London School of Economics and Political ScienceTuesdayWednesday April 56, 2022The aim of this workshop is to discuss how to overcome a lack of realism in many gametheoretic models. Game theory assumes that the interactive situation has commonly known welldefined rules, and that players act assuming rationality of other players. Even models of bounded rationality typically assume a given "game". In reality, many rules and expected ways of acting are implicit. In the workshop we hope to formulate new research directions as much as learning about new technical contributions, to find fresh perspectives on games and human behaviour.Organisers: Galit AshkenaziGolan, Olivier Gossner, Bernhard von Stengel, and Peyton YoungInvited Speakers:Françoise Forges, Penélope Hernández, Marco LiCalzi, Heinrich Nax, Christina Pawlowitsch, Rafael Veiel, Segev Wasserkrug, and Jörgen Weibull. 
Room capacity is limited.
Participants will need to register
with an email sent to Enfale Farooq (e.farooq@lse.ac.uk).
On the first day Tuesday 5 April at 7pm there is a
conference dinner which we need to reserve at the
restaurant. Please email Enfale Farooq by Monday 28 March at
noon if you would like to take part, with further
instructions from Enfale (selffunded unless you have been
invited, approximate cost per person £35, not including
drinks).
Click on a speaker name for title, abstract, and talk presentation (if available).
Tuesday 5 April 2022 
Wednesday 6 April 2022 
9:15  10:00


10:00  10:30 coffee break 

11:00  11:10 Welcome 
10:30  11:15

11:15  12:00

11:30  12:15

12:30  14:00

12:30  14:00

14:00  14:45

14:00  14:45

15:00  15:45

15:00  15:45

15:45  16:15 coffee break 
16:00  16:30 Discussion 
16:15  17:00

(End of Workshop) 
17:10  18:00


19:00 
Speaker 
Title, Links 
Abstract 
Françoise Forges

Identifying the game: some examples from classical and behavioral game theory

The definition of a game (say, in extensive form) is very precise and rigid: every detail of the agents' interaction is captured by the tree. It assumes complete information but allows for random events and imperfect information. The literature is full of examples in which some aspects of the environment (information, outcomes, ...) are given while others are flexible. I select some of them, with the following organization:

Olivier Gossner

The robustness of incomplete penal codes in repeated interactions

We study the robustness of equilibria with regards to small payoff perturbations of the dynamic game. We find that complete penal codes, that specify players' strategies after every history, have only limited robustness. We define incomplete penal codes as partial descriptions of equilibrium strategies and introduce a notion of robustness for incomplete penal codes. We prove a Folk Theorem in robust incomplete codes that generates a Folk Theorem in a class of stochastic games. 
Penélope Hernández

Leadership and peer effect

This paper analyzes the role of a Leadership as a shaper of preferences and studies in a formal dynamic setting its capacity and limits in influencing large and closed organizations where peer effects are operating. We study both a charismatic and a socializing Leaderships showing the common results and also the differences in their performance. We consider an organization composed by a Leadership and by a finite group of agents. The Leadership has some ideal organization composition or vector of preferred actions one for each agent. Each agent has as well her preference that evolves over time. When an agent makes a decision each period her behavior is driven by two competing motives: she wants her behavior to agree with her personal preference and she wants also her behavior to be as close as possible to the average behavior of her peers. The Leadership has also an effect over the agents' preferences. On the one hand, for the socializing Leadership, there exists a costly corporate socialization effort exerted by him trying to transform the preference of each agent into his own ideal preference. The charismatic Leadership affects directly the agents' preference by his charisma. We characterize the longrun outcomes of both situations and in particular in the ability of the Leadership to fully instil the corporate culture in the members of the organization. Finally we measure the price of charisma versus socialization. Joint work with Gonzalo Olcina and Raul Toral. 
Marco LiCalzi

Small games and cognitive discord

We consider players who use their frames to distill a strategic situation into a small game. When the frames concur, the small game may lead to better or worse payoffs than in the original (large) game. When the frames do not concur, players's joint choices may reveal a state of cognitive discord. We gather a few examples and discuss how and when cognitive discord may be repaired or mitigated. 
Heinrich Nax

Morgan's Canon or What's the point of "Black Box" experiments?

According to the socalled Morgan’s Canon “In no case is an animal activity to be interpreted in terms of higher psychological processes, if it can be fairly interpreted in terms of processes which stand lower in the scale of psychological evolution and development.” (C. Lloyd Morgan, 1903). Economists and game theorists, even behavioral ones, tend to favor extreme hyperrationality models, also for explaining human behavior. I shall talk about some novel approaches to economic experiments that could help bridging between such approaches. 
Christina Pawlowitsch

The sequencing of functions in time vs. the matrix: parallels in the debates in game theory and narrative analysis

Propp (1928) advances the hypothesis that all magical tales (tales classed under numbers 50151 in Afanasyev’s collection) follow the same sequence of basic events, which he calls functions. The French structuralists (LéviStrauss 1960, Barthes 1966, Bremond 1966, Greimas 1966a, 1966b) attempted to show that this sequence of functions, which anchors the narrative in time, can be reabsorbed by a “timeless” matrix. In this talk, I am going to shape out some similarities between the debate on “the sequence in time versus the matrix” in narrative theory and the debate on the relation between the extensive form and the normal form, the matrix, of a game. 
Rafael Veiel

Rationalizable Distributions in Games with Incomplete Information

We provide a finite characterization of all distributions of rationalizable outcomes that can arise from common priors in finite games of incomplete information. We characterize rationalizable distributions as Markov chains on a finite state space that generates hierarchies of bestreplies subject to a finite set of obedience constraints. 
What is the game? Thinking and overthinking
strategically

Game theory gives valuable insights into how the rules of
the game determine optimal play. Choosing the right model is
crucial in any applied setting, and often the main point of
debate in an economic theory paper.
 
Segev Wasserkrug

Game theoretical gaps for practical multi party decision making

Many realworld decisions in business settings such as business networks, cloud computing and cyber security involve situations in which there are multiple parties involved, each attempting to optimize their own objectives, and in which the actions of each party could influence the objectives of the other parties. Currently, however, even when advanced mathematical models, such as optimization, are used to make decisions in such settings, typically each party assumes that the environment is stationary in the sense that the impact of the decisions of each party on the other stakeholders is ignored. Game theory is the scientific basis intended to enable making decisions in such settings. However, game theoretic solution concepts often assume unlimited computational resources (unbounded rationality) and unrealistic assumptions made on the common knowledge of all participants, such as full common knowledge in full information games or common knowledge priors in Bayesian games. In addition, solution concepts such as equilibria may be difficult to justify and apply in real world settings due, for example, to the existence of multiple equilibria. Algorithmic game theory tries to overcome some of the computational limitations of game theory by combining both game theoretical and computational complexity considerations. However, its applications to scenarios such as the ones mentioned above has been very limited. Finally, multi agent reinforcement learning (MARL) can, in principle, be applied to many use cases. However, it lacks a strong theoretical basis providing theoretical convergence and performance results, making it difficult to guarantee its performance in practice. In this talk, I will give examples of such business scenarios where ignoring the multiparty objectives can lead to poor results, outline some of the gaps in the aforementioned fields, and discuss some of our current research directions in addressing these. The session will also include some time for discussion and brainstorming on how to make game theory more applicable for such settings. 
Jörgen Weibull

Game identification and solution invariance

Through the prisms of the three solution concepts: tenable strategy blocks (Myerson & Weibull, 2015), settled equilibria (Myerson & Weibull, 2015), and solid outcomes (Ritzberger, Weibull & Wikman, 2022), I will consider the following five tasks for game theorists:
