Millisecond Forums » Millisecond Forums » Inquisit 2.0 » Dot Probe Experiment Wanted Please!

Dot Probe Experiment Wanted Please!

Lucy and Shiv — Thu, 22 Jul 2010 07:10:19 GMT

Hi, We currently have our dot probe experiement on eprime but we want to switch to inquisit.

Does anyone have a sample experiement we could use please? Our study involves presenting two photos (one critical face and one compassionate face) on the screen for 500 milliseconds and then replacing one of the photos with either a : or ..

The particpant is required to click on on the dots as quickly as possible. And hopefully reaction times will show us where the attentional biases were.

Thank you!

Lucy and Shiv.

RE: Dot Probe Experiment Wanted Please!

Dave — Thu, 22 Jul 2010 07:10:19 GMT

[quote]

What email address shall I send it to?

[/quote]

Check your private forum messages.

RE: Dot Probe Experiment Wanted Please!

Lucy and Shiv — Thu, 22 Jul 2010 07:02:27 GMT

What email address shall I send it to?

RE: Dot Probe Experiment Wanted Please!

Dave — Thu, 22 Jul 2010 03:48:57 GMT

Sorry, I got lost somewhere along the way. As a last resort, can you send me the original E-Prime task that you're trying to recreate via email? If you do, please include all necessary supporting materials (pictures, etc.) and, again if possible, please send me the file in E-Prime 1.x format (*.es). Of course I won't make any of the enclosed materials available to any third party. I'll just reverse-engineer the original and re-implement it in Inquisit-lingo -- maybe that'll clear some of your questions up.

~Dave

RE: Dot Probe Experiment Wanted Please!

Lucy and Shiv — Thu, 22 Jul 2010 03:35:27 GMT

Hi,

Ok I'll just do two blocks then, one practice and then one for compassion and critical.

With regards to the correct column query in the output - How about if I double my trials and for each one add a '1' or '2' to the trial name. 1 = .. and 2 = : and define the correct response....

E.g. Instead of just one trial f1aCOMPneutral*, it is now:

/ pretrialpause = 500

/ stimulustimes = [1= focuspoint; 500 = rectangle_pic; 1000=f1compassionleft_pic,f1neutralright_pic; 1500 = rectangleright_pic, rectangleleft_pic; 2000= replace(target1left); 2500 = rectangleright_pic, rectangleleft_pic]

/ inputdevice = keyboard

/ responsetime = 2000

/ validresponse = ("a", "l")

/correctresponse = ("a")

/ posttrialpause = 180

</trial>

/ pretrialpause = 500

/ stimulustimes = [1= focuspoint; 500 = rectangle_pic; 1000=f1compassionleft_pic,f1neutralright_pic; 1500 = rectangleright_pic, rectangleleft_pic; 2000= replace(target2left); 2500 = rectangleright_pic, rectangleleft_pic]

/ inputdevice = keyboard

/ responsetime = 2000

/ validresponse = ("a", "l")

/correctresponse = ("l")

/ posttrialpause = 180

</trial>

This means I will have 224 trials in the trials document. I still want 112 trials to run in the experiment, and for each trial name just one out of the options trial f1aCOMPneutral*1 OR trial f1aCOMPneutral*2 for example.

Is this possible? (before I change all my trials!) Can I write in the </expt> bit to run 112 out of the 224 and choose either 1 or 2.

E.g. logically I'd guess it would be this (but obviously it won't be in inquisit language!)

/trials = [1-112 = noreplace (f1aCOMPneutral*1, f1aCOMPneutral*2) (f1bcompNEUTRAL1, f1bcompNEUTRAL2) then list all 224 trials....

Thanks again,

Sorry I don't understand any of it!

RE: Dot Probe Experiment Wanted Please!

Dave — Thu, 22 Jul 2010 00:51:02 GMT

[quote]If I do it how you've explained won't they all just be under the same block name?[/quote]

Yes. Look, a <block> is a self-contained unit. If you want to have both critical/neutral and compassionate/neutral trials run in random order, they need to be contained in the same block.

[quote]The idea of having two blocks was that in the output I'd have some trials with 'Compassion' in the block column and some with 'Critical', then I could easily sort them on Excel so I can look at the two conditions separately.[/quote]

Instead compute a new variable in Excel based on the trialcode column (which should tell you if a given trials was compassionate/neutral or critical/neutral along with a bunch of other things) and sort based on this new variable.

~Dave

RE: Dot Probe Experiment Wanted Please!

Lucy and Shiv — Thu, 22 Jul 2010 00:20:05 GMT

Thanks Dave

<block compassion>

/ preinstructions = (taskreminder)

/ trials = [1-56 = noreplace(LIST ALL COMPASSION/NEUTRAL TRIALS)]

</block>

and

<block critical>

/ preinstructions = (taskreminder)

/ trials = [1-56 = noreplace(LIST ALL CRITICAL/NEUTRAL TRIALS)]

</block>

would have to become

<block test>

/ preinstructions = (taskreminder)

/ trials = [1-112= noreplace(LIST ALL COMPASSION/NEUTRAL TRIALS, LIST ALL CRITICAL/NEUTRAL TRIALS)]

</block>

just like you've done it in the practice block.

So my two blocks would become <block test.> and have all 112 trials in them. The idea of having two blocks was that in the output I'd have some trials with 'Compassion' in the block column and some with 'Critical', then I could easily sort them on Excel so I can look at the two conditions separately. If I do it how you've explained won't they all just be under the same block name? i.e. test in the example you've given.

Sorry for my confusion, I haven't got a programming brain!

RE: Dot Probe Experiment Wanted Please!

Dave — Wed, 21 Jul 2010 06:50:17 GMT

[quote]

<expt>

/ preinstructions = (intro, task, taskreminder)

/ postinstructions = (end)

/ blocks = [1=practice; 2=critical; 3=compassion]

</expt>

Bit in bold - how do I get it to randomly take trials from the compassion and critical blocks (rather than just running them consecutively)

[/quote]

Not at all. This something that needs to be done at the <block> level. If you want a block to run both "critical" and "compassion" trials, you'll have to sample from these trials via the <block>'s '/ trials' attribute. So

/ preinstructions = (taskreminder)

/ trials = [1-56 = noreplace(LIST ALL COMPASSION/NEUTRAL TRIALS)]

</block>

and

/ preinstructions = (taskreminder)

/ trials = [1-56 = noreplace(LIST ALL CRITICAL/NEUTRAL TRIALS)]

</block>

would have to become

/ preinstructions = (taskreminder)

/ trials = [1-112= noreplace(LIST ALL COMPASSION/NEUTRAL TRIALS, LIST ALL CRITICAL/NEUTRAL TRIALS)]

</block>

just like you've done it in the practice block.

~Dave

P.S.: Your confusion suggests that you need to take a step back and review your knowledge about implementing experiments in Inquisit before proceeding. Maybe read http://www.millisecond.com/forums/Topic3712.aspx#3712 to get a better understanding of how Inquisit's language is structured.

RE: Dot Probe Experiment Wanted Please!

Lucy and Shiv — Wed, 21 Jul 2010 06:37:02 GMT

Hiya,

I'm getting really confused now - am tempted just to switch back to eprime! (Although I bet that would be just as hard)

Why will my trials have different number of stimuli? I thought that each one would be the same?!

I think I'm going to have manually score the correct/incorrect responses then. Because if I do it as you suggest - with each option being a separate trial - it will lead me to having too many trials (we don't want the experiment to last for more than 5 minutes really).

To keep it fair I would have to make sure all 8 possibilities from the 14 actors were presented. If i also added in the : and .. options I'd end up with an experiment twice as long. 224 trials. I just want the computer to choose whether it's a : or a .. and know which response is the correct one!

Does that make sense?! I really don't want to have to manually input though so not sure what I will end up doing....

I'm thinking about having 3 blocks - one for compassion/neutral and one for critical/neutral. Just so when I get the output I can easily sort the data and keep it separate. And also a practice. Will it work if I make these changes?:

/ trials = [1-10 = noreplace(LIST ALL OPTIONS)]

/ bgstim = (taskreminder)

</block>

/ preinstructions = (taskreminder)

/ trials = [1-56= noreplace(LIST ALL CRITICAL/NEUTRAL TRIALS)]

</block>

and then

<expt>

/ preinstructions = (intro, task, taskreminder)

/ postinstructions = (end)

/ blocks = [1=practice; 2=critical; 3=compassion]

</expt>

Bit in bold - how do I get it to randomly take trials from the compassion and critical blocks (rather than just running them consecutively)

Thanks again!

Lucy.

/ preinstructions = (taskreminder)

/ trials = [1-56 = noreplace(LIST ALL COMPASSION/NEUTRAL TRIALS)]

</block>

RE: Dot Probe Experiment Wanted Please!

Dave — Wed, 21 Jul 2010 05:07:58 GMT

[quote]

How would you do that in Inquisit 2? The ability to define "dynamic" correct responses became only available in Inquisit 3.0. That means post-hoc scoring then, I guess.

I thought that I had done this here:

/stimulustimes = [1=target2right]

/inputdevice = keyboard

/correctresponse = ("l")

</trial>

[/quote]

Looking at the data files you posted, this <trial target2right> is never run. So how would e.g. <trial f5gCRITICALneutral*>, which is actually run, know about this and know what to do with this? Note that

<trial mytrial>
/ stimulusframes = [1=target2left]
[...]
</trial>

does *not* reference <trial target2left>, but a *stimulus* element named 'target2left', e.g. <picture target2left>!

If you wanted to get the correct / incorrect scoring working in Inquisit 2, you'd again have to set up different <trial> elements for each possibility, e.g.

<trial f5gCRITICALneutral_target2left>
/ validresponse = ("a","l")
/ correctresponse = ("a")
[...]
</trial>

<trial f5gCRITICALneutral_target2right>
/ validresponse = ("a","l")
/ correctresponse = ("l")
[...]
</trial>

About upgrading to Inquisit 3: V3 is really cool because it allows you to do much more sophisticated stuff than V2, and as such it's worth considering. However, I don't think upgrading makes much sense as long as you haven't really gotten your head around the basics in V2. Anyway, you might just download the demo version and try it!

[quote]With regards to the columns it all makes sense up until Stimulus8. Then I am confused as to why some rows have a 0 and some have a 1. Whereas for the earlier Stimuli the whole column had 1's....[/quote]

Trials differ in the number of stimuli they display. Obviously you have a <trial> element in your script that displays 9 stims. Obviously, you also have <trial> elements that display less than 9 stims, so the 'stimulusnumber9', etc. columns will display 0s for these trials, since there's nothing there to record.

Cheers,

~Dave

RE: Dot Probe Experiment Wanted Please!

Lucy and Shiv — Wed, 21 Jul 2010 04:44:40 GMT

How would you do that in Inquisit 2? The ability to define "dynamic" correct responses became only available in Inquisit 3.0. That means post-hoc scoring then, I guess.

I thought that I had done this here:

/stimulustimes = [1=target2right]

/inputdevice = keyboard

/correctresponse = ("l")

</trial>

It makes sense to me logically that I have said for each of the 4 targets which is the correct response but obviously I am not making sense in Inquisit-language! Do you think I should get Inquisit 3.0 then?

With regards to the columns it all makes sense up until Stimulus8. Then I am confused as to why some rows have a 0 and some have a 1. Whereas for the earlier Stimuli the whole column had 1's....

Sorry I'm being a pain!

Thanks so much for all your help

Lucy.

RE: Dot Probe Experiment Wanted Please!

Dave — Wed, 21 Jul 2010 03:26:41 GMT

[quote]I assumed that if I defined the correct response for each of the 4 targets[/quote]

How would you do that in Inquisit 2? The ability to define "dynamic" correct responses became only available in Inquisit 3.0. That means post-hoc scoring then, I guess.

[quote]Can I just delete the columns where I've put a question mark? I can't see what use they have....[/quote]

The links I provided earlier explain what these are. For each stimulus displayed in a given <trial>, Inquisit records the stim's item number (stimulusnumber), the actual item (stimulusitem), its vertical and horizontal screen position (stimulusvpos, stimulushpos) and its actual onset (i.e. when the display command was issued; stimulusonset). Since this is done for every stimulus in a trial, these columns are numbered and appear in the same order the stims were displayed / listed in the '/ stimulustimes' or '/ stimulusframes' attribute. E.g. if you have a trial displaying three stims, you'll get 'stimulusnumber1, stimulusitem1, stimulusvpos1, stimulushpos1, stimulusonset1, stimulusnumber2, [...], stimulusonset3'. I suggest you leave these data alone, you never know if you might need them later for (re-)analysis purposes. Throwing these columns out in data preparation is easy enough.

~Dave

RE: Dot Probe Experiment Wanted Please!

Lucy and Shiv — Wed, 21 Jul 2010 03:24:59 GMT

Do I need to say something like I've put in bold? Obviously that's not right but is it something like that?

If not please could you help me find out how Inquisit can locate my previous definitions of the correct responses...

Just looked into writing the correct response column by hand but it will take me ages, 112 trials x 150 participants!

Thanks

/ pretrialpause = 500

/ inputdevice = keyboard

/ responsetime = 2000

/ validresponse = ("a", "l")

/ correctresponse = [(target1left, target1right = 'a'; target2left, target2right = 'l')]

/ posttrialpause = 180

</trial>

RE: Dot Probe Experiment Wanted Please!

Lucy and Shiv — Wed, 21 Jul 2010 03:08:39 GMT

Hi,

Also another output question,

Can I just delete the columns where I've put a question mark? I can't see what use they have....

I have highlighted some things I am still not sure about in yellow.

Thanks,

Lucy.

RE: Dot Probe Experiment Wanted Please!

Lucy and Shiv — Wed, 21 Jul 2010 03:00:04 GMT

Hi,

So is this not correct?:

(I can't specify which is the correct response in each trial because I don't know whether the : or the .. is going to be selected (I want it to be random)).

I assumed that if I defined the correct response for each of the 4 targets, then when they were called in the individual trial the correct response would be known....

/stimulustimes = [1=target2right]

/inputdevice = keyboard

/correctresponse = ("l")

</trial>

RE: Dot Probe Experiment Wanted Please!

Dave — Wed, 21 Jul 2010 02:55:41 GMT

[quote]I had a go at the experiment and answered incorrectly for the first 3 trials and then correctly for the next 3, however in the correct column in the output it just says 0 for each one.[/quote]

This really sounds like you haven't defined any '/ correctresponse' attribute for these <trial> elements. I suggest you review these parts of the script.

~Dave

RE: Dot Probe Experiment Wanted Please!

Lucy and Shiv — Wed, 21 Jul 2010 02:52:31 GMT

Hi,

I had a go at the experiment and answered incorrectly for the first 3 trials and then correctly for the next 3, however in the correct column in the output it just says 0 for each one.

I'll just have to go through by hand and write down if the response was correct or not if it's not possible to use the code I wrote defining the correct response in each instance.

Thanks

Lucy.

RE: Dot Probe Experiment Wanted Please!

Dave — Wed, 21 Jul 2010 02:38:29 GMT

Sorry, obviously I'm missing something. However, how about you just take the pragmatic way and generate a few test data sets to check if the correct / incorrect scoring, etc. is working as you intended? If so, perfect. If not, figure out in which cases it doesn't work and debug from there.

~Dave

RE: Dot Probe Experiment Wanted Please!

Lucy and Shiv — Wed, 21 Jul 2010 02:32:37 GMT

Hi Dave,

Is there no way Inquisit can know that whenever the target has a '1' in it the correct response is 'a' and when it's a '2' the correct response is 'l'?

If not is there any point me putting this in?:

/stimulustimes = [1=target1left]

/inputdevice = keyboard

/correctresponse = ("a")

</trial>

/stimulustimes = [1=target2left]

/inputdevice = keyboard

/correctresponse = ("l")

</trial>

/stimulustimes = [1=target1right]

/inputdevice = keyboard

/correctresponse = ("a")

</trial>

/stimulustimes = [1=target2right]

/inputdevice = keyboard

/correctresponse = ("l")

</trial>

/ validresponse = ("a", "l")

/ posttrialpause = 180

</trial>

RE: Dot Probe Experiment Wanted Please!

Dave — Wed, 21 Jul 2010 02:25:28 GMT

[quote]The problem is because I've asked it to randomly put a '..' or ':' on the left or right hand side I won't know if the correct response is 'a' or 'l'.[/quote]

Then you'll have to score correct vs. incorrect responses post-hoc (i.e. based on the dot position during data preparation / analysis in Excel, SPSS or whatever).

Running out of valedictions,

~Dave

RE: Dot Probe Experiment Wanted Please!

Lucy and Shiv — Wed, 21 Jul 2010 02:18:10 GMT

Thanks Dave,

Those links are really useful.

I changed each trial to show a response time at 2000:

<trial f1aCOMPneutral*>

/ pretrialpause = 500

/ inputdevice = keyboard

/ responsetime = 2000

/ validresponse = ("a", "l")

/ posttrialpause = 180

</trial>

I also specified earlier in my trials document that when it was target1left or target1right the correct response was 'a' and when it was target2left or target2right the correct response was 'l'. Do I need to add this into the the main trial section too? The problem is because I've asked it to randomly put a '..' or ':' on the left or right hand side I won't know if the correct response is 'a' or 'l'.

Here is where I defined the targets earlier in the file:

<trial target1left>

/stimulustimes = [1=target1left]

/inputdevice = keyboard

/correctresponse = ("a")

</trial>

<trial target2left>

/stimulustimes = [1=target2left]

/inputdevice = keyboard

/correctresponse = ("l")

</trial>

(same for right)

/ validresponse = ("a", "l")

/ posttrialpause = 180

</trial>

Thanks,

Lucy

RE: Dot Probe Experiment Wanted Please!

Dave — Wed, 21 Jul 2010 01:49:34 GMT

Hi Lucy,

[quote]There are a lot of columns that I am unsure about[/quote]

Inquisit's default data recording scheme is described in the documentation (http://www.millisecond.com/support/docs/v2/html/language/datarecording.htm; for details on each column see http://www.millisecond.com/support/docs/v2/html/language/attributes/columns.htm).

A few additional comments on my part:

[quote]I need to see the participants reaction times from when the dot appeared to when they gave a response[/quote]

Participant's reaction times are contained in the data file's 'latency' column. However, as I mentioned earlier, you should double-check if your <trial> elements start recording those reaction times at the right moment (-> my previous bit about the '/ responsetime' attribute), i.e. at the same time the dots appear (2000ms, if I remember correctly).

[quote]I also need to see if they gave the correct response (a or l).[/quote]

The key pressed is recorded in the 'response' column. However, note that Inquisit does not record the "literal" key value (a or l), but the corresponding keyboard scancode (http://www.millisecond.com/support/docs/v2/html/language/scancodes.htm). The 'correct' column would usually indicate whether the given response was correct, however you need to define the '/ correctresponse' attribute in your <trial> elements for this to work (otherwise Inquisit can't know which of the available, "valid" responses is to be scored as correct).

Regards,

~Dave

RE: Dot Probe Experiment Wanted Please!

Lucy and Shiv — Wed, 21 Jul 2010 01:26:39 GMT

Hi Dave,

Thanks for your help. Think I've done it now (well it appears to be working - no red lines when I try and run anymore!)

Having a bit of trouble with the output though. I've attached it with some queries.

I need to see the participants reaction times from when the dot appeared to when they gave a response. I also need to see if they gave the correct response (a or l). There are a lot of columns that I am unsure about so was hoping you could help me narrow it down to just get the information that I need.

Thanks,

Lucy.

RE: Dot Probe Experiment Wanted Please!

Dave — Tue, 20 Jul 2010 05:57:04 GMT

[quote]I don't really understand the different blocks/elements so thats why I just did it in one block[/quote]

Using one <block> is perfectly fine.

[quote]What does the responsetime attribute bit mean?[/quote]

By default, Inquisit will start "listening" for responses only when the complete stimulus sequence as defined by '/ stimulustimes' or '/ stimulusframes' has been completed. In the example from my previous post, that would be 2500 ms into the trial (i.e. after the dots have been erased / overwritten). However, you probably want your subjects to be able to respond earlier (e.g. as soon as the dots appear at 2000 ms). The '/ responsetime' attribute allows you to do this.

[quote]

and don't understand this:

Depends on how you've set up the <text> elements containing the dots and the respective <trial> elements. It's really important to map such stuff out before doing any coding (no offense intended). To me it seems like you have a whole bunch of factors (gender, facetype, face position, dottype, dotposition, etc.) you'd like have balanced (i.e. each possible combination of factors should appear equally often). As stated from the get-go, the easiest way to do this is to set up a separate <trial> element for each of these combinations (there may be a lot!), and then sample from these trials in equal proportions at the <block> level. This certainly isn't the only way to do it, though. However, all other approaches would probably require a little more experience with Inquisit / how Inquisit works (again, no offense intended).

Could you give me an example please?

[/quote]

Reciting from a previous response: In your design, you manipulates certain stimulus properties (gender of face -- male v. female, type of facial expressions -- compassionate v. critical, face screen position -- left v. right, dottype -- : v. ., dot position -- left v. right, maybe more). This yields the following combiantions (may not be exhaustive):

male-critical(left)-compassionate(right)-:(left)-..(right)
male-critical(left)-compassionate(right)-:(right)-..(left)
male-critical(right)-compassionate(left)-:(left)-..(right)
male-critical(right)-compassionate(left)-:(right)-..(left)

female-critical(left)-compassionate(right)-:(left)-..(right)
female-critical(left)-compassionate(right)-:(right)-..(left)
female-critical(right)-compassionate(left)-:(left)-..(right)
female-critical(right)-compassionate(left)-:(right)-..(left)

Given your limited experience (no offense, as you know...), I think you should simply set up a separate <trial> element for each single combination (let's call 'em t1 to t8 for brevity's sake) and then sample these in equal proportions at the <block> level

<block myblock>
/ trials = [1-32=noreplace(t1,t2,t3,t4,t5,t6,t7,t8)]
[...]
</block>

which would result in each trialtype being run 4 times (8 x 4 = 32). Forgive me if there's something I'm not catching.

~Dave

RE: Dot Probe Experiment Wanted Please!

Lucy and Shiv — Tue, 20 Jul 2010 05:33:23 GMT

Hi Dave,

Don't worry no offence taken! I don't really understand the different blocks/elements so thats why I just did it in one block. I prefer being taught things in 'real life' which is why i asked sean if there are any courses available for this. But unfortunately I think all learning has to be done online! In the end I found someone at the local university who had set up a similar experiment to me and went to see her and used her way of doing things (one block).

I'm happy to change if you think will work better though.

What does the responsetime attribute bit mean?

and don't understand this:

Could you give me an example please?

Thanks

Lucy.