Even in the Prisoner’s Dilemma, the classic thought experiment built to prove selfishness always wins, cooperation keeps showing up and sticking around.
A new study suggests this isn’t a fluke, and it might mean scientists have had the story of cooperation’s evolution backwards for a long time.
The work was led by Alexander Feigel from the Hebrew University of Jerusalem and Alexandre V. Morozov from Rutgers University.
The research offers a genuinely fresh answer to a question that’s nagged at biologists for decades: how does cooperation manage to survive at all in a world where natural selection is supposed to reward selfishness above everything else?
The puzzle of cooperation
In the classic setup, the “smart” move in a Prisoner’s Dilemma is always to defect, even though both players would come out ahead by cooperating instead.
That logic has underpinned a lot of evolutionary thinking for years: if selfishness is mathematically the better bet in any interaction, cooperation should eventually get squeezed out over time.
Theoretically, it should survive only in special cases propped up by family bonds, reputation, or some other external scaffolding holding it in place.
“Conventional evolutionary theory predicts that defectors – those who always act selfishly – should eventually outcompete cooperators,” Feigel said.
“Our work shows that this isn’t necessarily true. Once individuals recognize that not every opponent is the same, cooperation becomes surprisingly robust and can emerge spontaneously.”
How cooperation survives
What’s genuinely surprising is how little cooperation appears to require.
The researchers found it can emerge without close kinship, tracking who owes whom a favor, or rigid social hierarchies to keep individuals in line..
All the usual explanations biologists have leaned on for decades to explain why cooperation survives turn out not to be strictly necessary after all.
Instead, the researchers found something much simpler at work underneath all of it: individuals just responding differently depending on who they’re up against.
They did this without any need for memory, status, or complex social bookkeeping.
Responding to different opponents
Using mathematical models, evolutionary simulations, and populations of AI agents, the team found that cooperation repeatedly evolved when individuals adjusted how cooperative they were based on the specific opponent they faced.
Rather than treating every interaction the same way, this simple flexibility was enough to build stable populations where cooperation consistently outperformed pure selfishness.
“Individuals simply need to respond differently to different partners,” Morozov explained.
“That simple principle is enough to allow cooperation to survive – even under evolutionary pressures that were previously thought to favor selfish behavior.”
Populations that resist selfishness
Maybe the most unexpected part of the whole study is just how sturdy these cooperative populations were.
Across thousands of simulation runs, populations kept drifting toward highly cooperative arrangements.
Mranwhile, populations built entirely on selfishness often didn’t take over the way the old theory predicted they should.
Even when random mutations introduced fresh defectors into the mix, cooperative communities largely held their ground instead of unraveling into selfish chaos.
They absorbed the disruption without losing their overall structure.
This runs counter to a lot of intuition about how fragile cooperation is supposed to be. It suggests cooperation isn’t some delicate balance that falls apart the moment something goes wrong or a single bad actor shows up.
Once a population settles into a pattern of responding differently to different partners, it seems to shrug off a fair amount of internal disruption without collapsing – almost as though the system has built-in resistance to selfish invasion.
Broader implications of the study
This isn’t just relevant to cells, microbes, plants, and animals, the usual subjects of evolutionary biology research.
The researchers think similar principles could shape how people design more cooperative AI systems, autonomous robots, and distributed computing networks.
These are all places where getting independent agents to actually work together remains a genuinely hard engineering problem in its own right.
The ability to tell different partners apart and respond accordingly shows up everywhere in nature, from single cells all the way up to people navigating complicated social relationships.
The researchers argue this ability might have been one of the fundamental building blocks that allowed complex life to evolve in the first place.
“If cooperation can emerge without elaborate rules or sophisticated cognition, it becomes much easier to understand how complex biological systems evolved in the first place. Perhaps evolution has always been more cooperative than we gave it credit for,” Feigel concluded.
The study is published in the journal Proceedings of the National Academy of Sciences.
—-
Like what you read? Subscribe to our newsletter for engaging articles, exclusive content, and the latest updates.
Check us out on EarthSnap, a free app brought to you by Eric Ralls and Earth.com.
——