杏吧原创

IT鈥橲 a jungle out there. Beneath the thin veneer of civilisation, life for Homo sapiens is as cut-throat as it is for any other animal. Like them, we are in constant competition for the things that really matter: sex, material comforts and real estate. Inevitably, there are winners and losers. And our societies, like most groups of animals, are organised hierarchically, with a few big-hitters wielding most of the power and the vast majority of us somewhere further down in the pecking order.

But hold on. Surely this is one area where we are nothing like other animals. Don鈥檛 they simply rely on size and strength to fight their way to the top? Well, that鈥檚 what we used to think, but it鈥檚 becoming increasingly apparent that although other species can鈥檛 argue their case to get what they want, neither is brawn all that counts.

In many animals, the outcome of any clash is also shaped by subtle psychological factors鈥攍ike us, they have their winning and losing streaks. Past experiences matter, and the behaviour of members of a particular species is affected in characteristic ways by winning, losing, or simply by watching a fight.

What鈥檚 more, these psychological factors seem to have a drastic effect on animal societies and the levels of aggression within them. My own studies indicate that they can be used to make predictions such as whether interactions between group members will be autocratic or meritocratic, and whether there will be peacekeepers to break up fights before they get out of hand. The same rules may even help explain the dynamics of human societies.

Another researcher looking at these so-called winner, loser and bystander effects is Gordon Schuett from Arizona State University West in Tempe. While many of us shy away from studying animals that might kill us, Schuett is made of sterner stuff鈥攈e is studying the deadly copperhead snake, Agkistrodon contortrix. Most copperhead aggression is between males fighting for access to females, and to find out whether winning or losing affects a snake鈥檚 performance in subsequent fights, Schuett placed a female in the centre of an arena and then put a male at each end. The males had not been in a fight for between six and 12 months before the study, so they came into the ring as 鈥渂lank slates鈥� in terms of winner and loser effects.

At first, size mattered: in every one of 32 fights, the larger male copperhead emerged as the victor and won the female. But Schuett was more interested in what would happen next. Once the initial battles were complete he grabbed 10 winners and 10 losers and pitted each against a new male of the same size that had no prior fighting experience. He found first-round winners were no more likely to win than were their opponents鈥攊n other words, there was no winner effect. But males that had previously lost were very likely to lose again. What鈥檚 more, when two-time losers were given one last chance to prove themselves, the results were even more dismal鈥攏ot one of these males won a fight. Schuett had found a clear case of the loser effect in action.

In recent years, similar experiments have revealed that the performance of numerous species of insects, fish, birds and mammals is influenced by winning or losing. Many animals, including swordtail fish, blue gouramis, crickets and rats, display both winner and loser effects. Several, like the copperhead snake, are only affected by losing. Other examples are pumpkinseed sunfish, green sunfish and paradise fish. So far, we鈥檝e found no species where winner effects occur in isolation but time will tell.

There can be no doubt that winner and loser effects exist, and when animals square up to fight, these effects matter. But animals don鈥檛 fight in a social vacuum. Fights are the basis for dominance hierarchies鈥攖he status ladders that can determine who gets what resource, and just how much they get. Over the past five years, my colleagues and I have been using computer simulations to examine how winner and loser effects influence the formation and stability of social hierarchies. Do these effects have implications for how animals fight? And do they shape the sort of hierarchies we find in different groups?

In our digital arena, we start off with four players who are each assigned an 鈥渋nitial fighting ability score鈥�. Individuals are then paired up at random to fight. Each player has two pieces of information about itself: its initial fighting score, and how that score has changed as a result of subsequent wins and defeats. If a winner effect is operating, after each victory a player鈥檚 score goes up. Conversely, if a loser effect is in play its score goes down if it is beaten.

The player also knows the initial fighting score of the opposition鈥攂ut not the cumulative score that reflects their past performance. The assumption here is that initial fighting ability represents physical attributes that are obvious to any onlooker, such as size, whereas a player鈥檚 cumulative score is determined by experience, which is not apparent to anyone but the player. So individuals have more information about themselves than their opponents, which is usually how things are in the real world.

When two players square up inside the computer, one of three things happens. If both opt not to fight, they simply go their separate ways. This usually happens if both have low estimations of their own fighting abilities, and it represents the sort of posturing often seen among wild animals. If one individual chooses to be aggressive (because it has a high estimation of its own score compared with that of its opponent), but the other does not (for the opposite reason), then an 鈥渁ttack-retreat鈥� sequence is logged. This is the virtual equivalent of a pair of real animals settling a dispute by one charging and the other cowering. Finally, a fight may occur if both players think they have a chance of winning because each considers its score to be higher than the other鈥檚. In this case the winner is the one with the highest cumulative fighting score.

These simulations can be acted out with winner effects alone in place, loser effects alone, or both operating simultaneously. And the surprise is that winner and loser effects produce very different hierarchies. When winner effects are in play, a linear hierarchy emerges with a clear pecking order. For example, A dominates B, who dominates C, who dominates D. It is the sort of hierarchy you find among pigeons, many species of insects and group-living canids, such as wolves and hyenas. Watching such creatures interact, it鈥檚 obvious that each individual is aware of its rank, and the rank of every other animal in the group.

When loser effects alone are operating, a much more autocratic hierarchy surfaces, like those you find among gorillas and many species of fish and insects. In these social groups, the top-ranked or 鈥渁lpha鈥� individual is obvious, but the rank of the other members in the group is hard to work out because they seldom interact. This sort of hierarchy also emerges when both winner and loser effects are in play.

Even more intriguing is the finding that there are major differences between the types of social interaction in these two sorts of society. Fights are the name of the game when winner effects alone are in play. Here, each individual can only increase its estimation of its own fighting skills, so everyone is raring for a fight. The attack-retreat scenario, on the other hand, is most likely to arise where loser effects are in play or where both effects occur simultaneously. Here, many individuals have a low estimation of their fighting abilities and are likely to avoid aggressive interactions whenever possible.

Intrigued by these findings, we decided to extend our model to include 鈥渂ystander effects鈥�. In these, animals change their estimation of the fighting ability of others as a result of observing them in action. In other words, A changes its estimation of B鈥檚 fighting ability depending on how B does in a fight with C. If A notches up its estimation of B鈥檚 abilities when B wins, that鈥檚 called a bystander winner effect. If A lowers its estimation of B when B loses, that鈥檚 a bystander loser effect. Bystander effects seem to be involved in the original pecking order鈥攖hat of chickens. When birds have just lost a fight they are often attacked by onlookers, whereas this rarely happens when they have just won.

More rigorous evidence that bystander effects occur in nature comes from a recent study by my colleague Ryan Earley. He used trios of male swordtail fish to test the idea. First he allowed pairs of combatants to slug it out on one side of a partition while a third male sat the match out on the other side. Then he pitted the third fish against either the winner or the loser.

In some of the experiments the partition was transparent, in some it was opaque, and in others it was a one-way mirror that let only the observer view the combatants. Earley found that watching the initial fight affected the outcome of the second round. Bystanders were less likely to win against fish they had observed winning鈥攁 bystander winner effect. But they were also no more likely to win against a fish they鈥檇 observed losing鈥攖here was no 鈥渂ystander loser effect鈥�.

Confident that bystander effects are real, we put them into our computer model to see how they would affect the virtual pecking order. Once again, we assigned our four cyberfighters an initial fighting ability score, but instead of increasing or decreasing their own score, they would change their estimation of the scores of others. When bystander winner effects alone are in play a strange hierarchy emerges鈥攐ne that doesn鈥檛 seem to reflect any society in the real world. Only the lowest-ranked group member (omega) is clearly delineated, while the ranks of the other three are unclear. But we know they attack the omega whenever possible, and the omega always retreats.

When bystander loser effects alone are in operation, however, we get no discernible hierarchy鈥攁ll four players are equally likely to win or lose in a showdown. What鈥檚 more, these encounters are invariably aggressive, with both parties spoiling for a fight. According to our simulation, bystander loser effects are stronger than bystander winner effects, so if both are in operation the society is likely to be non-hierarchical and aggressive.

What happens when winner, loser and bystander effects are all in operation at the same time? As with the winner effect alone, group members are clearly ranked, but this time instead of fighting, most interactions take the form of attack and retreat. Sounds familiar? Unfortunately, no one has examined winner, loser and bystander effects in humans, but my bet is that when they do, we shall find that all of these are in play in our species. In the meantime, it is interesting to note that many human societies do seem to be organised in clear, linear hierarchies, but with not all that much real fighting going on鈥攋ust as our model predicts.

It seems amazing that something as simple as attitudes to winning and losing should have such huge implications for social organisation and the dynamics of aggression. But our models continue to surprise us with their predictive power. We are finding that these effects can explain other behaviours such as intervention鈥攚here an impartial bystander steps in to break up a fight. We鈥檙e used to this sort of behaviour in humans, but it also occurs in other primates and has even been observed in cichlid fish. Indeed, our model predicts that intervention is favoured whenever winner effects are in play. That鈥檚 because when an individual breaks up a fight between others, it prevents anyone from winning and getting 鈥渙n a roll鈥� which might later give them an advantage over the animal doing the intervening.

There鈥檚 no similar benefit to intervening when loser effects alone are in play, because here you鈥檙e breaking up a fight that will produce a loser who is likely to lose against you in the future. So we predict that these sorts of societies wouldn鈥檛 have peacemakers.

Of course, our cyberfighters and virtual peacemakers do not experience the myriad environmental factors that influence the behaviour of animals in the wild. But by paring social interactions down to basics we are starting to see that simple biases can drastically alter the shape a society takes and the levels of aggression exhibited by the animals within it. To get the full picture, we鈥檒l need to combine our findings with a lot more research on real animals鈥攁nd that鈥檚 not as simple as interacting with a computer screen.

But results are starting to emerge. For example, a recent study by Rui Oliveira and his colleagues at the University of Lisbon shed some light on the thorny question of how winning, losing or watching a fight affects an animal鈥檚 subsequent performance. Oliveira鈥檚 team found that when male cichlid fish observe a fight their androgen levels rise. The researchers suggest that this might fire up onlookers to attack losers鈥攁nd increase their chances of beating them.

In other words, these fish experience a surge of sex hormones rather like the surge of testosterone my friends from Brooklyn used to get from watching a good hockey fight.