2. Background• Memory for emotional stimuli is enhanced (e.g. Bradley & Lang, 2000)• Emotional stimuli: more arousing than neutral stimuli, of strong valence (positive or negative)H1: If we have a better memory for positive and negativearousing music than for neutral music…… then likely to make its way into our consciousexperience of involuntary musical imagery (INMI)
3. Emotional musical imagery?Voluntary musical imageryExperiment participants able to indicate the emotionexpressed in imagined music (Lucas, Schubert, & Halpern,2010) Q. transfer to INMI?Involuntary musical imagery (INMI)INMI during ‘affective states’ (Williamson et al., 2011)• Themes of ‘Mood’, ‘Emotion’, ‘Stress’, ‘Surprise’
4. Valence and ArousalValence• Positive tone of earworm music and words, experience described as ‘pleasant’ (Halpern & Bartlett, 2011)• Association between INMI frequency and its valence (Liikkanen, 2011)• Positive emotional engagement with music (Beaman & Williams, 2009) and musical preference (Hemming, 2009; Halpern & Bartlett, 2011) associated with subsequent INMI (level of processing?)• Music students sometimes attributed INMI to liking the particular tune (Bailes, 2007)Arousal• ‘Entertainment’ factor of INMI (Wammes & Baruss, 2009)• Mental relaxation and increased physical activity associated with INMI (Hemming, 2009)• INMI in ‘low attention states’ (Williamson et al., 2011)
5. AimsExplore the relationship between involuntary musicalimagery and emotion • Using findings from a follow-up of Bailes (2006, 2007) Caveat. Study not designed to test this relationship
6. MethodExperience sampling methods to observe the musicalexperiences of respondents from the general population(Bailes, 2006)Participants• N = 47 (21 male)• Volunteers from greater Western Sydney & undergraduate psychology students from University of Western Sydney• aged 18 to 53 years• Ollen Musical Sophistication Index range 39 – 944
7. Experience Sampling Form (ESF)• 2 sides of (A4) sheet of paper to be completed when messaged (Bailes, 2006)• Introductory section (date, time contacted, time filled out)
8. ESF ctd..Part BCompleted if hearing music at time of contact• Up-dated from Bailes (2006) to include laptops and mp3 players as possible sources of music• Stylistic categories updated to include trance/house/techno, country, blues, urban (rap, R&B, hip hop) and gospelPart CCompleted if imagining music at time of contact• Up-dated with style categories as in Part B• Questions adapted to accommodate respondents without musical training• Tempo/Rhythm added as a potentially important element of imagined music
9. Procedure• Briefing session: informed consent sought, distribution of background questionnaire & revised transliminality questionnaire• Participants received pack of 42 ESFs: 1 ESF to be filled out each time they receive an SMS• Bulk SMS provider scheduled sending of message “Please fill out your form” to participants 6 times a day, over 7 days, between 9am and 9pm• Quasi-random schedule, with one signal scheduled within each two-hour time period• On receipt of SMS, participants to fill out a blank ESF as soon as possible
10. Results1,415 ESFs returned (out of a possible 1,974) Imagining Music 13% No Music 52% Hearing Music 31% Both 4%
11. Musical State and MoodMultinomial logistic regression analysis DV: musical state at time of contact (hearing, imagining, neither hearing nor imagining music) Predictor variables: ratings along Part A mood pairs915 cases analysed, omnibus chi-square = 129.86, df = 68,p < .005Model accounted for 13.2% - 15.2% of varianceOnly Alert/Drowsy (p = .01) and Lonely/Connected (p = .05)reliably predicted musical state
12. INMI and MoodAlert/DrowsyBeing ‘drowsy’ or ‘neither alert nor drowsy’ significantly negativepredictor of imagining musicLonely/Connected‘Quite connected’ ratings significant predictor of imagining musicEnergetic/TiredBeing ‘neither energetic nor tired’ significantly negative predictorof imagining musicHappy/SadRatings DO NOT predict imagining musicNB. Similar model coefficients for odds of hearing music(‘drowsy’ and ‘neither energetic nor tired’ as negative predictors)
13. Imagining music from heard episodesDegree of choice in heard music• No correlation with the times subsequently imagined (rho(164) = .106, p = .17)• No difference between the reported degree of personal choice when hearing pieces that were imagined versus those that were not (U = 866.5, N1 = 152, N2 = 14, p = .226, two-tailed)Heard and imagined music mood congruencyAll mood pair ratings significantly correlated whenparticipants hear and imagine the same piece (except forAlert/Drowsy)
14. Reasons for imagining particular music Node References % of references Recently heard 61 37.7 Don’t know why 19 11.7 Stickiness 11 6.8 TV 7 4.3 Spontaneity 7 4.3 Recently imagined 6 3.7 Value judgement 5 3.1 Musical features 5 3.1 Favourite music 5 3.1 Visual cue 4 2.5 Recently sung/played 3 1.9 Imagery on waking 3 1.9 Intentional imaging 3 1.9 Sentimental/nostalgia 3 1.9 Other 20 12.3
15. Arousal and Valence When I exercise I usually listen to music on an ipod shuffle. I didn’tIt’s annoyingly cheesy have it with me today, so I usually just hear the same songs in my Like the song head I just love it Fun song The other girls at work love it, I dislike it very much Maybe because it’s a favourite song of mine As cleaning is boring – it is much easier to It was sentimental value Good song imagine some thing Bored in class. When Was the dance/music @ my I’m bored I imagine wedding music. Also at work.
16. Conclusions• Mood pairs that vary in arousal (Alert/Drowsy, Energetic/Tired) predict the likelihood of imagining music • Drowsy respondents, or respondents who are at neither end of the alert/drowsy and energetic/tired scales are not likely to have been imagining music• Mood pairs that vary in valence (e.g. Happy/Sad) do not predict the likelihood of imagining music• BUT mood congruence at level of specific piece, includes Happy/Sad• Relatively small percentage of affective reasons given for imagining music, comparable to percentage of codes for ‘affective state’ in Williamson et al. (2011)
17. Discussion• ‘Use’ of imagery as emotional self-regulation (comparable mood as when hearing)?• Low attention states & INMI in Williamson et al. (2011), but current respondents ‘quite connected’ • Diffuse attention? Herbert (2011) – absorption and everyday listening experience, trance and earworm characteristics (repetition)Q. Do affective associations with ‘real’ life influence ourmental jukebox?
18. Research DirectionsNeed to distinguish between…• Emotion/mood • Combine experience sampling methods with state-trait measures to explore interactions between mood, personality, and INMI• Perceived vs. induced emotion• Emotion of imagined music vs. self during episode• Emotion at encoding of music, of musical content, and at retrieval of music • Develop experiments to compare the induction of affective with neutral music
19. Acknowledgements Many thanks for the symposium organization.Thanks particularly to my stoic respondents, as well as topostgraduate diploma students Sarah Allen, Vicky Busuttil, Samar Dawidar and Asma Payara for data collection.