Re: Framsticks v2.0? 

Maciej and Brad,

I've had a great time with "Headless Chicken" as it like a comic drunk
constantly falling down!

I am trying to learn to "sight read" the .gen code (like reading a musical
score) and find this very difficult. (I get bogged down very quickly). I've
tried to format the list of "genetic" instructions to make it easier to
visualize what is connected to what but it is very difficult to do this for
complex figures and neural connections. I am replying to the thread on
"Framstiks v2.0?" because I'm hoping that perhaps there could be some
consideration of showing which stick correspond to which "X" in the listing
and also, which neuron corresponds to which "[]" in the listing. This could
be done by assigning a numeric value to the stick/neuron as its ordering
appears in the .gen listing. Ideally, the visual display of such assigned
numbers would toggle on/off.

Maybe I just don't have the proper attitude to figuring out this listing --
obviously many users do not have a hard time "reading" the specific structure
in this code. Maybe I'll discover the trick. For now, I think I'll just edit
the list to remove components and see which stick/neuron disappears. (in
neurophysiology research, this technique is called 'ablation').

Thanks for any comment or advice you can give. I really have been enjoying
my time with this and appreciate everyone's effort.

Bill Crandall

Maciej Komosinski wrote:

> > When will v2 be released? I don't need an exact date. Just a month or a
> > quarter or something...
>
> Seems like August.
>
> MacKo

Forums: 

>I am trying to learn to "sight read" the .gen code (like reading a musical
>score) and find this very difficult. (I get bogged down very quickly).

...
For now, I think I'll just edit
>the list to remove components and see which stick/neuron disappears. (in
>neurophysiology research, this technique is called 'ablation').

Hi Bill,

Instead of ablation, I taught myself to sight read the genome by staining
(another tried and true neurophysiological technique!). Take a creature,
like the four leg food finder, and put a bunch of modifiers, like AAAAAAA on
one of the sticks. Drop it into the world and view it with openGL. You
have just stained one of the sticks bright green. Do this for each of the
sticks in order, and it will soon become quite clear that there is a
straightforward and relatively intuitive correspondence between the serial
positions in the genotype and the placements of sticks and neurons in the
phenotype.

Cheers,

Pete

Thanks Pete and Maciej for the prompt reply.

I agree, Pete, that the "staining" technique is an option... a better one. In
fact, the "ablation" technique did not work so well for identifying the neurons
because the rewiring necessitated by this procedure left ambiguities as to which
particular neuron graphic was removed by a particular ablation (in some cases).
Given your good solution to the analysis problem, too bad the neurons,
themselves, are not "colored" to match the stick color they are associated
[with]. Relying on decoding the "relative addressing" of the neuron location
makes matching the listing and the graphics difficult.

Also, Maciej, I kept away from F1 because the "Introduction" gave me the
impression F0 was "easier" -- and I'm just beginning. However, I will now check
it out. "Fred" looks interesting. The following point may not be appropriate
given your recommendation to use F1, but I did think of an alternative to having
to graphically number ALL components -- that is to have each component's screen
location "tagged" so that it will display the number of that component (ordinal
number relating to its position in the listing) when the cursor points to it. I
don't mean to 'beat a dead horse' and I understand such things are always so
much more complicated than just suggesting them!

Thanks again for the suggestions...

Bill Crandall

Pete Mandik wrote:

> >I am trying to learn to "sight read" the .gen code (like reading a musical
> >score) and find this very difficult. (I get bogged down very quickly).
>
> ...
> For now, I think I'll just edit
> >the list to remove components and see which stick/neuron disappears. (in
> >neurophysiology research, this technique is called 'ablation').
>
> Hi Bill,
>
> Instead of ablation, I taught myself to sight read the genome by staining
> (another tried and true neurophysiological technique!). Take a creature,
> like the four leg food finder, and put a bunch of modifiers, like AAAAAAA on
> one of the sticks. Drop it into the world and view it with openGL. You
> have just stained one of the sticks bright green. Do this for each of the
> sticks in order, and it will soon become quite clear that there is a
> straightforward and relatively intuitive correspondence between the serial
> positions in the genotype and the placements of sticks and neurons in the
> phenotype.
>
> Cheers,
>
> Pete

Maciej Komosinski's picture

Dear Bill,

> I am trying to learn to "sight read" the .gen code (like reading a musical
> score) and find this very difficult. (I get bogged down very quickly). I've
> tried to format the list of "genetic" instructions to make it easier to
> visualize what is connected to what but it is very difficult to do this for
> complex figures and neural connections. I am replying to the thread on
> "Framstiks v2.0?" because I'm hoping that perhaps there could be some
> consideration of showing which stick correspond to which "X" in the listing
> and also, which neuron corresponds to which "[]" in the listing. This could
> be done by assigning a numeric value to the stick/neuron as its ordering
> appears in the .gen listing. Ideally, the visual display of such assigned
> numbers would toggle on/off.
>
> Maybe I just don't have the proper attitude to figuring out this listing --
> obviously many users do not have a hard time "reading" the specific structure
> in this code. Maybe I'll discover the trick. For now, I think I'll just edit
> the list to remove components and see which stick/neuron disappears. (in
> neurophysiology research, this technique is called 'ablation').

The f1 format you are talking about is very easy, so
you will surely discover the trick. Speaking about
visual presentation, we are developing Fred:

http://www.framsticks.com/dev/fred/fred

Best,

--------------------------------------------------------------
Maciej Komosinski, Ph.D. Maciej.Komosinski@cs.put.poznan.pl
Institute of Computing Science Fax: +48 61 8771525
Poznan University of Technology, Poland
--------------------------------------------------------------