Didier A. Depireux wrote:
>Make a little maze that looks like a tree graph, and put sugar at the end
>of one of the branches. Then see how long it takes for an ant to let other
ants
>at the bottom of the tree graph know that there is sugar at the end of one
of
>the branches. According to an article mentioned in Ballard's book "Natural
>Computation", it takes less time for the ants to communicate the position
of
>the sugar if there is a regularity to the path to the sugar (e.g. Left left
>left), than for the other situations. So there's a language between ants,
and
>it even uses a minimum description length kind of principle...
Not likely, ants communicate by means of traces of smell/taste, when you
follow
a regular path, this regularity must become visible before the end is
reached.
An ant can recognize and exploit it to speed up. This may seem weird, but of
course it works best on the most regular paths.
Mervyn
