The Future of Game Theory in Biology

4 Jul 2022, 16:00 → 8 Jul 2022, 16:00 Europe/Berlin

Game theory has played an instrumental role in developing concepts, hypotheses, and empirical studies in biology.

The workshop aims to discuss the old and new themes, mechanisms, and methods that could play a crucial role in defining game theory in biology as we look to the future of this theory.

Organisers: John McNamara (University of Bristol), Olof Leimar (Stockholm University), Chaitanya Gokhale (Max Planck Institute for Evolutionary Biology)

The future of game theory in biology

To identify how new theoretical approaches in game theory and novel applications of the theory can enhance our understanding of biology.

Although in use much earlier, game theory was popularly introduced to biologists by John Maynard Smith in his 1982 book Evolution and the Theory of Games. Drawing primarily from economics and several other disciplines, game theory in biology has been a melting pot of diverse ideas and concepts. While a swathe of research has focused on human cooperation, the applicability of games in biology is much broader and arguably as diverse as the different organisms it is applied to. This workshop aims to discuss this breadth of application of games in biology, the novel frontiers that are yet to be tackled, and how this will enrich the nature of game theory itself in the context of biological theory.

Invited Speakers

• Sigrunn Eliassen

• Yoh Iwasa

• Rufus Johnstone

• Nichola Raihani

• Franjo Weissing

• Peter Hammerstein

• Christian Hilbe

• Jens Krause

• Hans Metz

• Willem E. Frankenhuis

Monday, 4 July

16:00 → 18:00 Arrival & Check-In

18:00 → 19:00 Dinner

19:00 → 20:00 Welcome

Overview of the programme and workshop format

Speaker: Chaitanya Gokhale (MPI for Evolutionary Biology)

Tuesday, 5 July

09:00 → 10:30 Mechanisms and decision making: Sigrunn Eliasson and Willem Frankenhuis

09:00 Adaptive decision-making and animal interactions in social networks

Classical game theory has trained our intuition about important aspect of social interactions, making it possible to recognize general structures of games played out in complex natural settings. Moving beyond the basic models, we appreciate that games are dynamic in nature, cost and benefits are inherently linked to local interactions, and animal responses are based on previous experiences and adaptive decision-making. Information and the ability to acquire, integrate and restrict it, may alter individual strategies, and influence evolutionary dynamics. I will illustrate this with a model of coevolving mating strategies in social networks and propose a framework for adaptive decision-making in natural environments. Here I consider the cognitive machinery that enables animals to behave autonomously, make predictions about the future and adaptive decisions in real time. May viewing the animal as an agent, with goal-directed rather than purely stimulus-driven cognitive and behavioral control, help integrate our understanding of proximate and ultimate causation?

Speaker: Sigrunn Eliassen (University of Bergen)

09:45 Development in a frequency-dependent context

There is well-established theory exploring the conditions in which phenotypic plasticity is favoured by natural selection over non-plastic development. However, most of this work has assumed a two-stage life history: organisms sample a cue, or they don’t sample, relying instead on inherited information; and then develop a phenotype, either instantaneously or after a time lag. Recent modeling has conceptualized development as a sequential decision-making process, in which organisms learn about their environments and incrementally tailor their phenotypes. This representation of ontogeny, as unfolding, has created scope for new questions: In which conditions should we expect organisms to sample more or less at different times in their lives? When does natural selection favour sensitive periods (windows of heightened plasticity)? How do trajectories of skills development depend on environmental conditions across evolution and ontogeny? To date, this line of work has nearly exclusively explored ‘games against nature’. In this talk, we briefly describe this line of work and discuss ways in which it may be extended to include ‘games against others’; that is, development in a frequency-dependent context.

Speaker: Willem Frankenhuis (Utrecht University)

10:30 → 11:00 Coffee Break

11:00 → 12:30 Mechanisms and decision making: Discussion on three possible subtopics and group formation

12:30 → 13:30 Lunch

13:30 → 15:30 Mechanisms and decision making: Independent group discussion

15:30 → 16:00 Coffee Break

16:00 → 17:00 Games and Dynamics

I will speak basically on two topics.

[1] Coupled social and ecological dynamics:
I will discuss the issue of replicator dynamics versus best response dynamics. First I will speak two models related to illegall logging in tropical forests. One model discusses the action of profit sharing by the Indonesian government, which was effective in suppressing illegal logging of teak forests. In this model, we adopted the stochastic best response dynamics.
The second model is on the corruption — rule enforcer might receive bribe and neglect the illegal logging case. The work was in collaboration with Karl Sigmund and Ulf Dieckmann, and certainly it is based on replicator dynamics. The main conclusion was that the resurgence of illegal logging and corruption persists.
Then I ask what are the major differences between the two dynamics. Which is more appropriate? Are there any empirical evidence supporting one over the other? etc. These are just questions, and I do not have definite answers myself.

[2] Coupled behavioral and epidemilogical dynamics:
In 2020-2021, we have COVID-19 infection that exhibited several waves of infection, but the total number of infected stayed rather low (the cumulated number of infected was less than 1.5%). I think this was caused by the coupling of people’s behavioral responses and epidemiology.
There are many papers written on epidemiology models incorporating people’s behavioral responses, which are all quite high dimensional and were analyzed only numerically.
We constructed a very simples model of just two dimensions, and we can prove many bifurcations (homoclinic bifurcation, Hopf bifurcation, transcritical etc.). We concluded that both risk avoidance and conformity are needed to explain the observation.

Speaker: Yoh Iwasa (Kyushu University)

17:00 → 18:00 Break

18:00 → 19:00 Dinner

19:00 → 20:00 Mechanisms and decision making: Group Rapporteur Discussion and Conclusion

Wednesday, 6 July

09:00 → 10:30 Conflict and Cooperation: Nicola Raihani & Christian Hilbe

09:00 Evolution of cooperation -- what has happened, where are the gaps?

The study of cooperation is one of the landmark applications of evolutionary game theory. The respective literature explores why individuals may cooperate even when there are strong incentives to defect. In a broader sense, this literature also explores related questions, such as how social norms evolve over time, or why human behavior is subject to certain quirks (such as: why do we value modesty?) In this talk, I will first give a summary of the mechanisms for cooperation that have been proposed. This part will also highlight some of the wider literature that I find particularly interesting. In the end, I will speculate about possible future directions.

Speaker: Christian Hilbe (Max Planck Institute for Evolutionary Biology)

09:45 Punishment: one tool, many uses

Humans are outstanding in their ability to cooperate with unrelated individuals, and punishment – paying a cost to harm others – is thought to be a key supporting mechanism. According to this view, cooperators punish defectors, who respond by behaving more cooperatively in future interactions. However, evidence from laboratory and real-world settings casts serious doubts on the assumption that the sole function of punishment is to convert cheating individuals into cooperators. In this talk, I will discuss empirical work that has explored the motives underpinning punishment decisions, highlighting the important contributions of inequity aversion and other competitive motives in prompting punishment decisions. I will also talk about the consequences of punishment, both in terms of the targets’ responses and how punishers are viewed by others based on their investments. I will end by discussing how punishment in the lab differs from punishment we observe in the real world, focussing on gossip and centralised punishment institutions. I end by highlighting the need to to consider both legitimacy and rehabilitation if punishers are to safely wield an effective, yet potentially explosive, cooperation-enforcing tool.

Speaker: Nichola Raihani (University College London)

10:30 → 11:00 Coffee Break

11:00 → 12:30 Conflict and Cooperation: Discussion on three possible subtopics and group formation

12:30 → 13:30 Lunch

13:30 → 15:30 Conflict and Cooperation: Independent group discussion

15:30 → 16:00 Coffee Break

16:00 → 17:00 Multivariate adaptive dynamics and games

Adaptive dynamics (AD) is the natural biological embedding for evolutionary game theory. It looks how continuous traits, think e.g. of the frequency of playing hawk, develop over evolutionary time through subsequent mutant substitutions. This dynamic embedding allows to bring to bear various tools of dynamical systems theory, think e.g. of the development of a bifurcation theory for ESSes. In the talk I will give a short overview of the main tools and general results of AD, with particular attention to the biological perspective, with the goal to help with the further biologising of the mathematics. The stress will be on the multivariate case, as one inevitably ends up there when trying to increase the biological content of one's models, and it is where at this moment most open problems are to be found.

Speaker: Hans Metz (Leiden University)

17:00 → 18:00 Break

18:00 → 19:00 Dinner

19:00 → 20:00 Conflict and Cooperation: Group Rapporteur Discussion and Conclusion

Thursday, 7 July

09:00 → 10:30 Animal personalities: Franjo Weissing & Peter Hammerstein

09:00 Game theory confronted with facts: The analysis of variation in natural populations

Since its origin, evolutionary game theory has dealt with syndromes of behavioral and morphological traits in phenotypic models of adaptive evolution. Sneaky male mating
strategies, for example, require both the lack of masculine ornamentation and special behaviors to avoid fighting with dominant males. Due to an almost obsessive focus on mixed
strategy equilibria, many early attempts to explain the coexistence of alternative syndromes remained silent about the actual switching mechanisms that induce one or the other
alternative. If, however, one takes these mechanisms into account, this can dramatically change the perspective on adaptive evolutionary processes. My talk, therefore, describes
forward looking ways to empirically and theoretically investigate switching mechanisms that operate, for example, in the ecological context of structured populations. As part of this endeavor, I highlight the strong ties between evolutionary game theory and kin selection. In my view, the careful fusion of these two lines of evolutionary thought, as well as a genuine concern for mechanistic detail, seem to be essential for the future of game theory in biology. Needless to say that a deeper look at mechanistic detail will raise many new questions. Some of these questions will center around the problem of agency: who sets the switches and how?
Microbial symbionts are certainly involved in operating switching mechanisms of their hosts.
Some microbes can even reverse the process of sexual differentiation and turn genetic males into functional females. Conventional sex ratio theory did not capture this phenomenon and needed to be ‘patched’. As to future research, transplanting the gut microbiota from one
organism to another is a promising experimental technique to demonstrate the impact microorganisms have on behavioral syndromes and to provide evolutionary game theory with
new "food for thought".

Speaker: Peter Hammerstein (Humboldt University Berlin)

09:45 Animal personalities – a challenge for evolutionary game theory

No two individuals are alike. Even individuals of the same sex, age, size and social background tend to differ consistently in temperament, motivation, cognition and behaviour. In virtually all species studied, ranging from spiders to sticklebacks and from octopuses to chimpanzees, individuals differ systematically in whole suites of correlated behaviours. For example, one often finds stable differences in aggressiveness, and aggressive individuals tend to differ from non-aggressive conspecifics in a multitude of ways: they are bolder, more explorative, less philopatric, less ‘social’ (but not necessarily less cooperative), have a lower learning performance, and exhibit less parental care. The existence of such ‘personalities’ (also termed behavioural syndromes, coping styles or temperaments) provides a challenge for evolutionary theory. First, why do different personality types stably coexist? Should we not expect a single optimal phenotype? Second, why are behavioural differences consistent over extended periods of time? Should we not expect a more flexible structure of behaviour that is fine-tuned to the local circumstances? And third, why are quite different types of traits correlated, and why are similar correlations observed in very different taxa? Should we not expect pronounced differences in the behavioural architecture of molluscs, insects and mammals, reflecting the differences in their physiology, life history and ecology?
Evolutionary game theory can cope relatively easily with the first two challenges. Behavioural polymorphism can be explained by diversifying frequency-dependent selection or condition-dependent strategies (where individuals differ subtly in ‘condition’). Time consistency can be explained by positive feedbacks (such as training effects) that make regularly performed behaviours more efficient or less costly. I will focus on the third challenge and discuss various (adaptive and non-adaptive) explanations for frequently observed behavioural correlations. This challenge is a serious one, as it undermines the standard approach of evolutionary game theory. Game theorists are used to studying different types of behaviour in isolation. But does it make sense to investigate aggressive behaviour, foraging behaviour, mating behaviour, and parental behaviour in independent game models if these behaviours are intrinsically linked via the structure of personalities? To what extent is it possible to incorporate ‘personality thinking’ into the framework of evolutionary game theory?

Speaker: Franjo Weissing (University of Groningen)

10:30 → 11:00 Coffee Break

11:00 → 12:30 Animal personalities: Discussion on three possible subtopics and group formation

12:30 → 13:30 Lunch

13:30 → 15:30 Animal personalities: Independent group discussion

16:00 → 17:00 Boat Trip

17:00 → 18:00 Break

18:00 → 19:00 Dinner at Restaurant

19:00 → 20:00 Animal personalities: Group Rapporteur Discussion and Conclusion

Friday, 8 July

09:00 → 10:30 Collective Behaviour: Jens Krause & Rufus Johnstone

09:00 Fish waves as collective signal?

Collective signals which are used to attract mates, defend territories or deter predators have been observed in different species ranging from social insects, over fire-flies to fish and birds. In this talk I will show an example of fish schools which collectively produce waves that delay attacks by bird predators and reduce their success probability. Different mechanisms which result in such antipredator effects are conceivable such as predator confusion which blocks the information processing of birds or a perception advertisement which might signal to the bird that an attack is unprofitable. High degrees of relatedness between individuals (as, for example, in eusocial insects) has frequently been invoked as an explanation for the evolution of collective signals. However, it appears that collective behaviours which might be signals are also found in other taxonomic groups where individuals are clearly unrelated. I will explore conditions under which fish collective waves might have evolved a signaling function and discuss this in the context of other examples where collective signal production between unrelated individuals has been reported.

Speaker: Jens Krause (IGB Berlin)

10:30 → 11:00 Coffee Break

11:00 → 12:30 Collective Behaviour: Joint Discussion on the effect of previous topics on collective behaviour

12:30 → 13:30 Lunch

13:30 → 14:30 Closing Remarks & Feedback

14:30 → 15:00 Coffee