THE UNIVERSITY OF ADELAIDE

School of Psychology

Foundations of Perception and Cognition PSYCHOL 2006

PRACTICAL ASSESSMENT SUMMARY

Choice 2: Affective priming

To learn about cognitive psychology and how we study human thinking and behavior, we are going to conduct an experiment and then write up a prac report, in much the same way that a cognition researcher might do. This prac report is designed to examine the extent to which emotion affects the processing of stimuli that follow each other. Notably, we will investigate affective priming. In this sort of task, you see two stimuli that occur one after the other (typically words or pictures – we will use written words) but you only respond to the second one (the target word – the first word is called a ‘prime word’ or just a prime). We are typically interested in the extent to which the prime word affects the reaction times of the target word.

Perhaps the easiest way to understand this is to first look at semantic priming (Neely, 1991). With semantic priming, you see two words that are semantically related, like cat and DOG. You are told to only respond to DOG (the target word). What you almost always find is that if you respond to DOG after cat then you are faster than responding to DOG after (say) wall. So the semantic relationship appears to ‘prime’ you – that is, it makes you somehow expect DOG after cat. Typically this is attributed to semantic overlap. You might notice here that you need two different stimuli to compare. One is the related prime and the other is the control prime. If you think about it, each one of these stimuli only gives you one reaction time, so using two groups allows you to look at one group with respect to the other. This is known as the priming effect. If you didn’t have a control group, you would just have something like ‘a 653 ms average response time’ and so you wouldn’t be able to tell if the prime affected the target at all.

With affective priming, the type of priming we will be interested in here, you have a similar idea. However, ‘pure’ affective priming (if there is such a thing) looks only at the emotional connotation of a word. For example, sickness might prime loss even though sickness is only weakly semantically related to loss. If priming happens (i.e., your reaction times are affected by the prime word), then it suggests you are sensitive to affective priming. People often consider both valence and arousal so, for example, it might be that two words high in arousal prime each other regardless of the emotional category they are from (e.g., love-hate vs. kill-hate). Alternatively, one might need to use words of the same category to find a priming effect. This is an empirical question. You can see the design that I used from the picture below. Note that I didn’t explicitly look at arousal and this design can’t tell you everything!

What to do

This handout is designed to provide you with the information necessary to complete the FOPAC practical report. It is set out in the following sections: First, a brief overview of the previous empirical evidence that inspired this experiment is given (BACKGROUND INFORMATION). Next, in the ‘WRITING THE REPORT’ section you will find important information regarding due dates, structure, etc. Following this, I provide details regarding what you are expected to provide in each section of your report (ABSRACT, INTRODUCTION, METHOD, etc). With all of this information you will be able to put the report together. Of course, you are bound to have questions and if you do please post them on the discussion board.

Please do the task – there are two versions. Choose only one of them. There are quite a few stimuli, so it takes longer than I would have liked (about 20 minutes), and there is scheduled break in the middle.

Task link: https://run.pavlovia.org/ldt_emot/lexical-decision-task/html/

Task link: https://run.pavlovia.org/emotionldt/lexical-decision-task/html/

We will talk about this more in the Week 4. The tutorial will make vastly more sense if you have done the task (doing just one is fine).

SPECIFIC BACKGROUND INFORMATION

We are going to investigate the extent to which people use affective information in the lexical decision task, where you simply respond ‘yes’ or ‘no’ to a target word or nonsense word (e.g., splorf). This differs to most affective priming studies where people make some judgement about the word’s affective connotation. In these studies, priming has typically not be found (see Herring et al., 2013), unlike other experiments where people are asked to focus on the affective aspects of the words. Given that people’s attention seems to be quite malleable in terms of what they process based on affective priming studies using other types of responses, one possibility here is that people simply do not process that type of information. However, an alternative is that people do, but its integration is comparatively delayed compared to when compared to tasks which focus people’s attention on it. This means that in a typical LDT task, an effect of emotional affect may not be not found because the response is simply too quick to allow it to be seen. Another possibility is that task difficulty might change the type of information people use, but this is often not considered in LDT tasks.

To examine this, I created two conditions by modifying the filler nonwords and words in the task. Notably, apart from the affective prime-target pairs, I chose two groups of filler words: Those that were either hard to process or those that were easy to process. I did this by choosing words/nonwords from the English Lexicon Project (Balota et al., 2002), which is just a big database of reaction times for words and nonwords – i.e., the things you respond ‘is not a word’ too. I also added pseudohomopones (nonwords that sound like words e.g., keap) to the hard list, which makes the task very hard. The basic idea was to try and delay all responses you make. Using hard fillers is well known to do this (Stone et al., 1993). Basically, I asked the question: if we get people to respond more slowly, do we start seeing affective priming? To make life simpler for you, I am only going to look at the experiments where negative words were used. This means there are four conditions of interest. These can be seen below.

To examine the data, I used four t-tests (this is not how you would typically examine the data incidentally – you would typically use an ANOVA which most of you wouldn’t have done yet). You need to think about three hypotheses based on the pattern priming you might to expect.

This analyses is somewhat trickier than that which you would have seen so far in your statistics courses because there are really two things we need to examine. First, whether there is an overall effect of priming, and second, does the effect of priming differs across the two groups? This is called an interaction, because it means that something that differs in one group (i.e., the effect of priming) is a different size compared to another group (i.e., does the effect of priming differ in the easy condition compared to the hard condition?). I have broken these down into 3 hypotheses for you. Why you make these hypotheses needs to follow logically from your introduction.

Hypothesis 1: It is hypothesized there will be a significant overall effect of priming in the easy filler condition or It is hypothesized there will be no significant priming effect in the easy filler condition

Hypothesis 2: It is hypothesized there will be no significant overall effect of priming in the hard filler condition or It is hypothesized there will be no significant priming effect in the hard filler condition

Hypothesis 3: Will the effect be restricted to either the positive-negative or neutral-negative pairs? You need to work out the wording to this.

You don’t have to make directional hypothesis for all of these (e.g., “we expect X to be bigger than Y”), but you should have at least one. If you don’t, you can use the term research question rather than hypothesis. This means you don’t know what to expect, but are interesting in looking! If you want to think of your own hypotheses, feel free.

WRITING THE REPORT

Due Dates. The due date for the report is 9am on Friday the 30^th of April. There is no penalty for lateness if you submit before 9am on Monday the 2^nd of May. Reports received after this time will have the standard mark deduction for each working day late (5%). Any request for an extension must be made to the course coordinator Dr Conrad Perry before the stated due date.

Submission. Submissions will be made via the link in the Prac Information folder on My Uni. Your submission will be checked against Turn-it-in and by submitting it you are agreeing that the work is your own and that any work that is not yours is appropriately referenced. Please take care when selecting the file for submission, we WILL NOT reset any incorrect submissions. Please read the instructions on My Uni BEFORE submission of the assignment. DO NOT cheat. You will be amazed at what Turn-it-in catches now.

Structure. The report should be no more than 1500 words. This EXCLUDES the tables, references, anything I wrote (the method), figure captions and any appendices you consider necessary. It also excludes the method section. Please use double spacing and keep a copy of your report. It should be written up as you are asked to write up practical reports in psychology generally. Note: DO NOT submit your work with the method section in it – since everyone has the same section, it will cause the plagiarism dector to think you have cheated. Just put a placeholder in like [Method]. Also, do not put the DOI numbers in the references – it’s a pain and a waste of time – almost everyone uses automatic referencing software now which does this, so spend your time on the actual report. The report should be in the following order:

Abstract

Introduction

Method [This section is given to you]

Results

Discussion

References

For help and guidelines for writing up your report refer to the documents in the ‘Writing Information and Guidelines’ section of the ‘Prac report information’ page on My Uni. This page also contains the practical marking rubric which will give you an idea about the sorts of things we are looking for in a prac report. I would also recommend that you read the sample prac reports, which are also available on My Uni in the ‘Prac report information’page. There are many books which show you how to write these, as well as web sites.

Plagarism. You must write this report up individually, but it is fine if you discuss your ideas with other students. Standard policies for plagiarism apply. It is also plagiarism if you copy or simply paraphrase the information in the FOPAC PRACTICAL INTRODUCTION section of this document when writing your introduction. Your introduction should go beyond the basic information presented in this handout in some interesting way (see the sample pracs for examples).

FOPAC PRACTICAL INTRODUCTION

You can write your Introduction multiple ways, depending on how you want to focus it. Do you want to start with affective priming, or is there something more general we need to know about emotions first? You will need to think about models of emotions (i.e., continuous vs. categorical). We don’t do much of this in this course, but any cognitive psychology textbook should have some things on it. I will post a .pdf from Eysenck, Cognitive Psycholgy: A Student’s Handbook, which has a good introduction on it.

Next, you will want to provide a rationale for the current experiment. I have done this briefly in the BACKGROUND section above. You will need to make this focused on this experiment for your report. We are only interested in particular aspects of this – basically whether particular types of priming might occur and the role task difficuly plays in that. Think carefully about what sort of information is relevant to the current study. You only have 1500 words, so you need to be succinct!

You will want to finish your introduction section by stating your experimental hypotheses. You will need to select three hypotheses – I have more or less written them for you above. You need to write these – you need to think about what to say here, and this will be based on your introduction. See the information given above for what the hypotheses could be.

FOPAC PRACTICAL METHOD SECTION

The following is the method section for your FOPAC assignment. This section will not count towards your word-count.

Make sure you read this section. There are details in this section that may help with your interpretation of the results of the experiment. Please don’t put it in you lab report – just put a stub like [method], otherwise Turnitin will think you have plagiarized it.

——————————–start of method section————————————–

Method

Participants. 80 University of Adelaide undergraduates took part in this experiment. The data was collected over approximately 3 weeks and this data only represents those where the entire experiment was fully completed.

Stimuli. 80 matched pairs were chosen for each of the experiments. These were broken into two counterbalanced group such that each target word was paired with either a happy or neutral prime as well as another sad prime (i.e., Experiment 1: happy-sad, sad-sad; Expermient 2: neutral-sad, sad-sad). The stimuli were balanced across the conditions such that word frequency, length, and other important psychologinguistic variables were similar for both the prime words and the target words. Statistics for sad/happy were taken from the ANEW (Stevenson et al., 2007) database, and other statistics were taken from the English Lexicon Project (Balota et al., 2007). The stimuli were broken into two counterbalanced groups so that each target word was only presented once, with half the primes being congruent and the other half being incongruent (Experiment 1) or neutral (Experiment 2).

Design. Participants were randomly allocated into one of eight counterbalanced conditions. These were based on whether they did the easy or hard condition first, which group of possible prime-target pairs they were given, and whether easy or hard fillers were used first. Inspection of the data showed that the ordering did make any meaningful difference to the data, and so the results were collapsed and only results from the hard/easy filler conditions was used for the analyses.

Procedure. The study was run on the Pavolvia platform using jscript. An initial email was sent to a class of Adelaide University students asking them to use a html address to log onto the experiment. Once they did this, an instruction screen appeared telling them the details of the study and how to make responses. The details included that they should respond with the left and right arrow keys to signify ‘no, the answer is not correct’ and ‘yes, the answer is correct’.

In terms of the main task, the prime words first appeared in lower case for 200 ms. After that the target word appeared in upper case until the a response was given. After that there was a 1200 ms blank screen and the next stimuli appeared. Once participants had gone through half the stimuli (i.e., all of the stimuli in either the easy or hard condition), a screen appeared that told them to have a rest. Once they pressed the space-bar, the other half of the stimuli were presented.

FOPAC PRACTICAL RESULTS SECTION

The results section requires you to present the results of the experiment so that they address your hypotheses. We have produced the statistical comparisons for you, but you need to work out whether they support your hypotheses and how to write them. We expect that at a minimum you will describe what was found. Typically we would get you to do a graph, although since I have put messy versions in the R script I gave you, I don’t expect it here. If you want more information, you can get it by running the script.

Here is the data you need:

SAD vs POSITIVE condition Easy Fillers

The mean of the differences is the priming effect.

Paired t-test

data: df1$E_Kill_Sad and df1$E_Love_Sad

t = 1.0405, df = 39, p-value = 0.3045

alternative hypothesis: true difference in means is not equal to 0

95 percent confidence interval:

-12.86095 40.11095

sample estimates:

mean of the differences

13.625

There is no significant priming effect in this condition

SAD vs POSITIVE condition Hard Fillers

The mean of the differences is the priming effect.

> t.test(df1$H_Kill_Sad,df1$H_Love_Sad,paired = TRUE, alternative = “two.sided”)

Paired t-test

data: df1$H_Kill_Sad and df1$H_Love_Sad

t = -3.3711, df = 39, p-value = 0.001699

alternative hypothesis: true difference in means is not equal to 0

95 percent confidence interval:

-69.36024 -17.33976

sample estimates:

mean of the differences

-43.35

There is a significant priming effect in this condition

Here is a picture of the results – note error bars and so on represent measures calculated on each group separately– but the comparison was a repeated measures one. If you want to look at the individual difference data you can generate them from the script.

SAD vs NEUTRAL condition with Easy Fillers

The mean of the differences is the priming effect.

data: df1$E_Kill_Sad_N and df1$E_Wall_Sad_N

t = -0.51204, df = 39, p-value = 0.6115

alternative hypothesis: true difference in means is not equal to 0

95 percent confidence interval:

-36.13685 21.53685

sample estimates:

mean of the differences

-7.3

There is no significant priming effect in this condition

Let’s now look at the hard condition:

> t.test(df1$H_Kill_Sad_N,df1$H_Wall_Sad_N,paired = TRUE, alternative = “two.sided”)

Paired t-test

data: df1$H_Kill_Sad_N and df1$H_Wall_Sad_N

t = -2.9754, df = 39, p-value = 0.005003

alternative hypothesis: true difference in means is not equal to 0

95 percent confidence interval:

-63.70695 -12.14305

sample estimates:

mean of the differences

-37.925

There is a significant priming effect in this condition

SAD vs NEUTRAL condition Hard Fillers

FOPAC PRACTICAL DISCUSSION SECTION

Your discussion should focus on explaining what these results mean. In your discussion you need to answer the questions; Why we might have gotten that pattern of results? How do these results relate to the existing literature? and what possible confounds and limitations exist for this study? You should try to go beyond the general criticism that the study lacks psychological realism, we didn’t use a representative sample etc. Try to think in a critical way about the design of the experiment, how things were measured and the materials used and how this might have biased the results to certain findings. Since this is mainly a theoretically interesting prac, I don’t expect or want you to come up with some grandiose claim about how the results are massively useful. They aren’t. Just telling us how they might be useful or interesting in specific contexts is certainly going to be enough as long as you can relate what we found and those contexts.

References

Balota, D.A., Yap, M.J., Hutchison, K.A. et al. (2007). The English Lexicon Project. Behavior Research Methods, 39, 445–459. https://doi.org/10.3758/BF03193014

Herring, D. R., White, K. R., Jabeen, L. N., Hinojos, M., Terrazas, G., Reyes, S. M., Taylor, J. H., & Crites, S. L., Jr. (2013). On the automatic activation of attitudes: A quarter century of evaluative priming research. Psychological Bulletin, 139(5), 1062-1089. doi: 10.1037/a0031309

Stevenson, R., Mikels, J. & James, T. (2007). Characterization of the Affective Norms for English Words by discrete emotional categories. Behavior Research Methods, 39, 1020-1024. doi: 10.3758/BF03192999.

Stone, G. O., & Van Orden, G. C. (1993). Strategic control of processing in word recognition. Journal of Experimental Psychology: Human Perception and Performance, 19, 744-774. 10.1037/0096-1523.19.4.744

Note: Balota et al. and Stevenson et al. are unlikely to be very useful.