% by doing a within-subject study containing two immediate

% % Annual Cognitive Science Conference% Sample LaTeX Paper — Proceedings Format% % Original : Ashwin Ram ([email protected]

edu)       04/01/1994documentclass10pt,letterpaper{article}usepackage{cogsci}usepackage{pslatex}usepackage{apacite}usepackage{graphicx}usepackage{csquotes} itle{Working Memory Capacity for Images and Words} author{{large f Floris van Leeuwen ([email protected])}   Department of Artificial Intelligence, University of Groningen   9747 AG Groningen, The Netherlands}egin{document}graphicspath{ {img/} }maketitleegin{abstract}Past researched proposed that working memory capacity is dependent on multiple variables (Miller 1956). In an immediate recall test carried out by Erdelyi and Becker (1974) with mixed stimuli of images and words, more images were recalled than words.

This suggests that working memory capacity is higher for images than for words. In this study, we test this hypothesis by doing a within-subject study containing two immediate recall tests, one with words and the other with images as presented stimuli. Although the mean of recalled images was higher than for words, we did not find a significant difference between the number of stimuli recalled by the participants.

This concludes that there is no difference in working memory capacity for images or words in human participants. Due to some limitations of our experiment, we recommend doing more research on this topic. extbf{Keywords:} working memory capacity; images; words; immediate recall task; magical number sevenend{abstract}section{Introduction}The working memory makes humans able to reason and make decisions. It also plays an important role in the task of following instructions, as well as learning new information. How much humans are able to achieve using the human working memory, depends on the capacity of this cognitive system. Defining the exact capacity – and the variables influencing it – has been the aim of many researchers for decades.

Miller (1956) found in his paper the capacity to be 7, plus or minus 2. He states that this number 7 represents the number of stimuli a person likely will remember in an immediate recall test. However, he does not present a definition of the size of these stimuli. Because of this, the magical number 7 has endured much criticism. Cowan et al.

(2004) introduce the idea that the working memory capacity should be measured differently. The paper states capacity is not defined by the number of recalled stimuli, but the number of remembered chunks of these stimuli. These chunks represent a constant number of separable mental units.

As cited in Miller’s paper, Hayes (1952) measured the working memory capacity using bits, with every item displayed in an immediate recall test having a size in bits. He found that monosyllabic words have a lower recall rate than binary numbers. This supports the hypothesis that larger stimuli are less likely to be remembered.Considering this hypothesis, it should be harder to recall images compared to words in an immediate recall test, since the size of words in bits is just a fraction of the size of an image.

Erdelyi and Becker (1974) disproved this in their research. The participants were presented with a list of 80 items, varying between images and words. After studying the list, the participants were able to recall slightly more images than words. This research led us to the following research question:egin{displayquote}“Does presenting pictures rather than words affect working memory capacity in a recall test with human participants?”end{displayquote}In this research, we will examine the effect images have on working memory capacity compared to words by showing the participants two lists of stimuli, one containing words and the other pictures.

Erdelyi and Becker showed that images are more likely to be remembered when they are shown in one list. However, this does not conclude that the working memory capacity is higher for images compared to words.subsection{Hypothesis}In this research, we will make use of two lists where every stimulus describes an object. Considering the research mentioned above, we came up with the following hypothesis:egin{displayquote}“We expect that the subjects of the images shown will be better recalled than the presented words.”end{displayquote}This means that, according to our expectations, the amount of recalled stimuli in the immediate recall test will be higher for the image test than for the word test.

We expect the amount of recalled words to be around seven, following Miller’s (1956) findings. For the images test, we expect this to be about 9 remembered stimuli, following Erdelyi and Becker’s (1974) findings.subsection{Research Method}The within-subject study consists of two immediate recall tests. In this paper, the immediate recall test where the shown stimuli consist of words and the test where the shown stimuli consist of images will be referred to as the word test and the image test respectively. These lists used to test the hypothesis will consist of 20 stimuli, randomly picked out of a larger list of 40 stimuli. Every stimulus will be shown one after the other for a duration of 1.5 seconds each. After the list is shown, the participant will be asked to write down as many stimuli they can remember.

In between the two recall tests, there will be a short break.In the next section of the paper, I will elaborate on the method we used for conducting the experiment. The section after that contains information and analyzation about the results of the experiment.

Lastly, I will give the conclusion and some discussion points about the experiment.section{Method}subsection{Participants}In this study, 17 university students participated in two immediate recall tests. They were between the age of 18 and 28, with a mean age of 20.1. 10 participants were female and 7 male. Their Native Tongue varied, but participants were required to be fluent in the English language.

For participating, the participant received a chocolate cookie as a reward.We recruited the participants by asking friends from the university to participate in our experiment. Because of this, all participants were university students. None of these friends were told beforehand what the experiment was about. Before participating in the experiment, the participants all signed an informed consent form.subsection{Apparatus}For the experiment, we wrote a small computer program using XML, PHP, and javascript. This software was accessed using the Google Chromefootnote{Version 63.

0.3239.132 (Official build) (64-bits)} internet browser on a personal computer.subsection{Material and Stimuli}The task the participants had to do, was to recall as many items as possible in an immediate recall test. For the purpose of this experiment two types of stimuli were created, namely words and pictures. These stimuli were stored in two lists of 40 stimuli.

During the experiment, the program would pick randomly 20 stimuli out of these bigger lists and display them one by one at a rate of 1.5 seconds.All the images used in this research were found using Google image search. To be selected for use in the research, the images had to meet some criteria. First of all, the image needed to clearly represent one object, to avoid confusion. all images needed to have a transparent background as well.

The words used in this research were gathered using a random word generatorfootnote{https://www.randomlists.com/nouns} on the internet. Out of the lists generated online, we picked the words that met the following criteria.

The words had to be a noun that represents an object, as to not be too different in the type of meaning from the words the images represent.The words as well as the images needed to be or represent a word that is between the 1000 and 4000 most frequently used words. We followed the Corpus of Contemporary American Englishfootnote{https://corpus.byu.edu/coca/} to define the frequency of each word.

There was no correspondence between the meaning of the words and the images. During the test, the words and images had the same placement on the screen. The horizontal location was centered. Vertically, the stimuli were aligned in a way where the center of the image or the word had a distance of 30 percent of the height of the screen, from the top of the screen. When displayed, the images were shown on a white background with a height of 500 pixels.

The words were also shown on a white background with a black text color. The font size was 40 pixels.Since we wanted our amount of stimuli to be higher than could normally be remembered, we chose to display 20 stimuli in each test. This amount is based on research done by Erdelyi and Becker (1974), as they found that more than 15 images could be recalled in an immediate recall test.

The material and stimuli used for this research are publicly available via the Open Science Frameworkfootnote{Access the OSF via this URL: https://osf.io/kduv7/}.subsection{Procedure}For the experiment, the participants had to do an immediate recall test, which had a duration of about 5 to 10 minutes. The test was completely done using the computer program discussed in the apparatus. To make sure there were as few as possible unnecessary stimuli for the participants, the experiment occurred in a quiet and non-distracting environment.First, the participants had to fill in a form, where they were asked for their first name, gender, age, native tongue and other personal information regarding word recalling (e. g. dyslexia or color blindness).

The participants also signed an informed consent form where they agreed on participating in the experiment and the collection of data. The participants were told that the program will show them words or images and instructed them to try to remember as many they could. Since the participants had to do both the word test and the image test, they were assigned a test to start with by us.

Half of the participants started with the image test, the other half started with the word test.After filling out the form the participants were presented with the first list of stimuli, displayed as discussed in the material and stimuli section. Prior to displaying this list, a red indication mark appeared on the screen for a duration of three seconds at the location the stimuli will appear.After this list was displayed, the program displayed a form with twenty empty text fields. In this form, the participants were asked to fill in as many stimuli they could remember. The participants were not limited by a time limit and were allowed to recall the words in any order.

In between the two tests, there was a short break. This break was to make sure that before continuing to the other test, the participants had something else on their mind. During this break, the participants had to answer some very simple mathematical equations. This type of distraction was chosen because it keeps the mind busy, without using words or images. The break had an average duration of around two minutes.After the break, the participant had to do the other test. This went the same way as the first test.

After this, the experiment was over and there was time for the participants to ask questions about the goal of the tests.subsection{Design}The independent variable in this study was the stimuli presented by the program, either pictures or words. The dependent variable was the amount of these stimuli that were recalled by the participants in each test. This was measured by letting the participants write the stimuli down.

The study had a within-subject design because the number of participants was small and this way we could ensure that we would gather enough data.section{Results}To test the effect of presenting words rather than images in an immediate recall test, we compared the amount of recalled items between a word test and a test where images were shown.Due to a bug in the system used to gather the data, some data got influenced by other data. As a result of this, some participants results had to be removed from the dataset.

This bug influenced the data by rewriting some participants results with new data from another participant. These duplications were easy to notice and remove from the dataset.As expected, the participants recalled more images than words. In the word test, a mean of 8.4 stimuli (42.0\%) was recalled.

In the image test, the mean of recalled stimuli was 9.7 (48.5\%). There were 20 stimuli shown for each test.

We used the mean to define the average because there were no big outliers in the dataset.It is necessary to compare the average results in these tests because the difference in these percentages show whether pictures or words are more likely to be recalled. egin{figure}ht  includegraphicswidth=0.48 extwidth{graph}  caption{Frequency of amount of stimuli recalled}end{figure}Figure 1 shows the frequency of the number of stimuli that has been recalled by the participants for both the image and the word test. The mean of the number of items recalled is higher for images than for words, with a standard deviation of 2.

0 and 2.1 for the word and image test respectively. This results in a standard error of 0.49 for the word test and a standard error of 0.52 for the image test.Because of the size of the difference in the average amount of words remembered and the size of the standard error, a significant difference between the groups is not existing.

The standard error bars overlap. Also, we did not find a difference between the results of the test for participants that started with the image test or the word test.All results gathered by the program are publicly available via the Open Science Frameworkfootnote{Access the OSF via this URL: https://osf.io/kduv7/}. The database containing the dataset can be found as an OpenDocument Spreadsheet (.ods).

section{Discussion}The question researched in this research was: egin{displayquote}“Does presenting pictures rather than words affect working memory capacity in a recall test with human participants?”end{displayquote}An inspiration for this research question was the research done by Erdelyi and Becker (1974). Here they found that people can recall more images than words in an immediate recall test. We expected that the subjects of the images shown will be better recalled than the presented words.Out of 20 stimuli, a mean of 42.

0\% and 48.5\% were remembered by the participants respectively, so on average, more pictures were remembered than words. However, due to a standard deviation of around 2.0 for both tests, these results lead us to reject our hypothesis. This means that we did not find that working memory capacity is higher for images than for words.subsection{Limitations}There were a few problems in the study that may have influenced the results and so made the research less reliable.

First of all, there was a very small group of 17 participants partaking the experiment. Due to this, it is impossible to say the results give a general view of how humans behave. The average human was also not correctly represented because the participants were very similar in age and level of education.

The last limitation due to the group of participants was that some were not as fluent in English as we expected from international university students. It has occurred multiple times that a participant wrote the name of an object in their native language instead of English. Although in these cases the participants remembered the stimulus, it was not a correct answer.Secondly, the environment where this study was conducted was an empty classroom. Sometimes people in the hallway would make noise that might have distracted some participants. Besides this, there were two classrooms where this test was conducted. Because of this, the environment was not standardized. This standardization is important to ensure that there are few external stimuli influencing the data, and the few external stimuli that do occur are equal to every participant.

The third limitation of the research were problems with the program used to display the stimuli. In a few cases, the interval of 1.5 seconds used to display the stimuli one by one, was not executed properly by the program. It has happened that a stimulus was displayed for 1 second and another one for 2 seconds. This only happened in some cases.Lastly, there was a case of spreading activation. The list of words used for the research contained the word “station”.

More than half of the times this word was shown, the participants seemed to recall the word “train” as well. This effect has been discussed by Collins and Loftus (1975).subsection{Future research}As mentioned above, the use of the English language has been a barrier for some participants. It would be interesting to repeat this experiment with only native English speakers, to eliminate this effect.

Furthermore, it could be interesting to do research on the performance in this test between verbal and visual thinkers.  Visual thinking is the psychological phenomenon of thinking using visual processing. Verbal thinkers, however, think and remember more using words than pictures. It would be interesting to study the effect this might have on working memory capacity.The effect of spreading activation on working memory capacity is also an interesting point to research. Anderson (1983) showed the effect of spreading activation on long-term memory, however, it has not been researched yet what the effect can be on working memory capacity.Lastly, it is not clear from this study what technique the participants used to remember the stimuli. They could have translated the images into words and remembered them like that.

It is also possible that when they saw a word they made a visual representation of it and tried to remember the information like that. The effect that these different ways of using the working memory can have on working memory capacity is another interesting research topic.ocite{miller1956magical}ocite{cowan2004constant}ocite{hayes1952}ocite{ErdelyBecker1974}ocite{shepard1967recognition}ocite{collins1975spreading}ocite{anderson1983spreading}ibliographystyle{apacite}setlength{ibleftmargin}{.125in}setlength{ibindent}{-ibleftmargin}ibliography{CogSci_Template}end{document}