Alternative to response counters for selection from lists with fixed reward/loss?

Wed Jun 3 12:59:39 UTC 2009

Hey!

I couldn't let go of it and got another solution. I made a screenshot
of it which can be found here: [img]http://images.redial.net/kkat.bmp[/
img]

It's basically a rearrangement of the experiment that you have already
with a few elements copied and a couple of elements added. What it
does: it starts the experiment and a trial and then based on the
answer given it continues to load either one of the decklists. For
this the 'procedure' for each trial had to be split in two: the start
of the trialproc is just as it was allready, and everything that was
happening AFTER the answer is given is now moved to a separate
procedure called 'restproc'. (which, I realise now, makes sense in
dutch but not in english >.< rename it to somethign that's convenient
to you).  I realise that the whole thing looks rather messy on the
screenshot but it will make sense if you rebuild it. (again: i'd send
you the file but it won't work as I am in e-prime 2 and you're on 1).

The inline called respchoosedeck makes the program skip to one of four
labels based on the response given (labelrespA - labelrespD). These
labels are placed on the trialproc and each label is followed by it's
corresponding decklist. Still on the trialproc each of the decklists
is followed by another piece of inline that tells the program to skip
to the end of the trialproc (to endlabel) if that inline is
encountered.

Let's move to the next level: inside each of the four decklists you
should nest a new proc (restproc in my case, call it what you like),
and place the inline called "choose1234"and the slide called "FB" on
this proc. Make all the levels of all the four decklists refer to this
very same  'restproc' .

The last pivotal setting is to change the sampling criteria on each of
the decklists: a cycle should be all 100 samples, yet the list must be
exited after 1 sample (and selection stays sequential).  Now you're
set!

The only drawback is that your errorbar does no longer seem to work,
but that might also be due to something else that I tried before
founding this solution... I didn't notice when it stopped working.

Let me know how this works for you!

AW

On Jun 2, 9:54 pm, KKat <kathy.smolew... at gmail.com> wrote:
> Dear members:
>
> I posted a few weeks ago with regard to a modified version of the Iowa
> Gambling Task that I was working on and I received great initial
> advice from a fellow member on how to use nested lists to achieve my
> goal of selecting from four different lists on each trial. However, I
> now have another problem, so here it goes....
>
> Currently, I have a List object (named DesignList) which consists of
> one row and the following under the Nested column: DeckA, DeckB,
> DeckC, Deck D.
>
> It looks like this:
>
> ID    Weight     Procedure
> Nested                                    Attributes.....
> ---------------------------------------------------------------------------------------------------------------------------------
> 1         1          TrialProc     DeckA,DeckB,DeckC,DeckD
>
> Then, I have four separate lists (DeckA, DeckB, DeckC, DeckD) created
> under the DesignList.
> Each of these lists has the same structure, which looks like this:
>
> DeckA:
>
> ID    Weight     varwin1      varloss1
> ---------------------------------------------------------------------------------------------------------------------------------
> 1         1          +100            0
> 2         1          +100            0
> 3         1          +100            -150
> etc.
>
> DeckB:
> ID    Weight     varwin2      varloss2
> ---------------------------------------------------------------------------------------------------------------------------------
> 1         1          +100            0
> :          1          +100            0
> 10        1          +100         -1250
> etc.
>
> I am able to sample each list (i.e., the program is selecting from the
> correct list and all lists work) and the task works ok as it's coded.
> However, it's not doing what I need it to do. In this task, it is
> imperative that the
> losses/rewards in each deck are presented in a specific, non-random
> order so that the experimenter can assess for sensitivity to reward
> and punishment, as well as reinforcement learning.
>
> Although the sampling is set to "sequential" for each list, it does
> not start on ID_1 for each deck --- for example, if I start selecting
> from DeckA and then move to DeckB for the second trial, it selects the
> [varwin2] and
> [varloss2] values from row ID_2; if I then select DeckD, it selects
> the [varwin4] and [varloss4] from row ID_3.
>
> I would like the sampling for each deck to be independent from the
> others - the program should refer to row ID_1, then ID_2, then ID_3
> and so forth from EACH deck. In other words, the wins/losses should be
> fixed (i.e., if any participant chooses DeckX for the third time, they
> will all win/lose the same amount).
>
> I have been informed that when using nested lists, e-prime can only
> select one level from each list for each trial and  cannot select one
> of the lists based on the answer given. This means that for trial 1
> the four first levels of the four deck-lists are all four loaded by e-
> prime and based on the answer one of them is 'shown' to the subject.
> Because all four first levels are loaded e-prime will load the four
> second levels for the next trial etc. This means that a level for, for
> instance, deck B is 'used' whether or not the subject chooses this
> deck.
>
> Instead, it was suggested to me that I should delete the lists and
> code the wins/losses using response counters. My understanding is that
> I would would need to declare variables that count each response given
> (for instance 'respbcounter' counts how  often the response is b).
> Then, based on the respbcounter, the win/loss is
>  determined for a trial (if respbcounter is 8 then varwin2  = +100 and
> varloss2 = -0;  if respbcounter is 9 then varwin2 = +100 and varloss2
> = -1250). I would have to do the same for the three other decks.
>
> I have never used response counters or nested lists before this task -
> can anyone confirm whether or not response counters will solve my
> problem?? Or, does anyone have any other suggestions?
> Thank you!
> Kathy
--~--~---------~--~----~------------~-------~--~----~
You received this message because you are subscribed to the Google Groups "E-Prime" group.
To post to this group, send email to e-prime at googlegroups.com
To unsubscribe from this group, send email to e-prime+unsubscribe at googlegroups.com
For more options, visit this group at http://groups.google.com/group/e-prime?hl=en
-~----------~----~----~----~------~----~------~--~---