Re: Genetically Anticipated Environment

From: Stevan Harnad (
Date: Sat Dec 27 1997 - 23:54:16 GMT

On Sat, 27 Dec 1997, Mike Dale wrote:

> Hi Steven! I'm from your PY311 class but I'm at home right now and have
> been doing some work. I am really stuck on the concept of the GEA
> (genetically anticipated environment). I am not really sure on what it is
> - could you give me a hand?
> Hope your Xmas was good
> Thanks
> Becky

Hi Becky (no need to remind me! I remember who you are! this reply
is from Montreal, Canada):

Actually, if you look in the skywriting archive you'll see
I've already replied about the GAE, but here it is from another

The genetically anticipated environment (GAE) is those features of the
environment that the genome "assumes" (I'll explain "assumes" in a
second) or "anticipates" it will encounter, so it need not code it in
in advance.

Here is an example: The newborn duckling needs to follow its mother if
it is to survive, and it also needs to recognise its own species, so
that when it grows up it knows what the kind of thing it wants to mate
with should look like. Now it would be possible to code in a "duck
shape icon" that the duckling is born with an urge to follow
reflexively. That might even be the "best" (optimal) way to design a
duckling. But the Blind Watchmaker does not design things optimally. He
is an opportunist who makes use of shortcuts if they are available.
(Translate that into adaptive talk: evolution's slective effect is
guided only by SUFFICIENT success to survive and reproduce (or to
survive and reproduce better than, or at the expense of, the next guy:
not for doing it the best possible way.)

So evolution, instead of selecting for a specific duck-detector merely
piggy-backed some fast learning on the first moving thing the duckling
is most likely to see after it is born (that's the "anticipation" or
"assumption," in what is obviously a blind succeed/fail process). There
was no problem with this, because 99.9% of the time, that was the
duckling's mother (because if it was something else, like a fox or a
crocodile, then the duckling was done for anyway, end of evolutionary
story). So the mother's being the first moving thing the duckling
encounters is "anticipated" by its genes.

Other examples come from growth of the embryo: as it grows, some parts
of the embryo connect with other parts of the embryo, not because they
have a specific detector for their "target" destination, but because in
the normal geometry of growth, whose timing and direction is under
genetic control, it is safe to "anticipate" that after so much time has
elapsed growing in a particular direction, the target site will be
the one reached anyway; the fact that there is no specific code for
finding or recognising the target has been shown with embryological
experiments in which the growing region is moved, so it does not meet
up with the "target" according to genetic "anticipation": In that case,
it will connect with the wrong target and the embryo will usually not
be able to grow to term. (It's muscles might work backwords, or in
opposition to one another, for example.)

I have given one example involving rapid early learning (imprinting)
in place of an explicit genetic coding of the shape of the mother,
and another example of relying on geometry so as to code growth and
connections less explicitly. Not coding things rigidly in the
genome means more potential flexibility, and flexibility is adaptive
in the same way that the potential variability arising from the
recombination of genes through sexual reproduction is advantageous.
It's one of the ways the Blind Watchmaker, though not capable of
optimising, can hedge some of its bets.

The biggest piece of flexibility is human language and cognition itself.
There are all kinds of things that do not have to be encoded in our
genes because it is part of our GAE that we will learn to talk and
will TELL one another all those things we need to know to survive.

In a way, the GAE is the same as the EEA. The EEA is the environment
in which our brains (and bodies) were "shaped". It is the
environment to which we are adapted. Radical departures from the EEA
can be a handicap, because it throws off our adaptive mechanisms by
going against genetic anticipations.

Perhaps a non-human example would be the best one: In "domesticating"
animals for our own purposes (including to kill and eat them),
we have taken advantage of the fact that those domestic animals'
GAE "assumed" that whoever fed and took care of them till adulthood
would be its mother or father; there is nothing to fear from them;
and if it were not their parent but a predator, they'd be done for
from the start. We have managed to throw that genetic "anticipation"
off. (A similar GAE is the reason human newborns are not afraid of
strangers until they reach the age when, in the EEA, they would have
been able to start wandering off on their own.)

One of the puzzles of the evolution of cognition is much more
challenging than the bread-and-butter issues of survival and mating
strategies in the EEA: Just how much of our potential
cognitive/linguistic world could the Blind Watchmaker have
"anticipated"? It surely could not have anticipated anything like this
course on the sociobiology of cognition! So the cognitive/linguistic
GAE must have been some kind of a generic environment in which language
and reasoning were useful survival strategies, so useful that,
advertently or inadvertently (as a spandrel) they opened the door to all
of future human knowledge.

This archive was generated by hypermail 2b30 : Tue Feb 13 2001 - 16:23:09 GMT