Chapter Nine

Research Study: The Interconnection of Memory, Emotion and Valence

My inspiration for this research is in response and support of this discussion, and also due to the small amount of research that has previously been conducted to investigate emotional valence of music in association with memory. Studies have been conducted which look to anticipate future appreciation to music by looking at past emotional responses(147), or how musical emotions can be predicted by feelings of entrainment(148), but studies looking to observe correlations between familiarity and valence are few. Eschrich conducted a study to determine whether memory of musical pieces was influenced by emotional valence and arousal. Participants were asked to listen to 40 pieces of music twice, each time a week apart, and on the second listening choose whether the piece of music was heard the previous week, or if it was new. Physiological reactions such heart rate and breathing rate, and self indicated valence scores were correlated against each accurately or inaccurately chosen piece. Findings showed that the level of arousal didn’t strongly correlate with a higher accuracy score, but strong emotional ratings did. Therefore, listeners experiencing emotions is a strong indicator of predicting whether a piece of music will be seen as meaningful to us, and therefore remembered(149). 

However, this study and studies like it seek to find a relationship between if a piece being remembered is due to emotions playing a part in encoding of this memory and musical recognition, however, they do not look to observe a relationship between positive valence and music because of this memory. There is however, anecdotal evidence of this from the field of music therapy, which we will come to look at in the following chapter.

Therefore, I was interested to see if I could find any support for this and conducted a self- report questionnaire, where my interests were twofold. Firstly, I wanted to collect data to observe from the results how people perceived the emotion expressed within music, and what emotion they felt as a response. And secondly, I wanted to be able to see if there were any correlations between memory a piece of music, and the valence a person held for it due to this emotional familiarity.

Method:

    Participants:

The participants in my study were all students of the Koninklijk Conservatorium, or students from outside the conservatoire who also had a musical background. The questionnaire was sent to 132 people, and there was a 47% response rate; 63 people responded and are subsequently included in the study. All the participants are very familiar with classical music, and play instruments themselves. 5 people didn’t complete the second phase of the experiment, so their answers for the valence section of the questionnaire completed in the first week have not been included in results, as their scores would skew the data, but their answers to the Emotional Appraisal questions which were only asked in the first week have been included.

    Materials:

Musical stimuli: For responding to the questionnaire, participants were provided with a playlist which contained 22 excerpts. Excerpts were taken from the Naxos Music Library, and included publically available recordings from a variety of musicians.

Measures: I created a simple response sheet which participants were able to download and electronically send back to me. The sheet contained 45 questions, and was divided into the two separate parts of the study.

Procedure:

The questionnaire was sent out electronically, along with a consent form, to participants who agreed to take part in the study. They signed consent, and agreed to the anonymous use of their responses in this study.

The questionnaire in two parts (an encoding and recognition phase, testing their emotional appraisal and valence) were each completed a week apart so as to allow time for the excerpts to be known to the participants and remembered in the second part, but not not so much time as they would forget. The questionnaire completed the first time with 15 questions on emotional appraisal of musical excerpts, and 15 emotional valence questions of the same excerpts was completed by the participants, and the second time, 15 questions on emotional liking were asked with 8 repeated excerpts, and 7 new ones.

For the participants to respond to in the two part questionnaire, 22 musical pieces were selected from the flute repertoire, and the recordings were all performances with flute and piano, and of which the participants heard 30 second excerpts from each. The excerpts themselves were specifically chosen due to their differing styles, known emotive character storylines, and the differing emotive qualities due to varying harmonies, melodic styles, embellishments, tempi, key, articulation, era, etc. From this selection, I then created a playlist of these excerpts for participants to listen through to, the number to the correlating question on the response sheet was displayed on the screen, to allow for ease of filling in answers in the sheet. The participants were then also free to stop and start listening to the excerpts as they pleased, so they would not be rushed into providing an answer. The first week of the questionnaire presented the first 15 excerpts, and in the second week 8 recurring excerpts, and 7 new ones were presented.

Alongside the musical excerpts, the participants filled out a response sheet which was numbered, and divided into two sections corresponding to the two sections of the questionnaire. In the first section, the questions 1-15 were numbered so, and each correlated with a musical excerpt with the same open questions applying to each: What emotion do you think this piece is portraying? What emotions, if any, do you feel whilst listening to it? The second section also asked 15 questions, and participants were required to score the excerpt from 1-10 on how much they personally liked the music.

I specifically chose for the questions in the first part of the questionnaire to be open ones, and didn’t present participants with a pre- selected list of emotions to choose from as I wanted a true representation of what each person thought and felt, and didn’t want to influence their responses by giving pre- set emotions which they “should” feel, and create any demand characteristics. 

The pieces which I chose to take excerpts from due to their strong emotive qualities and clear use of harmonies, articulation, rhythm, tempo etc for expression were:

Heard on the first week for valence and appraisal questions:

Widor Suite, mvt iii

Jolivet, Chant de Linos

Takemitsu, Voice

Copland, Duo, mvt i

Martin, Ballade

Prokofiev, Sonata, mvt iv

Liebermann Sonata, mvt ii

Heard on the first week in the valence and appraisal questions, and on the second week in the valence questions (repeated excerpts).

Fernyhough, Cassandra’s Dream Song

Messiaen, Le Merle Noir

Bach, E Minor Sonata, mvt ii,

Hue, Fantaisie

Rouse, Concerto, mvt ii

Chaminade, Concertino

Taktakishvili, Sonata, mvt iii

Hindemith, Sonata, mvt i

Heard on the second week for valance questions (New excerpts)

Reinecke, Concerto, mvt ii

Burton, Sonatina, mvt iii

Mozart, Concerto in G major, mvt i

Poulenc, Sonata, mvt iii

Doppler, Hungarian Fantasie

Muczynski, Sonata, mvt 1

Dutilleux, Sonatine

Findings:

From the first part of the questionnaire, I assembled the most commonly given key words which the participants said they both understood and felt, if any at all, whilst listening to each piece. Often, there was general agreeance upon the expressed emotions; participants had very similar thoughts and feelings as to what emotions the piece held and was trying to express. There were larger differences in what individuals experienced in their own feelings however, due to a large number of individual and environmental differences I will come to discuss. Additionally, as Juslin suggests, emotion sometimes doesn’t come into the equation at all(150), and this is in line with what participants often responded in this study. It was frequently reported they didn’t actually experience any emotion themselves, but where they were experienced, the emotions were often of a similar category (but sometimes included extraneous emotions), and varying words to describe a particular emotion state were used. Selected responses are displayed in the table below.

Findings from Emotion Appraisal, week 1

It is interesting to observe the difference in emotions which were actually experienced in comparison to what participants recognised as being expressed. One example which stands out to me is that of the piece from Mozart. The piece itself was agreed upon by participants thought to be expressing happiness, confidence and hopefulness, however the felt emotions of listeners included those of dread, or felt they were analysing this more closely. A suggestion of why this may be perhaps lies in the range of people from the sample group. As mentioned earlier, all of my participants are or were students who are musically trained, and orchestral auditions and exams often require us to play pieces from Mozart. Therefore due to episodic memory or association, upon hearing Mozart, our own emotions due to these personal memories may drown out those which the piece expresses.

Another excerpt where felt emotions deviate from those perceived to be expressed is in the case of Chaminade’s Concertino. Here the emotions thought to be expressed by the music are of happiness and optimism, however one individual described themselves as becoming bored. This piece is technically virtuosic in the French romantic style, and incorporates many scalic passages and rapid technical playing. Perhaps again, as a music student, this individual had heard many pieces in a similar style before, and due to previous memories of hearing music such as this, became bored and subsequently gave a lower valence rating.

From the data of reported valence, I collated the average scores of each piece in the graph below, to show and allow for comparison of the scores upon both the first and second listening. Also included in a table which displays the range of emotional valence scores given by participants from the first to the second week.

As we can see from the graph, although the scores for the repeated pieces heard each week are not significantly higher or lower than scores given for the music heard solely in week 1 or 2. Also from this graph, we can clearly observe that emotions are positively valenced, and there is a generalisation that people did not feel negatively towards the music, with most of the ratings being around 6-7.One small exception to this in the case of Hindemith, show the distribution increasing by one point downwards due to two people experiencing a decreased liking for it in comparison to the first listening. However, this didn’t affect the mean, and the average liking score for this excerpt remained the same.

Aside from this, we can see that the valence scores when compared week 1 to 2 either stayed the same or increased; on average emotional valence ratings increased by 6.8%. This suggests that people continue to like a piece of music if they initially had positive feelings towards it (providing they don’t hear it too much), or in fact even more positive emotions are experienced upon further listening. One exception to this is in the case of Fernyhough.

When we take the example of Fernyhough’s Cassadra’s Dreamsong, we see that the range of scores given by respondents from week 1 to week 2 remains the same; participants give valence ratings from 2 to 7 both weeks, signifying that those people who strongly disliked the piece remained strongly disliking the piece, and the top score of 7 also didn’t increase for anyone higher than this. However, we can see that the average valence increased by one score point from 3 to 4, suggesting that there was a change in those people who gave scores in the middle of this range. This suggests that people who don’t have such strong negative or positive emotions are subconsciously more open to changing their views.

Additionally, in support of Juslin, who we saw earlier state that people generally experience positive feelings more than negative ones in response to music(151), the reported scores overall tend to be positively inclined; the highest score of personal liking was 9, whereas the lowest given was 3. This perhaps shows that participants didn’t want to appear negative, but also that even when listening to such music that is difficult or uncomfortable to listen to due to a loud dynamic, being high in pitch or extremely dissonant and non- melodic, by the second listening, liking ratings had either stayed the same or increased, due to this phenomenon where people have the desire to have positive emotional experiences.

Score Evaluation:

In support of Bradley and Lang, it can be difficult to know exactly what facets within the music we respond to(152). From the models and theories we have looked at so far however, looking to musical examples from this study, we can see how these same models and mechanisms function within a piece, and trigger our correlating emotional responses. I have chosen excerpts here which triggered our basic emotions and balanced them with the theories discussed earlier to help explain their occurence. 

To begin, the excerpt taken from the second movement of the Liebermann Sonata was seen to be expressive of anger, fury and fear. Defining musical characteristics of this excerpt include loud and explosive dynamics, driving rhythms, staccato articulation, fast tempo, high frequencies, high pitch and varied pitch, and upward phrase endings. Due to the combination of many of these musical features, this piece communicates both fear and anger. We can relate this back to Scherer and Oshinsky’s study we saw earlier, where all these features were represented in their study, and the additional feature of the composer’s use of high frequency short phrases add to our understanding that this piece also expresses urgency(153). Also, features of this piece echo back to Kotlyar and Morozov’s study, where they explain where fast tempo is connected with fear, and for us to use clear articulation is instinctual to express strong emotions, such as anger(154). Upon hearing all of these musical features displayed in a piece such as this, we can look to Juslin’s BRECVEM model to understand why we also feel the similar emotions to those which the piece holds. Our brain stem reflexes are triggered, as the urgent sounds combined here initiate responses in us which would allow us in a real life situation to respond(155). Additionally, due to the fast tempo of this piece, and our capacity for rhythmic entrainment, we may respond proprioceptively, and our heart rate increase, which in turn triggers us to feel “uncomfortable”, as respondents in this study suggest. In the emotional valence scores given in the questionnaire I conducted, this piece was rated highly, participants giving it a mean rating of 7. Due to the high potential for physiological and emotional responses here, it follows that participants liked it well even though the piece elicited “negative” feelings of uneasiness.

(156) 

Duo by Copland was perceived as containing sadness and loss, and subsequently elicited feelings of sadness in listeners. The piece does this by employing musical mechanisms such as a slow tempo, downward pitch contour, not clearly articulating but largely slurring the phrases, and using open fifths in the melody, and the pitch level is low and varies very little. Additionally, the tone a performer would likely use here is a soft, open and rich sound, which heard from afar resembles that of the human voice. Looking back to Goffman, he explains how our brains innately emotionally recognise non- verbal communications, and how others understand this expression(157). Due to the use of open fifths from the flute melody in the opening of the piece, we cannot distinguish whether we are in a major or minor key. This puts us into a state of emotional indeterminacy, and it is because of this we feel the sense of void and loss the participants described(158). The use of space and silence also add to this void, and the soft use of articulation, and downward facing phrases instigates an ill-defined sadness in us. As Scherer earlier forwarded, in sadness we are not trying to express an urgent emotion, but instead sink more into a negative feeling(159). 

This excerpts can also be connected to Juslin’s BRECVEM model, where the mechanism of contagion functions for us here; we hear the sounds of social loss and sadness, and we recognise these expressions due to our brains’ evolutionary trait to be able to understand emotional vocal communication, and mirror this to allow for greater understanding of each other, and closer social bonds(160). 

(161)

Takemitsu Voice- fear, use of voice shouting in the beginning, goffman scherer heimoltz- alerts our nervous system and brain stem directly. high pitched scream of the flute, high fundamental frequency, loud dynamic, for flute alone, so no security of piano harmony- alone and scared, without time signature, again insecurity, hard attacked articulation, large difference in dynamics

The excerpt from Takemitsu’s “Voice” is an interesting one, as in the opening phrases which the participants heard, the performer actually uses their voice as part of the piece. This piece was seen by participants as communicating fear, again closely tied with anger, and features the flute player strongly saying the words “Qui, Vala?”, which translates to English as “Who goes there?” which is closely followed by a vocal-like scream on the flute. This use of language adds another element of emotional expression to this piece, as aside from the feelings we understand from “screams” such as this one, due to the animalistic non- verbal expression we can understand as “response cries(162)”, as suggested by Goffman, but the language also gives us a direct verbal indication of emotion and meaning in the piece. It is also interesting to hear these primitive response cry sounds placed directly next to what these sounds evolved into; language as a means of emotional expression(163).

Relating to Goffman, this “response cry” utilises the function of harmonics, here asking for a sounding C natural on a D harmonic. This use of fundamental frequency connects to the idea that fundamental frequency allows us to recognise individual voices of people distinct from others(164,165), but also the use of high fundamental frequency, upward pitch contour, high pitch, large pitch variation and shallow tone tells us a lot of what the voice is expressing. As buy Scherer, an upward pitch contour is indicative of fear, a high fundamental frequency of anger, and large pitch variation suggests surprise(166). We can see all of these aspects reflected in Takemitsu’s piece, as fear and surprise are expressed by the speaker towards an unknown being, as anger follows from the character portrayed by the flute.

The lack of time signature also makes us feel insecure, adding to the mounting fear, and the hard attack and variety of articulation such as flutter tongue combined with the large degree of change dynamically over short periods of time further adds to our sense of fear, tension and surprise(167).

 (168)

Challenges and Evaluation:

Upon evaluation of these results, it is clear that there are some limitations with this form of study, and which could be adapted if I was to continue this research further. 

First, this study selected only classical music from the repertoire for the flute, and this restriction in genre means that my results are not accounting for emotional and valence responses of music from pop, rock or folk music for example, and these styles other people may have stronger or lesser feelings in response to. 

To encourage as many people to participate in this study as possible, participants were able to complete it in their own time, and wherever they wanted. However due to not being able to control the environment of all the participants and maintain this, results in felt emotions and liking may be due to factors such as time of day and tiredness, temperature or amount of other people in the room with them, light, motivation or hunger. 

To be ecologically valid, the size of the sample group could be made a considerable amount larger, the age range was also narrow; all the participants were within the ages 20-30, and this also could be expanded to increase the generalisability of results. Additionally, whilst many people were from a range of different countries, there is still a small cultural validity due to everyone being from the western world, and studying music from the western classical tradition. Also, I also asked for the questionnaire to becompleted in English, and due to the vast range of countries people originate from who took part in my study, their knowledge of  emotive words in English may have influenced the words they chose to describe their experience; they may for example have a more precise word in their own language to describe an emotion, which English doesn’t have. Also due to all of my participants being musically trained, this sampling bias further restricts the generalisability of my results.

Continuing with the study itself, whilst I used an open questionnaire which gave respondents freedom to express their feelings, reducing demand characteristics and allowing me to gain a truer insight into my points for study that a selection lost would have done, a questionnaire itself is hard to generalise to the larger population due to its sample size, even when this is ecologically valid.

In terms of how I presented the pieces, when hearing them one after the other, and in particular beginning listening to them often from a middle section, it is hard to gain a full picture of the piece as a whole, and the musical storyline and therefore emotional response may be influenced by which piece came before or due to the fast change of music; it may be difficult to shift emotional state. The participants also didn’t hear the whole piece, so weren’t able to understand the complete structure. When listening to the piece as a whole, the resultant emotions and response may therefore be different. 

The participants could only hear the music being performed on a recording, and are likely to have listened alone to the music. Therefore, other factors which contribute to experiencing emotion in a live performance, such as seeing facial expressions, perhaps physically feeling the vibrations of the sounds, sharing the experience with others were missing in this experiment, and the study questions only the musical content.

When looking at the results of valence in correlation with the music, because all of my sample group were musically educated people, there is the question also of liking due to them agreeing with the interpretation of the pieces presented, and especially in the case of Bach and Mozart, how the pieces were performed in a historically informed way. Therefore, due to the level of musical training, this may have had an influence on valence due to academic understanding. Additionally, if a piece is well known to the person, they may also like or dislike it more strongly due to having played it themselves or heard it many times before, due to associations with this.

To take this study further, I would look to increase my sample size if possible, and also incorporate music from different genres and people from a wider range of cultures within a further study. By increasing all my factors, it would be very interesting to gain an increased insight and wider perspective on how music is important to different cultural or demographic groups. It would also be interesting to speak to people and gain anecdotal evidence of their emotional response to musical features, and responses due to valence, to make the results richer.

Future Applications:

Through this study, I wanted to observe for myself if theories forwarded by researchers I have discussed here are applicable in real life, and if the harmonies, and other musical features are in fact influential in practice of evoking or bringing to the surface emotions. In knowing and understanding how harmonic structures, articulation and dynamics etc can affect people’s emotions towards a piece of music and their emotional understanding of it, this informs my personal performance practice and interpretations of pieces, such as those above. I can through this, make informed decisions about how to interpret a piece to make it more emotionally meaningful by utilising the functions which elicit emotion, thereby potentially creating a more meaningful performance for an audience. Through understanding the emotional meaning and purpose behind a composition, in collaboration with others, sharing these same ideas together as a group could create a stronger sense of connection within an ensemble.

I also wanted to test if having memory of a piece of music makes us like it more, or changes the type of emotion we feel in connection to it by associations from the past to be able to apply this to teaching, especially with teenage or adult students. If a person likes the music they are learning, they may have more motivation to study and improve. Alternatively, the results of this discussion may inform my decision in choosing pieces for a concert. If a performance has a particular target audience such as in an elderly people’s home or in a hospital, then having prior knowledge of this audiences’ preferences could help myself or others decide what kind of music, or from what era to choose music from, to provide an enjoyable and meaningful experience. 

In the following final chapter of this discussion, we can see the applications and how the usage of this theoretical knowledge can help us to be able to make these informed decisions in situations such as these, and music can be put into practice to be of great meaning and benefit to others through emotion. 

147- Woody, Burns, 2001

148- Labbé, Grandjean, 2014

149- Eschrich, 2008

150- Juslin, Västfjäll, 2008b

151- Juslin, 2008

152- Bradley, Lang, 2000

153- Scherer, 1984

154- Kotlyar, 1976

155- Juslin, 2010

156- Liebermann, 1987

157- Goffman, 1979

158- Solomon, 2004

159- Scherer, 1984

160- Juslin, 2010

161- Copland, 1971

162- Goffman, 1979

163- Scherer, 1991

164- Meyer, 1956

165- Seashore, 1967

166- Scherer, 1984

167- Scherer, 1984

168- Takemitsu, 1971