Lions and lambs—can you solve this classic game theory puzzle?

July 20, 2017 by Amirlan Seksenbayev, The Conversation
Credit: Shutterstock

How many lions does it take to kill a lamb? The answer isn't as straightforward as you might think. Not, at least, according to game theory.

Game theory is a branch of maths that studies and predicts decision-making. It often involves creating hypothetical scenarios, or "games", whereby a number of individuals called "players" or "agents" can choose from a defined set of actions according to a series of rules. Each action will have a "pay-off" and the aim is usually to find the maximum pay-off for each player in order to work out how they would likely behave.

This method has been used in a wide variety of subjects, including economics, biology, politics and psychology, and to help explain behaviour in auctions, voting and market competition. But game theory, thanks to its nature, has also given rise to some entertaining brain teasers.

One of the less famous of these puzzles involves working out how players will compete over resources, in this case hungry lions and a tasty lamb. A group of lions live on an island covered in grass but with no other animals. The lions are identical, perfectly rational and aware that all the others are rational. They are also aware that all the other lions are aware that all the others are rational, and so on. This mutual awareness is what's referred to as "common knowledge". It makes sure that no would take a chance or try to outsmart the others.

Naturally, the lions are extremely hungry but they do not attempt to fight each other because they are identical in physical strength and so would inevitably all end up dead. As they are all perfectly rational, each lion prefers a hungry life to a certain death. With no alternative, they can survive by eating an essentially unlimited supply of grass, but they would all prefer to consume something meatier.

One day, a lamb miraculously appears on the island. What an unfortunate creature it seems. Yet it actually has a chance of surviving this hell, depending on the number of lions (represented by the letter N). If any lion consumes the defenceless lamb, it will become too full to defend himself from the other lions.

Assuming that the lions cannot share, the challenge is to work out whether or not the lamb will survive depending on the value of N. Or, to put it another way, what is the best course of action for each lion – to eat the lamb or not eat the lamb – depending on how many others there are in the group.

N=3. Credit: Shutterstock

The solution

This type of problem, where you need to find a solution for a general value of N (where N is a positive whole number), is a good way of testing game theorists' logic and of demonstrating how backward induction works. Logical induction involves using evidence to form a conclusion that is probably true. Backward induction is a way of finding a well-defined answer to a problem by going back, step-by-step, to the very basic case, which can be solved by a simple logical argument.

In the lions game, the basic case would be N=1. If there was only one hungry lion on the island it would not hesitate to eat the lamb, since there are no other lions to compete with it.

Now let's see what happens in the case of N=2. Both lions conclude that if one of them eats the lamb and becomes too full to defend itself, it would be eaten by the other lion. As a result, neither of the two would attempt to eat the lamb and all three animals would live happily together eating grass on the island (if living a life solely dependent on the rationality of two hungry lions can be called happy).

For N=3, if any one of the lions eats the lamb (effectively becoming a defenceless lamb itself), it would reduce the game to the same scenario as for N=2, in which neither of the remaining lions will attempt to consume the newly defenceless lion. So the lion that is closest to the actual lamb, eats it and three lions remain on the island without attempting to murder each other.

And for N=4, if any of the lions eat the lamb, it would reduce the game to the N=3 scenario, which would mean that the lion that ate the lamb would end up being eaten itself. As none of the lions want that to happen, they leave the lamb alone.

Essentially, the outcome of the game is decided by the action of the lion closest to the lamb. For each integer N, the lion realises that eating the lamb would reduce the to the case of N-1. If the N-1 case results in the survival of the lamb, the closest lion eats it. Otherwise, all the lions let the lamb live. So, following the logic back to the base case every time, we can conclude that the will always be eaten when N is an odd number and will survive when N is an even number.

Explore further: South Africa approves export of 800 lion skeletons this year

Related Stories

South Africa approves export of 800 lion skeletons this year

June 28, 2017

Some 800 skeletons of captive-bred lions can be legally exported from South Africa this year, the government said Wednesday, meeting demand for the bones in parts of Asia while alarming critics who believe the policy threatens ...

Recommended for you

Cells lacking nuclei struggle to move in 3-D environments

January 20, 2018

University of North Carolina Lineberger Comprehensive Cancer Center researchers have revealed new details of how the physical properties of the nucleus influence how cells can move around different environments - such as ...

4 comments

Adjust slider to filter visible comments by rank

Display comments: newest first

MD1
Jul 20, 2017
This comment has been removed by a moderator.
julianpenrod
not rated yet Jul 20, 2017
First of all, proceeding from one case to extend to all cases is often referred to just as "induction". "Backward induction" can be said to be to start at an extensive case then work step by step to simpler ones.
As is so often the case with game theory, an unknown factor is pulled pout of the hat. In this case, which lion is closest to the lamb. If they're all the same distance, is the conclusion the same? Even when they're not all the same distance, the lions further away can be motivated by the idea of having less of a chance of getting to the lamb before the others and so decide not to try.
Not to mention the idea that all individuals necessarily go through all this planning ahead of time. It appears an eminent deception to claim that there is necessarily any majority of cases where "game theory" has predicted how things occur.
antialias_physorg
5 / 5 (1) Jul 20, 2017
First of all, proceeding from one case to extend to all cases is often referred to just as "induction". "Backward induction" can be said to be to start at an extensive case then work step by step to simpler ones.

Induction is - in mathematics (of which game theory is a subset) - the method of constructing a proof for all natural numbers starting from 1 and then extending to the general case of n+1
Backward induction -while not an official term- is just starting from any n to n-1.
(Note that you can usually reformulate the problem in such a way that you get a normal induction in the direction of n+1)

In this case, which lion is closest to the lamb. If they're all the same distance, is the conclusion the same?

Since the lions (and the lamb) aren't said to be static (nor pointlike) and do feed off the grass you'll eventually end up with a lion closer.
antialias_physorg
5 / 5 (1) Jul 20, 2017
Even when they're not all the same distance, the lions further away can be motivated by the idea of having less of a chance of getting to the lamb

According to the reasoning in the article only the closest one ever tries (only in the N = odd case). The motivation of the others is irrelevant.
Even for the induction reasoning when a lion turns into a lamb only the closest lion to that 'new lamb' matters.

Not to mention the idea that all individuals necessarily go through all this planning ahead of time.

Which is the premise of the game ('rational lions').

That this doesn't apply to real scenarios can be seen in economics. Conservative economic theories posit a 'rational market' (and these theories only work if all players are fully rational - and omniscient)...guess what. In reality they aren't.

In other areas this 'game' applies (e.g. when applied to material sciences where atoms have 'full knowledge' of the surrounding forces)

Please sign in to add a comment. Registration is free, and takes less than a minute. Read more

Click here to reset your password.
Sign in to get notified via email when new comments are made.