Dimensional Emotion Model Research Articles

Emotion models for textual emotion classification

This paper deals with textual emotion classification which gained attention in recent years. Emotion classification is used in user experience, product evaluation, national security, and tutoring applications. It attempts to detect the emotional content in the input text and based on different approaches establish what kind of emotional content is present, if any. Textual emotion classification is the most difficult to handle, since it relies mainly on linguistic resources and it introduces many challenges to assignment of text to emotion represented by a proper model. A crucial part of each emotion detector is emotion model. Focus of this paper is to introduce emotion models used for classification. Categorical and dimensional models of emotion are explained and some more advanced approaches are mentioned.

Angry Versus Furious: A Comparison Between Valence and Arousal in Dimensional Models of Emotions

ABSTRACTThe present study compared the roles of valence and arousal, proposed by the dimensional models of emotions, in recognizing emotional expressions. It was hypothesized that the dimension of valence, due to its functional significance, would be more salient in the recognition of emotional expressions than the dimension of arousal would be. The results of the current study supported this hypothesis. The participants in all age groups were more accurate and quicker in recognizing an emotion when the expression was paired up with another emotional expression that was different in the polarity of the valence dimension than with this similar polarity difference in the arousal dimension. The insignificant difference in recognizing the positive and negative emotional expressions in the group of elders also rejected the Socio-Emotional Selectivity Theory.

Tuned In Emotion Regulation Program Using Music Listening: Effectiveness for Adolescents in Educational Settings.

This paper presents an effectiveness study of Tuned In, a novel emotion regulation intervention that uses participant selected music to evoke emotions in session and teaches participants emotional awareness and regulation skills. The group program content is informed by a two dimensional model of emotion (arousal, valence), along with music psychology theories about how music evokes emotional responses. The program has been evaluated in two samples of adolescents: 41 “at risk” adolescents (76% males; Mage = 14.8 years) attending an educational re-engagement program and 216 students (100% females; Mage = 13.6 years) attending a mainstream secondary school. Results showed significant pre- to post-program improvements in measures of emotion awareness, identification, and regulation (p < 0.01 to p = 0.06 in the smaller “at risk” sample and all p < 0.001 in the mainstream school sample). Participant ratings of engagement and likelihood of using the strategies learned in the program were high. Tuned In shows promise as a brief emotion regulation intervention for adolescents, and these findings extend an earlier study with young adults. Tuned In is a-theoretical in regard to psychotherapeutic approach and could be integrated with other program components as required.

Perceived and Induced Emotion Responses to Popular Music

Music both conveys and evokes emotions, and although both phenomena are widely studied, the difference between them is often neglected. The purpose of this study is to examine the difference between perceived and induced emotion for Western popular music using both categorical and dimensional models of emotion, and to examine the influence of individual listener differences on their emotion judgment. A total of 80 musical excerpts were randomly selected from an established dataset of 2,904 popular songs tagged with one of the four words “happy,” “sad,” “angry,” or “relaxed” on the Last.FM web site. Participants listened to the excerpts and rated perceived and induced emotion on the categorical model and dimensional model, and the reliability of emotion tags was evaluated according to participants’ agreement with corresponding labels. In addition, the Goldsmiths Musical Sophistication Index (Gold-MSI) was used to assess participants’ musical expertise and engagement. As expected, regardless of the emotion model used, music evokes emotions similar to the emotional quality perceived in music. Moreover, emotion tags predict music emotion judgments. However, age, gender and three factors from Gold-MSI, importance, emotion, and music training were found not to predict listeners’ responses, nor the agreement with tags.

Expressivity, affective nuance, and presentation modality in a performer-to-audience communication.

In a music performance, both performers and audience experience emotions (Juslin, 2000). Performers attempt to communicate nuances of a piece in their own artistic performance: [I] communicate an emotion to audience; [I] transform emotions and moods into tones (Lindstrom, Juslin, Bresin, & Williamon, 2003, p. 30). Such emotions experienced by performers (and subsequently by audience) can be represented by dimensional (e.g., arousal, valence, Schubert, 2001) or discrete (e.g., happiness, sadness, fear, anger; Dahl & Friberg, 2007; Gabrielsson & Juslin, 1996; Ohgushi & Hattori, 1996; Juslin, 1997) models of emotion (Eerola & Vuoskoski, 2011). The projection of emotions in an artistic performance, however, derives from a performer's deliberate interpretation of a piece that often goes beyond dimensions of arousal and valence or limited number of emotion words. For example, pianist Kazuko Yasukawa has described Debussy's Jardins sous la pluie (Gardens in rain) as if it expressed the melancholy of children playing in rain as well as tranquil memories of happiness (Yasukawa, 1960, p. 6, translation and emphases by present authors). A critic also expresses characteristics of a piece with rich vocabularies: James Huneker described F. Chopin's Ballade No. Four in f minor, It is in his most reflective, yet lyric mood. Lyricism is keynote of work, a passionate lyricism, with a note of self-absorption, suppressed feeling-truly Slavic, this shyness! -and a concentration that is remarkable even for Chopin (Huneker, 1918, p. 286, emphases by present authors). These multifaceted emotions underlying a music performance have been studied extensively from listener's perspectives (Asmus, 1985; Hevner, 1936; Taniguchi, 1995), but it is still unclear whether affective nuances perceived by listeners are actually concordant with those interpreted by a performer. The main goal of present study was to explore a performer-to-audience communication of affective nuances of a piece by identifying how audience would perceive complex nuances of a piano performance through its acoustical and visual recordings.In Shoda and Adachi (2012), a pianist evaluated affective nuances of structurally contrasting pieces by S. Rachmaninoff- Tableaux Op. 39-1 in c minor (Etude) and Op. 32-5 in G major (Prelude)-by means of a set of 24 adjectives (see Figure 1), which was selected from Hevner (1936)'s adjective circle. For Etude (Allegro Agitato in 4/4 time), which consists of continuously moving figurations with contrasting block chords (Glover, 2003), pianist gave high ratings (i.e., 8 or 9 on 9-point scale) for emphatic, majestic, passionate, robust, and solemn. For Prelude (Moderato in 4/4 time), structurally much simpler than Etude and characterized by smooth melodic line and its accompaniment (Glover, 2003), adjectives such as bright, dreamy, graceful, joyous, longing, tender, tranquil, and passionate were rated high. Such interpretations were given only for artistic rendition of each piece (Shoda & Adachi, 2012), indicating that ideal level of expressivity1 (i.e., artistic rendition) elicited rich impressions in pianist but neither minimal (i.e., deadpan rendition) nor excessive level of expressivity (i.e., exaggerated rendition) did so.As for audience, previous studies focused primarily on affective nuances of musical performances represented as single target adjectives (Senju & Ohgushi, 1987), and to date their perception of more complex nuances through a combination of many adjectives has rarely been studied from a viewpoint of performer-to-audience communication. Nonetheless, basic emotions (e.g., happiness, sadness, anger, fear) can be portrayed by guitarists (Gabrielsson & Juslin, 1996; Juslin, 1997), violinists (Gabrielsson & Juslin, 1996), flutists (Gabrielsson & Juslin, 1996), percussionists (Laukka & Gabrielsson, 2000), and even children (Adachi & Trehub, 1998; Yamasaki, 2006), who manipulate tempo, timing, dynamics, and articulation of a simple melody or rhythm that otherwise conveys neutral emotion (Juslin, 2001, see Figure 14. …

High-Order Markov Random Fields and Their Applications in Cross-Language Speech Recognition

Abstract In this paper we study the cross-language speech emotion recognition using high-order Markov random fields, especially the application in Vietnamese speech emotion recognition. First, we extract the basic speech features including pitch frequency, formant frequency and short-term intensity. Based on the low level descriptor we further construct the statistic features including maximum, minimum, mean and standard deviation. Second, we adopt the high-order Markov random fields (MRF) to optimize the cross-language speech emotion model. The dimensional restrictions may be modeled by MRF. Third, based on the Vietnamese and Chinese database we analyze the efficiency of our emotion recognition system. We adopt the dimensional emotion model (arousal-valence) to verify the efficiency of MRF configuration method. The experimental results show that the high-order Markov random fields can improve the dimensional emotion recognition in the cross-language experiments, and the configuration method shows promising robustness over different languages.

Multivariate neural biomarkers of emotional states are categorically distinct.

Understanding how emotions are represented neurally is a central aim of affective neuroscience. Despite decades of neuroimaging efforts addressing this question, it remains unclear whether emotions are represented as distinct entities, as predicted by categorical theories, or are constructed from a smaller set of underlying factors, as predicted by dimensional accounts. Here, we capitalize on multivariate statistical approaches and computational modeling to directly evaluate these theoretical perspectives. We elicited discrete emotional states using music and films during functional magnetic resonance imaging scanning. Distinct patterns of neural activation predicted the emotion category of stimuli and tracked subjective experience. Bayesian model comparison revealed that combining dimensional and categorical models of emotion best characterized the information content of activation patterns. Surprisingly, categorical and dimensional aspects of emotion experience captured unique and opposing sources of neural information. These results indicate that diverse emotional states are poorly differentiated by simple models of valence and arousal, and that activity within separable neural systems can be mapped to unique emotion categories.

Where are emotions in words? Functional localization of valence effects in visual word recognition.

Where are emotions in words? Functional localization of valence effects in visual word recognition.

The neural correlates of the dominance dimension of emotion

Emotion has been conceptualized as a dimensional construct, while the number of dimensions – two or three – has been debated. Research has consistently identified two dimensions – valence and arousal – though ample evidence exists that three dimensions are necessary to describe emotion. One proposed third dimension, identified as dominance, is relevant in clinical syndromes, personality and consumer psychology. Dominance refers to an individual's sense of having an ability to affect the environment. Neuroimaging studies have generally focused on the two dimensions of valence and arousal, leaving the neural correlates of dominance unexplored. The current study used functional magnetic resonance imaging to explore the neural basis of dominance in 17 healthy male controls. Participants viewed images from the International Affective Picture System that were selected to represent high and low dominance conditions. Results indicated activation in paralimbic regions, including the bilateral anterior insula for high dominance and the right precuneus for low. The findings of this exploratory study support the consideration of dominance in dimensional models of emotion and suggest that further research is needed to understand the neural representation of dominance in emotional experience.

EEG Emotion Recognition Based on the Dimensional Models of Emotions

In this paper, we propose a method for EEG emotion recognition which is tested based on 2 dimensional models of emotions, (1) the rSASM, and (2) the 12-PAC model. EEG data were collected from 5 preschoolers aged 5 years old while watching emotional faces from the Radboud Faces Database (RafD). Features were extracted using KSDE and MFCC and classified using MLP. Results show that EEG emotion recognition using the 12-PAC model gives the highest accuracy for both feature extraction methods. Results indicated that the accuracy of EEG emotion recognition is increased with the precision of the dimensional models.

A Review of Music and Emotion Studies: Approaches, Emotion Models, and Stimuli

The field of music and emotion research has grown rapidly and diversified during the last decade. This has led to a certain degree of confusion and inconsistency between competing notions of emotions, data, and results. The present review of 251 studies describes the focus of prevalent research approaches, methods, and models of emotion, and documents the types of musical stimuli used over the past twenty years. Although self-report approaches to emotions are the most common way of dealing with music and emotions, using multiple approaches is becoming increasingly popular. A large majority (70%) of the studies employed variants of the discrete or the dimensional emotion models. A large proportion of stimuli rely on a relatively modest amount of familiar classical examples. The evident shortcomings of these prevalent patterns in music and emotion studies are highlighted, and concrete plans of action for future studies are suggested.

Classifier-based learning of nonlinear feature manifold for visualization of emotional speech prosody

Visualization of emotional speech data is an important tool for speech researchers who seek means to gain a deeper insight into the structure of complex multidimensional data. A visualization method is presented that utilizes feature selection and classifier optimization for learning Isomap manifolds of emotional speech data. The resulting manifold is based on those features that best discriminate between given emotional classes in the target space of specified embedding dimension. A nonlinear mapping function based on generalized regression neural networks (GRNNs) provides generalization for new data. A low-dimensional manifold of emotional speech data consisting of neutral, sad, angry, and happy expressions was constructed using prosodic and acoustic features of speech. Experimental results indicate that a 3D embedding provides the best classification performance. The manifold structure can be readily visualized and matches the circumplex and conical shapes predicted by dimensional models of emotion. Listening tests show excellent correlation between the organization of the data on the manifold and the listeners' judgment of emotional intensity.

Augmenting Weak Semantic Cognitive Maps with an “Abstractness” Dimension

The emergent consensus on dimensional models of sentiment, appraisal, emotions, and values is on the semantics of the principal dimensions, typically interpreted as valence, arousal, and dominance. The notion of weak semantic maps was introduced recently as distribution of representations in abstract spaces that are not derived from human judgments, psychometrics, or any other a priori information about their semantics. Instead, they are defined entirely by binary semantic relations among representations, such as synonymy and antonymy. An interesting question concerns the ability of the antonymy-based semantic maps to capture all “universal” semantic dimensions. The present work shows that those narrow weak semantic maps are not complete in this sense and can be augmented with other semantic relations. Specifically, including hyponym-hypernym relations yields a new semantic dimension of the map labeled here “abstractness” (or ontological generality) that is not reducible to any dimensions represented by antonym pairs or to traditional affective space dimensions. It is expected that including other semantic relations (e.g., meronymy/holonymy) will also result in the addition of new semantic dimensions to the map. These findings have broad implications for automated quantitative evaluation of the meaning of text and may shed light on the nature of human subjective experience.

Giving the Expectancy‐Value Model a Heart

ABSTRACTOver the past decade, research in consumer behavior has debated the role of emotion in consumer decision making intensively but has offered few attempts to integrate emotion‐related findings with established theoretical frameworks. This manuscript augments the classical expectancy‐value model of attitude with a dimensional model of emotion. An experiment involving 308 college students who face actual purchase decisions shows that predictions of attitudes, behavioral intentions, and actual behavior can be improved through the use of the augmented model for both hedonic and utilitarian products. The augmented model has theoretical implications for marketing scholars as well as practical uses for marketers.

Measuring Music-Induced Emotion: A Comparison of Emotion Models, Personality Biases, and Intensity of Experiences

Most previous studies investigating music-induced emotions have applied emotion models developed in other fields to the domain of music. The aim of this study was to compare the applicability of music-specific and general emotion models – namely the Geneva Emotional Music Scale (GEMS), and the discrete and dimensional emotion models – in the assessment of music-induced emotions. A related aim was to explore the role of individual difference variables (such as personality and mood) in music-induced emotions, and to discover whether some emotion models reflect these individual differences more strongly than others. One hundred and forty-eight participants listened to 16 film music excerpts and rated the emotional responses evoked by the music excerpts. Intraclass correlations and Cronbach alphas revealed that the overall consistency of ratings was the highest in the case of the dimensional model. The dimensional model also outperformed the other two models in the discrimination of music excerpts, and principal component analysis revealed that 89.9% of the variance in the mean ratings of all the scales (in all three models) was accounted for by two principal components that could be labelled as valence and arousal. Personality-related differences were the most pronounced in the case of the discrete emotion model. Personality, mood, and the emotion model used were also associated with the intensity of experienced emotions. Implications for future music and emotion studies are raised concerning the selection of an appropriate emotion model when measuring music-induced emotions.

Measuring music-induced emotion

Most previous studies investigating music-induced emotions have applied emotion models developed in other fields to the domain of music. The aim of this study was to compare the applicability of music-specific and general emotion models – namely the Geneva Emotional Music Scale (GEMS), and the discrete and dimensional emotion models – in the assessment of music-induced emotions. A related aim was to explore the role of individual difference variables (such as personality and mood) in music-induced emotions, and to discover whether some emotion models reflect these individual differences more strongly than others. One hundred and forty-eight participants listened to 16 film music excerpts and rated the emotional responses evoked by the music excerpts. Intraclass correlations and Cronbach alphas revealed that the overall consistency of ratings was the highest in the case of the dimensional model. The dimensional model also outperformed the other two models in the discrimination of music excerpts, and principal component analysis revealed that 89.9% of the variance in the mean ratings of all the scales (in all three models) was accounted for by two principal components that could be labelled as valence and arousal. Personality-related differences were the most pronounced in the case of the discrete emotion model. Personality, mood, and the emotion model used were also associated with the intensity of experienced emotions. Implications for future music and emotion studies are raised concerning the selection of an appropriate emotion model when measuring music-induced emotions.

A Study on Feature Analysis and Recognition of Practical Speech Emotion

Practical speech emotions as impatience and happiness are studied especially for evaluation of emotional well-being in real world applications. Induced natural speech emotion data is collected with a computer game, 74 emotion features are extracted, prosody features and voice quality features are analyzed according to dimensional emotion model, evaluation and selection of acoustic features are carried out for practical emotions in this paper, a method of practical speech emotion classification with rejection decision is proposed for real world occasions. The experiment results show, the speech features analyzed in this paper are suitable for classification of practical speech emotions like impatience and happiness, average recognition rate is above 75%, and the method of emotion classification with rejection decision is necessary for the proper recognition decision of ambiguous or unknown emotion samples, especially for the real world challenges.

A 12-point circumplex structure of core affect.

Core Affect is a state accessible to consciousness as a single simple feeling (feeling good or bad, energized or enervated) that can vary from moment to moment and that is the heart of, but not the whole of, mood and emotion. In four correlational studies (Ns = 535, 190, 234, 395), a 12-Point Affect Circumplex (12-PAC) model of Core Affect was developed that is finer grained than previously available and that integrates major dimensional models of mood and emotion. Self-report scales in three response formats were cross-validated for Core Affect felt during current and remembered moments. A technique that places any external variable into the 12-PAC showed that 29 of 38 personality scales and 30 of 30 mood scales are significantly related to Core Affect, but not in a way that revealed its basic dimensions.

A comparison of the discrete and dimensional models of emotion in music

The primary aim of the present study was to systematically compare perceived emotions in music using two different theoretical frameworks: the discrete emotion model, and the dimensional model of affect. A secondary aim was to introduce a new, improved set of stimuli for the study of music-mediated emotions. A large pilot study established a set of 110 film music excerpts, half were moderately and highly representative examples of five discrete emotions (anger, fear, sadness, happiness and tenderness), and the other half moderate and high examples of the six extremes of three bipolar dimensions (valence, energy arousal and tension arousal). These excerpts were rated in a listening experiment by 116 non-musicians. All target emotions of highly representative examples in both conceptual sets were discriminated by self-ratings. Linear mapping techniques between the discrete and dimensional models revealed a high correspondence along two central dimensions that can be labelled as valence and arousal, and the three dimensions could be reduced to two without significantly reducing the goodness of fit. The major difference between the discrete and categorical models concerned the poorer resolution of the discrete model in characterizing emotionally ambiguous examples. The study offers systematically structured and rich stimulus material for exploring emotional processing.

Personalized emotion space for video affective content representation

A personalized emotion space is proposed to bridge the “affective gap” in video affective content understanding. In order to unify the discrete and dimensional emotion model, fuzzy C-mean (FCM) clustering algorithm is adopted to divide the emotion space. Gaussian mixture model (GMM) is used to determine the membership functions of typical affective subspaces. At every step of modeling the space, the inputs rely completely on the affective experiences recorded by the audiences. The advantages of the improved V-A (Velance-Arousal) emotion model are the personalization, the ability to define typical affective state areas in the V-A emotion space, and the convenience to explicitly express the intensity of each affective state. The experimental results validate the model and show it can be used as a personalized emotion space for video affective content representation.