Millisecond Forums

Hi Dave,

We’re hoping to replicate a study with the following procedure:
Materials
Four lists of trait adjectives were created: three target lists of 24 words each to be presented during the learning phase, and one distractor list of 72 words to be used for recognition testing. Each of the three target lists constituted a block within a continuous 72-item study list, bounded by two primacy and two recency buffers. Words were selected from a pool of normalised personality trait adjectives (Anderson, 1968). Trait words were all moderate to highly meaningful with meaningfulness ratings ranging from 326 to 386 (M358). The three target lists were equated for word length, (i.e., mean number of letters8) and valence, such that each list was composed of half positive and half negative traits. In Ander- son’s (1968) list, words were ordered according to their likeability ratings. In the present study a word was considered positive if it was one of the first 252 words listed in the list and negative if it had a ranking between 253 and 555. The mean ranking for positive words was 97 and the mean ranking for negative words was 391. The distrac- tor list was matched on the same variables to the group of three target lists.
Procedure
Participants were assessed individually, and gave informed consent before participating in experimental procedures. There were two parts to the study, an incidental learning phase and an immediate recognition memory test phase. During the learning phase, the participants’ task was to answer a question about each of the target words. Each list was encoded under one of three condi- tions: SR, semantic, and structural encoding. In the SR encoding task participants judged whether trait adjectives were self-descriptive by answering the question ‘‘Does this word describe you?’’ In the semantic encoding task participants made valence judgements on a semantic dimension, answering the question ‘‘Is the dictionary defini- tion of this word positive?’’ Under the structural encoding task participants were asked, ‘‘Is this word typed in upper case letters?’’
Participants were presented with 72 words, consisting of the three target lists blocked by encoding task. Presentation was blocked by condition for two reasons: First, the constant switching between tasks might adversely affect performance in older people and second, a pilot study suggested that carry-over effects might occur in mixed lists, particularly in older adults. However, order of encoding tasks was counterbalanced so that each task appeared in each ordinal position an equal number of times, and across participants, target lists appeared equally often in each of the three conditions. Word order was randomised within each list, and each participant received a different random order.
Participants were seated in front of a computer where the procedure was explained to them. On each trial one of the questions defining the learning task (e.g. ‘‘Does this word describe you?’’) was presented on the computer screen for 2 s, after which an adjective appeared for 4 s, and partici- pants made a yes/no response by pressing one of two keys on the computer keyboard. A blank screen was then displayed for 1 s and the next trial appeared automatically.
Following the study phase, there was a 2- minute interval in which people engaged in an unrelated distractor task, and then a yesno recognition memory test was given. The recognition test consisted of 144 words, half targets and half distractors, randomly mixed (i.e., not blocked by condition), which were presented one at a time on the computer screen. Participants indicated whether each word was old or new. Each item remained on the screen until participants pressed one of two keys to indicate if they recognised the word as one that had been previously presented.

Would you mind providing a simple example of how this might be implemented？

Thanks very much for your help!

+x

 

Wasabi8888 - 7/11/2025

Hi Dave,

We’re hoping to replicate a study with the following procedure:
Materials
Four lists of trait adjectives were created: three target lists of 24 words each to be presented during the learning phase, and one distractor list of 72 words to be used for recognition testing. Each of the three target lists constituted a block within a continuous 72-item study list, bounded by two primacy and two recency buffers. Words were selected from a pool of normalised personality trait adjectives (Anderson, 1968). Trait words were all moderate to highly meaningful with meaningfulness ratings ranging from 326 to 386 (M358). The three target lists were equated for word length, (i.e., mean number of letters8) and valence, such that each list was composed of half positive and half negative traits. In Ander- son’s (1968) list, words were ordered according to their likeability ratings. In the present study a word was considered positive if it was one of the first 252 words listed in the list and negative if it had a ranking between 253 and 555. The mean ranking for positive words was 97 and the mean ranking for negative words was 391. The distrac- tor list was matched on the same variables to the group of three target lists.
Procedure
Participants were assessed individually, and gave informed consent before participating in experimental procedures. There were two parts to the study, an incidental learning phase and an immediate recognition memory test phase. During the learning phase, the participants’ task was to answer a question about each of the target words. Each list was encoded under one of three condi- tions: SR, semantic, and structural encoding. In the SR encoding task participants judged whether trait adjectives were self-descriptive by answering the question ‘‘Does this word describe you?’’ In the semantic encoding task participants made valence judgements on a semantic dimension, answering the question ‘‘Is the dictionary defini- tion of this word positive?’’ Under the structural encoding task participants were asked, ‘‘Is this word typed in upper case letters?’’
Participants were presented with 72 words, consisting of the three target lists blocked by encoding task. Presentation was blocked by condition for two reasons: First, the constant switching between tasks might adversely affect performance in older people and second, a pilot study suggested that carry-over effects might occur in mixed lists, particularly in older adults. However, order of encoding tasks was counterbalanced so that each task appeared in each ordinal position an equal number of times, and across participants, target lists appeared equally often in each of the three conditions. Word order was randomised within each list, and each participant received a different random order.
Participants were seated in front of a computer where the procedure was explained to them. On each trial one of the questions defining the learning task (e.g. ‘‘Does this word describe you?’’) was presented on the computer screen for 2 s, after which an adjective appeared for 4 s, and partici- pants made a yes/no response by pressing one of two keys on the computer keyboard. A blank screen was then displayed for 1 s and the next trial appeared automatically.
Following the study phase, there was a 2- minute interval in which people engaged in an unrelated distractor task, and then a yesno recognition memory test was given. The recognition test consisted of 144 words, half targets and half distractors, randomly mixed (i.e., not blocked by condition), which were presented one at a time on the computer screen. Participants indicated whether each word was old or new. Each item remained on the screen until participants pressed one of two keys to indicate if they recognised the word as one that had been previously presented.

Would you mind providing a simple example of how this might be implemented？

Thanks very much for your help!

Please break down what you don't know how to implement, then I can point you to relevant documentation, etc.

Pasting text from an article broadly describing a procedure is not useful.

Hi Dave,

I have a few questions about implementing the task and would really appreciate your advice:

Random Allocation of Target Lists:
What’s the best way to randomly assign the three target word lists to the different block conditions (SR, semantic, and structural encoding)? 

Primacy and Recency Buffers:
As each of the three target lists should be embedded within a continuous 72-item study list (bounded by two primacy and two recency buffers), how would you recommend assigning these buffers if the block order is randomised? Would it make more sense to have separate scripts that manually decide the order of block conditions, then insert the buffers accordingly?

Trial Setup for Valence Conditions:
Each target list includes both positive and negative adjectives. Do you think I should set up separate positive and negative trials, like in the Emotional Memory Task?

<trial phaseAImagePresentationNegative>
/ onTrialBegin = {
values.valence = "negative";
values.trialCount++;
values.itemNumber = list.negativeImagesPhaseA.nextValue;

values.valence1 = 1;
values.valence2 = 2;
values.valence3 = 3;
values.valence4 = 4;
values.valence5 = 5;
values.valence6 = 6;
values.valence7 = 7;
values.valence8 = 8;
values.valence9 = 9;
values.selectedRsp = 0;

values.rt = "";
values.startTime = "";    
}
/ stimulusFrames = [1 = clearScreen, negative]
/ onTrialEnd = {
    values.startTime = script.elapsedTime;
    values.image = picture.negative.currentItem;
    values.itemNumber = picture.negative.currentIndex;
}    
/ branch = {
    return trial.phaseARating;
}
/ recordData = false
/ trialDuration = parameters.imagePresentationDurationMs
</trial>

Note: trial.phaseA_imagePresentation_neutral
prepares each trial by resetting crucial values and presenting the next neutral picture
<trial phaseAImagePresentationNeutral>
/ onTrialBegin = {
values.valence = "neutral";
values.trialCount++;
values.itemNumber = list.neutralImagesPhaseA.nextValue;
values.selectedRsp = 0;

values.rt = "";
values.startTime = "";    
}
/ stimulusFrames = [1 = clearScreen, neutral]
/ onTrialEnd = {
    values.startTime = script.elapsedTime;
    values.image = picture.neutral.currentItem;
    values.itemNumber = picture.neutral.currentIndex;
}    
/ branch = {
    return trial.phaseARating;
}
/ recordData = false
/ trialDuration = parameters.imagePresentationDurationMs
</trial>

Note: trial.phaseA_imagePresentation_positive
prepares each trial by resetting crucial values and presenting the next positive picture
<trial phaseAImagePresentationPositive>
/ onTrialBegin = {
values.valence = "positive";
values.trialCount++;
values.itemNumber = list.positiveImagesPhaseA.nextValue;

values.valence1 = 1;
values.valence2 = 2;
values.valence3 = 3;
values.valence4 = 4;
values.valence5 = 5;
values.valence6 = 6;
values.valence7 = 7;
values.valence8 = 8;
values.valence9 = 9;
values.selectedRsp = 0;

values.rt = "";
values.startTime = "";    
}
/ stimulusFrames = [1 = clearScreen, positive]
/ onTrialEnd = {
    values.startTime = script.elapsedTime;
    values.image = picture.positive.currentItem;
    values.itemNumber = picture.positive.currentIndex;
}    
/ branch = {
    return trial.phaseARating;
}
/ recordData = false
/ trialDuration = parameters.imagePresentationDurationMs
</trial>

<block phaseA>
/ preInstructions = (phaseAIntro)
/ skip = {
    return (parameters.phaseSetting == 2);
}
/ onBlockBegin = {
    values.phase = "A";
    values.presentationPhase = "A";
    values.trialCount = 0;
    values.corrRsp = "";
}
/ trials = [
    1 = startiti;
    2-46 = noreplace(phaseAImagePresentationNegative, phaseAImagePresentationNeutral, phaseAImagePresentationPositive);
]
</block>

+x

 

Wasabi8888 - 7/11/2025

Hi Dave,

I have a few questions about implementing the task and would really appreciate your advice:

Random Allocation of Target Lists:
What’s the best way to randomly assign the three target word lists to the different block conditions (SR, semantic, and structural encoding)? 

Primacy and Recency Buffers:
As each of the three target lists should be embedded within a continuous 72-item study list (bounded by two primacy and two recency buffers), how would you recommend assigning these buffers if the block order is randomised? Would it make more sense to have separate scripts that manually decide the order of block conditions, then insert the buffers accordingly?

Trial Setup for Valence Conditions:
Each target list includes both positive and negative adjectives. Do you think I should set up separate positive and negative trials, like in the Emotional Memory Task?

<trial phaseAImagePresentationNegative>
/ onTrialBegin = {
values.valence = "negative";
values.trialCount++;
values.itemNumber = list.negativeImagesPhaseA.nextValue;

values.valence1 = 1;
values.valence2 = 2;
values.valence3 = 3;
values.valence4 = 4;
values.valence5 = 5;
values.valence6 = 6;
values.valence7 = 7;
values.valence8 = 8;
values.valence9 = 9;
values.selectedRsp = 0;

values.rt = "";
values.startTime = "";    
}
/ stimulusFrames = [1 = clearScreen, negative]
/ onTrialEnd = {
    values.startTime = script.elapsedTime;
    values.image = picture.negative.currentItem;
    values.itemNumber = picture.negative.currentIndex;
}    
/ branch = {
    return trial.phaseARating;
}
/ recordData = false
/ trialDuration = parameters.imagePresentationDurationMs
</trial>

Note: trial.phaseA_imagePresentation_neutral
prepares each trial by resetting crucial values and presenting the next neutral picture
<trial phaseAImagePresentationNeutral>
/ onTrialBegin = {
values.valence = "neutral";
values.trialCount++;
values.itemNumber = list.neutralImagesPhaseA.nextValue;
values.selectedRsp = 0;

values.rt = "";
values.startTime = "";    
}
/ stimulusFrames = [1 = clearScreen, neutral]
/ onTrialEnd = {
    values.startTime = script.elapsedTime;
    values.image = picture.neutral.currentItem;
    values.itemNumber = picture.neutral.currentIndex;
}    
/ branch = {
    return trial.phaseARating;
}
/ recordData = false
/ trialDuration = parameters.imagePresentationDurationMs
</trial>

Note: trial.phaseA_imagePresentation_positive
prepares each trial by resetting crucial values and presenting the next positive picture
<trial phaseAImagePresentationPositive>
/ onTrialBegin = {
values.valence = "positive";
values.trialCount++;
values.itemNumber = list.positiveImagesPhaseA.nextValue;

values.valence1 = 1;
values.valence2 = 2;
values.valence3 = 3;
values.valence4 = 4;
values.valence5 = 5;
values.valence6 = 6;
values.valence7 = 7;
values.valence8 = 8;
values.valence9 = 9;
values.selectedRsp = 0;

values.rt = "";
values.startTime = "";    
}
/ stimulusFrames = [1 = clearScreen, positive]
/ onTrialEnd = {
    values.startTime = script.elapsedTime;
    values.image = picture.positive.currentItem;
    values.itemNumber = picture.positive.currentIndex;
}    
/ branch = {
    return trial.phaseARating;
}
/ recordData = false
/ trialDuration = parameters.imagePresentationDurationMs
</trial>

<block phaseA>
/ preInstructions = (phaseAIntro)
/ skip = {
    return (parameters.phaseSetting == 2);
}
/ onBlockBegin = {
    values.phase = "A";
    values.presentationPhase = "A";
    values.trialCount = 0;
    values.corrRsp = "";
}
/ trials = [
    1 = startiti;
    2-46 = noreplace(phaseAImagePresentationNegative, phaseAImagePresentationNeutral, phaseAImagePresentationPositive);
]
</block>

> What’s the best way to randomly assign the three target word lists to the different block conditions (SR, semantic, and structural encoding)?

There is no one best way. If you want to achieve "across participants, target lists appeared equally often in each of the three conditions" as in the original study, you need to hardcode assignments anyway, not assign lists randomly to encoding condition. Combined with the counterbalancing of encoding conditions (below), that'd result in a large number of conditions then.

> Primacy and Recency Buffers:
> As each of the three target lists should be embedded within a continuous 72-item study list (bounded by two primacy and two recency buffers), how would you recommend assigning these > buffers if the block order is randomised? Would it make more sense to have separate scripts that manually decide the order of block conditions, then insert the buffers accordingly?

There's no explanation what exactly the authors mean by "primacy and recency buffers" and how those functioned in the study, so this question isn't answerable. Also, block order, according to the text you pasted in your original post, wasn't randomized in the original study, but counterbalanced (I assume in a Latin Square type fashion).

> Trial Setup for Valence Conditions:
> Each target list includes both positive and negative adjectives. Do you think I should set up separate positive and negative trials, like in the Emotional Memory Task?

Up to your preference.

Hi Dave,

I was wondering if there’s an easy way to implement the counterbalancing of both the three target lists and the three encoding conditions (using a Latin square design). If so, would you mind providing an example? Or would we need to write separate scripts to manually set the order?

Also, just to clarify, my understanding is that the primacy and recency buffers are extra words added to the start and end of the learning list to absorb position effects, since participants tend to remember the first and last items best. These buffers aren't analyzed, but rather serve to protect the 72 main target items. 

+x

 

Wasabi8888 - 7/11/2025

Hi Dave,

I was wondering if there’s an easy way to implement the counterbalancing of both the three target lists and the three encoding conditions (using a Latin square design). If so, would you mind providing an example? Or would we need to write separate scripts to manually set the order?

Also, just to clarify, my understanding is that the primacy and recency buffers are extra words added to the start and end of the learning list to absorb position effects, since participants tend to remember the first and last items best. These buffers aren't analyzed, but rather serve to protect the 72 main target items. 

> I was wondering if there’s an easy way to implement the counterbalancing of both the three target lists and the three encoding conditions (using a Latin square design).

No easy way.

> Also, just to clarify, my understanding is that the primacy and recency buffers are extra words added to the start and end of the learning list to absorb position effects, since participants
> tend to remember the first and last items best. These buffers aren't analyzed, but rather serve to protect the 72 main target items.

Maybe? The description isn't really clear, since throughout only the three 24 item encoding lists (72 items total) are referenced; . In any case, having a two-word block before at the very start (before any of the encoding blocks) and before the recognition phase (after the encoding blocks) is trivial.

+x

 

Dave - 7/11/2025

+x

 

Wasabi8888 - 7/11/2025

Hi Dave,

I was wondering if there’s an easy way to implement the counterbalancing of both the three target lists and the three encoding conditions (using a Latin square design). If so, would you mind providing an example? Or would we need to write separate scripts to manually set the order?

Also, just to clarify, my understanding is that the primacy and recency buffers are extra words added to the start and end of the learning list to absorb position effects, since participants tend to remember the first and last items best. These buffers aren't analyzed, but rather serve to protect the 72 main target items. 

> I was wondering if there’s an easy way to implement the counterbalancing of both the three target lists and the three encoding conditions (using a Latin square design).

No easy way.

> Also, just to clarify, my understanding is that the primacy and recency buffers are extra words added to the start and end of the learning list to absorb position effects, since participants
> tend to remember the first and last items best. These buffers aren't analyzed, but rather serve to protect the 72 main target items.

Maybe? The description isn't really clear, since throughout only the three 24 item encoding lists (72 items total) are referenced; . In any case, having a two-word block before at the very start (before any of the encoding blocks) and before the recognition phase (after the encoding blocks) is trivial.

You have three encoding conditions -- A, B, C -- and three target lists .. 1,2,3. So econding condition A can be paired with either list 1 (A1), list 2 (A2) or list 3 (A3). Same for encoding conditions B and C. Constructing Latin Squares for counterbalancing along the lines of



satisfies "order of encoding tasks was counterbalanced so that each task appeared in each ordinal position an equal number of times, and across participants, target lists appeared equally often in each of the three conditions." (I assume the authors constructed and used multiple such squares).

Hi Dave,

Do I need to manually write 9 separate scripts for each group? If possible, would you mind sharing an example script setup for one group? I think that would really help me get started.

Thanks very much!

Group    Block 1    Block 2    Block 3
1    List A – SR    List B – Semantic    List C – Structural
2    List B – SR    List C – Semantic    List A – Structural
3    List C – SR    List A – Semantic    List B – Structural
4    List A – Semantic    List B – Structural    List C – SR
5    List B – Semantic    List C – Structural    List A – SR
6    List C – Semantic    List A – Structural    List B – SR
7    List A – Structural    List B – SR    List C – Semantic
8    List B – Structural    List C – SR    List A – Semantic
9    List C – Structural    List A – SR    List B – Semantic

+x

 

Wasabi8888 - 7/11/2025

Hi Dave,

Do I need to manually write 9 separate scripts for each group? If possible, would you mind sharing an example script setup for one group? I think that would really help me get started.

Thanks very much!

Group    Block 1    Block 2    Block 3
1    List A – SR    List B – Semantic    List C – Structural
2    List B – SR    List C – Semantic    List A – Structural
3    List C – SR    List A – Semantic    List B – Structural
4    List A – Semantic    List B – Structural    List C – SR
5    List B – Semantic    List C – Structural    List A – SR
6    List C – Semantic    List A – Structural    List B – SR
7    List A – Structural    List B – SR    List C – Semantic
8    List B – Structural    List C – SR    List A – Semantic
9    List C – Structural    List A – SR    List B – Semantic

You don't need separate scripts at all. If you want separate scripts because you find that easier to handle, by all means, have separate scripts.

And examples you can find plenty already on the forums or in the library. Encoding-recognition or old-new paradigms have been discussed dozens of times. List learning tasks exist in the library.

Hi Dave,

If I don’t need to write separate scripts for each group, how would I implement the Latin Square counterbalancing within a single script?