the incentive to
change his mind unilaterally.
In the solution of a coordination problem each agent must do his part choosing one of the
possible equilibria because he expects the others agent to do their part so to choose the
equilibria too. The incentive to choose the equilibria stays in the expectation, in which he is
sufficient confident, that the others are going to do the same. The strong of reciprocal
concordant expectation is the measure of the possibility to reach coordination. So, if the aim
is reaching coordination: «I need to figure out what you expect me to do». This is precisely
the mechanism which construct higher order expectation. In make my expectation I attempt
to think what you will think and so on… The possibility for higher order expectation an
infinite. Coordination may be reached by having concordant mutual expectation on the other
behaviours.
The possibility to solve a coordination problem is by agreement. This type of coordination
is not an alternative to mutual expectation coordination. Agreement is simply a way for
construct concordant mutual expectations. It is also possible that agreement produce a
changing in the payoffs of the game e.g., a promise, but in this case the game would be no
longer a coordination game. So, it is still valid the previous claim that agreement is a way of
construct concordant mutual expectation. Explicit agreement is a particularly good way for
coordinate.
Is possible to show how the concept of salient elaborated by Shelling could be explain in
terms of concordant mutual expectation. Shelling was interested in situation in which agent
cannot communicate. The focal point, taken as the last resource for coordination, is the means
for creating concordant expectation capable of making reach an equilibrium.
Another typical form of coordination is the one with familiar problems where past soled
problem makes some equilibria particularly salient. This way crating the concordant
expectation.
Convention stays in the force of the precedent. If we have already reached an equilibrium
point, coordination, do not matter how we have reach it, we will use it in the future games.
Very interesting is that we could have reach coordination also by luck in the previous cases.
Precedents are the source of an important kind of salience: the uniqueness of the equilibrium
we have reach last time. So, if we have two similar and successive problems in the second
one, we have no reason to not follow salience of the precedent. The social importance of this
mechanism stays in the fact that mor or less all of us notice the same analogies within the
games and so our focal points are coincident and so we could make concordant mutual
expectations and so in the end reach coordination.
The mechanism of salience of the precedent work with only one iteration but nothing
prevents us to have many iterations. Several precedents are always better than one, more are
the precedent stronger is the connection. This way is how a regularity is constructed. If the
problems are analogous to the precedents every one want to conform to the norm if everyone
do. So, we construct a conditional preference of conformity. Every new action that conforms
to the regularity make stronger our experience of conformity and our expectation of future
conforming are the reasons that leads us to conform also in the present. Expectation of
conforming action produced conforming actions and conforming action produce expectation
of conforming actions.

A regularity R in the behaviour of members of a population P when they are agents in a recurrent situation S
is a convention if and only if, in any instance of S among members of P,
(1) everyone conforms to R;
(2) everyone expects everyone else to conform to R;
(3) everyone prefers to conform to Ron condition that the others do, since S is a coordination problem and
uniform conformity to R is a proper coordination equilibrium in S.

Handout on Solving Games

Sequential games are games with perfect information. We solve them using backward
induction. The solution is called sub optimal equilibrium.

Simultaneous games are game with imperfect information. We solve this games in two
ways: iterated elimination of dominant strategies or best response analysis. If is
impossible to find equilibria using these methods, it means that there exist only equilibria in
mixed strategy.
Schelling, T. C. (1957) “Bargaining, Communication, and Limited War”, Conflict
Resolution 1, 19-36.

The text states with many Examples of situation where mostly people practically reach
coordination (list of number both have to select one number). The common aim of
coordination is an important reason that leads to the final positive result. The core is finding a
key of coordination, a focal point. Practical result leads at result very different from those at
which would leave random probability. This type of coordination works when the interests
are common.
When the are divergent, or only partially divergent interest, things change. After many
examples of game of this type (hands and tails with asymmetric payoffs) is shown as
participant reach coordination in a proportional number of cases, many more than on a
random hypothesis. Tax problem shows the tendency for an equal split.
Both bargaining situations analysed in the previous part, with and without common
interest (symmetrical or asymmetrical payoffs) of the paper are tacit bargaining, now the
author is interested in explicit bargaining. The previous concept of coordination is not
directly applicable to explicit bargaining. In this case is noted a particular attention at status
quo antes and at natural bounds. Important becomes the formulation of the problem, the
obvious outcomes depend on framing the problem. There is a remarkable difference in
coordination in tacit and in coordination in explicit bargaining, explicit one seams to require
coordination of participants expectation. In the end is argued that explicit and tacit bargaining
are different but that the core of coordination is the identification of focal points (exactly
what argues Lewis).
The general aim of the paper is to show how is possible to reach coordination with tacit
bargaining, but the fine outcome is not necessarily different from the outcome available in
situation of explicit bargaining.
The author attempts also to transport the reflections made about tacit bargaining in tacit
bargaining in cold war. Keeping communications channel open seems to be the obvious
solution. Shelling argues that also unilateral negotiation could provide the necessary
coordination for saving both parties.
Final considerations:

Our analysis of tacit bargaining contains two suggestions on this problem. One is to adopt criteria that are as
qualitative and discrete as possible, avoiding reliance on matters of degree or judgment. The other is simply
that unilateral suggestions, even if not tested for “acceptability” to the Russians in any sense analogous to
explicit agreement, may serve the desperate purpose when there is no alternative but mutual destruction. In
fact, the analogous feature of explicit negotiation suggests that negotiated agreement might be more quickly
achieved if there were some prior suggestion around which mutual expectations could crystallize when quick
agreement was required.

Dilemmas of cooperation

Lecture 11/10
 We have already briefly spoken about those games like the coordination games, but there
is not an individual interest for coordination, there is a social interest for cooperation
(interesting possible observation on the difference between particular interest and social
interest; social interest is not the sum of individual interest). These problems are called
problems of cooperation. The prisoner dilemma game (PD) is the most important game
where the individual interest is not in line with the collective interest. To decide if the game is
a prisoner dilemma game, we have to look at the structure of the payoffs not at their absolute
value.
In the analysis of the prisoner dilemma game, there are two important concepts that we are
going to use, which are elaborated by the Italian economist W. Pareto: 1) Pareto Efficiency:
an allocation (outcome) is Pareto-efficient (optimal) iff it is not Pareto-inferior to any other
possible allocation. 2) Pareto-inferior: an allocation Y is P-inferior to X iff at least one
individual prefers X to Y, and no one prefers Y to X. The idea is that no one is damaged if
the allocation is changed. The switch from an allocation is useful for avoiding waste. There is
no necessity for unanimity, there could be also indifference. In the prisoner dilemma, there is
only one allocation that is Pareto inefficient DD and there are three Pareto efficient
allocations. CD and DC are not comparable in the Pareto efficiency frame because someone
loses something. The allocation that is not Pareto efficient is the only Nash equilibrium DD.
The notion of Pareto efficiency is not a strategic notion, it is possible to give it from outside
of the game.
The paradox of the PD is that if all the agents act rationally, we end up in an allocation
that is not Pareto efficient. The example of the PD could be used as a counterexample of the
invisible hand that governs the market in Adam Smith’s theory. In real life, PD helps us in
the comprehension of situations like public goods problems. When something could be used
by the community, it is not the property of anyone.
In general, is important to remember that we are considering only one-shoot PD. There are
two lines of analysis of the PD problems: 1) attempt to solve PD, to find a solution and 2)
discussion of the PD. The first solution that comes to mind is that a rational individual would
not behave this way, but this line of reasoning put into question the notion of rationality in
game theory and in RCT. The attempt that pretends to solve the prison dilemma usually fails
in the sense that many attempts solve the problem but change it to another problem. So, a
solution in a strict sense seems to be impossible. All solutions are destructions.
Example of solution of the PD: the symmetry argument. The argument sound like this:

Both players are facing the same problem

2. They are both rational
3. Therefore, they will choose the same action
4. They will not choose CD or DC
5. The only possible solutions are CC and DD
6. But both players prefer CC to DD
7. Therefore, both players will choose C

The problem with this argument is that 5 implies that if P1 chooses C, then P2 cannot choose
D. The argument violates causal independence: it assumes that P1 and P2 are not autonomous
agents. The argument forgot about the independency of the players.
 Example of (di)-solution of the PD: Policing. The approach is that is necessary to put an
addiction cost on the defection strategy to make it unattractive. The pro is that it works. The
cons are that it simply changes the payoffs matrix so that PD is no longer a PD. It requires a
central authority (primitive society used to cooperate without a central authority so maybe
this is a reason for thinking that there could be other solutions). It also adds a cost, and we
enter the problem of police as an enforcer of rules of cooperation: we can enforce only a
minority of people if freerides are too many is impossible to make something.
The idea of the construction of society and the importance of the dispositive of fear and
obligation is well represented by this quotation from D. Hume:

No man would have any reason to fear the fury of a tyrant if he had no authority over any but fear; since, as
a single man, his body force can reach but a small way, and all the further power he possesses must be
founded either on our opinion or on the presumed opinion of others.25

Lecture 12/10

The di-solution of the PD usually constructs a coordination game with one equilibrium
Pareto superior to the others: changes cooperation games in coordination games. But in the
present lecture we have considered only one-shoot PD not reputed Pd. What about the
possibility of repetition? Just intuitively thinking there could be many interesting elements.
There is a remarkable link with the Lewis concept of the convention as a regularity, only
repetition could create regularities. Is very different to cooperate with people that we already
know. This difference stays in the possibility that repeated iteration of the game gives us the
possibility to construct an opinion on others and to construct ourselves a reputation. A good
reputation is an element that highly increases cooperation. All of us try to have a good
reputation exactly, of for this reason, we like to cooperate with who has a good reputation of
co-operator. This description clearly resends us to the concept of conditional cooperation. We
would like to cooperate with who has a good reputation because this means that we have
more probability to encounter a co-operator and so we can more safely paly cooperation.
The first try we make is to play the game finitely repeated times, so knowing perfectly
how many times the game would be played. We solve the game using the concept of Nash
equilibrium and the method of backward induction and the conclusion is that the dominate
strategy would always be D. the fact that the game is played definitely repeated times is what
gives us the possibility to apply backward induction. In general, the iteration of a game by
backward induction tries us to the same conclusion. So, this first conclusion leaves us with
two options: 1) people do not behave this way or 2) is unrealistic to think that we will know
exactly how many times the game would be played.
So, we try to apply the second option: indefinitely repeated times. Without a definite
number of repetitions, we could not apply backward induction so we could just imagine that
the result is going to be very different. The fact that the iterations are indefinite complicates
the possibility of representation. We will try to classify strategy by looking at the average
payoffs. The same average payoffs could rise from a different strategy. Every pair of strategy
25
Essey on government, 1742.
generate a pair of average payoffs e.g., if I always defect and you always cooperate, the
average payoff profile is (5,1)26 If we both cooperate on even days and defect on odd days,
the average payoff is (2.5, 2.5) …
The surface of polygon CD, DD, DC, and CC
contains all average payoffs. We call this surface F.
Each point in F is the result of a ‘plan of
action’(rule?) that tells each player what to do at each
round. There is in accordance with the plan of action
among the players. I stick to the plan if also you
follow the plan otherwise, I will defect. This is called
the Trigger strategy: If you do your part of the plan
of action, I will do my part; otherwise, I will defect.
Trigger strategies are useful to support a plan of
action but only some plans could be supported by a
trigger strategy. Folk Theorem: Every plan of action
with average payoffs inside N can be supported by a
trigger strategy. The strategies in N are all Nash
Equilibria of the indefinitely repeated game. The
folk theorem transforms the PD into a coordination
game with multiple equilibria. The part of F that
does non belongs also to N and contain all strategies
that are not Pareto efficient. There are infinite Nash Equilibria, not all desirable, in some
someone earns more than others, but in any case, all Nash Equilibria are in the interest of
both players. The fact there are infinite Nash equilibria makes us fall another time into the
problem of the coordination games with multiple equilibria, the problem of equilibrium
selection.
There are some conditions that the Folk theorem must satisfy to work: 1) information: in
fact, coordination works better in small societies where information is better (cfr. gossip
phenomena studied by anthropologists). 2) Mistakes: if the information is not perfect is easier
to make mistakes. 3) probability of continuation (horizon): if the probability of non-encounter
the other is high also the incentive ad defection is higher. This type of game works well when
there is a high probability that there would be longs. When the game is ending players have
less to loos of the fruits of cooperation. 4) temporal discounting (patience): the rate of
discount of the future is determinant in the incentives to cooperation. If we have a high-grade
of discount, future rewards are valued less, and the incentive to cooperate is lower. In practice
is important to remember that humans are the animal with the lower grade of discount about
the future.
So, the general conclusion that we could surely make is that the defined and finite
repetitions of the game take us to a very different conclusion from the indefinite number of
repetitions.

26
For this example we have used a matrix with DD 33, CC 22, CD 15 and DC 51.
Peterson, M. (ed. 2015) The Prisoner's Dilemma. Cambridge University Press
(Introduction).

cooperat defect
e
cooperat B;B D;A
e
defec A;D C;C
t

The core structure of a PD is A>B>C>D for technical reasons we also assume that
B>(A+D)/25. In its classic form, the PD is a two-player, non-cooperative, symmetric,
simultaneous (without information), and with only one Nash equilibrium. It is a non-
cooperative game in the sense that there is no incentive for the player to cooperate. There
could be also a sequential version of the PD (cfr. figure 0.4).
Fundamental in the study of the structure of the game are its repeated versions. If the game
is played a finite number of times the dominant strategy is still defection (cfr. backward
induction). The practical experiments have reached different results. People cooperate more
that the theoretical result that would be nothing.
The most interesting version of the PD is the indefinitely repeated PD. The key point is
no longer how many times the game is played but that the player does not know this. Now
players have reasons to take the future behaviour of other players into account. The
possibility of punishment in the future incentivizes cooperation in the present. Now there are
incentives for cooperation.
The most famous strategy for indefinitely repeated PD is called tit-for-tat. The strategy is
always to cooperate in the first round and then adjust the decision on what the other player
does.

Handout on Repeated Games

The “grim” trigger strategy says that a single mistake is sufficient to trigger defection
forever. There are more forgiving trigger strategies e.g., tit-for-tat. The “grim” strategy is
simpler to analyse.
Two trigger strategy is a Nash equilibrium if the average payoffs are inside N (cfr. figure
in the lecture). In the two average payoffs out of N at least one of the two is smaller than
mutual defection so at least one player has the incentive by deviating from the plan.
Mutual cooperation is a Nash equilibrium of the infinitely repeated version of PD. This
can explain the real behaviour of the people.
The number of equilibria is infinite so now we have the problem of justifying why some
equilibria are played more times than others. This new question generates another time a
problem similar to the problem of equilibrium selection. On the other hand, this problem can
account for the enormous diversity of societies. If the PD is a game that really tell us
something about human nature, so this discovery justifies the differences across societies.
 In this analysis of the indefinitely played PD, we have assumed that players are indifferent
about time: it does not exist discounting future tendency. This is obviously false in the real
world. This means that in the real-world people prefer satisfaction now than after, we can
wait only if the reward increase with the distance in the time of receiving. The main reasons
for this phenomenon are: 1) the risk of termination, 2) the psychological propensity (our
psychology is used to an unpredictable environment), and 3) investments (the possibility to
make the reward bigger, this only in the case of money and in the case of capitalist societies).
To account for this phenomena economist have introduced the temporal discount factor that
represent the present value of a unit earned tomorrow. The value of this factor is always
between 0 (no matter for a unit gain in the future) and 1 (the unity in the future has exactly
the same values as if I’m earning it now).

Kuhn, S. (2019) “Prisoner’s Dilemma”, Stanford Encyclopaedia of Philosophy

Speaking generally, one might say that a PD is a game in which a “cooperative” outcome
obtainable only when every player violates rational self-interest is unanimously preferred to
the “selfish” outcome obtained when every player adheres to rational self-interest.

How people cooperate

Lecture 18/10

Is important to remember that the theory that we are constructing is a theory, a model. So,
has all the problems that models and theories have. But it reflects really how people really
behave? In these two lectures we are going to see empirical studies on cooperation. The
experiments are usually theoretical games of game theory played by real agents and real
payoffs (usually monetary payoffs but in societies that do not use money other payoffs).
The first experiment is: Andreoni, J., & Miller, J. H. (1993). Rational cooperation in the
finitely repeated prisoner’s dilemma: Experimental evidence. Economic Journal, 103(418),
570-85. Here we can see the decline of cooperation in both conditions, partners and strangers.
Normann, H. T., & Wallace, B. (2012). The impact of the termination rule on cooperation
in a prisoner’s dilemma experiment. International Journal of Game Theory, 41(3), 707-718.
In the second experiment, we can see the difference between finitely repeated games and
indefinitely repeated games. In the unknown condition, there are much more cooperative
choices.
 In many experiments PD is replaced by its complicated version the Public Goods game: a
PD where there are more than two subjects and in which the subjects are strangers. A
resource must be allocated in the private account or in the public account. What goes in the
public account is multiplied and then shears among the participants irrespectively to their
contribution. The rational allocation for self-interest agents is to allocate all of the resources
in the private account (the phenomenon of free-riding).
Isaac, Walker & Thomas (1984) “Divergent Evidence on Free Riding: An Experimental
Examination of Possible Explanations”, Public Choice43: 113-149. There are two anomalous
things that we must account for: over-contribution (in theory there would be no contributions)
and the decade of over-contribution (why contribution decade and are not). These two
anomalous things make rise the question: do people cares only about monetary payoffs? The
explanation is people do not really understand the logic of the game, but they learn a bit each
new round. The first part explains the over contribution and the second one the decrease of
over contribution.

This experiment shows the restarting effect. Every time that the experiment restarts
contributions in the first games is higher than in the last’s ones. This put in doubt the
explanations made before: what are people learning? Maeby the problem says in the self-
interest assumption. So, another possible explanation could be altruism.

It’s easy to build other-regarding utility functions, for example:

Ui= α(πi) + β(πj)

i,j = payoff of player i, j
α, β = parameters weighing self-interest vs. altruism

But also with this new function: why altruism should decay? Maybe the solution is
conditional altruism: reciprocity27. The concept of reciprocity has been a fundamental
concept history of ethics systems. This attempt is an attempt to formalize strong reciprocity.

Ui = α πi+ ρ [κ(.) σ(.)]

ρ= reciprocity parameter
κ= kindness parameter (+ if actions is kind, -if it’s nasty)
σ= reaction parameter (+ if kind, -if nasty)

27
Cf. Rabin, M. (1993) “Incorporating Fairness into Game Theory and Economics,” American Economic
Review83: 1281-302.)
 Reciprocity could be negative or positive so, it could increase or decrees the value of
utility. An important distinction in the concept of reciprocity is the one between strong and
weak reciprocity: 1) Strong reciprocity is when the players care about kindness, fairness,
etc. The reciprocity is ‘written’ in their preferences, they are willing to reciprocate at a cost.
2) Weak reciprocity is when the players want to maximize their own payoffs. Preferences
are selfish. They reciprocate only if it’s not costly e.g., in indefinitely repeated games (Folk
Theorem).
Summarising the structure of the previous argument there could be 3 possible solutions to
the anomalous results of the Public Goods game (PG): 1) error and learning. 2) Altruism. 3)
reciprocity.
Interesting in showing the difference between strong and weak reciprocity is the sequential
version of PD. If both agents are self-interested the result is DD. If there is strong reciprocity
(Both players are strong reciprocators) the result is CC, because when P 2 has to make his
choice, he would not choose D because is a strong reciprocator, and P 1 will not choose D
because he is a strong reciprocator this works in any case. But if P 2 is a weak reciprocator he
will choose D. The argument could run also saying that if are both strong reciprocators they
can communicate and that if they are weak reciprocators they would not communicate and so
P1 has to make conjectures about P2.

Lecture 19/10

The previous analysis on altruistic behaviour let us with a possibility open: to impose a
cost for antisocial behaviour. Within this line of thought we can see the model of altruistic
punishments is social dilemmas. This model has two main features: 1) is a normal PD. 2)
Each player can take away from other payoffs an amount x at a cost c<x (the important thing
to remark is that this decision is voluntary, not like where there is the implementation of a
police system where the cost is an obligation). This mechanism useful to punish freeriders.
But a perfectly self-interest agent will not pay any cost, he will only enjoy the benefits of the
other paying the cost.28
The following graph shows a PD with or without punishments is known and unknown
situation. Had also been seen how the difference between the known and unknown situation
is influenced by cultural differences. The same results are observed in the two situations.
28
Cfr. Boyd& Richerson (1992)“Punishment Allows the Evolution of Cooperation (Or Anything Else) in
Sizable Groups”, Ethology and Sociobiology; Fehrand Gachter (2000)“Cooperation and Punishment in Public
Goods Experiments”, American Economic Review; Gintis, H. (2000)“Strong Reciprocity and Human Sociality”,
Journal of Theoretical Biology
Both results are in line with the theoretical predictions of the model. Punishments increase
cooperation.

Always in line with the reflection on the concept of altruism and reciprocity are the
experiments with the ultimatum game. The structure of the game is very simple. The
problem is the same of dividing a cake with the difference that without an agreement there is
no transaction. P1 make an offer and P2 can accept the offer or refuse the offer. If the offer is
refused there is not any transaction. For a rational agent if P 1 offer 0 he would be indifferent,
if P1 offers more than 0 he would accept because something is better than nothing.

Camerer, C. F., & Fehr, E. (2004) “Measuring social norms and preferences using
experimental games”, in Henrichet al. (eds.) Foundations of Human Sociality, Oxford
University Press. In real experiments the previsions of the model are not respected. Almost
all offers are of 50%, only some under 50%, and almost nothing over 50%. Most of the offers
under 40% are refused. It seems that the dimensions of the cacks counts (many experiments
in different countries with different culture and different value of money).

J Henrich et al. (2006) Science 312: 1767-1770. Experiments around the world shows
important differences between cultures. The black circle represents the probability of receive
a refuse as an answer. Some paradoxical result could be explained with the theory of
competitive presents elaborated by anthropologists (cfr. Marcell Mauss). What happens is
intrinsically related at social norms.

Sanfey et al (2003) “The Neural Basis of Economic Decision-Making in the Ultimatum

Game. Science, 300, 1755-1758. Interesting studies had been made about the cerebral
reaction in the ultimatum game. The main areas that are activated are Insula (automatic
emotions, disgust, pain), Dorso-lateral Prefrontal Cortex, (rational cognition, attention), and
Anterior cingulate cortex: (conflict resolution). The relation between these three activations
can give us an idea of which the response will be.
Example of application of game theory in the solution of conflicts:
 In the last part of the lecture discussion about emotions as commitment devices (cfr.
registration)

Camerer, C. F., & Fehr, E. (2004) “Measuring social norms and preferences using
experimental games: A guide for social scientists”, in J. Henrich et al (eds.)
Foundations of Human Sociality, Oxford University Press.

Aim of the paper is to describe useful experiments for measuring social norms and social
preferences. The working assumptions are that people are self-interested but sometimes they
can behave as having altruistic (in a strong sense) preferences of preferences for equality and
reciprocity. These preferences have an important role in the enforcement of social norms.
 The main theoretical uses of game theory are 1) taxonomy of social world. 2) construct
preferences. In experimental economics there are many convections, one of these is that is a
taboo to lying about the experimental conditions.

The matrix above describes the characteristics of the main games used for measuring
social preferences.
PD and PG: model situations like pollution of the environment. In PD self-interest
subjects have an incentive to defect and in PG when for each unit of contribution returns m<1
the self-interest hypothesis predicts 0 contribution. A mechanism to prevent the decay of
cooperation is social ostracism. Experimental evidence shows that the introduction of
punishment works for increase cooperation. So, also reject the prediction for social ostracism.
If agents are perfectly self-interested, they would not punish since it is costly. But this do not
happen. The PG with the opportunity of punishment can be viewed as a paradigmatic
example of the enforcement of social norms. Another mechanism that works vey well in
increasing cooperation is communication. These two games are useful for capture general
social behaviours but cannot distinguish between purely self-interested agents and other
agents. The next game that we are going to analyse solve exactly this problem.
Ultimatum games are a set of games in the form take-it-or-leave-it. It measures the
disposition of P2 in punish P1, losing himself monetary something, for having been unfair.
The second player sacrifice his own money to punish the first player, the proposer. It measure
the preference for reciprocity. We do not know if the proposer offer average of 45% because
of his fear of rejection of because his altruistic payoffs. There is an interesting variation of
ultimatum game that makes it competitive with multiple player.
Dictator game has been thought exactly to respond the question that ultimatum game
cannot respond. The general idea is to measure pure altruism. It consists in a one player game
where the player has to divide a cake with another player and the other player if receive
something must accept. Is like a unilateral ultimatum game. The comparison between the two
games seams to show that part of the big contribution of the ultimatum game are linked to the
fear of rejection. The offers are lower when the others could not reject. The problem of the
dictator game is that is a “weak situation” because little changes in experimental design
change significantly results (has been argued that this could be an advantage in the use of this
game).
Trust and gift exchange games is an interesting companion game is the “trust game” (cfr.
Berg, Dickhaut and McCabe 1995). In a trust game an Investor and Trustee each receive an
amount of money S from the experimenter. The Investor can invest all or part of her money
by sending any amount y, between zero and S, to the Trustee. The experimenter then triples
the amount sent, so that the Trustee has 3y (in addition to her initial allocation of S which is
hers to keep). The Trustee is then free to return anything between zero and 3y to the Investor.
Self-interest trustees will keep all the money and self-interested investor will give 0 to the
trustees. In practical experiment invest about half and trustees reply with les then y. But the
results are very variable. Interesting is that what trustee resend grow with the growth of the
amount that is send to them. This phenomena has been interpretate as positive reciprocity or
feeling of obligation in replay more to an investor that trust the trustee.
The gift exchange game is a variation of the trust game thought to comprehend the role of
reciprocity in the enforcement of contracts. Subjects are in the role of employers or buyers
and of workers or sellers, respectively. An employer can offer a wage contract that stipulates
a binding wage w and a desired effort level ê. If the worker accepted this offer, the worker is
free to choose the actual effort level e between a minimum and a maximum level. The
employer always has to pay the offered wage irrespective of the actual effort level. The level
of effort of the worker could be higher or lower than the level of effort required by the
employer. Both of the player could choose. Also, the employer could choose the level
required and the gain of the worker do not change Perfectly self-interested workers in theory
would not go over the minimum level of effort. In practical experiments we see that there is a
phenomenon of positive reciprocity: workers level is similar to the lever required as this lever
becomes lower. Respond effort to generosity.
The third-party punishment games, the game between player A and player B is just a
dictator game. Player A receives an endowment of 100 tokens of which he can transfer any
amount to player B, the Recipient. Player B has no endowment and no choice to make. Player
C has an endowment of 50 tokens and observes the transfer of player A. After this player C
can assign punishment points to player A. For each punishment point assigned to player A
player C has costs of 1 token and player A has costs of 3 token. Since punishment is costly a
self-interested player C will never punish. However, if there is a sharing norm player C may
well punish player A if A gives too little.
The theories that try to explain the result described above are called Theory of social
preferences, the main of these is that agents have social preferences; take in account other
payoffs and preferences.
Ultimatum game with reputation, high reputation in refuse low offers increase the
probability to receive less lower offer. So, the majority of responder increase the thresholds
of acceptance.
The difficulty for theory of social preferences have to explain a lot of different results with
one model. There has been proposed two main models: 1) Model of inequality-aversion 2)
Model of reciprocity.
The iniquity-aversion model uses two constant that modify payoffs of the game: one
measure how much a player dislike disadvantageous inequality and another that measure how
much the player dislike advantageous inequality. The particularity of this model is that
population is considered to be heterogeneous. Inequality-adverse players are conditional co-
operators.
Reciprocity theories includes others player actions. Every player forms a judgment about
whether another player has sacrificed to benefits him and reciprocate kindness with kindness
and negative behaviour with negative behaviours.
There are also hybrid models that combines the two precedent illustrated approaches.
There are also experiments to test each theory and hybrid theories. This experiment since
now have the clear result that there is clear evidences for the model of reciprocity. People
care about other’s intentions and actions; they do not care only about their self-material
payoffs.
There are two main advantages in the use of games: replicability and comparability. The
possibility to collect data are big for each game, as big as the possible design of the game.
There are marked differences with the anthropologist method despite these games could be
very useful tools for empirical anthropology.
In the conclusion there is a fundamental evidence, players systematically deviate from the
standard model of self-interest. This deviation had always been interpretate as the evidence
of existence of social norms and social preferences.

Frank, R. H. (1988) Passions within Reason, Norton (Ch.3: “A Theory of Moral

Sentiments”)

The discussion starts with Adam Smith’s theory of moral sentiments: good moral
sentiments (overall sympathy) are a natural disposition of humans. In modern terms, we
interpret Smith’s theory with natural selection. Modern natural selection explains this with
the selfish gene29. Frank thinks that the evolutionary model is useful for the analysis of
29
Cfr. Dawkins.
human behaviour. The aim of the chapter is to sketch an alternative evolutionary model for
the evolution of non-self-interest behaviours such as altruism. The author is perfectly
conscious of the risks that he is dealing with. He does not pretend to reduce the appearance of
altruistic behaviours only to evolutionary factors but simply claims that also without other
factors evolutionary patterns seem to be sufficient. Moral sentiments evolved because they
help us in solving important problems of social interaction: the commitment problems.
In general commitment, problems are those problems that require that individuals make a
commitment for their interest that commits them to behave in a way that seems contrary to
self-interest e.g., Cheating, deterrence, bargaining, and marriage. These problems have two
features: 1) everyone benefits from the binding commitment not to cheat. 2) the proposed
solutions to the problem of cheating, and control, are vague or impractical.
Some of the contracts proposed to solve commitment problems change the material
payoffs of the player. But is not always easy to alter material payoffs. Fortunately, material
payoffs are not the only thing that governs our behaviour. Material payoffs do not play a
direct role in motivation. Behaviours are guided by a complex psychological reward
mechanism. This mechanism overcome also the intention motivated by rational assessment.
But this claim does not mean that these intentions are not fundamental to human victory in
biological competition. Rational calculations play an indirect role. They are inputs in the
reward mechanism. There are feeling triggered by rational calculation and original feelings.
These two types of feelings can compete and, in this competition, help people to solve
commitment problems.
Our task is to explain how moral sentiments (altruistic) might have evolved. For evolving
in the real world, they must have competed in natural selection, and necessary for this
competition is that they must have material payoffs.
The question that rise is: why if moral sentiments are so useful so many people in the
world are dishonest? Let us model the situation using PD and the average expected payoffs.
There are different assumptions that imply very different results: 1) When co-operators and
defectors are indistinguishable. In this situation, the expected payoffs depend on the
proportion of the two groups in the population. Suppose also that each individual reproduces
in proportion to his average payoffs (like in evolutionary game theory). In this situation,
independent to the starting proportions, the population will be all invaded by defectors. 2)
When co-operators are well distinguishable from defectors. In this case, are defectors that
are destinate to extinction, cooperators interact only with co-operators and so defectors could
have only mutual, disadvantageous, interactions.
This basic model could be useful because with few more elements have allowed us to
introduce the problem of scrutiny cost. The only addition is thinking that in an initial situation
of indistinguishability is possible to change at a situation of distinguishability with a cost of
scrutiny (the cost would be obviously only in the interest of cooperators). The decision to
pay the cost of scrutiny depends on the average payoffs that depend on the distribution of the
two groups in the population, on the relationship between the payoffs and the cost of scrutiny
itself. There is a defined proportion of the two groups in which is in the cooperators interest
to pay the cost of scrutiny over this proportion is no longer convenient and so defectors starts
to grow, but this grow refine the proportions and when the proportion return the one in which
co-operator have interest to pay the scrutiny cost the mechanism restart. When the average
payoffs of the two groups are equal, we can say that there is an ecological niche for the two
groups.
So fundamental for altruistic behaviour to help to solve the commitment problem is that
others are able to recognize that you will have this behaviour: the distinguishability
condition. So moral sentiments predispose us in a certain way. The relationship between the
biological dimension and cultural dimension, the cause of innumerable discussions, has to be
interpreted as a relation of inter-necessity. The diffusion of moral sentiments has been
necessary in both dimensions.
Two conceptions of rationality: 1) The present aim theory, rationality is efficient if it
pursued the aims in the moment of deliberation. This standard permits every behaviour to be
rational. 2) The self-interested theory, this concept adapts a more general concept of interest
without specifying it. This seems to be a more adequate definition because it considers also
temporal dimension and previsions on the future. But on the other hand, is very difficult to
define what is in the general interest of an individual.

The evolution of altruism and cooperation

Lecture 25/10

Evolutionary game theory is the application where a decision-maker act automatically

(biologically). The founder of the game theoretical approach in game theory is J Maynard
Smith. The groundings ideas are that the larger part of behaviours is non-rational, there would
be required too much energy to be always rational. We can try to explain some automatic
behaviours with the mechanism of natural selection.
Basic concepts of evolutionary game theory are: variation (possibility of the born of a
new variation in the population, without this possibility all the theory would not have any
sans), transmission, selection, phenotype (behavioural), genotype, fitness (reproduction
success), mutation can invade a population, and a stable state cannot be invaded by any
mutant. We would not assume the existence of a gene of cooperation, for cooperation we will
speak only about phenotype. The fitness measure also the horizontal transmission, how many
times the trait is copied and reproduced in the population. A stable strategy is when a
population could not be invaded by a mutation.
There are some differences with the standard game theory. These differences are not in the
structure (is sufficient to think the scientific purpose of standard game theory) but are mainly
conceptual. In evolutionary game theory there are no rational players, the strategies are the
phenotypes, and the payoffs represent the relative fitness (the tendence that the phenotype has
to reproduce). The utility of game theory in biology lies in the fact that the diffusion results of
a modification in a population depends on the interactions with the others member of the
same species e.g., cows with a larger stomach increase their fitness.
We can see these ideas in practice applied to High-Low game. We assume that there is
random matching, that the situation is a one shot encounter and that all the population have
only one trait. All the individuals have H like phenotype because who has L as phenotype
have 0 fitness and so he will not reproduce.

H L
H 2,2 0,0
L 0,0 1,1

So HH is an evolutionary stable strategy. S is evolutionarily stable iff either (I) S does

better playing against S than any mutant does against S or (II) Some mutant does just as well
play against S, but S does better playing against the mutant than the mutant itself does. Now
we are going to see some examples where this concept could be useful.
Suppose that there is a heterogeneous population of L and H. Suppose that if p(H) 1/3
then H will do better than L on average, until L is driven to extinction. The difference in
fitness of the two strategies will drive L to extinction. In general, in time the phenotype with
higher payoffs is the one that will spread all the population. This happens in dependence also
of the starting distribution of the mutation is a population (problem: think at the fact that a
mutation appeared initially one time in the population).
A related question is: can cooperation evolve?

C D
C 2,2 3,0
D 0,3 1,1

We will distinguish three cases:

Case one: we assume random matching and one-shot encounter. In this situation C can be
invaded by D but not vice versa. In general, is valid that evolutionary stable strategies (ESS)
are a proper subset of Nash equilibria. So is not necessary that CC is an ESS and in this case
it is not. If a strategy is a ESS is also a Nash equilibria but the contrary is not true. We could
imagine a population of C with a mutant D, mutants spread very well at the start and decrease
their rate of growth as the probability of matching another mutant increase.
Case two: we assume repeated encounters, definite number of times. The situation is the
seme that happens with definitely repeated PD in standard game theory. Nothing in change,
the payoffs’ structure remains the same. Defection is still better than cooperation.
Cooperation is still not an ESS.
Case three: in the third case we try conditional strategy, tit-for-tat (TFT). This strategy
works very well because it his: optimistic, not exploitable and forgiving.
Can TFT evolve? TFT strategy could be invaded by an unconditional cooperation strategy
but if they play one against each other is impossible two distinguish the two strategy because
TFT starts with cooperation an always reply cooperation at cooperation. On the other hand a
unconditional cooperation strategy could be invaded by free riders, TFT no. TFT is not an
ESS because of an initial fluctuation in the population.
In the real world obviously, we find much more strategies. Populations are defined as
polymorphic: with different strategies. Irrational strategies are better from an evolutionary
point of view than fully rational ones. Other reason for this difference is the energy economy
principle and the limited computational capacities of brains. Strong reciprocity e.g., TFT, is
not an ESS but it can coexist with other strategies in a polymorphic population.

In this lecture we have reasoned using the idea of random encounter but in the next one
we are going to put this idea in question.

Lecture 26/10

In this lecture we put in question the assumption of random matching between defector
and co-operators. In the real world there are strong mechanisms of selective interaction.
The biologists were interested in altruism because they were interested in behaviour that
increases fitness of the other animals. So, they are interested in altruism in a purely biological
sense. In this biological sense cooperation is an altruistic behaviour. With cooperation
strategy a species is paying an opportunity cost (the cost is CC DC, and the risk is CD DD).
Biological sense of altruism is very different from psychological sense of altruism.
Altruism is a puzzle from a biological point of view, although there are many altruistic
behaviours in nature. Why an individual decrees his fitness? Is sems to not have any
biological sense. Paradigmatic in this sense is the example of the Vervet monkeys: an
individual shout when he sees predators, in this way he attracts the attention on him and not
on the rest of the group. The individual is also losing precious time escape from the predator
risk. If this behaviour is genetic these monkeys should be always less. But in this part of the
genotype is not advantageous for the individual is it is advantageous for the group. So, the
presence of this altruistic behaviour benefits the group. Using the logic of groups selection
groups with altruistic behaviours of this type grows faster that is they would not have had it.
So, it seems that groups with these behaviours do better that free rider groups. But there is a
problem with the groups selection logic. In each group there is a selection against co-
operators. if you follow the logic of freeride, increase personal fitness, the free riders should
grow faster than co-operators and so the group should be composed of free riders.
The problems that group selection rise could be solved by Gene selection (Dawkins).
Genes are the real elements of natural selection. This theory has been built for animals but is
possible also to apply is in humans. But there is another possible solution that altruistic
behaviour is directed at specific individuals not at all the group. The commonest
characteristic is that these altruistic behaviours are directed at our relatives (kin selection).
interesting in this sense are two examples. Velvet spiders where the mother let herself be
eaten by the children and the Angler fish where the mail attach his self to the female and died
in order to increase the probability of passing his genes (Dawkins).
 In Dawkins prospective this relatedness mechanism could be explicated by the fact that
our relatives are the individual with we share the largest part of our gens. This behaviour so
increases the fitness of the genes, so Dawkin’s theory of gene selection is an explanation of
altruistic behaviours in a reductionistic sense. This explanation leaves a puzzle: most
cooperation in humans is outside the family.
Important theory in this field is Hamilton’s rule that formalized the necessary condition
for survive for biological altruism behaviours. C<br: C is the cost of cooperation (difference
in fitness between cooperative and non-cooperative), b is the value of the benefits of
cooperation, and r is the relatedness coefficient (measures the degree of relatedness that you
have in your in in dependence of the subject of interaction). There is also the generalized
version of the Hamilton’s rule where r is replaced with : probability that you are matched
with a similar phenotype/strategy.
So, correlation is the key for cooperation. Returning at the group selection problem with
this background we can observe that the problem of co-operators is soled if the group is able
to expel free riders. The type of correlation which we are speaking about is assortativity
(mechanism that match the same persons). Reciprocity is an important part of this process
because it increase the probability of matching between operators. As we have seen in
Camerer paper social ostracism is the technique used by humans to increase cooperation and
punish free riders. With the concept of assortativity we can define a assortativity group
selection theory that sems to be the final solution of aur problems with group selection.
There are many ways to increase , the variable that represents assortativity. humans make
in in many ways: communication, social ostracism… In animals the are other important
factor: reciprocity (is human is represented by sharing norms), geography disposition, and
signalling (communication in humans).
In humans prospective in small societies is easier to increase assortativity. small societies
are more trustworthy than our society. In human societies initiation rites are a paradigmatical
example of the signalling aspect of the growth of . If you want to be part, so enjoy the
benefits of cooperation, you have to give a clear signal of the fact that also you are a
cooperative person in the sense that you are ready to cooperate. If you pay a big cost this sign
il much more credible. Is for this reason that initiation rites are frequently very painful. These
phenomena are another evidence in favour of the thought that cooperation is not universal but
is within a group. In this sense we could justify the differences in game theoretical
experiments in the results of some experiments: a PD played with a stranger or with a relative
is totally different. In general, the belonging is very important in practical cooperation.

Lecture 02/11

Lecture of linking between the discourse about evolutionary game theory and social
norms. The biological model is interesting for mainly two reasons: they are very flexible and
is possible to use them for many phenomena. In the attempt of finding humans peculiarity
between animals we could look in the direction of social dimension. Humans are hyper
social, and there is a unique attention for social transmission. These mechanisms are very
important in the evolution of our specie. Mabey these mechanisms are what difference us
from other animals.
In the selfish gene Dawkins crate the concept of meme (thinking in Greeks mimesis). This
concept represents the idea of the possibility of use the same models of transmission in
biology and in sociology. This happens only at some condition and mainly this is an empiric
hypothesis. There are many authors that develop this approach. Sperber put his attention on
cognitive factors and on the difference between cultural replication and biological replication.
He thinks that the idea of meme is useless. Cavalli Sforza is also interested in studying
cultural transmission and evolution. He makes statistic studies on genes diffusion. Starting
from these studies he shows as there are situation where the genetic evolution and the cultural
evolution pull in different direction and also situation where cultural evolution and
transmission is more important than genetic one. A paradigmatic example of puzzle that
derives from the cultural evolution is the
observation that in some society fertility decrees as
the progress increases, this is difficult to explain
within a genetic frame. So in situation like this
cultural explanation are more adequate.
The graph shows the phenomena of the decrees
of number of children given the increase of the tax
of instruction. This could be explained as the
vertical passage of decreasing fertility between
social classes and thinking in the gradual change of the condition of women.
There are four types of transmission: 1) genetic inheritance, biological transmission is
only vertical. Between parents and children. 2) vertical cultural transmission, cultural
transmission withing the relationship between parents and children. 3) oblique cultural
transmission, when cultural transmission is between generations but not with the parents.
Paradigmatic examples are teachers. 4) horizontal cultural transmission, when cultural
transmission happens within the same generation.

The idea of imitating the successfully behaviours is not always the solution. The imitation
of antisocial behaviours is understandable from and individual point of view. Humans
intervene this situation with the construction of norms. Also, norm are part of the cultural
process of transmission. In sociology there are many, maybe too much, ideas on what a norm
is. Intuitively norms are necessary to make humans do what they would not individually do.
How can we incorporate norms in the RCT? Maybe the norms could be useful to change
rules of games. There are two similar theories about social norms (Bicchieri and Sugden).
The starting point of both theories is the theory of social convention by D. Lewis: regularity
in a behaviour are underline by a set of consistent beliefs.
Bicchieri starts from the definition of the subject of the theory. Social norms are
distinguished to social conventions because conventions solves coordination games and
social norm mixed-motives games. Social norm are also distinct from personal norms, moral
norms and legal norm, that are a subset of the social norms. As said above the starting point
of Bicchieri in the Lewis definition of convention30. The definition of social norm is:

Let R be a behavioural rule for situations of type S, where S can be represented as a mixed-motive game. We
say that R is a social norm in a population P if there exists a sufficiently large subset Pcf ⊆P such that, for
each individual I ∈ Pcf:
Contingency: I knows that a rule R exists and applies to situations of type S;
Conditional preference: I prefers to conform to R in situations of type S on the condition that:

(a) Empirical expectations: I believes that a sufficiently large subset of P conforms to R in situations of type
S;
and either
(b) Normative expectations: I believes that a sufficiently large subset of P expects I to conform to R in
situations of type S;
or
(b’) Normative expectations with sanctions: I believes that a sufficiently large subset of P expects I to
conform to Rin situations of type S, prefers I to conform, and may sanction behaviour.

With these norms we can explain how cooperation evolves and how some people become
cooperatives.

Absent to the last part of the lecture.

Alexander, J. M. (2019) “Evolutionary Game Theory”, Stanford Encyclopedia of

Philosophy, 1,2,3,5.

30
A regularity Rin the behaviour of members of a population P when they are agents in a recurrent situation S is
a convention if and only if, in any instance of Samong members of P,
(1) everyone conforms to R;
(2) everyone expects everyone else to conform to R;
(3) everyone prefers to conform to R on condition that the others do, since S is a coordination problem and
uniform conformity to R is a coordination equilibrium in S.
 Evolutionary game theory can be used to treat not only biological evolution but also
cultural evolution (intend in the sense of changing of believes and norm over time).
Rationality assumptions of evolutionary game theory seams to be more appropriated for
social systems than the assumption of traditional game theory. Evolutionary game theory is
explicit dynamical so it provides important elements that are missing in traditional game
theory.
R. A. Fisher developed the idea that the individual fitness depends on the proportion of
male and female in the population and this introduce a strategic element into evolution.
Maynard Smith introduces the concept of evolutionary stable strategy. Have been developed
two approaches to this theory: 1) based on the concept of ESS 2) construct an explicit model
of the process by which the frequency of stages changes in the population.
Maynard Smith uses the example of the Howk-Dove game (aggressive strategy vs
unaggressive) to construct the definition of ESS. A strategy is an ESS iff is has the property
that if almost all members of the population adapt it the population would not be invaded by
mutants. The fitness of the strategy must be superior to the fitness of the mutant strategy. One
of the two conditions must be satisfied (σ is the ESS): 1) σ does better playing against σ than
any mutant does playing against σ, or (2) some mutant does just as well playing against σ as
σ, but σ does better playing against the mutant than the mutant does. Important is to
remember that when two strategy that are different face each other the result represent the
amount of resource that is shared so for example in the Hawk-Dove game DH has a different
payoff from HD, in one case H win and in the other D loose. In this example the strategy D is
not an ESS because a population of D would be easily invaded by a population of H.
H D
H ½ (V-C) V
D 0 V/2

If V the value of the resource is grater than C cost of injury H is a ESS.

Interesting is the analysis of a population that play the PD. Cooperation is not a ESS, the
population, no matter the starting proportion tends to the ESS (defection). This analysis of PD
is called PD in evolutionary dynamics. Also the other approach to evolutionary game theory
leads to a similar conclusion

A selection of strategies by a group of agents is said to be in a Nash equilibrium if each

agent's strategy is a best response to the strategies chosen by the other players. By best-
response, we mean that no individual can improve her payoff by switching strategies unless
at least one other individual switches strategy as well. Also, evolutionary game theory has
fallen into the problem of equilibrium selection. Many scholars have tried to solve this
problem and so there are many refinements of the concept. Now the problem is to choose
between the refinements. The failure of the strong assumption of rationality in experimental
economics may suggest the weak senso of rationality of evolutionary game theory as a
possible solution. Evolutionary game theory explains many behaviours without the
assumption of strong rationality, in this sense maybe it could be useful to experimental
economics and traditional game theory.
Which is the meaning of fitness is cultural evolutionary interpretation? Depending on the
problem modelled there could exists different interpretation of fitness (money, amount of
land, cakes slice…)
Evolutionary game theory sems to have a very limited power in explaining why social
phenomenon came to be. It is much more appropriate to explain normative phenomenon
attached to social phenomenon.

Okasha, S., (2013) “Biological Altruism”, Stanford Encyclopedia of Philosophy

A behaviour is sad to be altruistic if one individual benefit another organism at a cost for
itself, this cost is measured in terms of reproductive fitness. The unit of measurements is
fundamental because only looking at this that we can determine if the behave is altruistic or
not. So intention do not count in the biological sense of altruism. From a Darwinian poi of
view the altruistic behaviour is a puzzle. How we can solve this puzzle?
An attempt to solve the problem of altruism pass through the problem of the level of the
selection. Is the level of selection is the individual one it seems impossible to the altruism to
emerge. There is the possibility that altruistic behaviours are advantageous at a group level
selection. Also, Darwin himself speaks about the altruistic behaviour in humans. But in many
years of debate, its semes that it is not the solution. The most plausible solution is the Kin
selection or inclusive fitness theory.
The individual within this theory act in a way that decrease his personal fitness but he
increases the fitness of his relatives and so also of his genes. It seems that empirical works
have confirmed Hamilton theory. Kin selection could be understood both from the point of
view of the gene and of the organism. With the altruistic behaviour the organism decrees his
personal fitness but be increase his inclusive fitness. The inclusive fitness of an individual is
his fitness plus the sum of the weight effects on his fitness of other organisms. So we can
think at altruistic behaviour as the individual trying to maximise his personal inclusive
fitness. The kin selection prospective is the most obvious solution that takes seriously the
fundamental condition for altruism to evolve: the recipients of altruistic actions must a grater
possibility to be altruistic themselves. The necessity of this condition should also be shown as
a short analysis of the average fitness in the PD. The donator-recipient correlation, rather then
the genetic relatedness, is the key for the altruism to evolve.
There are many conceptual issues related to the concept of altruism. For altruism to exist
there must be a decrease of personal fitness (in Hamilton definition) so, how we call when a
behaviour increases the fitness of the relatives and also the personal fitness? Cooperation or
mutual benefits. A crucial difference that these concepts have with the concept of altruism is
that in the latter case the beneficiary of the behaviour could be also a member of another
specie.
Another conceptual issue is the distinction between weak and strong altruism, strong
altruism is the precedent definition of altruism. Weak altruism only decrees the relative
fitness of the doner individual. These behaviours have to be classified as altruistic or as
selfish? The two behaviours seems to evolve from different mechanism so they would not be
co-classified.
The last important conceptual issue is the timescale. In which scale we have to measure
fitness: long term or short term? Some action could decrease short term fitness but increase
the long term one. A shared solution is to consider as a parameter the lifetime fitness.
A theory related to the altruism theory is the reciprocal altruism theory. Is aim to explain
case of altruism among unrelated individuals or among different species. Necessary for this
theory is the capacity to recognize the individual with the individual has been in interaction.
This theory is very related to the TFT strategy in the PD. An example of reciprocal altruism
in real world are the cleaner fishes in the coral rife.
All the theories about altruism attempt to solve the Darwinian puzzle so to reconduct the
altruistic behaviours at Darwinian principles. One question is: is this real altruism? Here we
could not give a scientific answer. The other question that rise is: can human altruism be
explained by evolutionary approach? There have been many reflections on these themes. we
can claim without any risk that we have to be particularly prudent to apply evolutionary
approach to humans being. The importance of cultural dimension in our societies is not to
underestimate. It is also possible that natural selection has formed humans with a real
psychological altruism.

Social norms

Lecture 08/11

Cfr. lecture 2/11 for the definition of a social norm by Bicchieri.

In this lecture we will see the evidences in favour of the theory of social norms. This
theory attempts to remain in the framework RCT. It attempt to explain also the deviation
from social norms with social norms. It helps to explain behaviours that are difficult to
explain with the traditional theory. This because is changes the payoffs of the games
transforming the structure a game in another game. Another important change with the others
theories is that here people care about the conformation to others expectation. So, the other
expectation now counts, those can influence my utility function. I cooperate if I think that the
others will cooperate and if the others expect from me to cooperate. This theory of not
presuppose an explicit calculation.
The practical results of the experiments thought to support the theory gave very strange
results. We can see many cultural differences. These differences could be explained starting
from the observation that peoples attempt to link the game with their social reality. How
anthropologist teach us there could be very specific interpretation of pretend natural games.
The anthropological analysis suggests the presence of huge methodological problems (cfr.
cultural variation in the ultimatum game).
Liberman, V., Samuels, S. M., & Ross, L. (2004) “The name of the game: Predictive
power of reputations versus situational labels in determining prisoner’s dilemma game
moves”. Personality and Social Psychology Bulletin30:
1175-1185. This experiment shows clearly the
importance of the framing effect. The manipulation of the
game shows well the importance of social norms. The
game is a common PD played with different frames:
community game or wall street game. The results are
very different despite the agents are playing the same
game.
Quercia& Kolle (2021) “The influence of empirical
and normative expectations on cooperation”, J Econ
Behaviour & Organization.
This experiment uses
Public good game to show
how there exist a norm of
conditional cooperation.
Cooperate if there is
expectation that the other
will do the same. First is
played the standard PG and
secondly a sequential one. So
first is played in situation of
strategic uncertainty and
secondly is a situation without strategi uncertainty. There are important differences in
contribution in relationship to the expected others contribution. Is the situation of uncertainty
there is a remarkable difference between people opinion on other behaviours and the effective
behaviour of people. This experiment shows also the difference between normative
expectation and empirical expectations.
Another important experiment is this sense is the variation of the ultimatum game the
Dictator game. In dictator game there is no strategic interaction. The other player has a
passive role. In general offers are much little than in ultimatum game. This results seams to
suggest that a port of the apparently altruistic results of the ultimatum game are caused by the
fear of being refuse. The framing of the game influence results in a significative way (if we
frame it as a leaving or a taking game the result are different. Cfr. Guala, F. & Mittone, L.
(2010) “Paradigmatic Experiments: The Dictator's Game”, Journal of Socio-Economics 39:
578-84).
In order to see how is possible to neutralize expectations has been constract another
variant of the dictator game: Dana, Cain & Dawes (2006) “What you don’t know won’t hurt
me: costly (but quiet) exit in dictator’s games. Org Behav& Human Dec Proc.

Lecture 09/11

The previous lecture suggests us the important of studying framing effects. The first
studies on framing effects were made by the gestalt psychologists. Framing effects sems to be
something co-existent with our perception. If we play a game with the individual frame the
result will be very different from the result of a game paled with a communitarian frame. The
explosion of the studies on framing effect is started at the end of the second world war (crf.
the banality of evil in seeing the potential force of collective framing effects).
There had been made two paradigmatic experiments: Sherif’s “Robbers Cave” experiment
and Tajfel’s “minimal group” experiments. In these experiments has been observed an
important effect that underline the difference between in-group and out-group. The
behaviours within the group are very different from the behaviours with the out-group
(mainly allocation of different amount of money). We can define this type of change in term
of group thinking. There has been observed also an interesting relationship between group
identity and group norm. about this Bicchieri says:

It is entirely possible … to view group identity as a trigger for norm-abiding behaviour. When we represent a
collection of individuals as a group, we immediately retrieve from memory roles and scripts that ‘fit’ the
particular situation, and access the relevant empirical and normative expectations that support our
conditional preference for following the appropriate social norm, if one exists.

Yamagishi, T., Mifune, N., Liu, J. H., & Pauling, J. (2008) “Exchanges of group ‐based
favours: Ingroupbias in the prisoner's dilemma game with minimal groups in Japan and New
Zealand”. Asian Journal of Social Psychology11: 196-207. This experiment halps us in the
identification of a connection between norms and groups. the object of study is the relation
between belonging to the group the expectation. The experiment is conduced under two
conditions: 1) reciprocal knowing of the belonging to the group. I know that you know. 2)
every player knows his belonging but do not know if the other player belongs to the game
group. In these two situations has been discovered a difference in the expectation of
cooperation. So, is seams that we can confirm that there is a relation between group identity
and norms.
 In this field there are two possible approaches: first norms and after groups (Norms are
ontologically primary; groups are just ‘bundles of norms’) or first groups and after norms
(Groups are primary; norms are group-phenomena). The first approaches are the one of
Bicchieri the second one is the one of the theories of collective intentionality.
There are four different and similar theories of collective intentionality: Margaret Gilbert,
Raimo Tuomela, John Searle, Michael Bratman. Gilbert critics Lewis’s theory of
conventions. She is also critical on the possibility the individualistic theories can explain
everything (RCT). She argues in favour of a total distinction between groups and individuals,
they are not reducible one in another. She claims:

The game-theoretical scheme in general concerns a set of individuals who face each other, so to speak, while
each sees himself as faced with a certain problem of his own: what should I do now? … Meanwhile, the
conceptual framework of plural agency is different and the questions and answers relevant within it are
different. … the group ranking of the outcomes may be distinct from the ranking of either individual.
Moreover, from the point of view of ‘us*’ it may not even be obvious that the utilities of the individual
elements of the plural subject are relevant for what we should do. … I assume that some notion of ‘our*
preferences’ is available, that is, coherent and intelligible. I presume also that it finds instances in the real
world.

Here we could make an interesting comparison with Rousseau he also faces the problem
of aggregation (cfr. social contract distinction between general will and the will of all).
Gilbert argue in favour of the importance of the plural subject:

When a goal has a plural subject, each of a number of persons (two or more) has, in effect, offered his will to
be part of a pool of wills which is dedicated, as one, to that goal. It is common knowledge that, when each
has done this in conditions of common knowledge, the pool will have been set up. Thus what is achieved is a
binding together of a set of individual wills so as to constitute a single, ‘plural will’ dedicated to a particular
goal.
 The main elements of Gilbert’s theory are: 1) offered his will, may be, but need not be,
explicit. In On Social Facts, she speaks of ‘quasi-agreements’. 2) common knowledge, it is
not enough that Jill believes they are going for a walk, and Jack believes the same. They must
also believe that the other believes … etc. 3) binding together, there must be a joint
commitment: a commitment by two or more individuals that cannot be unilaterally cancelled.

Bicchieri, C. (2006) The Grammar of Society, Cambridge University Press, Ch. 1

(“The rules we live by”).

There is general confusion of what is a norm, and this influence the comprehension of the
power that norms could have on direct humans actions. the theory that we are going to build
is a constructivist theory of norms, it explains norms in terms of preferences and expectation
of the follower of the norm. people have conditional preferences about norms, these
preferences are conditional on having expectations about other people conforming to the
norm.
Social norms are usually against self-interest. These norms solve usually mix-motived
games transforming them in coordination games. In social norms there is not necessary to
exist an incentive to follow the norm. the prospective that this theory adopts is to emphasise
the automatic component of this process.
There are two ways to reach a decision: 1) calculate expected utility of the outcomes and
weight them on the probability. 2) following behavioural (deliberational) rules that regulate
that situation. This process is a process of categorization; fill pre-existing categories through
the use of the principle of saliency. We call this way the heuristic rout. Here is emphasised
the automaticity of the decision-making.
Usually, we combine the two routes. Rational deliberation is better conceived as an ideal-
type, and it implies a conscious process of information and evaluation. But if the beliefs and
the desire are conscious that is impossible to take them as playing a role in the heuristic. In
line with this problem heuristic account of decision account for the dispositional tradition that
argues that beliefs and desire are disposition, they are so functional to the heuristic way.
The idea that is possible to manipulate social norms is very attractive. We have to find in
which situation which norm is triggered. Is we could make these types of prediction we will
be able to manipulate social norms.
 What are social norms? social norms are public, sheared and not necessary formal (cfr.
distinction with legal norms). conformity to social norms is conditional conformity. Is
conditional on expectation about other people behaviour. A norm could be prescriptive or
proscriptive. The aim of a definition of norm is to point out what differences a social norm
respect to others social facts. A social nor is a behavioural rule, something that say us how to
behave in a situation. Important for a useful definition is to define the condition in which a
behavioural rule is a norm:

Let R be a behavioural rule for situations of type S, where S can be represented as a mixed-motive game. We
say that R is a social norm in a population P if there exists a sufficiently large subset Pcf ⊆P such that, for
each individual I ∈ Pcf:
(1) Contingency: I knows that a rule R exists and applies to situations of type S;
(2) Conditional preference: I prefers to conform to R in situations of type S on the condition that:

(2.a) Empirical expectations: I believes that a sufficiently large subset of P conforms to R in situations of
type S;
and either
(2.b) Normative expectations: I believes that a sufficiently large subset of P expects I to conform to R in
situations of type S;
or
(2.b’) Normative expectations with sanctions: I believes that a sufficiently large subset of P expects I to
conform to Rin situations of type S, prefers I to conform, and may sanction behaviour.

Pcf is the set of the conditional followers. If this set is sufficiently large the behavioural
rule could be considered a social norm, and a social norm is followed if the set of followers is
sufficiently large. The contingency condition is the very constitutive part of a norm. The
condition 2.a means that the rule is practice. The expectations could be empirical but also
normative, in the sense that people belief it is an obligatory commitment to conform to the
norm. condition 2.b and 2.b’ tell us that we have different reasons to conditional conformity
to the norm. for some individual condition 2.a and 2.b are sufficient for conformity. Someone
needs also condition 2.b’. The conditions just described are necessary but not sufficient.
Following a social norm may be contrary to the self-interest.
Example of benevolence in PD used to clarify what we intent when we are saying that
motives to follow a norm should be distinguish from other motives. Benevolent motives for
purely altruistic ones. in this example benevolent motive are also self-interested, the self-
interest to be a benevolent agent.
The norm follower’s preferences are similar to the benevolent preferences with some
differences: the benevolent prefers CD to DC. Condition 2 make a distinction between social
and personal norms. moral norms demand for an unconditional preference. There are
independent motives to believe in it.
 What distinguish social norms from norm of justice is our commitment to these norms
stays in the legittimity of normative expectation required by condition 2.b. What really
distinguish between a social and a moral norm is our attitude toward it.

Newson, L., Richerson, P. J., & Boyd, R. (2007). “Cultural evolution and the shaping of
cultural diversity”, in S. Kitayama & D. Cohen (eds.) Handbook of Cultural
Psychology, London: Guilford Press, pp. 454-476.

Kölle, F. & Quercia, S. (2021) “The influence of empirical and normative expectations
on cooperation”, Journal of Economic Behavior & Organization 190: 691-703

Guala, F. & Mittone, L. (2010) “Paradigmatic Experiments: The Dictator's Game”,

Journal of Socio-Economics.

The DG, on its own, is probably too unusual and too abstract to trigger any real-life
normative behaviour. Indeed, the variability observed in experiments is probably due to the
fact that all sort of norms (even conflicting ones) can be triggered by adding small cues to the
basic design. The DG then is likely to be mainly useful to study the effect of contextual cues
that determine the framing and triggering of social norms.
Economist can use it in the right way because they do not have an adequate theory of
norms.

Collective intentionality

Lecture 15/11

Cfr. lecture 9/11 for the introduction to collective intentionality and to Gilbert position.
The collective intentionality framework distinguishes between two important questions:
what should I do? And what should we do? The theory adopts the secondo question as a
starting point. Given this starting point is interesting to ask which is the relation between
RCT that clearly starts from the other question.
RCT is conceived as an individualistic research program. This does not mean that we
could not apply it to groups but simply means that we assume that the basic units behave as
individuals. The collective intentionality theory (CIT) refuses the individualistic reduction of
RCT. It assumes that there are phenomena that could not be reduced to individuals. To go a
little much deeper in this comparison we will se an CIT explain cooperation and coordination
problems.
RCT deny that is rational to conform to a social convention. For RCT focal point make us
abandon rationality, precisely for this reason CIT works where RCT do not work. A
paradigmatic example is Hi-Low game:

H L
H 2,2 0,0
L 0,0 1,1

RCT solve this game using the concept of NE and of PE. This solution sems to be pretty
ad hoc because the two players have the same problem: the question about the others choice.
Another paradigmatic example is the four-card game that could be translated in an Hi-Low
game. If we manged to solve Hi-Low with CIT better than RCT we could solve many others
coordination games. Can CIT propose a more satisfying solution to Hi-Low than RCT?
The solution that CIT propose is team reasoning. Using the «we frame» we can reframe
Hi-Low. The payoffs are now the payoffs on the team. We could no longer apply strategic
reasoning, but we must apply parametrical reasoning. The change in the payoffs imply a
change in reasoning.

The new inferential scheme works like this:

1.Our goal is to maximize the group’s payoffs

2.The best way to maximize the group’s payoff is for us to choose HH
3.We will choose HH
4.Therefore, I will choose H and you will choose H

This new inferential scheme introduces two new questions: 1) Is it rational to pursue a
collective goal if you identify as a member of a plural subject (team)? 2) Is it rational to
identify as a member of a plural subject (team)? It seems that the questions have to be
answered in a sequential structure. First I have to solve the belonging question and than the
act question. In general team reasoning pull of a meta coordination problem similar to the one
that CIT critics at the RCT solution of Hi-Low:

It’s rational to see it as a collective problem only if the other player sees it as a collective problem. But how
do I know that she will see it as a collective problem? She is asking the same question, and she cannot
answer unless she knows what my answer is.
 Bacharach solution the second problem is the following:

Group identification is a framing phenomenon: Among the many different dimensions of the frame of a
decision maker is the ‘unit of agency’ dimension: the framing agent may think of herself as an individual
doer or as part of a collective doer.

and

The reasons why group identifiers team reason do not depend on what causes the identification. It does not
have to be endogenous, caused by features of the decision situation itself. In particular, it does not have to be
generated by the agents’ individual preferences. When two English explorers meet by chance in the African
wilderness, they stand ready to die for each other not because of strong interdependence but because they
went to the same school. The group identities that people have when they make decisions, and the group
payoffs that go with them, can be, and in fact typically are, quite exogenous to the decision problem they
face.

and

“…a joint commitment binds each of the parties in at least the following way: It gives the parties sufficient
reason to act in conformity to it unless and until the joint commitment has been satisfied, has been jointly
rescinded, or has otherwise jointly been put to rest

Here is the reconstruction of PD in the CIT:

Rember that CC is not an equilibrium so there is always the temptation to defect. So why
we should reason as a team? For Bacharach it is not something one chooses to do. It just
happens. Differently for Gilbert one is committed to do it. To arrive to CC is necessary a
commitment. In the team reasoning framework, we have already reached the commitment
deciding to stay in the group. So now we need to decide what is better for us as a group.
We could make an analogy between collective intention and individual intention.
Intention/intentional state are mental state with propositional content e.g., belief, …Intention
in action is a plan or purpose to do something; a mental state that directs action toward a goal.
We could make an example of intention for individual action:
 Preferences change in time the solution is a commitment that could be public or personal.
Many times, sheared commitments are stronger because we put within the variables our
reputation. This could be a reason because of the fact that we need a collective intention to
pursue a common goal (Cfr. McClennen, E. F. (1997). Pragmatic rationality and rules.
Philosophy & Public Affairs, 26(3), 210-258.)

In the first example, the individual case I can always change idea because the commitment
is private. In the second case the commitment is public so I cannot abandon it unilaterally.
The commitment in team reasoning could be only collective. For Gilbert:

“…we can see how agreements or other joint commitment phenomena—such as specific collective
intentions or goals or rules with appropriate content—can lead to relatively good outcomes for all in
collective action problems of all kinds, even when one otherwise consults only one’s own inclinations. That
is, one need not allude to morality.
Included here is the prisoner’s dilemma. If (for instance) we intend to do as well as possible for each
individually—and each is therefore committed through the underlying joint commitment to do what he can
in conjunction with the others to achieve this aim—we have to work out what doing as well as possible for
each individually in this situation amounts to. Quite likely this will mean that each must ‘cooperate’ with the
other, doing his part in a combination of actions that do not give him what he is most inclined to get.
Summarizing the two positions: for Bacharach rationality applies after the unit of agency has
been defined. The choice of unit is a non-rational matter. For Gilbert the agreement creates a
reason to think as a plural subject. People may decide (rationally) to reach an agreement.

Gilbert, M. (1990) “Walking Together: A Paradigmatic Social Phenomenon”, Midwest

Studies in Philosophy 15: pp. 1-14.

Two open problems of the philosophy of social sciences: the method problem and the
relational problem. In this paper, the central concept is the concept of social group or
collectivity in general. The core of the relational problem. The methodology used is the one
of G. Simmel, investigating small-scale phenomena, sheared actions. precisely in this sense
«going for a walk together» could be considered a paradigmatic social phenomenon.
When are we going for a walk together? There must be common knowledge of strong
sheard personal goals. This imply some obligation and the right to remember it to the other,
these are reciprocal. Which type of obligation are these? Moral obligations?
There is an important distinction between being obligated and having an obligation. In our
case we are dealings with an obligation of the second case. There is also a entitlement to
rebuke that adds something to the obligation. The disobedience to the obligation gives the
other person the entitlement to rebuke. But where the obligation really stays? Is seems that it
does not stay only in the sheared personal goals, not in the moral obligation, not in the self-
interest.
The process we are investigating is the one of the construction of a plural subject (if the
goal is a goal of the plural subject, we can say to have the condition to say that we are going
for a walk together??). the formation of a plural subject is a set of individuals will that
generate a collective will. The wills are bounded simultaneously and interdependently, there
is not unilaterally entrance of the individuals. The plural subject has a single will. The
individual commitment is conditional to the commitment of the others. Once formed it imply
obligation to rest inside. We use «we» to account for this new subjectivity, it refers to a pull
of wills that are oriented on one goal. «We» is the first person, it is auto-implicative, it could
has a goal exactly as «I» could have «my» goal. «Our goal» in the same that «my goal». The
subject of the «we» refers to each other as «we», at least in relation to the common goal. So
the paradigmatic phenomena of going for a walk together involve a special type subject and a
special type of goal, the goal of a plural subject that is different from the sheard goal of the
individual constituting the plural subject.
 Human social groups are plural subjects. All social groups must have a joint goal as plural
subjects has one. The claim of the paper has been that if we search for the nature of social
groups we can find it in the paradigmatic example of going for a walk together. Now we can
say that for going for a walk together to subjects must constitute a plural subject so, the
concept of plural subject is the key to understand what a social group is.
The problem of group constitution is also a problem of classicals political theories (cfr.
Rousseau and Hobbes). These argumentations support the idea of a strong link between plural
subjects and social groups. for G. Simmel there are important distinction between two person
groups and larger groups.
The theory exposed in the paper has clearly some consequences on the two fundamental
question that constitute the first part of the inquiry of the paper.

Gold, N. & Sugden, R. (2007) “Theories of Team Agency”, in F. Peter & H.B. Schmidt
(eds.) Rationality and Commitment, Oxford University Press. [online]

Bacharach, M. (2006) Beyond Individual Choice, Princeton University Press, especially

Chapter 2 ("Groups")

Entries on Collective Intentionality in the Internet Encyclopedia of Philosophy

Many philosophers argue that lonely individual intention could not take in account
collective action. This presupposition is the starting point for all collective intentionality
analysis. The ascription of intentional states to groups. there are many possible interpretation
and justification of this.
Instrumentalism position in the one that these ascriptions are simply metaphor. The
justification of this metaphorical dimension is a pragmatic one: the explanatory power of
ascription of intentional states to groups.
Another account of collective intentionality is the summative account. The ascription of
an intentional state to a group is justified by the diffusion of this state in the group: «the
group G believes that p iff all or most of the members of G believes that p». There are also
more complex summative accounts that add to the previous definition the notion of
common knowledge. The problems with the summative account stay in the sum, the
irreducible difference between the whole and the parts. Collective intentions are not reducible
to individual intentions so, none of these accounts could be a good explanation of collective
intentionality phenomena.
 J. Serle argues in favour of a non-summative account but still individualistic account of
intentions. He argues that collective intentions exist only in individuals’ brains
(individualistic feature). In the brains could exist the collective intention «We intent to do X».
The possibility that individual has to form this type of intention is guaranteed by the sense of
the other, a primitive feature of human’s biology. Collective intentionality is fundamental in
Serl’s construction of social ontology because for him collective intentions are the origin of
social facts. Serl’s approach have been criticised for many reasons: the main point of critique
is the fact that his account fails to capture the normative relation co-existing in the nature of
collective intentionality. The formation of collective intentions also creates obligation and
collective expectation.
M. Gilbert aim is to explain collective intentionality, which is a normative phenomenon,
without any other postulation of normative phenomena. This attempt of explanation is in
relation with the account of social groups: the collective subjects who can has collective
intentions. The individual will be fundamental in the formation of collective subjects but
when the commitment is made appears the irreducible dimension of plurality. Each individual
is now obligated to do his part. Obligation that rises in the joint commitments are the
irreducible dimension of social groups and so of social facts. The individual commitment and
the collective commitment are not indiscernible, is possible to brake the individual
commitment without braking the collective commitment. Joint commitments create reasons
to act. These reasons remain despite the agent change his mind. We can now formalized the
joint commitment:

Individuals A1…An… form a plural subject of believing that p if and only if A1…An form a joint
commitment to believe that p as a body.

The difference with summative accounts is that members do not have to believe that p. so
for Gilbert when individuals form plural subject there are no more individuals that believes
that p but there is only a group that believe that p. The particular feature of this account is that
the sufficient condition for a joint commitment to exist is the existence of persons that
collectively believe that p. Gilbert’s account is less individualistic that Serl’s account. The
main critique to Gilbert account of collective intentionality are the charge of circularity and
the charge of limitation. Gilbert’s account is considered to explain only dynamics in small
groups and so there is not the necessity of a formal decision method. The difficulty in big
organisation is that is seams the not all the subjects have to express their willingness to do
their part.

Constitutive rules

Lecture 16/11
 The capacity of doing things together is a very important human skill. No doubt that
human cooperation and coordination works at higher levels. Structured heavily by
complicated systems of rules. Cases when we say that our behavior is ruled by institutions.
Introduction to a theory of a philosopher that tried to explain the step from low level to higher
and more complex behavior when coordination and cooperation is included.
John Searle is one of the four philosophers that introduced the concept of collective
intentionality. In (1990) “The construction of social reality” he developed some ideas that
have important implications for philosophers of social scientists. His work constructs a theory
that allows us to pass from collective intentionality to institutions. From social facts to
institutional facts.
He is trying to give an account of the unique capacity proper of human beings to build
social institutions. The flexibility of institution is a proper human characteristic. In his book
he tries to capture this human specificity making a distinction between social facts and a
subset of them, institutional facts.

Collective intentionality is the basis of all society, human or animal. Humans share with many species of
animals the capacity for collective intentionality and thus the capacity to form societies. Indeed, I will define
a social fact as any fact involving the collective intentionality of two or more agents . Our problem, then,
is to specify what is special about human collective intentionality that enables us to create special forms of
social reality that go beyond the general animal forms. Both the Supreme Court making a decision and a
pack of wolves hunting a sheep are engaged in collective intentionality and, thus, are manifesting social
facts. Our question is, what is the difference between the general class of social facts and the special
sub-class that constitute institutional facts?31

The fact that collective intentionality can also be observed in animal species is
controversial. Today ethologists (ex. chimpanzee) do not agree that the behaviour observed is
coordinating behaviour. It is not clear if it is deliberative or not. Very controversial also the
capacity of animals of attributing functions.

“Human beings have a capacity that they share with some, though this time with not very many, other
species of animals, the capacity to impose functions on objects where the object does not have the
function, so to speak, intrinsically but only in virtue of the assignment of function.
… [There] is a special kind of assignment of function where the object or person to whom the function is
assigned cannot perform the function just in virtue of its physical structure, but rather can perform the
function only in virtue of the fact that there is a collective assignment of a certain status, and the object or
person performs its function only in virtue of collective acceptance by the community that the object or
person has the requisite status.32

Members of supreme court can do what they do because they have been assigned a certain
status-function. They can do so through the concept of constitutive rule, introduced by Searle.
We have to introduce the concepts of a constitutive rule.

31
Searle, 2005, « What is an Institution? », pp.6-7
32
Searle, 2005, p.7.
 Searle was a student of J. L. Austin famous for theory of performative speech acts: we
often use language to do things, not only to describe how the world is and to communicate
with others about.

“You are an idiot!” --> we are offending someone

“I promise to give you ten euros tomorrow” --> We’re making a promise
“You are now man and wife” --> We’re creating a marriage,
“I name this ship Titanic”

He clarified that in order for speech acts to be successful certain background conditions
have to be in place e.g., if I say you’re an idiot during a play in theatre doesn’t work. Searle’s
idea was that an essential part of this background conditions are constitutive rules, which are
rules that specify what is the effect of a speech act e.g.:

a constitutive rule can have this form: Saying “you are now man and wife” to a couple, in a Church or public
place, in front of witnesses, during a ceremony (etc.) counts as a wedding. We’re creating something which
didn’t exist before.
An individual who is born in the USA, is at least 35 years old, has won the majority of delegates in a
national election (etc.) counts as President of the USA

It can be extended beyond performative speech acts: the creation of certain social facts is
regulated by constitutive rules even if they don’t necessarily involve any speech act, ex. the
Alps constitute the boards between France and Italy. There’s no speech act involved. The
only thing necessary is that we all agree with this rule and behave coherently with its
consequences.
Searle made a contrast between regulative rules and constitutive rules, to say what
constitutive rules are not. This distinction was inspired by J. Rawls paper «different
terminology, similar distinction». Regulative rules regulate pre-existing forms of behaviour
e.g., moving a piece of wood on a checkerboard or driving on the left/right. Those are things
that we do independently of the existence of the rule. The grammatical form of regulative
rules is: make this or make that. On the other hand, constitutive rules make possible new
forms of behaviour e.g., putting the other player’s king under a direct attack from which it
cannot escape counts as checkmate in the game of chess. Unless we have a set of rules that
constitute the game of chess, certain facts are not even possible.
Comparing the grammatical form of the two types of rules could be especially useful:

Regulative rules --> do this, do that, drive on the L/R; if X then Y

Constitutive rules have this grammar: X counts as Y in C.
X= a pre-institutional term. Ex. an individual born in the USA, older than 35…
Y= a status function. Ex. President
C= specifies what is the domain of application.

Over years Searle had to make the formula more complicated. This new formulation
specify the conditions that identify x as y. The conditions identifying X must be made
explicit. The formula becomes: If P, X counts as Y in C e.g., if an individual is at least 35
years old, was born in the USA, has won the majority of delegates in a national election
(etc.), then he/she counts as President in the USA.
 the relation between constitutive rules and collective intentionality is a key point in
understanding institutions. Collective intentionality for Searle is more fundamental for the
existence of all social facts, not also institutional facts. Without collective intentionality we
could not have institutional facts because constitutive rules must be accepted collectively by
the member of a community in order to be effective. Different members of a community are
going to formulate this intention which is in each individual mind (individualistic ontology).
It takes the classic form of “we” as a collective intention. The «we» frame is grounding. This
acceptance is necessary in order for constitutive rules to do their job. Is by collectively
accepting that something is the case we make it the case. Relation between social and
institutional reality:
This formula can be applied repeatedly:
you can make sense of the fact that social
facts and in particular institutional reality
are made of layers and layers of
institutional facts. Institutional facts are a
subset of social facts.

With the iteration of constitutive rules we can reach the ground of explanation. We can
reach physical facts e.g., when we speak about the president of United states we make
reference to objects that are pre institutional e.g., being born in USA presupposes that we
already know what USA are and that exist, that have certain physical boundaries… At the
bottom we have very simple institutional facts that are constructed directly from physical or
neutral objects.
Going on with the analysis of institutional facts a central concept is the concept of status
function:

Because the creation of institutional facts is a matter of imposing a new status and with it a function on some
entity that does not already have that status-function, in general the creation of a status-function is a
matter of conferring some new power. … In the simplest case, the Y term names a power that the X term
does not have solely in virtue of its X structure.
 When we use a constitutive rule, we assign a function. A status function in Searle theory is
associated with the notion of the deontic power. The status function assigns deontic power to
something. Deontic powers are irreducible. A corkscrew can open bottle because of his
physical structure, but we can use also a lighter. But here the powers are different, they are
social powers, recognised by the collectively and the imposition of a function is effective
because the members of the group accept that a certain object or person has a certain power
(ex. money). The reason why I can do things depends on the recognition of certain powers
that are associated with money. According to Searle the fact that money or institutional
entities in general have deontic power makes them irreducible to non-deontic facts or entities.
There is a gap between institutional facts and other kind of facts which prevents the
reducibility of institutional facts to non-institutional ones. These facts are for example the
non-reducibility of non-institutional facts to behavioural facts. Institutional facts can only be
explained if we postulated the existence of collective intentions, which explain the deontic
power of status function. Deontic means normative (from Greek deon that means duty).

… such deontic phenomena are not reducible to something more primitive and simpler. We cannot analyse
or eliminate them in favour of dispositions to behave or fears of negative consequences of not doing
something.

There is something special about these deontic powers and they have to do with collective
intentionality. There is a strong correlation between collective intention and cooperative
attitude.

The reason that we-intentions cannot be reduced to I-intentions, even I-intentions supplemented with beliefs
and beliefs about mutual beliefs, can be stated quite generally. The notion of a we-intention, of collective
intentionality, implies the notion of cooperation. But the mere presence of I-intentions to achieve a goal
which happens to be believed to be the same goal as that of other members of a group does not entail the
presence of an intention to cooperate to achieve that goal. One can have a goal in the knowledge that others
also have the same goal, and one can have beliefs and even mutual beliefs about the goal which are shared
by the members of a group, without there being necessarily any cooperation among the members or any
intention to cooperate among the members.

There seems to be a tight relationship between the collective intentionality, deontic powers
normativity and cooperation. So tight that we cannot explain one without the other. We
cannot get out of this circle because if we try to analyse one of the notions (collectve
intentionality) using some simpler notions, we are not going to succussed because we’re
going to lose some important elements of the notion. Why?
Example of «the business school case». It comes in two parts, one in which there is a
collective behaviour which involves only individual intentions and mutual knowledge of
individual intentions; on the other hand, there is another version of the business school case
where we have a genuine collective intentionality. He wants to understand by intuition that
there is something that is special to collective intentionality and has to do with cooperation
and normativity.

Suppose a group of businessmen are all educated at a business school where they learn Adam Smith's theory
of the hidden hand. Each comes to believe that he can best help humanity by pursing his own selfish interest
and they each form a separate intention to this effect, i.e. each has an intention he would express as, "I intend
to do my part toward helping humanity by pursuing my own selfish interest and not cooperating with
anybody". Let us also suppose that the members of the group have a mutual belief to the effect that each
intends to help humanity by pursuing his own selfish interests, and that these intentions will probably be
carried out with success. That is, we may suppose that each is so well indoctrinated by the business school,
that each believes that their selfish efforts will be successful in helping humanity.

This is the case of individual intentions and common knowledge of these individual
intentions.

This case has to be distinguished from the case where the business school graduates all get together on
graduation day and form a pact to the effect that they will all go out together and help humanity by way of
each pursuing his own selfish interests. The latter case is a case of collective intentionality the former case is
not.

Collective intention requires an agreement jointed with a commitment, key aspect which
provides the normative force of collective intentionality (Gilbert). Similar idea.

Structure of the “Business School” argument:

1. The alumni are indoctrinated in Adam Smith’s invisible hand theory. Each seeks to help humanity by
pursuing his own interests, and this is common knowledge.
2. The alumni “form a pact the effect that they will all go out together and help humanity by way of each
pursuing his own selfish interests”.
Claim: case 2 is a genuine example of collective intentionality, 1 is not.

The idea is that there are these 4 concepts related to one another: collective intentionality –
agreement/pact – deontic power – cooperation.
In the case of constitutive rule the mechanism is this: we collective accepts that an object
or a person has certain deontic power in virtue of a collective intention, which is not merely
descriptive. It also involves an attitude of cooperation towards the implementation of the
consequences of status facts. President of USA: we accept that he has a certain power and
that we ought to respect and follow whatever piece of legislation that he decides to create.
The attitude is not nearly instrumental, it is not that I follow a piece of legislation because
I’m afraid of the consequences of not following, but because I have the cooperative attitude
towards constitutive rules and the power associated. Idea that these four notions seems to
constitute a circle. So tight that in order to define one of them you need the other and we
cannot define any of them without the other three concepts. We cannot define them through
other simpler facts. Irreducibility of institutional reality to individualistic facts.
There are two problems with this account. First problem with this formula is that is too
simple; seems to take for granted important mechanisms. It is not clear what exactly are the
implications of status functions (what is a president of USA? Unless we specify what are the
implications e.g., the things that he can or cannot do. How can we try to make the constitutive
formula more informative?). Second problem is that this formula seems to take for granted a
very high level of []in a society; we also know that there are societies that continue to exist
despite the fact that the some of their fundamental institutions are not accepted by the
authority of [members?] ex. societies based on slavery.

Searle, J. (2005) “What Is an Institution?”, Journal of Institutional Economics 1: 1-22.

The aim of the article is to try to answer the question about institutional ontology. The
most important error of all the tradition is that takes the language for grounding and if you
presuppose language, you are presupposing institutions. The right way is to analyse the role
of language in the construction of institutions. we start this analysis by not defining what is
an institution but from which and what are institutional facts. Institutional facts exist only
given some human institutions. A further distinction is a distinction between the statement of
a fact and a stated fact. The fact stated is obviously not dependent on the institution of
language.
There are others general distinctions to made. The distinction between features that exists
independently from humans’ perception and features that do not depend on that. The social
sciences are conceived with the human dependent phenomena. We need a second distinction,
the one between kinds of objectivity and subjectivity. There are two senses of the distinction
between object and subject, empirical sense and ontological sense. The question that we are
facing in the paper is a very small part of the bigger question about the relationship between
physical reality and institutional reality. Could exists institutional reality in a world governed
only by physical laws?
We start explaining social reality starting from three primitive and irreducible notions:
collective intentionality, the assignment of a function, and status functions. Collective
intentionality is the base of animal societies. The inclusion in the collective intentionality
changes the motives of an action I am doing something as a part of our doing something.
Collective intentionality is collective intentions but also collective beliefs and desires.
Humans have the capacity of assign function to parts of reality. This capacity is not only
humans but para excellence human. This feature of humans if fundamental in the solution of
the question about institution looking at the next fundamental feature.
The status function is a particular type of function which can be performed not in virtu of
the physical structure but in virtue of a collective assignment of status. The process of
assignment takes this form: X count as Y or more precisely X count as Y in context C. A
status function is a function that can be performed in virtue of the collective acceptance of the
object as having a status. This structure is different from the structure of regulative rules
those are constitutive rules. Status functions are the glue that holds human society together.
The status function carries deontology.
Humans’ institutions create deontic powers, but not all deontic powers (rights, duties,
obligations…) are a matter of institutions.
The error of the tradition in studying institution is that they had taken language for granted
and grounding, language is also an institution. language do not only describe institution, but it
also partially constructs institutions. The role of the language in that it allows us to represent
status function. For this reason, animals cannot have an institutional reality. In the perception
of institutional facts, we have to look in the same time from the brute facts to the institutional
level (cfr. possibility of iteration). Language performs four functions in the construction of
institutional facts: 1) facts can exist only if represented as existing, and language offer us this
means of representation. Language takes with its deontic powers that could not be taken by
brute facts. 2) language in the only possibility to represent deontic powers. 3) language is also
important in the possibility of having a deontology. 4) Last function of language is the
recognition of the institution as such particular instances of institutional facts are instances of
general institutional phenomena. With language we can identify the distinct levels.
Importance of status indicators in the construction of society because they allow the
recognition of status functions. There is also institution that has the «free-standing Y term».
This means that there is a status function but there is not an object on which the function is
imposed. This could work because status functions are matter of deontic power and so
deontic powers goes directly on the individual e.g., having hundred dollars is not to have
them physically but to have the right to buy something that cost hundred of dollar.
In «the constitution of social reality» the fundamental piece was «We accept (S has a
power (S does A))». In which relation is this formulation with the new formulation of this
paper? We can accept S having a power because we have already accepted S=X and X count
as Y is C. status function are not necessary in general, non-humans’ animals do not have
them. They are necessary to institutional facts.
There are many kind of institution, the aim of the paper is not analysing the common
usage of the term institution. The facts that carry deontology are the glue that holds together
society.
Important is remember the presupposition of economic sciences. This way is to avoid all
misunderstandings about the approach we have used in the text. Economist uses model and
the importance of the model is evaluated through its predictive powers. Economist use
thought experiments and it is not always easy to understand their role. Methodological
individualism is the base of all the analysis: all independent mental observer must exist
separately in individual minds.
An institution is any collectively accepted system of rules (procedures, practices) that
enable us to create institutional facts. The creation of an institutional fact is, thus, the
collective assignment of a status function. The typical point of the creation of institutional
facts by assigning status functions is to create deontic powers.
We have to look at institution form and individual point of view and society have logic
structure. These are the last general consequences of this analysis.

Searle, J. (1990) “Collective Intentions and Actions.” In Intentions in Communication,

edited by P. Cohen, J. Morgan, and M. E. Pollack. Cambridge, Mass.: MIT Press, pp.
401–415.

Smit, J. P., Buekens, F., and du Plessis, S. (2011) “What is Money? An Alternative to
Searle’s Institutional Facts.” Economics and Philosophy 27: 1–22.
Rules and equilibria

Lecture 22/11

In this lecture we are going to discuss some of the fundamental points of Serl’s theory of
institution. There are three grounding problems: 1) sometimes there is no X. 2) acceptance
sems too strong for some institutions. 3) constitutive rules are not necessary.
There are some institutions that sems to be base-less. A counter example to Serl’s theory is
money. There is no physical instantiation of money or at least there is not a single physical
instantiation. Money allows us to do certain things. It do not fits Serl’s formula. It is a power
that we gave something (cfr. Smit, J. P., Buekens, F., and du Plessis, S. (2011) “What is
Money? An Alternative to Searle’s Institutional Facts.” Economics and Philosophy 27: 1–
22). Serl reply to this critic in «What is an institution» arguing in favour of the fact that the
core of the argument is that we accept the base having a power of doing something. This
argument could still be criticised because the core of the critic is the real absence of a base.
Are all institutions really collectively accepted? There sems to exist many institutions
without collective acceptance. A paradigmatic example in this sense is apartheid. In Serl’s
view what gives normativity (deontic powers) to institution is the collective acceptance and
there are many cases where this collective acceptance does not exist. There are some counter-
counterexamples to the apartheid counterexamples. Is possible to argue that apartheid
precisely for this reason is not an institution or at least a legitim one. Is possible also to say
that it is supported only by someone, and this is considered sufficient of that the non-
acceptance in reality is an acceptance in virtue of the existence of the institution but this last
argue sems to fall in a vicious circle. We could also refuse the acceptance condition but then
where normativity came from? Infibulation is a paradigmatic example of the difficulty to
change norms.
Constitutive rules sems to be non-necessary. Is possible to reduce constitutive rules in
regulative rules. The example through this could be analysed is the equilibrium in the Sorbat
valley (cfr. Evans-Pritchard). The river shoal and so now is possible for the Nuer and the
Dinka to enter in relation.
 The obvious solution sems to be to follow the old rule also if there is no more physical
constriction.
(I) Graze if North, do Not Graze if South (for the Nuer)
(II) Graze if South, do Not Graze if North (for the Dinka). This could be sad also
introducing the term « property *»:

[B] If a piece of land lies north of the river, then it is Nuer’s property.
[S] If a piece of land is Nuer’s property *, then the Nuer graze it.
[B]: basic rules
[S]: status rules

The same works for the Dinka. The idea is that BS=CR

[CR] If a piece of land lies north of the river, then it’s Nuer’s property*; and if it’s Nuer’s
property* then the Nuer graze it.

In Searle’s formula: If P, then X counts as Y [B]; and if Y then Z [S]

[S] specifies the actions Z associated with status Y.

P= a certain condition apply
X= piece of land
So with a basic logic trasformation it is possible to show that CR could be reduced at B and S
with the use of a conditional.
(A→B and B→C) → (A→C)

To make the argument clearer we could make an analogy.

A constitutive rule « is a theory of sorts »:

[CR] introduces a theoretical term (Y) naming an institution, makes explicit its conditions of
applicability (X, C, P), and the regulative rules Z) associated with the institution. But
property * does not add anything new, over, and above the
regulative rules (I) and (II).

So constitutive rules are not fundamental for the existence of institution. the deep nature of
institution could be reduced to simply regulative rules. We use many times institutional terms
and concept for economic thought reasons. It would be too complicated to say very time that
we speak about institution all the regulative rules directly entailed. As institutions gets more
complicated, we have to construct more complicated terms.

Lecture 23/11
 The reduction constructed in the previous lecture leaves us with the idea that regulative
rules make institutions. The rule approach to the ontology of institution is not new. There
is a line of thought starting from Weber that is based on this idea. Interesting in this line of
thought are the consideration of D. North:

“Institutions are the rules of the game of society or, more formally, the humanly devised constraints that
shape human interactions. [. . .] They are a guide to human interaction, so that when we wish to greet friends
on the street, drive an automobile, buy oranges, borrow money, form a business, bury our dead, or whatever,
we know (or can learn easily) how to perform these tasks.

This approach has an intuitive problem, the problem of ineffective rules. An idea
important idea of this approach is that rules are useful for cooperation, they are beneficial to
society. This problem leads us the make the distinction between nominal rules and effective
rules. Institution could be the rules that people have incentive to follow. Institution could be
only effective rules. Effective rules imply a situation of equilibria between the enforcers and
the enforced of a rule. In general, the rule approach let us a little unsatisfied.
There is another approach to institution, the equilibria approach. Equilibria are intended
in a game theoretic sense Equilibria are regularities in behaviour where none as an incentive
to change his behaviour. Institutions are «regularities in behaviour which are agreed to by all
members of a society» (Schotter, 1981). Out of equilibrium actions are unstable, hence
unlikely to become regularities in repeated games. In explanation how institution could be
seen as equilibria useful is the Evens-Pritchard example. The game could be represented by
the Howk-Dove game, the grazing game. A game with to equilibria. Immediately rise the
problem of equilibrium selection, it seams difficult that the Nuer and the Dinka could reach
an agreement between NG G or G NG. institution are the solution to the problem of
equilibrium selection. Institution adds to the game new conditional strategies:

(i) G if N (north), NG if S
(ii) G if S (south), NG if N.

This new equilibrium is clearly the solution of the game, is the efficient equilibria. This
idea of institution as creating new equilibria account for Serl’s idea that institutions have a
big creative power. Institution are not only equilibria.
This type of equilibria exists also in nature e.g., Specul wood butterfly. This type of
solution in animals seems to be genetically pre-determinate. Also, animals play cooperative
games the fundamental difference with us is that there is not new solution. If we were
naturally social, we would not need institutions. The problem that let us doubt about the
equilibria approach is the human uniqueness problem. The unique characteristic of humans is
the ability in finding new solution that allow new forms of coordination. Our creativity and
flexibility are our uniqueness. In nature there is only natural evolution in humans there is also
cultural evolution. We need institutions in virtue of our cognitive capacities.
Regulative rules could be correlated equilibria of coordination games. The solution of the
institution ontology is that we have to keep the two theories. From an external point of view
institutions seem to be equilibria and from the internal point of view they seem rules.

Guala, F. and Hindriks, F. (2015) “A Unified Social Ontology”, Philosophical

Quarterly 165 (2015): 177-201.

Definition of correlated equilibrium: e.g. the toss of the public coin that determine the
decision of the two tribes. A correlated equilibrium of a game G is a Nash equilibrium of a
larger game G∗ obtained augmenting G with the addition of new strategies. The conditional
prescription of the new strategy is about the occurrence of an event external to the game.

The payoff of the augmented grazing game are constructed considering that there is a 50%
probability that a land lies on the north and 50% that a land lies on the south of the river.

Why the river? Because of saliency cfr. Shelling.

Conventions are a Nash equilibrium of the augmented game but a correlated equilibrium of
the starting game.

The reduction of constitutive rules is regulative rules is functional to make possible for the
unified ontology theory and Serl’s theory to coexist.
The introduction of an institutional term give the possibility to translate constitutive rules in
regulative rules.

From conventions to norms: adding normative power. The attempt is to introduce deontic
powers in the unified theory. They could be considering as cost that modifies payoffs. But in
general, the authors suggestion is to be neutral about deontic powers.
There are many ways to represent deontic powers. One of these is by means of the delta
parameter. A cost that adds something to the payoffs of a cooperation problem in a
coordination problem. Normative powers change the game itself. In this way normative
power fulfilled two fundamental features of institutions: stabilize human behaviour make it
more predictable and constructing new equilibria. In general the unified theory could account
for different account of normativity.

Searle, J. R. (2015) “Status Functions and Institutional Facts: Reply to Hindriks and
Guala.” Journal of Institutional Economics 11: 507-514.

Hindriks, F., & Guala, F. (2015) "Understanding Institutions: Replies to Aoki, Binmore,
Hodgson, Searle, Smith, and Sugden". Journal of Institutional Economics, 11, 515-22.