Big problem with selectmode


Author
Message
akifurai
akifurai
Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)
Group: Forum Members
Posts: 5, Visits: 43
I've just tried to replicate the Valent/Neutral Categorisation Task (Werner & Rothermund, 2012) in Inquisit 3, and encounter con with the following troubles:
first, I used stimulustimes attribute to present the trial. I know the LCD refresh rate might have something to do with the accuracy of time, but it doesn't mean a lot to me. Well I run the Monkey and everything seems all right. Then I checked the data, and suddenly discovered that there's something wrong with the stimulusonsets. The trial should present in the sequence of fixation(500ms) - blank screen(200ms) - priming(150ms) - blank screen(50ms) - target(200ms) - interval(200ms). But what I found was:
stimulusonsets1 stimulusonsets2 stimulusonsets3 stimulusonsets4 stimulusonsets5
5 855 705 855 905
My script:
<text prime_PN>
/items =prime_PN
</text>
<text target_PN>
/items =target_PN
</text>
<text fixation>
/items=("+")
/fontstyle=("Arial",48px)
</text>
<text blank>
/items=("  ")
</text>
<trial PN_PN_L>
/validresponse=("D","K")
/correctresponse=("D")
/stimulustimes=[1=fixation;400=blank;700=prime_PN;850=blank;900=target_PN]
/posttrialpause=250
</trial>
Maybe anyone can tell me what's wrong with the second stimulus?

Second problem is about the blank screen or the mask according to what I actually put into use. If I set the text stimulus for the blank trial with no string, one of the priming text item will take the place. So I have to set the stimulus with several spacebars. It works, but I would prefer any better alternative if it is possible.

Third, and the most critial problem is, I couldn't set the assignment perfectly balanced as the original paper done.According to the paper, amount of positive and negative primes are both 20, and neutral primes are 40. The assignment of targets are the same. Each target will be paired with one positive, one negative and two neutral primes. That will make the total pairs(trials) upto 320. The valence relations and affective congruency will be perfectly balanced (valent prime - neutral target 25%,neutral prime - valent target 25%, valent prime - valent target 25%, neutral prime - neutral target 35%; of valent - valent trials, affective congruency and affective incongruency are both 50%)
Well I cut down the amount to half, so there will be 160 trials totally. And I tried different ways to put down my scripts but all failed in getting the balanced assignment. It's interesting that when I used 8 items (positive and negative primes and targets are all 2) to create a simple evaluative priming task, every target was presented two times while the selection of primes were always unequally. And I just switched the prime and target items, I still found the same result. I post my scripts down here, and beg anyone to help me find out what the problem really is.
Trials are all in the same form as I post above. And I've also tried to change the stimulustimes to stimulusframes, and the result is the same.
<block pr1>
/trials=[1-8=noreplace(PN_PN_Lp,NN_NN_Lp,PN_NN_Lp,NN_PN_Lp)]
</block>
And the following script is one block of the valent/neutral task. PN_NN means postive prime paired with negative target (n stands for noun, what I set for further design of experiments), the other trial names are in the same way, and UN means neutral. I set 5 affective congruency and 5 incongruency in one block in order to make the balance.
<block L1>
/trials=[1,9,17,25,33=noreplace(PN_PN_L,NN_NN_L);5,13,21,29,37=noreplace(PN_NN_L,NN_PN_L);2,6,10,14,18,22,26,30,34,38=UN_UN_L;4,8,12,16,20,24,28,32,36,40=noreplace(UN_PN_L,UN_NN_L);3,7,11,15,19,23,27,31,35,39=noreplace(PN_UN_L,NN_UN_L)]
</block>

Thank you a lot for reading my poor description.
Dave
Dave
Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)
Group: Administrators
Posts: 11K, Visits: 67K
#1: There is nothing technically wrong with those stimulusonsets. Since you display the 'blank' stimulus twice -- once at ~400, once at ~850 - both stimulusonset2 and stimulusonset4 reflect the onset time of the stim's *last* instance.

You can either (a) use different 'blank' stims, i.e.,

/ stimulustimes = [...; 400=blank1; ...; 850=blank2; ...]

or explicitly log the stims multiple onsets to the data file

<data>
/ columns = [..., text.blank.stimulusonset.1, text.blank.stimulusonset.2, ...]
[...]
</data>

#2: Specify an appropriate /size attribute for your 'blank' <text> element to have it fully overwrite any preceding stimuli.

#3: Your third question cannot be answered based on the information you provided -- it is incomplete. Let me point out a couple of general things, though:

(a) How you sample trials at the <block> level has *nothing to do with* how the respective <trial> elements sample items from stimulus elements they display (e.g. <text> elements). Information on the latter is completely missing right now.

(b) Unless you specify otherwise, every stimulus element's selection is completely independent from every other stimulus element's selection. I.e. a selection occurring for

<text mytext1>
/ items = myitems
</text>

<item myitems>
/ 1 = "A"
/ 2 = "B"
/ 3 = "C"
</item>

does *not* in any way affect the available items / selection for

<text mytext1>
/ items = myitems
</text>

Both elements have 3 items, if "A" is selected for 'mytext1', that *does not* remove "A" for selection in 'mytext2'. You can think of them as two separate decks of cards.

(c) At the block-level, trial selection occurs according to the proportions given. I.e.

<block myblock>
/ trials = [1-8=noreplace(a,b,c,d)]
</block>

will result in 2 a-trials, 2 b-trials, 2 c-trials and 2 d-trials in random order.

<block myblock>
/ trials = [1-8=noreplace(a,a,b,c)]
</block>

will result in 4 a-trials, 2 b-trials and 2 c-trials in random order.

(d) The selection pools you specify in the /trials attribute are *independent*. I.e.,

<block myblock>
/ trials = [1-3=noreplace(a,a,b,c); 4-8=noreplace(a,a,b,c)]
</block>

what happens for trials 1 to 3 *does not* affect selection for trials 4-8. The two noreplace() define completely separate and independent selection pools. If that's not what you want, don't do it!

akifurai
akifurai
Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)
Group: Forum Members
Posts: 5, Visits: 43
Thanks for replay quickly, Dave.

#1: I take your advise to use two different 'blank' stims, and it really works. thx a lot

#2: Yep, that's exactly what I do ;P

#3: Sorry that I couldn't describe more clearly. The text elements are also mentioned above, like this
<text prime_PN>
/items =prime_PN
</text>
You mention by using
<block myblock>
/ trials = [1-8=noreplace(a,b,c,d)]
</block>
will result in 2 a-trials, 2 b-trials, 2 c-trials and 2 d-trials. Well, I check the data file and find the trials distribution is right. But what I really expect is that items will be evenly select, i.e.

<item prime>
/1="a"
/2="b"
</item prime>
<item target>
/1="1"
/2="2"
</item>

<text prime>
/item=prime
</text>
<text target>
/item=target
</text>

<trial my1>
/stimulustimes=[...;400=prime;500=target;...]
</trial>
<trial my2>
/stimulustimes=[...;400=prime;700=target;...]
</trial>

<block my_prime>
/trials=[1-4=noreplace(my1,my2)]
</block>

There will surely result in 2 my1-trials and 2 my2-trials. My question is: if it's possible to presnt a-1, a-2, b-1, b-2?

Dave
Dave
Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)
Group: Administrators
Posts: 11K, Visits: 67K
> My question is: if it's possible to presnt a-1, a-2, b-1, b-2?

Of course, but that's not what you instructed your code to do!

Think of <trial my1> as your *left hand* and <trial my2> as your right hand.
Think of <text prime> as a deck of two cards, labelled A and B respectively.
Think of <text target> as another deck of two cards, labelled 1 and 2.

As per your <block>'s /trials, you will draw cards from the two decks twice with your left hand (my1) and twice with your right hand. Let's say the random (trial) order is left-right-right-left or my1-my2-my2-my1. What happens?

- In the 1st trial (my1), your left hand draws one card from the "prime" deck and one card from the "target" deck, lets say B and 1 respectively.
- In the 2nd trial (my2), there's only one card left in every deck, so A and 2 are going to end up in your right hand. Guaranteed.
- At this point no cards are left at all. Both decks will be reshuffled and *all cards become available again*.
- In the 3rd trial then (my2), you might again end up with A and 2 in your right hand.
- etc.


akifurai
akifurai
Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)
Group: Forum Members
Posts: 5, Visits: 43
Now I understand the mechanism of selection.
Just follow your metaphor. My *left hand* pick B and 1, thus the *prime* deck only leaves A while *target* leaves 2. So it's only possible for my *right hand* to pick A and 2.
Since no more cards in both decks, they are reshuffled and *now all cards become available again*.
Now here's the point. How to guarantee that my *left hand* will pick B and 2 this time?




Dave
Dave
Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)
Group: Administrators
Posts: 11K, Visits: 67K
In keeping with the metaphor: Draw from separate decks with your left and right hand. I.e. define additional text elements and have different <trial> elements draw from different <text> elements.

akifurai
akifurai
Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)
Group: Forum Members
Posts: 5, Visits: 43
Dave (2/28/2014)
In keeping with the metaphor: Draw from separate decks with your left and right hand. I.e. define additional text elements and have different <trial> elements draw from different <text> elements.

Finally I achieved the expected assignment. Thanks a lot Dave.
Another question has come up with me. I want the posttrialpause of odd and even trials variant. Is there any simple solution?



Dave
Dave
Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)
Group: Administrators
Posts: 11K, Visits: 67K
The /pre- and /posttrial attributes accept expressions and the like, so you can make it vary between trials quite easily.

akifurai
akifurai
Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)Partner Member (742 reputation)
Group: Forum Members
Posts: 5, Visits: 43
Thanks, Dave.

And another question about radiobuttons element. I've already set the txcolor white, screencolor black and txbgcolor transparent as default.
The option labels of radiobutton are invisiable on the black screen. I've tried set txcolor white in the radiobutton element and it doesn't work for the option labels.
Any suggestion?

Dave
Dave
Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)Supreme Being (922K reputation)
Group: Administrators
Posts: 11K, Visits: 67K
The text color for survey question options cannot be changed. It will always be black. You thus need to set the screen color to some light color (white, some shade of grey) for them to be visible.

GO

Merge Selected

Merge into selected topic...



Merge into merge target...



Merge into a specific topic ID...




Reading This Topic

Explore
Messages
Mentions
Search