Dans le cadre du Séminaire “Systèmes complexes en sciences sociales”, organisé par Jean-Pierre Nadal (référent principal), Henri Berestycki (EHESS) & Annick Vignes (ENPC & INRAE), nous accueillerons Joshua Epstein , Professeur invité à l’EHESS pour le mois d’octobre, pour une série de quatre exposés. Joshua Epstein est Professeur à NYU, directeur du NYU Laboratory on Agent-Based Modeling .
Les deux premières conférences de Joshua Epstein , Professeur invité à l’EHESS pour le mois d’octobre, seront finalement données ce vendredi 6 à l’ Institut des Systèmes Complexes , une le matin, l’autre l’après-midi :
10h-12h - Lecture 1: Generative Social Science and Agent_Zero
14h30-16h30 - Lecture 2: Agent_Zero and Cognitive Epidemiology
vendredi 6 octobre
Institut des Systèmes Complexes Paris Île-de-France (https://iscpif.fr/) :
113 rue Nationale, Paris 13è
métro Nationale (M6) ou Olympiades (M14), bus 83 (arrêt Olympiades) ou 27 (arrêt Clisson)
Première étage, Salle 1.1
Pour entrer à l’ISC, le plus simple est de composer le code 0506 sur le boîtier de gauche de la première porte (113 rue Nationale). Pouis composer le 0507 sur le gros boîtier quand vous êtes devant la seconde porte.
La salle 1.1 (salle du séminaire) se trouve au 1er étage.
Joshua Epstein
Professeur invité à l’EHESS
Department of Epidemiology, School of Global Public Health,
Director, the NYU Laboratory on Agent-Based Modeling,
New York University
Série de quatre exposés.
Lecture 1: Generative Social Science and Agent_Zero
Lecture 2: Agent_Zero and Cognitive Epidemiology
Lecture 3: The Epstein Civil Violence Model and Endogenous Inequality
Lecture 4: Inverse Generative Social Science: Backward to the Future
Série de conférences par Joshua Epstein ,
Professeur invité à l’EHESS
Department of Epidemiology, School of Global Public Health,
Director, the NYU Laboratory on Agent-Based Modeling,
New York University
Lecture 1: Generative Social Science and Agent_Zero
Abstract: A Formal Alternative to the Rational Actor. In the epistemology of generative social science, to explain a macroscopic pattern, it does not suffice to demonstrate that it is a Nash equilibrium. Rather one must show how the pattern could emerge on time scales of interest in a population of cognitively plausible agents. Despite numerous deep anomalies, the rational actor model dominates the social sciences for lack of explicit formal alternatives. Although minimal and provisional, Epstein’s Agent_Zero is one candidate. Based on cognitive neuroscience, Agent Zero’s behavior results from the interaction of an affective module, a boundedly rational deliberative module, and social interactions with other emotionally driven and statistically hobbled agents. The model generates important macro-phenomena and individual behavior in groups that violate Rational Choice Theory.
Lecture 2: Agent_Zero and Cognitive Epidemiology
Abstract: lassical mathematical epidemiology excludes behavioral adaptation, which has shaped pandemics from the 1918 Flu, to COVID. Endogenous multiple waves in particular are mathematically precluded by the classical differential equations. In Epstein’s Coupled Contagion framework, the disease and fear of the disease are both contagious and interact to produce endogenous waves. In Triple Contagion: A Two Fears Epidemic Model, there is also fear of the vaccine, resulting in richer dynamics, where waves grow in amplitude. An Agent_Zero version with fear, distrust of government, and psychic numbing produces spatio-temporal wave dynamics. Applications to financial panics and crashes are discussed.
Lecture 3: The Epstein Civil Violence Model and Endogenous Inequality
Abstract: Epstein will present his agent-based civil violence model, its counterintuitive results, several extensions, and the related Agent_Zero ‘parable’ of the Arab Spring. Economic hardship is central to the civil violence model, but is exogenous. An extremely elegant model of endogenous emergent inequality is the Boltzmann Economy. Following its MAXENT derivation, we show that the equilibrium Power Law distribution in not robust to minor redistributive departures from the trade rule, suggesting that, as equilibria, power law distributions of wealth, while ubiquitous, may be quite fragile.
Lecture 4: Inverse Generative Social Science: Backward to the Future
Abstract: The agent-based model is the principal scientific instrument of generative social science. Typically, we design completed agents—fully endowed with rules and parameters–to grow macroscopic target patterns from the bottom up. Inverse generative science (iGSS) stands this approach on its head: Rather than handcrafting completed agents to grow a target—the forward problem—we start with the macro-target and evolve micro-agents that generate it, stipulating only primitive agent-rule constituents and permissible combinators. Concrete examples of iGSS and outstanding foundational issues surrounding it are discussed. An important goal of iGSS is to evolve cognitively plausible formal alternatives to the Rational Actor, with Agent_Zero as one possible point of evolutionary departure.
http://cams.ehess.fr/systemes-complexes-en-sciences-sociales/