# Speed of Common Learning with Biased Memory
**Date de l'événement :** 02/08/2024
* Publié le 02/08/2024

### Image(s)
![Page de garde du mémoire](https://firebasestorage.googleapis.com/v0/b/memory-sp-pr.appspot.com/o/prod%2F1y7CgDOTakHDiB3dgoi0%2FprojectsMedias%2FkjiDx92ORiKt2GFALens%2Fthumbs%2FCapture%20d'%C3%A9cran%202026-02-16%20173241_diapd_1600x900.png?alt=media&token=3cc015ea-868b-4bae-ab81-010badea736e) 

## Auteur(s)
Ruichao Pan

## Direction
Franz Ostrizek

## Description
**Ce mémoire est issu de la sélection des meilleurs mémoires du master d’économie de l'année 2023-2024 :  
  
Abstract**  
  
When multiple agents have access to many draws of private signals, results from Frick, Iijima, and Ishii (2023) show that we can characterize the speed of individual and common learning away from the limit. However, in a data-rich setting like this, it is challenging for agents to remember and utilize precisely all of the signals they have received in the past. In this paper, we consider a setting where agents experience biased memory when making inferences, meaning certain signals are more likely to be remembered than others. Our first main result characterizes the speed of individual and common learning under this novel setting. We then show that, contrary to classical wisdom, under certain information structures, common learning occurs slower than individual learning; that is, higher-order uncertainty vanishes slower than first-order uncertainty. We also provide an intuitive necessary condition for this result: the state that agents individually find difficult to distinguish from the true state has to be different from the state that agents expect others to find difficult to distinguish.  
  
\[...\]

**Accéder au mémoire en ligne :** [https://www.sciencespo.fr/ecole-recherche/sites/sciencespo.fr.ecole-recherche/files/PAN_Ruichao.pdf](https://www.sciencespo.fr/ecole-recherche/sites/sciencespo.fr.ecole-recherche/files/PAN_Ruichao.pdf)

### Type(s) de ressource
`#Texte` 

### Discipline
`#Économie` 

### Thématique(s)
`#Sciences / connaissance` `#Méthodologie` 

### Langue(s)
`#Anglais` 

### Famille(s) de contenu
`#Recherche` `#Production étudiante` 

### Type(s) d'accès
`#Accès libre` 

## Droits
Les ressources et les images visibles sur Sciences Po Sources sont susceptibles d’être protégées par un droit de propriété intellectuelle (comme par exemple un droit d’auteur ou une marque) qui peuvent appartenir à Sciences Po ou à des tiers. Pour plus de précisions notamment sur les usages autorisés et non autorisés, consulter les [Conditions Particulières d’Utilisation](https://sources.sciencespo.fr/p/cpu).



---
### Navigation pour IA
- [Index de tous les contenus](https://sources.sciencespo.fr/llms.txt)
- [Plan du site (Sitemap)](https://sources.sciencespo.fr/sitemap.xml)
- [Retour à l'accueil](https://sources.sciencespo.fr/)