Choice under social influence: effects of learning behaviors on the collective dynamics
1
Departamento de Economía [Buenos Aires]
2 TIMC-IMAG - Techniques de l'Ingénierie Médicale et de la Complexité - Informatique, Mathématiques et Applications, Grenoble - UMR 5525
3 AMA - Analyse de données, Modélisation et Apprentissage automatique [Grenoble]
4 CAMS - Centre d'Analyse et de Mathématique sociales
5 LPS - Laboratoire de Physique Statistique de l'ENS
6 GEMAS - Groupe d'étude des méthodes de l'analyse sociologique
2 TIMC-IMAG - Techniques de l'Ingénierie Médicale et de la Complexité - Informatique, Mathématiques et Applications, Grenoble - UMR 5525
3 AMA - Analyse de données, Modélisation et Apprentissage automatique [Grenoble]
4 CAMS - Centre d'Analyse et de Mathématique sociales
5 LPS - Laboratoire de Physique Statistique de l'ENS
6 GEMAS - Groupe d'étude des méthodes de l'analyse sociologique
Mirta B. Gordon
- Fonction : Auteur
- PersonId : 828892
Jean-Pierre Nadal
- Fonction : Auteur
- PersonId : 11356
- IdHAL : jean-pierre-nadal
- ORCID : 0000-0003-0022-0647
- IdRef : 032824300
Denis Phan
- Fonction : Auteur
- PersonId : 1117320
- IdHAL : denis-phan
- ORCID : 0009-0005-2118-9839
Résumé
We consider a simple model in which a population of individuals with idiosyncratic willingnessesto pay must choose repeatedly either to buy or not a unit of a single homogeneous good at a givenprice. Utilities of buyers have positive externalities due to social interactions among customers.If the latter are strong enough, the system has multiple Nash equilibria, revealing coordinationproblems. We assume that individuals learn to make their decisions repeatedly. We study theperformances along the learning path as well as at the customers’ reached equilibria, for differentlearning schemes based on past earned and/or forgone payoffs. Results are presented as a functionof the price, for weak and strong social interactions. Pure reinforcement learning is shown tohinder convergence to the Nash equilibrium, even when it is unique. For strong social interactions,coordination on the optimal equilibrium through learning is reached only with some of the learningschemes, under restrictive conditions. The issues of the learning rules are shown to depend cruciallyon the values of their parameters, and are sensitive to the agents’ initial beliefs.
Domaines
Sciences de l'Homme et Société Complexité [cs.CC] Apprentissage [cs.LG] Système multi-agents [cs.MA] Economies et finances Microstructure des marchés [q-fin.TR]Format du dépôt | Notice |
---|---|
Type de dépôt | Chapitre d'ouvrage |
Titre |
en
Choice under social influence: effects of learning behaviors on the collective dynamics
|
Résumé |
en
We consider a simple model in which a population of individuals with idiosyncratic willingnessesto pay must choose repeatedly either to buy or not a unit of a single homogeneous good at a givenprice. Utilities of buyers have positive externalities due to social interactions among customers.If the latter are strong enough, the system has multiple Nash equilibria, revealing coordinationproblems. We assume that individuals learn to make their decisions repeatedly. We study theperformances along the learning path as well as at the customers’ reached equilibria, for differentlearning schemes based on past earned and/or forgone payoffs. Results are presented as a functionof the price, for weak and strong social interactions. Pure reinforcement learning is shown tohinder convergence to the Nash equilibrium, even when it is unique. For strong social interactions,coordination on the optimal equilibrium through learning is reached only with some of the learningschemes, under restrictive conditions. The issues of the learning rules are shown to depend cruciallyon the values of their parameters, and are sensitive to the agents’ initial beliefs.
|
Auteur(s) |
Viktoriya Semeshenko
1, 2
, Mirta B. Gordon
3
, Jean-Pierre Nadal
4, 5
, Denis Phan
6
1
Departamento de Economía [Buenos Aires]
( 97466 )
- Facultad de Ciencias Económicas - UBA - Av. Córdoba 2122 - CABA
- Argentine
2
TIMC-IMAG -
Techniques de l'Ingénierie Médicale et de la Complexité - Informatique, Mathématiques et Applications, Grenoble - UMR 5525
( 707 )
- Domaine de la Merci, 38706 La Tronche, France
- France
3
AMA -
Analyse de données, Modélisation et Apprentissage automatique [Grenoble]
( 405357 )
- Laboratoire LIG - Bâtiment IMAG - 700 avenue Centrale, CS 40700 - 38058 Grenoble cedex 9
- France
4
CAMS -
Centre d'Analyse et de Mathématique sociales
( 1318 )
- 54 boulevard Raspail 75006 Paris
- France
5
LPS -
Laboratoire de Physique Statistique de l'ENS
( 4 )
- 24 rue Lhomond 75231 PARIS Cedex 05
- France
6
GEMAS -
Groupe d'étude des méthodes de l'analyse sociologique
( 1348 )
- 54 Bd Raspail 75006 PARIS
- France
|
Audience |
Internationale
|
Titre de la collection |
Contributions to Economic Analysis
|
Langue du document |
Anglais
|
Titre de l'ouvrage |
Cognitive Economics: New Trends
|
ISBN |
978-0444522429
|
Date de publication |
2007
|
Volume |
280
|
Page/Identifiant |
178-203
|
Lieu de publication |
Amsterdam
|
Vulgarisation |
Non
|
Voir aussi |
|
Financement |
|
Domaine(s) |
|
Éditeur commercial |
|
Éditeur scientifique |
|
Loading...