To what extent can the

Working Memory model

predict behaviour during
data processing

By: Alejandro Mendoza

Daz de Len

1
Introduction

Memory is one of the principal mental processes that cognitive psychology

studies. It has often been classified into two categories: long-term memory (LTM)
and short-term memory (STM). STM is the most active type of memory, processing
data every moment that we are conscious, probably the most influential cognitive
process in humans since it constantly redefines the focus of our minds, thus having
a massive influence on our behaviour.

Many theoretical models have been proposed to describe how memory works.
One of the first important theories proposed was the multi-store model, by
Atkinson and Shiffrin (1968) dividing memory into three separate stores:
sensory, short-term, and long-term memory. However, Baddeley and Hitch (1974)
argued that the multi-store model was too simplistic and that STM should not be
treated as a single store. Thus, they created the working memory model (WM).

This model argues that STM consists of three slave systems that are monitored
and coordinated by the central executive. These three slave systems are the
phonological loop (with its two sub-systems: the articulatory control system
holds information in verbal form- and the phonological store retains information
phonologically (i.e. as it sounds)), the visuospatial sketchpad, and the episodic
buffer. This last system serves as a temporary display that does not process
information and is therefore irrelevant to this study, which concentrates on
information processing.

Baddeley (1968) asked his participants to remember a sequence of zero to eight

digits while at the same time realizing a reasoning test. The test consisted in
classifying 32 sentences into true or false categories. The study found that the
more complicated the sentence or the greater number of digits they had to
memorize, the more time participants took to answer. 1 This shows that the
articulatory control system can rehearse the series of numbers while
simultaneously the visuospatial sketchpad processes the visual information from
the sentences.

On a second study conducted by Baddeley, participants listened to passages taken

from a reading ability test and then answered questions designed to test if they had
comprehended the passage. During the presentation of the passage, the
participants had to watch a screen where series of digits appeared. In the control
condition, participants had to write down the digits as they appeared. In the other
conditions they had to recall and write down the digits until they had accumulated
three or six digits. The number of correctly answered questions decreased as
participants had to recall larger groups of digits. 2 This supports the findings from
Baddeley (1968) regarding the separation of the visuospatial sketchpad and the
phonological loop.

1
Baddeley. Alan D., Memoria humana: teora y pra ctica, Madrid: McGraw-Hill, 1999, p.59
2
Baddeley, Alan D, The Psychology of memory, New York: Harper and Row.,1976, p.173

2
 Baddeley also wanted to investigate the effects of articulatory suppression on
short-term memory. He conducted a study where participants performed the same
32 reasoning tasks from the study of Baddeley (1968) but under four different
conditions: saying nothing, repeating the word the 4-5 times per second,
repeating the a familiar sequence, or repeating a random six-digit sequence (both
also at 4-5 utterances per second). The random sequence was the condition that
diminished STM capacity the most. 3 This suggests that the two sub-systems within
the phonological loop can process information separately but interfere slightly
with each other.

The present study is a modification of this last study by Baddeley, integrating the
study mentioned before it, since these studies capture the essence of how STM
works according to WM. The focus of this experiment is to treat the working
memory model as the theory that dictates how humans process different forms of
information; this will be achieved by time-tempering participants (i.e. measuring
each individuals notion of time by allowing them to count freely until they reach a
minute) and putting them under four different conditions while they count up to a
minute (this idea was inspired by an experiment carried out by the physicist
Richard Feynman).

The aim of the study is to investigate the extent to which the predictions based on
the working memory model developed by Baddeley and Hitch in 1974 can
accurately describe behaviour during data processing.

H0 null hypothesis: There will be no change in the duration of each personal

minute between the different conditions. Any variations will be due to chance.

H1 research hypothesis: The mean time called out by participants will be affected
by the type of task they are asked to perform. The conditions are: doing sit-ups
and counting the number of sit-ups performed, reading a passage and answering
questions on it, uttering hi-yah, and counting to a minute in silence (control).

Methodology Section
3
Bower, Gordon, Human memory: basic processes : selected reprints, with new commentaries from
"The Psychology of learning and motivation, New York: Academic press, 1977., p.207

3
Design

A repeated measures design was used in order to identify how the different
independent variables affect the way a person processes data. One of the main
strengths for this type of design is that it eliminates problems of participant
variability. A controlled environment (the same isolated and quiet room) was
used, participants were given the same briefing, performed the same tasks and
used the same reading comprehension test to avoid any effect from extraneous
variables. To avoid order effects different orders of the conditions were applied.

The independent variable is:

The conditions imposed while people count up to a minute.

The dependent variable is:

The time it takes for people to count up to what they consider to be a minute.

Ethical guidelines were followed by giving participants a consent form as well as

being debriefed when the experiment was finished. They were reminded that they
have the right to withdraw at any time and that their results will be confidential.
The participant did not suffer any harm, either psychological or physical. The
intensity of the sit-ups depended solely on the participants, having the freedom to
stop if they wished. They werent asked to eat or drink anything during the
experiment. They werent pressured into finishing the experiment (see Appendix
1).

The brief and consent form were handed out before the start of the experiment,
allowing them time to read it and sign it if they wish to participate. At the end of
the experiment, they were given a debrief form fully explaining the purpose of the
experiment. The participants were deceived only minimally since they were not
told exactly what the experiment measures. This slight deception was necessary to
avoid demand characteristics. However, this deception didnt harm any of them in
any way and they were fully debriefed at the end (see Appendix 2).

Participants

The sample consisted of 10 people. They were chosen using opportunity

sampling because its the easiest way to collect participants and people who were
willing to do sit-ups were needed. They were tested independently to avoid any
distractions. All the participants were aged 16 to 18 years old, both men and
women, from a private school in Mexico City, studying the IB and who speak fluent
English as a second language.

The target population consisted of bilingual IB students studying in a private

school in Mexico City. Therefore, the results can only be generalized to this specific
group of people.

The materials needed were:

Consent form (Appendix 1)
Standardized debriefing note (Appendix 2)
Reading comprehension test (Appendix 3)

4
 Results sheet (Appendix 4)
Stopwatch

Procedure
Select participants using opportunity sampling. Include men and women from the
target population.
Give them the Standardized brief note to sign.
Give out the consent form for them to sign.
Take one participant to an isolated room where he/she isnt distracted by external
sounds or actions.
Time-temper participant; that is, ask him/her to count out loud and stop him/her
when he/she has reached a minute. Participants should count at their normal
pace. Repeat 3 times to get an average of how long their personal minute is.
Ask participant to count up to a minute in silence. This will be the control
condition.
Apply the following conditions in a random order:
1. Ask participant to count up to a minute whilst doing sit-ups; the
participant can perform any number of sit-ups, calling out whenever he
finishes counting up to a minute and stop doing sit-ups at that point.
He/she will be asked to recall the number of sit-ups performed. Record
time taken for participant to finish counting.
2. Participant must count up to a minute while he/she repeats the word
hiya at a rate of 3-4 utterances per second. Participant must stop when
he/she finishes counting. Record time taken to complete count.
3. Participant must count up to a minute while reading a passage. He must
stop reading when he finishes counting. Take the time taken for him/her
to stop. When finished, participant must answer a series of reading
comprehension questions.
Record all results in the results sheet.
Give participants a full debrief, remind them of their right to withdrawal and
confidentiality.

Results

The present study used ordinal data, from which a mean was calculated, as well as
percentage differences (see Appendix 5). Using the means allowed a

5
straightforward answer to the research question. The results do not support the
WM model: the visuospatial sketchpad condition yielded the greatest average
deviation in time from the control condition even though our prediction was the
exact opposite. The condition with the least average deviation from the control
condition was the one for the phonological loop, which wasnt expected to deviate
either the maximum or minimum amount out of the three conditions. The sit-ups
condition, which, according to the model, is supposed to use the articulatory
control system for two simultaneous tasks (counting the time and counting the
number of sit-ups), was predicted to be the condition with the most deviation.
However, this was, once again, not the case.

As seen from the bar graph, our highest deviation from the control condition was
from the condition that required the use of the visuospatial sketchpad, whilst the
least deviation was obtained from the phonological loop condition.

The error bars represent the deviation if the time obtained through the
percentage difference of sit-ups recalled from the actual number (from the first
condition) and the percentage difference of questions correctly answered (for the
second condition). In order to obtain the percentage difference of the amount of
questions answered incorrectly, we used the study of Schafer (2008) that a reading

6
comprehension test in normal conditions usually yields a 64% accuracy. 4 Thus,
the maximum amount of correct answers was taken to be 6.4 and a percentage
difference from this was obtained using the raw data. The phonological loop
condition did not require an error bar due to the lack of an external factor that may
have been used to measure this.

Discussion

Since the results do not support the WM model, the research hypothesis is
rejected and the null hypothesis accepted.

The results for the visuospatial sketchpad support the findings from Baddeley
(1968) since the percentage of questions answered correctly was 89.53%. The fact
that the percentage of correctly answered questions was so high means that both
slave systems can work independently at the same time. However, since it was the
condition that yielded the greatest time deviation from the control condition, this
would prove that the WM models claim that these two systems are completely
independent is an erroneous one because their simultaneous use created the
biggest time distortion.

The results from the phonological loop condition also support the other two
experiments by Baddeley mentioned in the introduction since the simultaneous
use of the phonological loop resulted in only a slight deviation. However, when the
entire WM model is considered, the relative deviation should be higher than the
deviation from the visuospatial sketchpad; therefore, these results would not
support the theory.

The results from the sit-ups condition does not support the WM model because
the standard deviation of the number of sit-ups counted was of only 18.65% from
the real number of sit-ups done. This would mean that people were able to count
the time and the number of sit-ups done at the same time with a slight deviation in
both numbers (the average time deviation was 24.53% less than the visuospatial
sketchpad condition), and, according to the WM model, this should actually be
impossible.

One possible explanation for the results from the articulatory control system
condition could be that the participants, instead of actually trying to count the time
and number of sit-ups performed, assigned a time value for each sit-up (e.g. 2
seconds per sit-up) and just counted the number of sit-ups required to reach one
minute at their estimated rate. Thus, this condition runs the risk of being invalid
since the exact internal processes of each individual cannot be known.

The unusually high deviation for the reading condition might be due to the
pressure felt by the participants to answer correctly the greatest number of
questions possible in order to avoid seeming stupid. Thus, they may have
concentrated more on memorizing the data from the passage rather than focus on
counting up to a minute.

4
"ATLAS | Practitioner Research Resources." Hamline University | Saint Paul, Minnesota. 04 Oct.
2010. <http://www.hamline.edu/education/adult/atlas/practitioner_research/resources.html>.

7
 The distortions of these two conditions is what might have created the apparent
distortion of the third condition when put into context; the failure of these
conditions determined the position of the third one, making the WM models
predictions seem askew. Since all our results agreed with the individual studies,
we can assume that the disagreement with the model may be due to these
distortions and lack of proportionality rather than actual mistakes of the model.

In order to clarify such ambiguities, a similar study could be carried out using a
larger and more representative sample. Moreover, a survey asking participants
how they carried out each task could be given in order to gain deeper insight and
check for validity.

The experiment lacks ecological validity since it was carried out in a room where
the participants were observed, thus demand characteristics may have influenced
the results. However, the experiment was highly controlled.

These results can only be applied to bilingual students from Mexico City, aged 15-
18 years old.

In conclusion, the present study does not support the WM model. However, it
does support the findings of Baddeleys experiments. This incongruence may be
explained through the lack of numerical values from the experiments that, when
combined under the WM model, may result in disproportional results.

8
Appendix 1 (Consent Format)

I agree to take part in this psychological

experiment and I am conscious that I will be required to read, count and do sit-ups.
I am fully aware that I have the right to withdraw from the experiment at any time.
I understand that my results will be anonymous, and I agree to the publishing of
the results of this experiment. I also understand the implications of this
experiment and that I have the right to remove my results from the study.

Signature

Appendix 2 (Debrief)

9
I understand that the purpose of this
experiment is to study how our brains process information. I did not find the study
offensive, harmful or stressful in any way. I have been fully informed about the aim
of the experiment. I am conscious that my results will be anonymous and I
agree/disagree for their publication and processing according to what the study
requires.

Signature

Appendix 3 (Reading comprehension test)

10
Extract

Saturday is the day I share with my grandpa, and we use it to clean, but we also
have fun.

We like to dance while we dust. We like to sing while we mop. We clean his car. We
do the washing together. We both wear our leather jackets. Mine is brown and
grandpas is black.

When we are done, we hurry to the store. We shop for the meals. Grandpa lets me
pick the fruits that I like best, my favorite fruit are papayas, pears and mangos but
when there are rot, he changes them while Im not seeing. He carries the food
home in his van. Grandpa tells me stories while we walk to the van and before I go
to sleep.

At night, we are done. His friends come over. He cooks dinner and sometimes
includes fruit in the salad, and then, when we start to feel tired, we go to sleep at
around 11.

Questions
1. What happens on Saturdays?
2. What do the boy and his grandpa both wear while they clean?
3. What do they do while mopping?
4. What kind of car does his granpa have?
5. Whos friends come over?
6. At what time do they go to sleep?
7. Whats the boys favorite fruit?
8. What happens when the fruits are rotten?
9. What does the grandpa include in the salad?
10. What colour is my jacket?

Appendix 4 (Results sheet)

11
 Name Age

Number when Number when Number when Average number

counting up to a counting up to a counting up to a when counting up to a
minute (1st try) minute (2nd try) minute (3rd try) minute

Control Condition Time taken to Number of Sit-ups The number of sit-ups

count up to a done by the which the participant
minute while doing participant recalled
sit-ups

Time taken to count up to a Number of questions

minute while reading the answered correctly
passage

Time taken to count

up to a minute while

repeating the word

hiya

3-4 times per second

12
Appendix 5

Difference in seconds from time in control

variable/s
Articulatory Difference # of questions
Phonological Visuospatial control in # of sit- answered
loop sketchpad system ups recalled correctly
IR (m) 14 3 10 0 3.33
FS (m) 5 44 17 3 9
SA (m) 6 9 0 4 8
BF (m) 17 2 13 5 4
RV (m) 9 14 7 0 3
DB (f) 2 4 4 8 8
MH (f) 23 30 5 8 5
AA (f) 8 13 10 1 9
MC (f) 28 10 3 13 3
KV (f) 8 30 6 16 5
Avg. 12.0 15.9 7.5 5.8 5.73

13