Cognition and Self-Organization: The Role of MAS

Minnesota AAMAS AAMAS 2013 - Castelfranchi

InMind and OutMind InMind and OutMind Societal OrderSocietal Order

Cognition & Self-Organization: The role of MAS

_____________________

Cristiano Castelfranchi

Institute for Cognitive Sciences and Technologies - Roma


Premise and Issues

““Socio-(Cognitive-)Technical Systems”Socio-(Cognitive-)Technical Systems”

What we are unavoidably building with computer networks, AI, and Ag technologies are Socio-Cognitive-Technical Systems:

Socio-Technical System in fact means that any new technology implies/requires/introduces not only new skills and competences, but new expectations, goals, beliefs; new "scripts", with their roles, norms; new form of interaction and conventions among the social actors. So we have to specify the "cognitive" and interactive side of the new system.

You are Social EngineersSocial Engineers; are you aware of that?


Premise and Issues

““Socio-Cognitive-Technical Systems”Socio-Cognitive-Technical Systems”

BUT….. this new complex Socio-Technical (and mental) System

cannot be just planned and designedcannot be just planned and designed.

It is dynamically emerging and self-organizing: it is a spontaneous Social Order spontaneous Social Order (von Hayek); a dynamic equilibrium not necessarily "good" for the goals of the actors.

What we need is not just a top-down organization and control.


Premise and Issues

““Socio-Cognitive-Technical Systems”Socio-Cognitive-Technical Systems”

in orderto support and mediate human interaction and organization and to emulate them in efficient open MAS systems, we have to (partially) "understand" and to reproduce features of human social mind (like commitments, norms, mind reading, power, trust, "institutional" effects, ...) and of social macro-phenomena.

In particular we have to model "immergence" "immergence" and "cognitive cognitive emergence"emergence"; the mental "mediators" of societal phenomena and the partial understanding and awareness in the actors


Our general PERSPECTIVE - 1

The “Cognitive Mediators”“Cognitive Mediators” of Social Phenomena

Social and cultural phenomena cannot be deeply accounted for without explaining how they work through the individual agents’ minds (mental “counterparts” or “mediators”).

This requires a richer cognitive model (architecture) for “Agents,” moving from formal and computational AI and ALife models, closer to those developed in psychology, cognitive science, and in cognitive approaches in economics, sociology, organization studies.

“The most important fact concerning human interactions is that these events are psychologically represented in each of the participants”

(Kurt Lewin, 1935)


““COGNITIVIZING”COGNITIVIZING”

Cooperation, Conflict, PowerPower,, Social ‘Values’, Commitments, Norms, Rights, Social Order,Rights, Social Order,

Trust, , …- AIJ paper- AIJ paper- Social Commitment paper- Social Commitment paper____________________________________________________________

Von Hayek, Pareto, Garfinkel, … the aim of Von Hayek, Pareto, Garfinkel, … the aim of founding the Social Sciences as Autonomous founding the Social Sciences as Autonomous from Psychologyfrom Psychology





Social phenomena are due to the agents’ behaviors, but…but…the agents’ behaviors are due the the mental mechanisms controlling and (re)producing them. (Castelfranchi, Conte, Miceli, Falcone,…)

For example:

My Social Power Social Power lies in, consists of, the others’ Goals & Beliefs!!!!

That’s why we need Mind Reading! Not for adjusting ourself, but for manipulating and exploiting the others or for helping or punishing them.




Social phenomena are due to the agents’ behaviors, but…but…the agents’ behaviors are due the the mental mechanisms controlling and (re)producing them. (Castelfranchi, Conte, Miceli, Falcone,…)

For example:

• How the normnorm should work through the minds of the agents? How is it “represented”?

? Which are the proximate mechanisms underlying the normative behavior?


Our general PERSPECTIVE – 2

However,

Mind is not enoughMind is not enough!!

the “individualistic + cognitive” approach is not sufficient for the social theory and processes (even when modeling joint and collective attitudes and actions).

The social actors do social actors do notnot understand, negotiate, understand, negotiate, and plan and plan for all their collective behavior and cooperative activity.


Premise and Issues““Socio-Cognitive-Technical Systems”Socio-Cognitive-Technical Systems”

We have to "understand" and to reproduce also

hhow humans do socially construct something ow humans do socially construct something

without understanding without understanding itit! !

How is it possible that intentional agents do not intend the functions of their collective behavior?

Which the relationship between emergent functions and intended goals?

CAN WE SUPPORT HUMAN ORGANIZATIONS & BUILD EFFECTIVE SOCIAL SYSTEMS WITHOUT UNDERSTANDING AND GOVERNING THAT!?


Mind is not enough Mind is not enough Emergence, Self-Organization

Functions and Cognitions

1


INDIVIDUAL MIND

COLLECTIVE STRUCTURES

& BEHAVIOURS

Bel --> G --> action

Mind is not enough

emergenceemergence & immergenceimmergence

not only knowledge, mutual beliefs, reasoning,shared goals and

deliberately constructed social structures and cooperation


Agents in a common world (INTERFERENCE)

r

or

q

q p

q

andp

G

objectiveDEPENDENCEnetwork

Cognitive Cognitive emergenceemergence:awareness

Levels ofemergence...


For a (Pessimistic) Theory of

Spontaneous Social Spontaneous Social OrderOrder

A critical homage to F. von Hayek


I will examine:

• the crucial relationships between the intentional nature of the agents' actions and their explicit goals and preferences, and the possibly

unintended 'finality' or 'function' of their behavior.

• in favor of 'cognitive architectures' in computer simulations.

• propose some solutions about the theoretical and functional relationships between agents' intentions and non-intentional 'purposes' of their actions.

• 'Social order' is not necessarily a real 'order' or something good and desirable for the involved agents; nor necessarily the best possible solution.

• It can be bad for the social actors against their intentions and welfare although emerging from their choices and being stable and self-maintaining. How to TRUST it?

• Hayek's theory of spontaneous social 'order' and Elster's opposition between intentional explanation and functional one will be criticized.


Agent-based Modeling & Social Simulation will be crucial for the solution of one of the most hard theoreticaltheoretical problems of economic and social sciences:

the spontaneous organization of a “dynamic social the spontaneous organization of a “dynamic social order” that cannot be planned, but emerges out of order” that cannot be planned, but emerges out of intentional planning agentsintentional planning agents guided by their own guided by their own choices.choices.

This is the problem that Hayek assumes to be the real reason for the existence of the Social Sciences.


SOCIALSCIENCES

A

B

MULTI-AGENT

SYSTEMS

B: Not only an ‘experimental method’ and experimental platforms

CONCEPTS, MODELS, THEORIES

The new COMPUTATIONAL SOCIAL SCIENCES


Agents & MASAgents & MAS paradigm

Just a Technology?I stress those aspects still needing a theory (that we cannot just buy and import from the cognitive and social sciences) also because it is important not reducing “Agents” (and MAS) to a technology. This is not only an impoverishing move but even a risky move (remember the serious mistake that AI did with the “expert systems”).

““Agents”Agents” are an intellectual (and formal-computational) framework; a way of thinkinga way of thinking and of analyzing dynamic and complex phenomena that involve active, partially independent, distributed but interfering and interactive entities, producing common (either planned or unplanned) collective results, for individual or collective advantages.


Only MAS can fully deal with this problemOnly MAS can fully deal with this problem>> Up & Down>> Up & Down


“THE core theoretical problem of the whole social science”

(Hayek )


“THE core theoretical problem of the whole social science” (Hayek )

"This problem (the spontaneous emergence of an unintentional "This problem (the spontaneous emergence of an unintentional social order and institutions) is in no way specific of the social order and institutions) is in no way specific of the economic science.... it doubtless is economic science.... it doubtless is THE core theoretical THE core theoretical problem of the whole social scienceproblem of the whole social science" " (von Hayek, Knowledge, Market, Planning)

the problem is not simply how a given equilibrium or coherence is achieved and some stable some stable orderorder emerges emerges

To have a "social order" or an "institution", spontaneous emergence and

equilibrium are not enough. They must be "functional".


Adam Smith’s "invisible hand"

Adam Smith’s original formulation of “THE problem” is much deeper and clearer

The great question is how:

"(the individual) - that does neither, in general"(the individual) - that does neither, in general, intend to , intend to pursuepursue the the public interestpublic interest, , nor is aware nor is aware of the fact that he is of the fact that he is pursuing it,... pursuing it,... is conduced by an invisible hand to is conduced by an invisible hand to pursue an pursue an endend that is not among his that is not among his intentionsintentions"" (Smith, ). (Smith, ).

Hayek like Smith in acknowledging the teleological nature of the invisible hand and of spontaneous order, cannot avoid attributing to it

a (positive) value judgment, a providential, benevolent, a (positive) value judgment, a providential, benevolent, optimistic visionoptimistic vision of this process of self-organization of this process of self-organization ((ideologismideologism).).


In the “Invisible Hand”:

1) there are intentions and intentional behavior

2) some unintended and unaware (long term or complex) effect emerges from this behavior

3) but it is not just an effect, it is an end we “pursue”, i.e. its orients and controls -in some way- our behavior: we "necessarily operate for" that result (Smith).

- how is it possible that we pursuepursue something that is not an intention of ours; that the behavior of an intentional and planning agent be goal-oriented, finalistic (‘end’‘end’), without being intentional;

- in which sense the unintentional effect of our behavior is an "endend”??


Theory of “Function”Theory of “Function”

This problem appeared in other social sciences as the problem of the notion of "functions" (social and biological) impinging on the behavior of anticipatory and intentional agents, and of their relations with their "intentions".


Social Functions Social Functions and CognitionCognition

a) no theory of social functions is possible and tenable without clearly solving this problem;

b)without a theory of emerging functions among cognitive agents social behavior cannot be fully explained.

Moreover: we have to we have to buildbuild social functions social functions and and spontaneous ordersspontaneous orders (conventions, conformity, (conventions, conformity, …) in Agent supported human organizations and …) in Agent supported human organizations and in open MASin open MAS

not only good not only good intentionallyintentionally cooperating/competing systems cooperating/competing systems



Functions install and maintain themselves Functions install and maintain themselves parasitical to to cognition: cognition:

functions install and maintain themselves thanks to and through agents' mental representations

but not as mental representations: i.e. without being known or at least

intended.



While Social NormsSocial Norms emergence and functioning require also a (partial) "cognitive emergencecognitive emergence",

Social FunctionsSocial Functions require an extra-cognitiveextra-cognitive emergence and working

For a Social Norm to work as a Social Norm and be fully effective, agents should recognize and treat it as a Social Norm.

On the contrary the effectiveness of a Social Function is

independent of agents' understandingindependent of agents' understanding of this function of their own behavior:

a) the function can rise and maintain itself without the awareness of the agents;

b) if the agents intend the results of their behavior, these would no more be mere "social functions" of their behavior, but just "intentions".


The problem: Emergence and Functions should not be

what the likes or notices,

(“just in the eye of the beholder”)

but should be indeed observer-independent, based on self-organizing and self-reproducing phenomena, >>> "positive”, “good” can just consists in this.

We cannot exclude "negative functions" (Merton) (kako-functions) from the theory: perhaps the same mechanisms are responsible for both positive and negative functions.

>> Two kinds of finalistic notionsfinalistic notions:

- evolutionary finalities, adaptive goals; and

- mental ends (motives, purposes, intentions).


Intentional Intentional behavior Vs. functional functional behavior

Finalistic systems:

There are twotwo basic types of system having a finalistic (teleonomic) behaviour:

Goal-oriented systems - (Mc Farland, 1983),

Goal-governed systems

a specific type of Goal-oriented system based on representations representations that anticipate the results


MAIN PROBLEMSMAIN PROBLEMS

• If a behavior is reproduced thanks to its good effects, that are good relatively to the goals of the agent (individual or collective) who reproduces them by acting intentionally, there is no room for "functions” (Elster).

If the agent appreciates the goodness of these effects and the action is replied in order to reproduce these effects, they are simply "intended".

• How is it possible that a system which act intentionally and on the basis of the evaluation of the effects relative to its internal goals reproduces bad habits thanks to their bad effects?


>> ?? a behavioristic reinforcement layer (van Parijs) together with

>> a deliberative layer (controlled by beliefs and goals) ???

the deliberative layer accounting for intentional actions and effects,

the behavioristic layer (exploiting conditioned or unconditioned reflexes) accounting for merely "functional" behaviors??

Are “functions” and “roles” just impinging on ‘habitus’ ???(Bourdieu), while

intentions would just be for personal purposes??

Our problem is indeed that:

intentional actions intentional actions have functionsfunctions!

Goals and beliefs of the agents have functions.


The fundamental problem is how to graft teleological but unintentional behaviours precisely on intention-driven behaviours.

[WE HAVE TO BUILD THIS KIND OF MAS-BASED THIS KIND OF MAS-BASED SYSTEMSSYSTEMS]

What answer can be given to Elster according to whom the idea of intention makes that of the function of behaviour impracticable and superfluous.

How can intentional acts also be functional, How can intentional acts also be functional, that is,that is, unwitting but unwitting but reproducedreproduced precisely as a result of their precisely as a result of their unintentionalunintentional effectseffects.


Why alsoKako-functions?Kako-functions?

How is it possible?


Why also kako-functionskako-functions?

- thethe mechanism that install a bad function can be exactly mechanism that install a bad function can be exactly the same installing a good onethe same installing a good one

- to definitely separate a functional view of behavior and society from any teleological, providential view (functions can be very bad and persist although bad)

- kako-functions cannot be explained in a strictly behavioristic framework of reinforcement learning: the result of the behavior can be disagreeable or useless, but the behavior will be "reinforced", consolidated and reproduced.


Unexpected evil effects exist, or evil effects combined with good individual intentions (Boudon, 1977) in which

the intended good effects reproduced the intended good effects reproduced in spite ofin spite of the negative consequences the negative consequences.

This is true, - both in the case in which the evil effects are not perceived or are not

attributed correctly, - and in the case in which they are perceived

(in the second case the good effects must be subjectively more important and in any case preferred (for instance, be closer in time), or else are more conditioning/reinforcing than the evil effects)


But there are also

harmful effects capable of self-reproductionharmful effects capable of self-reproduction (through the action) precisely because of their negative naturebecause of their negative nature (Castelfranchi, 1997; 1998b; 1998d).

a long line of automobiles and the slowing down due to the simple individual intention of rapidly glancing at an accident that has occurred in the other lane


The notion of ‘function' as

an effect selecting and reproducingan effect selecting and reproducing its own causeits own cause

How is it possible for a system that acts intentionally on the basis of an evaluation of the effects vis-à-vis its own goals, to reproduce bad habits precisely as a result of their bad effects?

And even more crucially - if a behaviour is instead reproduced thanks to its good effects with respect to the (individual or collective) goals of the agent who reproduces them by acting intentionally, then there is no room for the "functions".


It is necessary to have complex reinforcement learning forms not merely based on classifiers, rules, associations, motor sequences, etc. but operating on the cognitive representationsoperating on the cognitive representations governing the action, that is, on beliefs beliefs and goalsgoals.

In this view "the consequences of the action, which may be more or less consciously anticipated, nevertheless modify the probability of the action being repeated the next time in similar stimulus conditions " (Macy, 1998). More exactly:

the functions are simply effects of behaviour which the functions are simply effects of behaviour which go go beyond thebeyond the intended effectsintended effects but which can but which can successfully be reproduced because they reinforce successfully be reproduced because they reinforce the agent's beliefs and goals that give rise to this the agent's beliefs and goals that give rise to this behaviour.behaviour.


How Social Functions are implemented through

cognitive representations

The basic model


There are two Cognitive "reinforcement" principles:

1. Belief Reinforcement: two different mechanisms can be postulated:

association (accessibility) : the association between the belief and that context or scenario is strengthened: the believe will have more probability to be retrieved next time in similar situations; it will be more activated, more available and accessible (accessibility bias);

confirmation (reliability) :some of the action's effects are perceived by the agent (even if not necessarily understood and causally connected to its actions) and they confirm the beliefs supporting the action: they give new evidence for that belief, increase its "credibility", and reliability: they augment its “truth” or the subjective probability of the event.


2. Goal Reinforcementtwo different mechanisms can be postulated (analogous to the beliefs reinforcement mechanisms):

association (accessibility) : the success of the chosen goal, plan, action is memorized in the sense that the association between the goal-plan and that problematic context or scenario is strengthened: the goal/plan (solution) will have more probability to be retrieved next time in similar situations; it will be more activated, more available and accessible;

confirmation (reliability) :the success of the chosen goal, plan, action is memorized; it increments a "successfulness index" relative to that choice; or better some meta-cognitive evaluation of the value of the action. This memorized behavioral choice is "confirmed": next time the probability to choose the same way (goal, plan, strategy, action) will be greater: it will be more preferable and reliable (we will trust more it).

The reinforcement of both the belief and the goal/plan will determine a reinforcement of that behavior


An example: dirty and clean streets

A social (kako)function based on social conformity and imitation.

The agent assumes (B1) that this is a bad behavior or even a forbidden one; he assumes (B2) that a lot of other people behave this way; that (B3) this can be quite practical and easy sometimes; he assumes that (B4) his contribution to the garbage is quite marginal and small (that its true). He has the goal (G1) to do as others do and until others do so (Bicchieri, 1989); or at least, to do as others do and until others do so if this is useful and practical for his goals. Goal G1 on the bases of beliefs B2, B3, B4 will generate a goal (G2) to leave small garbage in the street, which overcomes the possible goal (G3) -based on B1- of not dirtying the city. Now the result of such a behavior is that streets are dirtier; this is perceived and then it will confirm the supporting beliefs (B2, B4) and the goal G2.

Everybody reinforces the behavior of the others. The global effect is not wanted and intended by anybody; the

reinforcement effect is also unattended and unintended.

The behavior is (reciprocally) reinforced by its effects. These effects are self-maintaining and reproducing through the reinforcement of their own causes. This passes through the mind of the agents (their beliefs and goals) but not through their consciousness and intention.

It is quite interesting to observe that exactly the same kind of beliefs, and an identical goal (G1) can generate in this case an eu-function: to maintain the city clean.


Interpersonal and collective level: vicious circlesvicious circles

Hubbub in a restaurant or at a party

(The example given is merely -on a small scale- the model followed by the arms race).

Hostility leads to hostility


Institutional level: vicious circlesvicious circles

Prisons <==><==> Delinquency


What distinguishes function from non function is not not that the unintentional (collective) effect is good but that it is self-organizing and self-producing by means of positive feedback, that is, by reinforcing, selecting, and reproducing the behaviour that generated it:

unintended effects that select their own causesunintended effects that select their own causes.

>> functionality must be kept distinct from goodness (and that is from the subjective goals of the agents),

>> good and bad functions (exactly like unintended good and bad effects) are on the same plane: both may be self-organizing.

>> the function is not reproduced or maintained or repeated by virtue of its good effects (a risky approach owing to the boundary with intention)


Goals Goals vs. “Functions” “Functions”

An example:

The ADOPTION of a NORMADOPTION of a NORM

without the UNDERSTANDINGUNDERSTANDING (sharing) of its END


Delegation, Proactivity,Delegation, Proactivity,

Order, Norms, ViolationOrder, Norms, Violation

3


“We have to maintain the control”

We can delegate “how” to achieve a given goal, but not “what”: goal decision and choice should remains our power (a general worry; ex. Müller-Schloer)

However…. There are different kinds and level of “delegation” useful for cooperation, and autonomy, and iniciative or pro-activity.

Over-helpOver-helpNot always our objective is the right one (ex. expert systems’ advices).

Not always we understand our own interest.


Goal

action

subGoalsubGoal

PLAN

Des.

action action

Levels ofLevels of Delegation Delegation

Closed-Delegation:

mere execution

A plan-based approach


Goal

action

subGoalsubGoal

PLAN

Des.

action action

Goal

action

subGoal

?

Open-DelegationOpen-Delegation:

• Delegation of meta-tasks (searching, planning, deciding, monitoring, ...)

• Intelligence, autonomy (agent modelling)

Levels ofLevels of Delegation Delegation


Social-AgentSocial-Agent’’s Architectures Architecture and Multiple Goal-Multiple Goal-

SourcesSources

ACTIVEGOALS

ACTIONS

INTENTIONS

ADOPTEDADOPTEDGOALSGOALS

EMOTIONS

BODILYNEEDS

DESIRESDESIRES

PRACTICALREASONING

EXOGENOUS GOALS


non shared plannon shared plan Goal

action

subGoal subGoal

action action

subGoal

GoalPLAN

action

subGoal

PLAN Goal

action

subGoal subGoal

action

Unaware CooperationUnaware Cooperation

•Reconciling EmergenceEmergence and CognitionCognition


Goal

action actionaction

subGoalsubGoal

PLAN

Des.

Goal-DelegationGoal-DelegationVs Goal-AdoptionGoal-Adoption

• Over- HelpOver- Help• Critical-HelpCritical-Help

Levels of collaboration


Degrees of AutonomyDegrees of Autonomy

There are degrees of Autonomy, but also domains of Autonomy;And we also need “adjustable” “adjustable” autonomyautonomy

Possibly based on “negotiation”: both sides (delegator and delegated) might ask for a broader or more limited autonomy and initiative; and for more or less control.


Monitoring PeopleMonitoring People(and Autonomous Agents)?

We cannot monitor and control people

without understanding without understanding “what” they are doing (goal) and “why”,

and why sometime they have to have to violate violate norms or roles

? the possible danger of a computer-based formalization and enforcement of rules in organizations


Functional violations of norms and commitments

For "functional disregard" I mean

The violation of a norm/rule/prescription/commitment finalized to make the required/expected work (also) more functional to the organisation's aims (over- and critical-help). (““functionalfunctional”” here means here means ““good for ..good for ..”!)

“Finalized” can mean either “intentional”(goal-governed) or “functional” (goal-oriented)



Two different forms of “functional disregard or violation”:

(““functionalfunctional”” here means here means ““good for ..good for ..”!)

Deliberated functional violationsDeliberated functional violationsIn order to be deliberated this behaviour has

to be free (one could have done differently), and aware of its consequences and on purpose, thus the agent cannot be merely self-interested.

• Deliberated is not only the violation but also its functionality to the organisation

I focus on felicitous deliberated deliberated violations (first kind)



The Sociological Description and Definition of the Phenomenon

"The repeated and systematic violation of criteria, prescriptions and dispositions included in the normative apparatus that defines and regulates (through "norms", in a broad sense) a collective organised activity. This violation favours an acceptable level of functioning of the organisation itself, while the literal and punctual application of the normative will produce levels of functioning either unacceptable or less efficacious" (Manciulli et al., 1986, p. XI).

"grève du zéle" - "forms of sabotage consisting just in a rigorous application of rulements" (Mottez, 1971).



" it is well known that the best way for sabotaging an organisation is that of literally obeying to all its rules and to refusing to use our own judgement capacity. Beyond what is obtainable by commands, beyond what is controllable by supervision, beyond what is inducible by incentives or preventable by punishment, even in the execution of the more humble jobs there is a bit of discretionality... This "discretion" can be used both to This "discretion" can be used both to allow or to subvert the aims of the organisation"allow or to subvert the aims of the organisation". . (Bendix, 1959/72).

WILL WE WILL WE ““SABOTAGESABOTAGE”” ORGANIZATIONs ORGANIZATIONs

via our COMPUTATIONAL CONTROL?via our COMPUTATIONAL CONTROL?

During the Tayloristic period in the industrial organisation one of the main points in the dominant paradigm of the "scientific management" was the monito to the worker: "you are not paid for thinking, but just for executing orders and instructions".



Social conventions to violateSocial conventions to violate

Fly controllers in an airport control tower should use the interphone , also in communication among people within the same control room. There is a rule on this. However, the interphone has some inconvenience (for ex. voices overlaps, misunderstandings, ... ) and the controllers consider its use not only less natural for them, but “potentially safety critical”. This is why they in fact do not use interphone for messages within the same room, and communicate “directly”.

In this real example we clearly have:

a systematic violation for functional, cooperative reasons


Trusting Trusting & Delegating & Delegating

without without

UnderstandingUnderstanding

4


The a-technical, non-rational nature of the deontic “ought”deontic “ought”

The “alienated” “alienated” nature of norm adoption

&


The a-technical, non-rational nature of the deontic “ought”

In a sense the deontic “ought” “have to” is a de-technicalized “ought”: no longer a necessary means means for.... for.... something that you have to want, to chose.you have to want, to chose. “If you like/want to ..., you should, have to.....” (technical ought)“You have to” for what? why? (deontic ought) In the mind of the “issuer” the N is supposed to be a means, a solution for some problem, a way of achieving a higher goal.


Citizens like childrens

We are not supposed to ‘cooperate’ but to ‘obey’!


The a-technical, non-rational nature of the deontic “ought”the deontic “ought”

Both the ‘ideal’ and the ‘sub-ideal’ (for avoiding sanctions) obedience share a fundamental core, crucial for the real nature of the deontic ‘norm’, ‘ought’.

A core that differentiate the mind of the normative ‘subject’ S from the mind of the ‘issuer’ or legislator.S is NOT supposed to (have to) adopt the N (to ‘adhere’ to the imperative) because s/he understands or agrees about its funtion, aim.

On the contrary, S is supposed to have to obey even if s/he to obey even if s/he does not understand the meaning of the N, or disagrees does not understand the meaning of the N, or disagrees about it.about it.

A normative education is precisely an education to obey in any case, and even to not wondering and worry about the validity of the N. (That’s why Socrates takes the poison)


The “alienated” “alienated” nature of norm adoption and on

institution and social powers

We do not intend the ends of our intentional behavior!


Multi-Agent SystemsMulti-Agent Systems&

Emergence, Self-OrganizationFunctions and Cognitions

5


Not only the MAS supported social order cannot just cannot just be top-downbe top-down, but it must allow autonomy and flexibility, and true delegations and commitments, and "conventions" and "norms" (top-down & bottom-up) with their intrinsic possible “violation”.

We need cooperation/support by “autonomous”“autonomous”, “pro-“pro-active”active”, really helping , really helping Agents.

Not stupid executors.


Compliance to norms and commitments is assumed to be and to have to be absolutely guaranteed in E-institutions. Why?

N-violation (commitments, rules, roles, ...) can be adaptive, functional. But to really be adaptive should not be just accidental; it should be based on some "understanding" and giving-priority to the "goal" of that norm or task or request or prescription. It should be based on some Over-help.

That’s why we need “Trust”That’s why we need “Trust”

However, how to support "trust" within those self-organizing, changing, and normatively not-rigid system?


Only MAS can solve this theoretical and technical problems; by changing the Social Sciences methods, data, and modeling tools. By modeling Emergence Emergence but also Immergence Immergence and Cognitive Cognitive EmergenceEmergence, and different levels and kinds of order; how minds become coordination artifacts; different kinds and levels of cooperation. And in particular the relation between mental representations and learning and self-organizing autonomous systems.

Can we Program with the "Invisible Hand"? Will the "Invisible Hand" - governing human society - be implemented in the emergent intelligence of Self-Self-organizing open MA systemsorganizing open MA systems?

How to combine Control & Self-Organization?


MAS will play several crucial roles:

- Experimental methods and platforms for the Social and political sciences;- New conceptual, theoretical and modeling tools;- Infrastructure for Self-organizing but monitored and governed organizations: MA implementation, Ag mediation and support;- Participation, transparence, information; mediamorphosis of knowledge institutions;- Simulation for understanding, predicting and deciding about real social dynamics;- MA coordination/interpretation of feedbacks from intelligent sensors, stakeholders, people, ... to authorities and control institutions.

Let me stress just one:

Artificial Social ImaginationImagination“The best way to predict the future .. is to invent it” Allan McKay


Artificial Social “Imagination” like human imagination/simulation guiding both human executive actions and long-term decisions.

Anticipatory Simulation in action control: ex.

Comparison between purely stimulus-response systems (a) and those endowed with anticipatory capabilities, which run an ‘internal looprun an ‘internal loop’ on-line with action (b), or off-line (c) (Pezzulo).


Social Simulation, Social

Innovation, and

System Management

6


Governement and Simulation

Social Simulation will be Agent-based Agent-based and will be crucial for planning and governance of complex systems.

The most serious problem of the Social “Sciences” is that they cannot do real experiments.

This is even more dramatic for Social Policies! Simulation provides this.


… in fact, learning (and adjustment) requires trials and errorserrors! You cannot do this on purpose in real context and people. But we can systematically explore “what “what would happen if….” would happen if….” in the computational model of the given system.

Even more simple for/in those (future) systems and organizations that will be fully networked and based on agents. Their working not just they simulation!

Governement and Simulation


“Simulation” is so important and crucial because it finally provides to the social sciences a truly “experimental” method, for the validation and adjustment of the models (and of working “architectures” not simply “formal”).

But:“Simulation” is more than “experimental method”; it is a modeling method (synthetic approach; proximate mechanisms

The challenge is much more serious: to develop a new form a new form of collective intelligenceof collective intelligence; of imagination, prediction, experimental design and planning, finally able to support collective strategic decisions and policies.

76

The SimulationSimulation Revolution

Minnesota AAMAS AAMAS 2013 - Castelfranchi77

political issues and challengespolitical issues and challenges

What will happen in a few decades is the systematic use of Simulations as the basis for any big decision about strategies and policies in any domain: military, environmental, financial, economic, urban, demographic, energetic, educational, logistic, sanitary,....

No political/managerial complex decision will be taken without a grounded prevision prevision about possible effects and trends; thanks to the modeling and simulation of the phenomenon.


given the power of future computational scientific models and platforms, and

given a society fully connected in real-time, on-line, reactive and ‘virtualized’, with “smart environments and ambient intelligence”,

with continuous simulation and re-simulation of what is happening,

there would be the cognitive collective power the cognitive collective power and intelligenceand intelligence for better governing future and complexity



A fundamental challenge for the future is - in my view - the following one:

"How" to systematically integrate -the simulationssimulations of social phenomena with -the real-time feedbackreal-time feedback from the "playground” dynamic processes?



A "cyclic" model

Planning/design may not just remain Planning/design may not just remain top-down top-down and and centralizedcentralized, and be just "applied" "applied" to the field.

One will need to combine:- simulative models and their running and predictions

and the consequent decisions and choices, with possible timely feedbacks from the territory, due to intelligent sensors or witnesses; then

- run time readjusting the simulations and then the policy or intervention; and

- so on, cyclically (Figure).



A "cyclic" model


ENVIRONMENT


Information and computer technology gives us the tools and the cultural conditions for those run time, decentralized, high qualified feedbacksrun time, decentralized, high qualified feedbacks.

Different kinds of "agents" might be exploited: (i) local sensors and detectors, of various kinds (possibly learning and adjusting), even interconnected and locally "cooperating" for a more reliable and global information; (ii) local computational elaborations and intelligences, for primary elaboration and reaction (additional data, additional "agents", alarm smart systems, ...); (iii) active voluntary people communicating via sms, call centers, ... (iv) local or domain "groups", “communities”, associations interested in monitoring and giving advice on the phenomenon: bloggers, consumers unions, ..;Elaboration of big-data, sentiment analysis, specific alert, …. (v) local professionals in charge: policemen, bombers, medical staff, teachers, social workers, specialized observatories (economic, demographic, environmental,..), ...... (vi) local administrations, with their "research/documentation centers" or their managers and administrators;...



Obviously, the solution (and the design) cannot be the same in every domain.

For example, in traffic monitoring and planning (for instance, in view of a transportation strike or of a big political demonstration) this is quite trivial: one should design a fast interaction between the simulated intervention "plans" (with their predictions) and the various feedbacks from the actual event: indications from sensors; telephonic advices from people; advices from policemen and toll booths; recommendations and requests from users' blogs, or consumers’ associations, or local authorities.

Rather different feedbacks are needed in case of an epidemic, or for the leakage of toxic substances.



Much less simple is the case of social policies and interventions on population welfare ..:

Who are the stakeholders to privilege stakeholders to privilege and which role/weight should they be given? Which (objective or subjective) wellbeing indicators should be chosen? And which conflicting “interests”? (Friedman’s talk)Which instruments should be adopted for the surveys? (expert evaluations? interviews? groups? ......).

The decision and re-planning is not technical; it is politicalThe decision and re-planning is not technical; it is political! And requires the right subjects and participation and negotiation.



To play such a role however computer Social Simulation must become stronger and much more reliable. Not biased by internal arbitrary parameters, that might produce whatever result we want, guided by the values of a given group of interest.

And perhaps we even will need a “participatory” simulation (like in scientific work) with different stakeholders providing and comparing their results.

85

The SimulationSimulation Revolution


The computer modeling and simulation is a possible revolution of the

"collective" cognitio"collective" cognitionn:

its “anticipatory” and “imagination” power

… or such anan unbelievable cognitive powerunbelievable cognitive power should be only at disposal of the army, of the “big brother”, or of the word corporations??



Concluding Remarks Concluding Remarks


??


Can We……?

Can we engineering and manage our society (in part) preserving

self-organization self-organization and

individuality and (semi-)autonomy (semi-)autonomy of peopleof people?


the new synthesisnew synthesis


• Reconciling EmergenceEmergence and CognitionCognition

1) Cognition as a level of emergence:

- from objective to subjective- from implicit to explicit ex. dependence

2) Beyond cognition: emergent unaware, functional social phenomena

(ex. cooperation, problem solving) among cognitive agents

the new synthesisnew synthesis



Social and cultural phenomena cannot be deeply accounted for without explaining how they work through the individual agents’ minds (mental “counterparts” or “mediators”).

This requires a richer cognitive model (architecture) for “Agents,” a richer cognitive model (architecture) for “Agents,” moving from formal and computational AI and ALife models, closer to those developed in psychology, cognitive science, and in cognitive approaches in economics, sociology, organization studies.


WILL WILL THIS “COGNITIVE MEDIATED” VIEW THIS “COGNITIVE MEDIATED” VIEW of social of social phenomena ad dynamics, and of Agent-based modeling phenomena ad dynamics, and of Agent-based modeling WIN??WIN??

Not so sure at all: short cut of statistics, big data, correlations, probability, … Ex. of current robust trend

PREDICTING WITHOUT UNDERSTANDING, without PREDICTING WITHOUT UNDERSTANDING, without modeling the modeling the ““proximate causesproximate causes””.

:


“AlienationAlienation”:

We are dominated by our own ‘delegated’ (emergent) (social and intellectual) powers, we are not aware of;

we do not realize and we do not decide/intend to create such collective or such institutional powers impinging on us.


Can We Overcome our AlienationAlienation?

Will the Leviathan become

a giant connected and informed connected and informed

community of agentscommunity of agents,

managing their collective Power?

1. I’m skeptical about that (also for cognitive reasons)

2. I also worry about possible net-Demagogy


Engineering a new SocietyEngineering a new Society?

Are we aware that we are in fact building, engineering a new SOCIETY?

A society where we shape H-H, C-C (Ag-Ag), H-C & C-H interactions and organizations,

with their mental and behavioral counterparts.


ENDENDThank you for your attention!

And sorry for my “English”

Cognition and Self-Organization: The Role of MAS

Technology

cognitive sciences

minnesota aamasaamas

social theory

social macrophenomena

social actors dosocial

technical systemswhat

social powersocial power

technical systemsbut