The primary goal of this study was to examine the effect that colour which appertains to the emotional valence of presented images has on recall for words. Secondary goals of the study examined effects of emotional valence, congruence and gender on recall for words, and to examine whether any main interactions within these variables was evident. It was predicted that there would be greater recall for words when colour was introduced that appertained to the emotional valance of the images. Furthermore, it is predicted that greater levels of recall would be obtained when the words were relating to the images. Moreover, it is also predicted that recall for words presented with arousing emotional valence would have significantly greater recall for words than neutral emotional valence and that there would be significantly greater recall for words for the female gender. The research was carried out by presenting subjects three sets of images; each image was followed by a word slide. All stimuli were taken from universally standardised systems. The study employed a 3 X 2 X 3 mixed factorial ANOVA. Sixty three subjects participated, consisting of thirty one males and thirty two females. The overall results of the first ANOVA indicated that there was a highly significant effect on overall word recall when the word was presented with a colour back-drop, colours appertained to the emotional valence of the image. Furthermore, there were significant effect’s within levels of emotional valence as well as between genders for word recall. However, there were no main interactions between other variables. The overall results of the second ANOVA suggested a highly significant effect on word recall when words were introduced that related to the images. Furthermore, emotional valence along with gender effects supported the results of the initial ANOVA. There were no main interactions between other variables.
Colour is an inseparable part of our lives presenting itself in everything that we perceive. It has been widely recognised that colours have a strong impact on our emotions and moods (Hemphill, 1996; Lang, 1993; Mahnike, 1996).
Researchers have examined how colours affect emotion. Lawler and Lawler
(1965, cited in Boyatzis & Varghese, 1994) found that children “coloured with a
yellow crayon after hearing a happy story and with a brown crayon after hearing
a sad story.” Clearly there is a close link between colours and emotions.
Furthermore, Herlitz & Rehnman (2004) studied colour affectivity characterise emotional profiles of colour in terms of certain emotional dimensions. Moreover, the significant influence of chroma (or saturation) of a colour onto human emotions has been researched and found to be consistently evident. Studies by Valdez (1994) have used a variety of colour stimuli (including coloured objects, rooms, or clothing). Valdez, tentatively concluded that their results could be generalized to colour stimuli encountered in everyday situations. A meter analysis conducted by Valdez 1994 resulted in a surprisingly accurate description of the effects of colour brightness and saturation on emotions. This study investigated colour reactions as functions of psychopathology and physiological reactions to colour, colour preferences, colour effects on emotions, colour effects on behaviour, and reactions to colour concepts. Valdez 1994 found that dominance reactions were greatest to green-yellow while reactions to red-purple elicited submissive feelings. More saturated colours elicit greater feelings of arousal. The study showed that brighter colours are more pleasant, less arousing, and less dominance inducing than are the less bright colours. Arousal increased linearly and strongly with colour saturation. However, arousal can be explained on many different levels Valdez 1994.
Arousal may be explained as levels of emotional stress, pleasant images may increase our arousal however our subjective impression is that they are highly enjoyable and well worth experiencing. It is concurred that stress can be beneficial by evaluating what happens when we are subjected to too little excitement and activity. However, when a person is over-aroused, he or she may experience unpleasant emotions rather than positive. Clearly too little arousal can be just as bad as when it’s excessive. Somewhere between is an optimum – a point at which adrenalin levels are just enough to make the brain sharper and quicker and improve performance, which is the beneficial side of stress.
Despite rapidly growing literature on the impact of colour and emotional arousal, many studies have failed to use colour samples from a standardised system of colour notation (Boyatzis & Verghese, 1994; Hemphill, 1996; Terwogt & Hoeksma, 1995). Moreover, several studies have used colour – emotion matching tasks (Zentner, 2001); matching colours (e.g., red, yellow, blue) to a certain number of emotions (e.g., happiness, sadness, anger), which results in limited assessments of reactions to colours. Of the numerous colour systems that exist, one in particular is noted internationally for its precise identification process this is known as the Munsell Colour System (Ballast, 2002; Valdez & Mehrabian, 1994). According to the Munsell colour system, individual colours have three basic attributes which include hue, value (brightness), and chroma (saturation).
A substantial body of evidence has suggested that colour background had influence on cognition and emotion; further studies have examined the affects emotion has on memory. Research into this area has been extensive, studies such as Dolcos and Cabeza (2002), suggest that stimuli which induce emotions are remembered better than neutral stimuli. Dolcos and Cabeza’s (2002), study on memory and emotion utilised sixty pleasant, sixty unpleasant and sixty neutral pictures (These images were rated previously by the participants and were not universally standardised). Subjects were instructed remember images for a later recall test. After examining recall performance, the data indicated that there was a significant difference between neutral and non-neutral stimulus. These results suggested that arousal is the main factor in memory. The results also suggested that emotional images are encoded faster than non-emotional images. A subcortical structure; the amygdala, has been related to emotional arousal. Dolcos and Cabeza, 2002 study indicated that activity in this subcortical structure modulates the hippocampal and medial temporal lobe activity, thus causing memory effects. Other experiments suggest ventro-medial / medial prefrontal region involvement in memory and learning (Windmann and Kutas, 2001). However, it has also been suggested that emotion not only enhances memory, it may also distort memory recall performance by inducing a ‘recognition bias’ meaning that subjects become more likely to react ‘old’ to a negative stimulus than that of a positive or neutral stimulus (Windmann and Kutas, 2001).
For a greater physiological perspective on emotion and memory, Wiswede et al 2006 studied memory effects in conjunction with electroencephalography (EEG) recording and event related potential (ERP) analysis in an attempt to localize the region related to emotional memory and quantify its effects. This research demonstrated a relationship between memory recall for list learning tasks and p300 amplitude which is an ERP which is recorded via EEG as a positive deflection in voltage at a latency of approximately 300ms. This signal is measured by positioning electrodes in the parietal lobe region. The presence, magnitude and timing of the signals are often used as metrics of cognitive function in decision making processes hence its use in list learning tasks. Wiswedes participants were asked to recall stimuli from a list consisting of twelve words. These were presented in front of pictures taken from the International Affective Picture System (IAPS) The International Affective Picture System (IAPS) is widely used in studies of emotion and has been characterized primarily along the dimensions of valence, arousal, and dominance. One word in each list was made distinct by either font colour or by presenting a distinctive background IAPS picture. This isolation process was initially utilised by Von Restorff (1933).
Wiswedes’ findings indicated that recall performance was enhanced for colour but not for emotional isolates. Wiswedes’ findings suggested that event-related brain potentials displayed a greater positive -P300 component for recall of non- isolated words and colour- isolated words, when compared to respective non- remembered words, however not for words isolated by an arousing background. Suggesting that highly arousing environments may force the cognitive system to disrupt rehearsal processes in the working memory, which might benefit transfer into more stable memory systems. Effects of distinct properties of arousing background stimuli may also be a factor in this process, as distinctiveness affects memory performance. It is suggested that subjects recall a greater number of items from a list that are distinct in at least one dimension. This phenomenon is known as the von- Restorff or VR effect. Distinctiveness of an item can be created by altering the colour, size, the meaningfulness, background colour or any other aspects of a stimulus.
The current study aims to further delineate the relationship of IAPS images and word recall for words presented between item distinctiveness and recall performance by introducing emotional as well as physical attributes for the words to be recalled.
The relationship between images and memory has long fascinated psychologists, yet the evidence available is somewhat inconsistent. Psychologists such as Allan Paivio and Gordon Bower suggested that when participants were instructed to link items together through images, their recall was considerably improved, however, the work of some researchers, (Sheehan, 1971, Shaughnessy, Zimmerman and Underwood, 1972, Richardson, 1972) contested these findings.
The work of Morris and Stevens (1974) set out to investigate these inconsistencies in the evidence. They set out to investigate whether recall is facilitated only when the images which are formed link together the items. If it’s the image that improves recall in memory tasks then what is it about the image that facilitates this? Morris and Stephen (1974) proposed four different ways in which this effect would be achieved. Firstly, and perhaps most obviously is that imagery provides a means of connecting and organising material. Secondly imagery may enhance the ‘distinctiveness’ of an item, thus rendering it more memorable. Thirdly, storing the item as both a visual and a verbal memory, may aid recall.The results of Morris and Stephen’s (1974) experimentation appears to suggest that only the first factor, the connecting and organising of items has any effect on improving memory.The current experiment extends beyond this theory, the study examines whether there is a significant effect for the recall for words that are congruent to the IAPS images over that of non-congruent words, to see whether a direct relationship between word and image (word relating to image) would significantly affect word recall.
Herlitz and Rehnman (2004); mentioned on page two, also proposed a theory that human sex influences the ability of memory recall in their 2004 study. Their findings indicated that there was a significant gender difference in episodic memory, in the direction of females. Results suggested that females excelled in verbal episodic memory tasks, such as remembering words or pictures, and males outperformed females in remembering symbolic, non-linguistic information, known as visuo-spatial processing.
Therefore the current study tests some of the findings previously mentioned. However, the researcher has found no evidence of past studies utilising universally standardised measures in regards to colour as well as IAPS images and recall for words.
The predictions for the first ANOVA are that words made distinct by introducing a colour backdrop appertaining to the emotion of the IAPS images would be significantly higher than those without a colour backdrop due to the distinctive effect of the colour along with the theory that colour further evokes the emotional valance of the image. Furthermore, it is also predicted that recall for arousing IAPS images would have significantly greater recall for words than neutral emotional valence. Moreover, greater recall for words between genders is predicted in the direction of the female gender.
The predictions for the second ANOVA are that words made distinct by being congruent to the IAPS images would be significantly higher than for non congruent words As it is theorised that by presenting words congruent to an image helps store the stimulus as both visual and verbal memory, and thus may aid word recall. It is also predicted that emotional valance effects along with gender effects on word recall would be similar to results found in the initial ANOVA.
This study employed a 3 (words corresponding to IAPS images (congruent),words not corresponding to IAPS (non-congruent) and words congruent to IAPS images with appropriate colour backdrop appertaining to emotional valence of the IAPS images – between) x 2 (gender – between) x 3 (valence – within) mixed subjects ANOVA. Congruence was examined on two factors; words congruent to image and words non-congruent. Gender examined on two factors; male and female, and valence examined on three factors; neutral, pleasant and unpleasant.
The IAPS images were presented in an identical procedure for each of the three between subject variables, each test consisted of three within subject variables in valence (IAPS images provoking a neutral emotion IAPS images provoking a pleasant emotion, and IAPS provoking an unpleasant emotion).
63 participants were employed to carry out the study (31 males, 32 females) 36 were randomly assigned members of the public, 27 were randomly assigned students from the University of Central Lancashire. The mean age was 30 years of age (range = 18 – 64 years). None of the participants had known defective colour vision or dyslexia. No other personal information was collected.
The study employed 36 International Affective Picture System images (IAPS) presented via a 19? lCD PC monitor, the IAPS images were categorised by valence scores; unpleasant, neutral and pleasant there were 12 IAPS images within emotional valence levels. All IAPS images were controlled within a range of; unpleasant valance 1.45 – 1.91, neutral valance 5.01 – 5.53 and pleasant 7.24 – 8.34 (see Appendix section for image lists).
Three colours taken from the Munsell colour system were utilised; pleasant = yellow (7.5y (hue), 9 (value/ 10 (chroma), unpleasant = yellow/green (2.5g (hue), 5(value)/10 (chroma) and neutral = grey (n/s). These colours were chosen as they had been examined by Naz Kaya, Helen H. Epps in their 2004 study ‘relationship between colour and emotion’. The colour samples were compared to the Munsell colour system via Microsoft colour programme, a full page view of each colour was presented with the words presented centrally. These colours were chosen from the findings of Helen H Epps PH.D in her study ‘Relationship between colour and emotion’ as they were supported by studies carried out by Ballast, 2002, Linton 1999, Davey, 1998, Saito, 1996, Mahnke, 1996, Wexner, 1982.
The word stimuli were extracted from ‘common five letter words’ based on TWI-2006&CSW-2007 Prepared in 2008 by John Chew. (Word list for the 3 between subject variables are presented in the Appendix section). The incongruent words were arbitrarily chosen by the researcher to ensure they were not related to the emotion of the pictures. Congruent words were chosen with best fit to the content of the IAPS image (i.e. image of a car accident was presented with the word ‘crash’, image of a beach presented with the word ‘beach’).
The materials were put together in power-point and comprised of standardised verbal instructions, identical IAPS slides (one variable with colour backdrop), followed by a word screen and all research stimuli were presented for 2 seconds (measured using PowerPoint slide timings). A verbal debrief was presented at the end of each presentation. Also a typed debrief was available for the subjects to take with them. All subjects undertaking the relevant variable were presented with an identical presentation. (The study presentation disc can be found in the appendix section).
Samples of the unpleasant IAPS images were shown to the subjects prior to the study taking place (These sample images can be seen in the Appendix section), it was explained that they may leave the study at any time.
A data recall sheet, containing three boxes appertaining to each of the three test blocks (valence) was employed to collect the raw data. (A copy of this is illustrated can be seen in the Appendix section).
Although the brief and the debrief were presented on the lCD monitor as well as verbally recorded a hard copy of each was also presented. (This can be seen in the Appendix section).
All participants gave informed consent when taking part in this study, which was approved by the University of Central Lancashire ethics procedure. Stimuli were presented visually in three differing tests: One for recall for words incongruent to IAPS images within three study test blocks (pleasant, neutral and unpleasant), the second, recall for words congruent to IAPS images (with colour backdrop appertaining to valence) of the three within subject variables (valence) and the third; words congruent to the IAPS images in the three within subject variables (valence).
A preliminary test was compiled based on SAM ratings, this manipulation check tested that subjects rated the IAPS images with the correct emotional valence; a satisfactory level of internal consistency was gained. Thus, the emotional profiles of pleasant, unpleasant and neutral colours were deemed as consistent throughout the study.
The main study was carried out in a quiet lab room at the University of Central Lancashire; room size was approximately 3m x 2m, all participants executed the study under daylight with lights on, window closed and all presented on an identical LCD screen to control the natural pigment variation between monitors. All stimulus including instruction and brief/debrief were presented via LCD screens. Currently, with the development of visual technology, extensive research into the effects of colour onto human emotion has to this day been operated through LCD screens. This method of presenting colour stimuli is thus valued.
Participants were tested individually, all participants were shown three examples of unpleasant IAPS images due to the nature of the images used, it was explained to all participants that they were not under any obligation to undertake the study and could leave at any time. The instructions, and debrief were displayed via PowerPoint slides alone. Participants were provided with a raw data sheet for memory recall. The order of presentation of valence for each between subject variable was standardised across all participants. Each experimental session lasted approximately ten minutes.
The participants were instructed to view twenty four slides (twelve IAPS image slides each followed by a word slide) in each of the within subject variables (pleasant, neutral and unpleasant)). After the welcome slide, brief and task instructions, participants viewed each IAPS slide for two seconds followed by a word slide for two seconds at the start of each variable an introduction slide was presented. This was carried out across all variables.
Following each of the three study test blocks, the participants were given sixty seconds recall period during which they were instructed to recall as many of the five letter words as they could recall by filling in the relevant box provided on their recall sheet. Once the test was complete the participants were thanked for their time and debriefed via a pre-recorded verbal presentation and via a hard copy to take away when exiting the laboratory.
The raw data consists of total word recall and was analysed by gender, word congruence, valence levels and colour/no-colour presentation. (This can be viewed in the Appendix section). The data was analysed using Statistical Package for Social Sciences (SPSS) software programme version 17. Descriptive statistics were used to summarise data, based on the results obtained from the participants’ responses.
The CI graphs appertaining to the above tables can be viewed in the appendix section.
The initial 2 tailed mixed factorial ANOVA examined the main effects of between and within subject factors on the dependent variable. Between subjects factors = colour; appertaining colour and no colour, and gender; male and female. Within subjects factors = emotional valence of IAPS images; neutral, pleasant and unpleasant. The dependent variable = recall for words.
The test results for within subject factors indicated that there was a significant main effect of valence on word recall (F (2,118) = 4.52 p< .05 indicating that word recall was affected by the valence of the IAPS images. Word recall in the neutral valence was significantly greater than word recall for pleasant valence. Bonferoni post hoc test results indicated that there were no significant interaction between unpleasant valence and neutral valence p = .62 or pleasant valence and unpleasant valence p = .21.
The results also indicated that there was no significant interaction within valence and colour (F(2,118) = .78 p> .01 indicating that word recall was not significantly affected when valence and colour interacted. Moreover, results indicated that there was no significant interaction within valence and gender (F(2,118) = .12 p> .01 indicating that word recall was not affected when valence and gender interacted. Furthermore, there was no significant main interaction with valence, colour and gender (F(2,118) = .27 p> .01 indicating that word recall was not significantly affected when valence, colour and gender interacted.
The test results for between subject factors indicated that there was highly significant main
effect of colour on word recall (F(1,59) = 16.13 p< .001 Thus suggesting that participants
presented with words with a colour backdrop relating to the mood of the IAP significantly
recalled more words overall than those without colour. Results also indicated that there was a
significant main effect of gender on word recall (F(1,59) = 10.28 p< .01. The gender effect
suggests that females recall significantly more words than males overall.
(Females = 7.48, Males = 6.08).
Results also indicated that there was no significant interaction between colour and gender
with word recall (F(1,59) = .16 p> .01 suggesting that word recall was not significantly
affected when colour and gender were examined.
A second 2 tailed mixed factorial ANOVA examined the main effects of between and within subject factors on the dependent variable. Between subjects factors = gender; male and female and congruence; words relating to the IAPS images, words not relating to IAPS images. Within subjects factors = emotional valence of IAPS images; neutral, pleasant and
unpleasant. The dependent variable = recall for words.
The test results for within subject factors indicated that there was a significant main effect of valence on word recall (F (2,118) = 5.44 p< .01 indicating that word recall was affected by the valence of the IAPS images. Word recall in the neutral valence was significantly greater than word recall for unpleasant valence. Bonferoni post hoc test results indicated that there were no significant interaction between pleasant valence and neutral valence p = .79 or pleasant valence and unpleasant valence p = .08.
The results also indicated that there was no significant interaction within valence and gender (F(2,118) = .71 p> .05 indicating that word recall was not significantly affected when valence and gender were tested. Moreover, results indicated that there was no significant interaction within valence and congruence (F(2,118) = 2.09 p> .05 indicating that word recall was not affected when valence and congruence were tested. Furthermore, there was no significant main interaction with valence, gender and congruence (F(2,118) = 1.13 p> .05 indicating that word recall was not significantly affected when valence, gender and congruence were tested.
The test results for between subject factors indicated that there was a highly significant main
effect of congruence on word recall (F(1,59) = 20.29 p< .001. Indicating that words
congruent to emotional valance of the IAPS images affected word recall. Results
also indicated that there was a significant main effect of gender on word recall(F(1,59) = 7.82 p< .01. The gender effect suggests that females recall significantly more words than males
throughout the study. (Females = 7.48, Males = 6.08).
Results also indicated that there was no significant interaction between gender and
Congruence with word recall (F(1,59) = 3.45 p> .05 suggesting that word recall was not
significantly affected when colour and gender were tested.
Results of the current research indicated that there were four main effects, two of which were highly significant. These were the effects of colour on word recall and congruence on word recall. The remaining significant main effects were that of valence and gender on word recall, there were no main interactions found.
The initial mixed factorial ANOVA indicated that there was a highly significant effect between words with a colour backdrop appertaining to the emotional valance of the IAPS and no presented colour. The mean scores for word recall between colour and no colour indicated that words presented with a colour backdrop had greater recall overall.
The mechanisms underlying this highly significant effect are concluded to be due to colour creating a VR-effect as proposed by Von Restorff in 1933, by giving distinctiveness to the words to be recalled.
However it is theorised that the highly significant effect may have been due to human emotions being induced by colours, supported by Valdez 1994 whose study resulted in an accurate description of the effects of colour brightness and saturation on emotions. In the current study it is proposed that by deepening emotion by presenting colours appertaining to the emotional valance of the IAPS images created greater emotional arousal. It is proposed that this was the key factor for the highly significant recall for words presented with colour and that by creating a stronger distinctive dimension to the valance of the IAPS images, stimuli was encoded faster than less emotional stimuli, supported by Dolcos and Cabeza 2002.
Multiple studies have shown that yellow relates to pleasant emotional arousal and has been perceived to be highly preferred regardless of race, culture or age. Studies such as that carried out by Birren 1978, Valdez 1994 and Hemphill 1996 examined colour and emotion. Results indicated that bright colours elicited mainly positive emotional arousal, hence the bright yellow colour used in the current study for pleasant stimuli. While green-yellow colours elicited 71.4% negative emotional arousal, confirming the results obtained by Boyatzis and Varghese (1994). In their child study they found that children used a bright coloured crayons after hearing a happy stories and a dull crayon after hearing a sad stories. In Hemphills’ study; green-yellow, elicited the feelings of sickness and disgust, hence the green-yellow used in the current study for the unpleasant stimuli. However, as colour always appertained to the IAPS images during the current study then it can’t be concluded that it was the inherent emotional characteristics of colour that was the cause of a highly significant word recall between colour and no colour. Therefore, it is suggested that a further study should utilise a full factorial design examining how colours appertaining to emotional valence of images and colours that do not appertain to emotional valence effects word recall.
The present results, have not answered the proposed question entirely, as it can only be
concluded that colour effects recall for words. However it has pathed the way for further
research which would answer the question proposed, entirely.
Moreover, there was a significant main effect between neutral and pleasant valence, results
indicated that a greater number of words were recalled when presented with neutral IAPS
image. There was no main effect with neutral and unpleasant or pleasant and unpleasant
emotional valences. This does not support Dolcos and Cabeza’s (2002) study on memory and
emotion. They found a significant difference between neutral and non-neutral stimuli, with
non-neutral stimuli having greater recall. It may be concluded that subjects may have been
distracted by the pleasant emotional content of the IAPS images thus paying less attention to
the words. Another theory is that arousing emotional stimuli within an image is processed
differently than that of colour invoked emotion, this may be supported by the current
approach to the organisation of the visual cortex, that certain brain regions are solely
dedicated to particular aspects of perception.
Furthermore, results indicated a main effect between gender and word recall. The mean scores for overall word recall indicated a significant main effect for recall directed towards females. This supports the initial prediction that there would be a significant difference between genders. As research looking into memory performance for male and female revealed that females did better at verbal episodic memory tasks, such as recall for everyday events, words or objects, and males outperformed females in remembering symbolic, non-linguistic information, known as visuo-spatial processing (Herlitz, 2008). The probability of genetically based differences between the quality of male and female memory remains unknown, however the results indicate that females currently hold the advantage in episodic memory.
The secondary mixed ANOVA results showed that there was a highly significant effect between congruent words and non- congruent words. The mean scores for word recall between words congruent to the IAPS image and words non-congruent indicated that recall for congruent words was overall highly significant than non congruent words.
The mechanisms underlying this highly significant effect are theorised to be due to the relationship of the images and words improving memory recall. Morris and Stephen (1974) proposed four different ways in which this effect would be achieved and their experimentation supported their initial view, that images provide a means of connecting and organising material. The results of this experiment extends on this theory, recall for words congruent to the IAPS images was highly significant over that of non-congruent words, thus also supporting Morris and Stephen third hypotheses. As presenting a word congruent to an image helps store the stimulus as both visual and verbal memory, and thus may aid word recall as the words are distinct in another dimension creating a VR-effect.
Furthermore, the significance of the main effect of valance and gender for the secondary mixed ANOVA equals that of the initial mixed ANOVA (See above).
The confounding variables in the study were consciously controlled throughout, however it is Impossible to control for all extraneous variables, a number of suspected extraneous variable