Skip to content

Evolutionary Stability Checking

Why It Matters

Nash equilibrium identifies stable outcomes in games between rational players, but it doesn't explain which equilibrium will be selected or whether equilibria are stable against invasion. Evolutionary stability adds a dynamic selection criterion: strategies that reproduce more successfully spread through populations. This explains which equilibria we should expect to see in the long run and which cooperative arrangements resist free-rider invasion. It's particularly powerful for analyzing situations where "rationality" is implausible but selection pressure exists (animal behavior, cultural evolution, market competition).

The Key Move

  1. Identify a candidate strategy S used by most of the population.
  2. Introduce a small proportion of mutants using alternative strategy M.
  3. Calculate payoffs: S vs. S (population members playing each other), S vs. M (population vs. mutants), and M vs. M (mutants playing each other).
  4. Check ESS condition: either (a) S vs. S gives higher payoff than M vs. S (S is better against itself than M is against S), or (b) if payoffs are equal, then S vs. M must exceed M vs. M (S does better against mutants than mutants do against each other).
  5. If both conditions fail, S is invadable - mutant strategy can spread.
  6. If ESS condition holds, small proportions of M will be selected against.

Classic Application

Hawk-Dove game in animal conflict. Animals compete for resources. Hawk strategy: escalate, fight if opposed. Dove strategy: display, flee if opponent escalates. Payoffs: Hawk vs. Dove (Hawk wins), Dove vs. Dove (split resource peacefully), Hawk vs. Hawk (both injured, net negative). Is "always Hawk" evolutionarily stable? No - in a Hawk population, being Dove avoids costly fights when meeting Hawks. Is "always Dove" stable? No - in a Dove population, a single Hawk takes all resources. ESS: mixed strategy (probabilistic Hawk/Dove) or conditional strategy (Hawk if opponent is Dove, Dove if opponent is Hawk). This explains observed behavioral polymorphism in nature.

Surprising Application

Workplace norm persistence. Consider "work-life balance culture" versus "overwork culture" in organizations. If most employees work reasonable hours (balance strategy), is this invadable? A few workaholics (overwork mutants) might gain visibility/promotions (overwork vs. balance favors overwork). But if many switch to overwork, competition intensifies and everyone is worse off (overwork vs. overwork is worse than balance vs. balance). ESS analysis: balance is stable only if rewards don't disproportionately favor overwork, or if overwork becomes self-defeating when common (burnout, errors). This explains why some company cultures sustain healthy norms while others spiral into overwork equilibria - depends on the payoff structure and whether the culture has ESS properties. Non-obvious because it treats culture as evolved strategy rather than chosen policy.

Failure Modes

Assuming single ESS - many games have multiple evolutionarily stable strategies or none. Ignoring frequency-dependence - ESS analysis assumes payoffs depend on strategy frequencies, but this isn't always modeled correctly. Not checking for neutral stability - sometimes mutants neither grow nor shrink (drift). Applying to non-replicating populations - ESS requires differential reproduction/success, doesn't apply when strategies don't spread. Missing environmental changes - ESS is relative to fixed payoff structure, but real environments shift.

Go Deeper

Maynard Smith, Evolution and the Theory of Games, Chapter 1; Weibull, Evolutionary Game Theory, Chapter 2