We have all seen this kind of situation.

The crack in the wall.

The ants coming into our kitchen or our bathroom,

searching for food,

and when they find it,

being able to attract all their little friends

so quickly so that they can collectively devour

whatever we dropped on the floor.

You have probably seen the vast trail systems

that they can create in your backyard or in public park somewhere.

Ants are really amazing in their ability

to collectively forge for food in a very efficient way.

We talked about this in a little bit in unit 1.

And we are going to revisit here in this unit

along with some amazing behaviors

of ant colonies.

So the basic idea of ant foraging,

as we talked about in unit 1,

is that ants come out of a nest

and move around randomly in different directions.

When an ant encounters a food source ,

it brings some food back to the nest.

but leaving a pheromone trail.

Where pheromone is a kind of chemical

excreted by the ant

that attracts other ants.

Ants encountering the trail are likely to follow it

in the abscene of reinforcement

the pheromone will dissipate

but as long as ants can continue

to follow the pheromone trail

and continue to find food,

then they will continue to reinforce it.

Here is a video of a famous experiment

that has been repeated many times.

in which ants are given the choice

of two different paths to food

a long path like this one.

and a shorter path.

Well the shorter path

ends up attracting more ants,

finally all of them

because the pheromone is able to

remain reinforced on the shorter side

more than it is on the longer side.

These ants going down this way

Their pheromone evaporates

often before other ants get to follow it.

So here we go.

You can see the ants very reliably

following this trail.

A few of them manage to go over here.

There is some randomness in what they do.

But for the most part,

they are following this shorter path

to the food.

You probably remember this model,

ants new.

It concise of ants nests

and three food sources

and it models ant foraging along the lines

of the algorithim I just described to you.

So if we do go,

we slow it down a bit.

The ants go out from the nest

if one of them encounters food,

they are moving around randomly,

if one of them encounters food

they take it back to the nest

laying a pheromone trail.

Let's run that again,

so we can see it more slowly

and other ants are attracted

to the pheromone trail.

They will follow it,

but if the trail is not reinforced by ants

like if the food source dries up

the trail evaporates eventually

to this trail evaporating.

But if there is still food along the trail,

the ants will find it and they will reinforce the pheromone trail.

So in this way the ants are able to build

the kind of model using their trails

to represent what they know collectively

about their food environment.

Using this model

we can do an experiment

to see what is actually the role

of the pheromone in helping the ants find the food.

So, to do that

we are going to do a simple experiment.

Where I'm going to turn off the pheromone.

To do that I'm going to

put the evaporation rate

at its highest level

which means that the pheronome will evaporate.

so fast the ants can't sense it.

So let's see what happens then

So here the ants have no pheromone to sense.

They are randomly finding food

They are not interacting with one another

and we will see how long it takes them.

I just speeded it up

This whole simulation stops when the food is gone.

ok and it took them 538 ticks to find the food

When there were five hundred ants.

I'll mark that down here.

We will say no pheromone, 538 ticks.

To find the food.

So know let's turn the pheromone back on again.

We'll put it down to two.

A very low evaporation rate.

We will see how that effects the behavior.

Of course because there is alot of randomness.

If I really wanted to get accurate numbers here

I'd have to this many times.

I'm actually going to let you do that in your own homework

and average the results.

I'll do this informal test here.

So we will see if there is a change in the time it takes

for all the food to be gone

and indeed it only took 415 ticks

so we will say with pheromone 415 ticks.

so this simplified model

Let's have some intuition of the role

of pheromone in helping ants

efficiently forage for food

Just to reiterate.

for ants that are foraging at any given time

the existing trails and the concentration of pheromone on those trails

encode the colonies collective information

about its food environment

moreover the information is able to adapt

to changes in environmental conditions

once the food is eaten for example

the trails will evaporate

and if new food is found then

new trails will be formed

this way the ants are able to

adaptively forage

even in a changing environment.

Now i want to briefly mention

another kind of self organizing behavior

in ant colonies and that's the notion

of task allocation..

Deborah Gordon is a biologist

at Standford University

who has done extensive work on many behaviors

in ant colonies.

But in particular we are going to talk about her work

on task allocation.

According to her,

talk allocation is the process that adjusts

the number of workers engaged in each task

in a way appropriate to the current situation

task allocation operates without central

or hierarchical control

as a self organizing activity

Gordon has studied task allocation in harvester ants

particular kind of ant

In this kind of ant

workers in a colony divide themselves

among a number of tasks

Nest maintenance, looking around cleaning up the nest.

here is a picture of a nest

Patrolling, foraging, and refuse sorting

and so on

So here are some representative pictures

of different kind of ants doing different kinds of tasks

but the question is how it at any given time

that ants decide tasks they should be doing.

and it turns out that ants actually switch tasks.

according to what is going on in their environment

so the number of workers pursuing each kind of task

adapts to changes in the environment

due to weather,

food availability,

predators

and so on.

For example Gordon showed that

the if the nest is distrubed

the number of nest maintenece workers

will increase

and the number of foragers will decrease

more ants are taking on the task of nest maintenaince.

when it is needed

and fewer are taking on the task of foraging

and she was able to do this experiment

by putting different toothpicks

at different places around the nests

and the ants really didn't like this evidently

and so the number of nests maintence

to remove the toothpicks increased.

She also found

that is the food supply is large

and high quality

the number of foragers will increase.

In short ants are able to choose tasks according to

what tasks need the most attention

and they also look at how many other ants

are already doing these tasks

How do they do this?

What is evident ants are not able

to get the big picture of what is going on

in their environment

They only interact locally

with other ants

and have to learn about what is going on globally

from limited interactions

similar to the other self organizing systems we've seen

so here is the question in a nutshull.

How does an individual ant decide which task to adopt

in response to nestwide environmental conditions

even though no ant directs

the decision of another ant

and each ant interacts

with a small number of other ants

Well here is what Gordon found

doing extensive experiments

She found that ants decided to adopt

a particular task such as foraging

or patrolling

or nest maintence

as a function of two things

1. What they encounter in the environment.

2. Their rate of interaction with ants

performing different tasks.

An Ant might encounter a toothpick

or some other debris on their nest.

and that would possibly make them switch

to do nest maintenece

but also they are influenced by whatever

ants they come in contact with

Gordon found that ants could actually tell

what job another ant is doing by

sensing chemical residue on its antennae.

of the other ant

here is a little video that Deborah Gordon's

research group took using a video scope

inside an ants nest.

and here you can see

ants coming out of the nests and into the nests

and actually touching each other

with their antennae

and that way they can sense

what job another ant has been doing.

if ants coming out of the nest sense.

many foragers coming back with food.

then they will be more likely to adopt the task

of being a forager.

So ants act as little statisticians.

they measure the rate of interaction

of ants they encounter doing different tasks

those rates are one of the things

they use to decide what tasks to adopt

it's not known how the brain

of ant is able to do this

but its seem from the data

that Gordon and her research collaborators

have taken is exactly what the ants are somehow doing.

So here is a little Puzzle

Gordon and her colleagues observed that

larger ant colonies are more deterministic

and more consistent in task allocation

than smaller ant colonies

Why do you think this is?

Pause the video for just a minute

and think about it.

Why would a larger aunt colony

be in sense better at optimally allocating tasks

in a consistent and deterministic way

than smaller colonies

Well the hypothesis

that Gordon and her collegues put forth

that ants in larger colonies

basically can get better statistics

on interaction rates

because they have more samples

each ant has on average more sample

of interactions with other ants

and so they get better statistics

so next time you see ants

in your kitchen, in your bathroom, or out in your garden.

or even crawling on your leg.

Just remember that

ants are very brilliant, little statisticians.

Think how great it would be

take advantage of their statistical abilities.