scholarly journals When to (or not to) trust intelligent machines: Insights from an evolutionary game theory analysis of trust in repeated games

2021 ◽  
Vol 68 ◽  
pp. 111-124
Author(s):  
The Anh Han ◽  
Cedric Perret ◽  
Simon T. Powers
Author(s):  
Arij Michel

The article uses evolutionary game theory analysis as the research object, which is the most commonly used research method of institutional change, and summarizes some methods in the research of institutional change, and points out the advantages and disadvantages of evolutionary game analysis in the research of institutional change and through the comparison of cutting-edge methods and evolutionary games to see the development direction of future research institutional changes.


Author(s):  
Cristina Bicchieri ◽  
Giacomo Sillari

Game theory aims to understand situations in which decision-makers interact strategically. Chess is an example, as are firms competing for business, politicians competing for votes, animals fighting over prey, bidders competing in auctions, threats and punishments in long-term relationships, and so on. In such situations, the outcome depends on what the parties do jointly. Decision-makers may be people, organizations, animals, or even genes. In this chapter, the authors review fundamental notions of game theory and their application to philosophy of science. In particular, Section 1 looks at games of complete information through normal and extensive form representations, introduce the notion of Nash equilibrium and its refinements. Section 2 touches on epistemic foundations and correlated equilibrium, and Section 3 examines repeated games and their importance for the analysis of altruism and cooperation. Section 4 deals with evolutionary game theory.


2021 ◽  
Vol 8 (5) ◽  
pp. 202186
Author(s):  
Masahiko Ueda

Repeated games have provided an explanation of how mutual cooperation can be achieved even if defection is more favourable in a one-shot game in the Prisoner’s Dilemma situation. Recently found zero-determinant (ZD) strategies have substantially been investigated in evolutionary game theory. The original memory-one ZD strategies unilaterally enforce linear relationships between average pay-offs of players. Here, we extend the concept of ZD strategies to memory-two strategies in repeated games. Memory-two ZD strategies unilaterally enforce linear relationships between correlation functions of pay-offs and pay-offs of the previous round. Examples of memory-two ZD strategy in the repeated Prisoner’s Dilemma game are provided, some of which generalize the tit-for-tat strategy to a memory-two case. Extension of ZD strategies to memory- n case with n ≥ ̃2 is also straightforward.


2018 ◽  
Vol 218 ◽  
pp. 104-113 ◽  
Author(s):  
Kristof Coninx ◽  
Geert Deconinck ◽  
Tom Holvoet

Sign in / Sign up

Export Citation Format

Share Document