The document summarizes research on how pictures and words are encoded in human memory. It discusses several studies that have found: 1) Pictures are typically better remembered than words (the "picture superiority effect"); and 2) Concrete words are more memorable than abstract words (the "imageability effect"). The author proposes an experiment to examine if pairing low-imageable words with picture associates can boost memory performance for those words, reducing the imageability effect. The results showed pairing pictures with high-imageable words improved memory, but not for low-imageable words.

1. Running head: THE ENCODING OF PICTURES AND WORDS 1
Imageability and the Picture Superiority Effect: How do we encode pictures and words?
Aaron R. Peterson
University of Nebraska at Omaha

2. THE ENCODING OF PICTURES AND WORDS 2
Abstract
This document is a review of some of the major findings of studies done on the picture superiority
effect, the imageability effect, and related studies. Previous studies have shown that concrete words
such as key are more memorable than abstract words such as trust. Other research has shown that
pictures are typically recognized better than words. Here, I propose an experiment to examine the
relationship between these variables and recognition memory. I predict that the imageability effect will
be reduced when high and low-imageable words are accompanied by picture associates at study
followed by a recognition memory test. This effect might be reduced due to the use of both a verbal and
a visual code in the study process. This experiment sought to yield results to improve the use of word
and other visual stimuli in the field of education. The results showed that pictures were better
remembered than words, low imageable words were no better remembered than high imageable words
in isolation.
Keywords: Picture Superiority Effect, Recognition, Encoding, Imageability, Memory, Dual Coding

3. Imageability and the Picture Superiority Effect: How do we encode pictures and words?
When it comes to the human memory, encoding helps people learn and retain useful
information. Every human sense has some level of encoding; you might remember the last time you
burned your mouth with a hot drink or the smell of a fresh pot of coffee. The question here is how can
one better encode useful information and how can one encode useful information on several levels?
The point of all this is to demonstrate that how you encode information matters, and a conscious effort
to encode information in several ways can extend or reinforce memory. A good mechanism for this
study will compare the encoding of words and pictures and their implications on memory.
Peoples' past experiences may influence what they think is the best method for remembering
details. In reality there are several environmental, visual and kinesthetic factors that alter the
effectiveness of our memory. In the following section I discuss various experiments that have been
conducted on this very subject. The question here is, how can one better encode useful information and
how can one encode useful information on several levels? I hope to demonstrate that how one encodes
material matters, and making conscious effort to encode information that has been presented in a
certain manner may extend or reinforce recognition memory. This study will compare learning
mechanisms that are related to the encoding of words and pictures.
In a recent article, Fawcett, Quinlan, and Taylor (2012) described three experiments examining
both the production effect and the picture superiority effect. The production effect refers to the
phenomenon whereby producing a word aloud improves explicit memory performance as compared to
simply reading a word silently. The picture superiority effect refers to the phenomenon whereby
pictures are better remembered than words. The production effect was also tested by mouthing the
word, which is considered a produced classification. Across all of these experiments, a yes-no
recognition task was used to test overall memory scores. The results of the experiments all
demonstrated a memory advantage for words that were produced relative to words that were read

4. THE ENCODING OF PICTURES AND WORDS 4
silently. In all cases, the experiment that examined the production effect showed better memory for
pictures as opposed to words. This study reinforces the notion that encoding through imagery is
stronger than the encoding of words.
In five experiments, Dewhurst and Conway (1994) observed memory performance for visual
and abstract stimuli. This was done by studying the picture superiority effect and imageability in
separate experiments with recall, reaction time, and recognition memory tests. In the first experiment,
they used images and words with a recognition memory test to examine the picture superiority effect.
They found that the images were in fact more memorable than words. In their fifth experiment, they
studied the imageability effect using another recognition memory test. Twenty subjects studied 30 high
and low-imageable words and were then tested on 60 words (30 studied, 30 unstudied). As expected,
high-imageable words were more memorable than low-imageable words.
Encoding information is processed and stored for later use, this includes both pictures and
words. They also had reason to believe that viewing images as opposed to words had a more profound
effect on memory, this effect illustrates the picture superiority effect. The evidence behind this supports
the idea that images are encoded into our memory in a more sophisticated manner and therefore have
different implications on recall and learning. They describe recollective memory as an experience that
is based on a specific event from the past that has been encoded into long-term memory. They
simplified this experience as a feeling known as “pastness.” Another form of memory discussed was
this feeling of familiarity, which was more closely associated with known information as opposed to
simply remembered information. The encoding of images and recollective memory go hand-in-hand to
form the original basis for the picture superiority effect.
In one experiment, Dewhurst and Conway had participants view a slide-show that included one
word or one image at a time and were then tasked to quickly write down the words or images they
could remember in a free recall task. The results showed that the pictures were more memorable than

5. THE ENCODING OF PICTURES AND WORDS 5
the words, thus supporting the picture superiority effect. It should be noted that imagery when
associated with past experience seems to be linked to the recollective effect of memory, that is to say
that pictures helped participants retain memory of a past learned experience. The picture superiority
effect can be found in a plethora of research, including research that examines both retrospective and
prospective memory as well as within various age groups (see Fink, 2014).
Defeyter, Russo, and Partlin (2009) studied items related to word and picture recognition in
conjunction with the picture superiority effect. They used four different age groups to study words and
images in a time sensitive study context. The picture superiority effect occurred in all age groups
except for the youngest of participants. For the youngest groups the results were skewed in a reversed
manner, such that words were more memorable than pictures. Factors such as familiarity and
recollection were studied within each age group. Despite their results, this further supports the picture
superiority effect and how it improves memory even in a time-sensitive scenario. To truly understand
the picture superiority effect, it is important to understand how the effect can be increased or reduced
when studied under the contexts of age groups, reaction times, and other variables.
The imageability effect is a phenomenon whereby high-imageable words (e.g. car) are better
remembered than low-imageable words (e.g. faith) (Fink, 2014). In a mega-study by Cortese, McCarty,
and Schock (in press, also see Cortese, Khanna, and Hacker 2010) nine variables including
imageability were tested with roughly 3,000 words to measure memory performance. Imageability was
found to be the strongest variable in predicting performance during the memory recognition tests. Dual
coding theory (Paivio, 2011) explains why mental images have such a powerful effect on our memory.
He hypothesized that abstract words were coded verbally, whereas concrete words were coded both
verbally and non-verbally, perhaps through an image code. It was theorized that this coding happening
twice elicited a strong memory representation for high-imageable words thus providing a boost in
memory performance.

6. THE ENCODING OF PICTURES AND WORDS 6
Glenberg and Grimes (1995) wanted to know how pictures of political candidates affected
memory related to political standpoints. Photographs of political candidates were used along with
political standpoints. Political positions of candidates whose message was attached to a picture of them
were remembered better than political messages with no picture. They found that once an association
was made between the picture of a fake political candidate and one or more of their political views, it
was no longer necessary to include the picture with the memory test (because the picture in conjunction
with their political view helped to dual-encode the message from the original studied message).
However, this phenomenon was only noticeable when the picture in question was the political
candidates face as opposed to an arm, leg, or other body part. This outcome may be due to the encoding
specificity effect (Tulving & Thomson, 1973), which states that understanding contextual information
affects memory and recall. When asked about the political positions of various candidates, participants
in a study were more likely to recall information about a candidate if the information had been
accompanied with a picture of the candidate during encoding as opposed to a political paragraph alone.
They also noted that pictures helped to identify some personal characteristics of candidates (e.g. kind,
bold, happy, etc.). The participants selected and rated these personal characteristics. There was also
some discussion about the effect of distinctiveness on memory, as opposed to pictures. This
distinctiveness in images was strongly correlated with the political message when the pictures were of
human faces. That is to say that pictures of faces were more memorable than images of other body parts
that were used in the experiments due to the distinctiveness of the human face. There may be additional
encoding that accompanies the viewing of the human face as opposed to other body parts. People may
not only encode image information more elaborately than textual information, but there may also be a
more significant emotional reaction to a face in conjunction with a message. The experiment by
Glenberg and Grimes (1995) not only demonstrates the picture superiority effect, but provides
information about what is in an image and how it effects memory.

7. THE ENCODING OF PICTURES AND WORDS 7
Through these various studies it is clear that the picture superiority effect and the imageability
effect play a role in the encoding of many aspects of memory. The way we view both images and words
undoubtedly affects memory. Other variables such as age, what is in an image, whether or not you read
aloud, whether or not a word is concrete or abstract, and the planning or spontaneity of learning all
affect our memory. More specific research is required to understand imageability and the picture
superiority effect and their implications on learning and retaining information.
I investigated the possibility of a relationship between the picture superiority effect and the
imageability effect. I paired high and low-imageable words with picture associates with the hope that
these words would be coded on both a verbal and a visual level. I had hoped that this may boost
memory performance, especially for low-imageable words. A PowerPoint presentation was presented
with either a word in isolation or a word with a picture associate. The imageability rating of the words
and whether the word was paired with a picture was analyzed using a memory recognition test. The
results of the test calculated the proportion of words that were remembered as well as the words that
were mistakenly remembered. I predicted that by combining low-imageable words with picture
associates the memory performance for these words would be significantly improved. High-imageable
words already have both a verbal and visual code, so these words would likely be easily remembered.
Method
Participants
This study was done using students from the University of Nebraska at Omaha who have
registered for the study through SONA, the online extra-credit resource. Fourteen participants took part
in this study, five males and nine females. The large majority of the participants were psychology
majors, with a few majoring in exercise science and pre-nursing among other fields. The average age of
the participants was about 21. Only two participants had previous knowledge of the picture superiority
effect. All participants were required to have normal vision and English as their first language.

8. THE ENCODING OF PICTURES AND WORDS 8
Materials
Fourty eight monosyllabic words were studied during a PowerPoint presentation and 96 words
monosyllabic words were tested. Participants used a desktop computer to view all of the one-syllable
words within a PowerPoint presentation that was selected from an already existing database of a total
of 3,000 words (Cortese & Fugett, 2004). The same database was used to select words of varying
imageability based on their imageability score. A low imageable word is less concrete and more
abstract or conceptual, and is associated with an imageability score of 2.5 or less. A high-imageable
word is more concrete or thing-like and is associated with an imageability score of 5 or higher. Picture
associations were utilized in the PowerPoint presentation that were found using Google images. Only
one word or one word with a picture associate appeared on each slide. All picture associates were found
using Nelson, McEvoy, and Schreiber's (1998) database of association norms. This database was used
to find the strongest word associate to a high or low imageable word, for which a picture associate is
possible A paper and pencil test was also administered.
Procedure
The experiment consisted of three phases. The first phase included a PowerPoint presentation
that displayed 48 slides of single-syllable high or low imageable words either with or without a picture
associate. The PowerPoint presentation allowed a participant to manually control the slides for
viewing. The participant was instructed to spend no more than a few seconds viewing each individual
slide. There was a second phase that consisted of a five-minute break, during which time demographic
information was gathered along with some brief socializing. The third phase consisted of a paper
recognition memory test complete with a writing utensil. The memory test had 96 words, 48 of the
words were new unstudied words and the other 48 were previously studied words as seen during the
PowerPoint presentation.
In the first phase, participants were given 10 minutes to study the words and images displayed

9. THE ENCODING OF PICTURES AND WORDS 9
on the PowerPoint presentation. A single high or low imageable one-syllable word or high or low-
imageable picture associate with a word appeared on a slide. The pictures were not be direct images of
the words, but rather associations. This was done to reduce confounds in the low-imageable picture
slides, that is to say that high-image words do have direct images (e.g. cloud) however low-imageable
words do not have a direct image associate (e.g. trust). Participants studied 48 slides in total. There was
12 slides for each condition of the experiment. Immediately after this phase the participant was given 5
minutes of time in which demographic information was gathered. The third phase consisted of a paper
and pencil test which took approximately 10 minutes. The participants circled words on the test that
they recognize as studied items from the PowerPoint presentation. Following the paper test, there was a
brief explanation of the study. The explanation will included a note about why the pictures from the
slide show were associations and not direct representations of the words. The overall experiment took
no more than fifteen minutes per participant. It should also be noted that there were four different
version of the PowerPoint presentation, but only one version of the paper test. Each version of the
PowerPoint presentation included the same words for study and recognition, but in a different order as
a counterbalancing measure.
Design
In this experiment the independent variables were imageability (high or low) and picture status
(accompanied with a picture or not accompanied with a picture). The dependent variable was the rate of
recognition. The rate of recognition is the proportion of hits (correct words circled) minus false hits
(incorrect word circled) within the paper test.
Results
The mean proportion of hits minus false alarms is presented in table 1. A two-way within
subjects ANOVA was performed on the proportion of hits minus false alarms. The main effect of
picture status was not significant, F (1, 13) = 2.51 p = .138. The main effect of imageability was not

10. THE ENCODING OF PICTURES AND WORDS 10
significant, F(1, 13) = 2.11 p = .17. The interaction between picture status and imageability was
significant, F(1, 13) = 7.94 p = .015. This interaction was due to the large imageability effect for
pictures, T (13) = 2.96 p = .011. In contrast, the imageability effect for words went in the opposite
direction, but was not significant, T(13) = 1.79 p = .097.
Table 1
High-Imageable Low-Imageable

11. THE ENCODING OF PICTURES AND WORDS 11
In isolation .38 (.21) .48 (.10)
With picture .59 (.25) .40 (.19)
Discussion
This research shed light on the relationship between the picture superiority effect and the
imageability effect, however the original hypothesis did not yield the expected results. It should be
noted that pairing a word with a picture associate increased the hit rate for high-imageable words but
not low-imageable words. Despite the fact that results for conditions such a high-imageable words and
the condition of low-imageable words with pictures scored lower than expected. There may be a
reasonable explanation for this finding. Bleasdale (1987) studied the concreteness and abstractness of
words in an associative priming experiment. He used a lexical decision task to test the reaction time of
concrete and abstract words with associative words using semantic priming. For example, a word
pairing such as key with the word door should be recognized as a word (as opposed to a non-word)
faster than a word pairing like key and trust. This is because the word trust is more abstract than the
word key. This is also because these words are further apart in semantic space. Words that are related
are more quickly recognized when paired as compared to non-related words. However, this categorical
boost in reaction time did not occur when abstract words were paired together.
Solving the problem of how to better learn and encode abstract words and concepts seems to be
a nearly impossible feat, although future research might be able to extrapolate more information with
more participants and a better choice of words based on frequency along with a better selection of
pictures. In he long run, however I believe that the results of using a recognition memory test, lexical
decision task, and semantic priming have all shown that low-imageable words are simply more difficult
to recall or recognize.

12. THE ENCODING OF PICTURES AND WORDS 12
References
Bleasdale, F. A. (1987). Concreteness-dependent associative priming: Separate lexical organization for
concrete and abstract words. Journal Of Experimental Psychology: Learning, Memory, And
Cognition, 13(4), 582-594. doi:10.1037/0278-7393.13.4.582
Cortese, M.J., Khanna, M.M., & Hacker, S. (2010) Recognition memory for 2,578 monosyllabic words
Memory, 18, 595-609. DOI: 10.1080/09658211.2010.493892.
Cortese, M.J., McCarty D.P., & Schock, J. (in press). A mega recognition memory study of 2,897
disyllabic words. Quarterly Journal of Experimental Psychology.
Cortese, MJ, Fugett, A. (2004). Imageability ratings for 3,000 monosyllabic words. Behavior Research
Methods, Instruments & Computers. US, 36, 3, 384-387, Aug. 2004. ISSN: 0743-3808.
Defeyter, M. A., Russo, R., & McPartlin, P. L. (2009). The picture superiority effect in recognition
memory: A developmental study using the response signal procedure. Cognitive Development,
24(3), 265-273. doi:10.1016/j.cogdev.2009.05.002
Dewhurst, S. A., & Conway, M. A. (1994) Pictures, Images, and Recollective Experience.
Journal of Experimental Psychology: Learning, Memory, and Cognition 20(5) pp. 1088-1098.
doi:10.1037/0278-7393.20.5.1088
Fawcett, J. M. Quinlan, C. K., Taylor, T. L. (2012) Interplay of the Production and Picture Superiority
Effects: A Signal Detection Analysis.
Memory 20(7) pp. 655-666 doi:10.1080/09658211.2012.693510
Fink, N. (2014). Picture superiority effect in prospective memory: Examining the influence of age and
attention load. Dissertation Abstracts International, 74,
Glenberg, A. M., & Grimes, T. (1995). Memory and faces: Pictures help you remember who said what.
Personality And Social Psychology Bulletin, 21(3), 196-206. doi:10.1177/0146167295213001

13. THE ENCODING OF PICTURES AND WORDS 13
Nelson, D. L., McEvoy, C. L., & Schreiber, T. A. (1998). The University of South Florida word
association, rhyme, and word fragment norms. http://www.usf.edu/FreeAssociation/.
Paivio, A. (2013). Dual coding theory, word abstractness, and emotion: A critical review of Kousta et
al. (2011). Journal Of Experimental Psychology: General, 142(1), 282-287.
doi:10.1037/a0027004
Tulving, E. Thomson, D (1973). Encoding specificity and retrieval processes in episodic memory.
Psychological Review 80 (5): 352–373. doi: 10.1037/h0020071