Millisecond Forums

Actually, belatedly, I  think I understand, why the not operation fails.

It is operating on the  x,y co-ordinates separately, hence not pic1 x coordinate if it was 25% will only allow 75% in the grid and so on. I suppose the x,y combined coordinate is what must be unique for each picture, but I still cannot work out how that can be done.

Perhaps use the position parameter with an enclosed expression. I'll try that ,,,,

Yes.

You'll want to define a dependency between the respective x and y counters *and* make the values unique:

<counter type1xpos>
/ items = (25.0000%,25.0001%,75.0000%,75.0001%)
/ select = noreplace
</counter>

<counter type1ypos>
/ items = (25.0000%,75.0000%,25.0000%,75.0000%)
/ select = current(type1xpos)
</counter>

etc.

Thank you for this Dave, but I am going to need a bit more guidance if  you have the time.

I have tried to implement your suggestion here but still get clashes so I am obviously not making the selection unique.

I have tried to understand the select current (countername) but the documented example of presidents and vice-presidents did not help as I cannot see how that works without seeing the item list.

In my case if the current x coordinate selected is 25.001 then a choice of any of the others seems possible.

Reading the documentation on the /not property has not helped me either.

Your comment of defining dependency and making unique values seemed clear and a guiding light for me, but implementing seemed as difficult as understanding recursion.

Any other tips gratefully received.

Latest relevant code:

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 Candidate positions for pic2 stimuli (all positions but *not* the
 one already chosen for pic1  stimulus)
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

<counter type2xpos>
/ items = (25.0000%,25.0001%,25.0002,75.0000%,75.0001%,75.0002%)
/ not = (type1xpos,type1ypos)
/ select = noreplace
/ selectionrate = always
</counter>
<counter type2ypos>
/ items = (25.0000%,75.0000%,25.0000%,75.0000%,25.0000%,75.0000%)
/ not = (type1ypos,type1xpos)
/ select = current(type2xpos)
/ selectionrate = always
</counter>

BTW when the display of the current position of X is created, sometimes the x co-ordinate may be something like 25.00000000000000097344445555.

How does that work?

With great appreciation for your help as always.

Patrick

Patrick,

note the differences between your previous example and your current code. The above counters effectively contain the same position multiple times, so of course things will clash.

Computers have no concept of real (in the mathematical sense) numbers. They deal in binary. Floating point representations of reals are thus subject to rounding error and imprecision.

Thanks again Dave.

I feel you are teaching the village idiot how to fish rather than throwing him a fish and I appreciate this.

I assume you indicate with the red text that these are the redundant entries.  I just don't see what you mean by the same position multiple times.

If I give my narrative of how I understand it to work, you may see where my thinking is awry.

Using your hints, my current code runs correctly over ten tests for just two of the images, however as soon as I extend it to three pictures collisions abound.

<counter type1xpos>
/ items = (25.0000%,25.0001%,75.0000%,75.0001%)
/ select = noreplace
</counter>
<counter type1ypos>
/ items = (25.0000%,75.0000%,25.0000%,75.0000%)
/ select = current(type1xpos)
</counter>

+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++
 Candidate positions for pic2 stimuli (all positions but *not* the
 one already chosen for pic1  stimulus)
+++++++++++++++++++++++++++++++++++++++++++++++++++++++++++++

<counter type2xpos>
/ items = (25.0001%,25.0001%,75.0001%,75.0001%)
/ not = (type1xpos,type1ypos)
/ select = noreplace
/ selectionrate = always
</counter>
<counter type2ypos>
/ items = (75.0000%,75.0000%,25.0000%,25.0000%)
/ not = (type1ypos,type1xpos)
/ select = current(type2xpos)
/ selectionrate = always
</counter>

CORRECTION: Testing again, there were collisions.

If I just leave the items  in red, not understanding why they don't create the same position  multiple times, I get collisions even more frequently.

I am wandering blind and just throwing my line in without really knowing how to bait the hook as I am simply not getting this basic concept of creating unique positions.

 I have tried introducing new positions to strive for 'different' positions, eg 25.0003 and 25.0004 etc., but this doesn't appear to work either.

Also I am not getting how the select = current(type2xpos) works.

When I look at the example in the documentation,

<text vicepresidents>
/ items =
vicepresidentitems
/ select = current(presidents)
</text>

The following first selects Jefferson, then Adams, and then Washington:

<text presidents>
/ items = ("George Washington",
"John Adams", "Thomas Jefferson")
/ select = sequence(3, 2,
1)
</text>

my understanding is that the vicepresidentitems would look something like:

<text vicepresidents>

<items = ( "Adams", "Jefferson","Burr")

</text>

Hence at first choice, current(presidents) returns 3, this then is the selection from vicepresidentlist, #3 -  Burr ?

So extending this to picking a vposition after choosing a hposition, if say 25.001 was chosen at the second position, then

/select = current(type2xpos), looks for the second choice in the list, say 75% in the above example, ignoring for the moment the red text.

Is that correct?

I can't rationally proceed until I know if I am on the right track with this or totally lost at sea.

Any life lines handy Dave?

Patrick

(Thanks for the explanation on the floating  point representations, that made sense.)

Late Press:

This code produced no collisions over 20 tests:

<counter type2xpos>
/ items = (25.0001%,25.0001%,75.0001%,75.0001%)
/ not = (type1xpos,type1ypos)
/ select = noreplace
/ selectionrate = always
</counter>
<counter type2ypos>
/ items = (75.0000%,25.0000%)
/ not = (type1ypos,type1xpos)
/ select = current(type2xpos)
/ selectionrate = always
</counter>

However, extending what I thought I had "seen" to the third image co-ordinates resulted in multiple collisions again. :-(

Here's a short, fully functional snippet that should hopefully be helpful. It's just a straightforward extension of the original example script to two dimensions (x and y coordinates instead of x coordinates only):

<block myblock>
/ trials = [1-4=mytrial]
</block>

<trial mytrial>
/ pretrialpause = 500
/ stimulusframes =  [1=type1,type2a,type2b,type2c]
/ validresponse = (57)
/ ontrialend = [reset(counter.type2xpos)]
</trial>

<text type1>
/ items = ("Type 1")
/ hposition = counter.type1xpos.selectedvalue
/ vposition = counter.type1ypos.selectedvalue
</text>

<text type2a>
/ items = ("Type 2a")
/ hposition = counter.type2xpos.selectedvalue
/ vposition = counter.type2ypos.selectedvalue
</text>

<text type2b>
/ items = ("Type 2b")
/ hposition = counter.type2xpos.selectedvalue
/ vposition = counter.type2ypos.selectedvalue
</text>

<text type2c>
/ items = ("Type 2c")
/ hposition = counter.type2xpos.selectedvalue
/ vposition = counter.type2ypos.selectedvalue
</text>

<counter type1xpos>
/ items = (25.0000%, 25.0001%, 75.0000%, 75.0001%)
/ select = noreplace
/ selectionrate = trial
</counter>

<counter type1ypos>
/ items = (25.0000%, 75.0000%, 25.0000%, 75.0000%)
/ select = current(type1xpos)
</counter>

<counter type2xpos>
/ items = (25.0000%, 25.0001%, 75.0000%, 75.0001%)
/ select = noreplace
/ not = (type1xpos)
/ selectionrate = always
</counter>

<counter type2ypos>
/ items = (25.0000%, 75.0000%, 25.0000%, 75.0000%)
/ select = current(type2xpos)
</counter>

Thank you so much Dave. Of course it works perfectly for my situation. Sorry I was so  obtuse about getting to grips with the principles here. If you were British, I would recommend you for a knight hood.

Patrick

Absolutely no need to be sorry. I wish I could have made what I meant clearer from the get-go, but I couldn't quite put my finger on where the actual problem / misunderstanding was (I'm still not sure, really). Anyway, if you need me to elaborate on anything, let me know.

That would mean I'd have to rescue maidens and slay dragons in the process, wouldn't it? Thanks, but I'll give this a pass. I'm not good with swords... :-)

Cheers,

~Dave