Current Organizations / Institutes

The organizations/institutes from the scholar's latest article


Cameron J. Anderson has a total of 29 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    The influence of audience presence in expert classical piano performance: Reliable reproduction with greater expressive variability

    by Emily A. Wood, Rory Kirk, Cameron J Anderson, Martin Alejandro Miguel, Laurel Trainor

    How the presence of an audience shapes skilled performance has been studied for over a century, largely through laboratory tasks testing whether being watched improves or impairs. Music offers a richer test case: a demanding skill performed for others in which how the performance changes can be measured, not only whether it improved. We compared an expert pianist’s performance of the same six-piece programme alone in a private rehearsal and a public concert, measuring her motion, beat-level timing, sound intensity, and pedalling. The two performances were remarkably similar, with the pianist reproducing her expressive interpretation with high fidelity across every feature. Further, the passages she had marked as most expressive in the score were no more affected by the audience than the rest. Despite this overall consistency, audience presence introduced small but reliable changes: greater variability in her motion, timing, and soft pedal use, and at the least intense moments in the music (as rated by listeners) she drew out the tempo and shaped the quietest passages more with the soft pedal in concert. For this performer, an audience did not impair her performance but facilitated it, supporting a faithful execution of her prepared interpretation while increasing expression through enhanced variability.

  • Open Access

    A Meta-Analysis of Music Emotion Recognition Studies

    by Tuomas Eerola, Cameron J. Anderson

    This meta-analysis examines music emotion recognition (MER) models published between 2014 and 2024, focusing on predictions of valence, arousal, and categorical emotions. A total of 553 studies were identified, of which 96 full-text articles were assessed, resulting in a final review of 34 studies. These studies reported 290 models, including 86 for emotion classification and 204 for regression. Using the best-performing model from each study, we found that valence and arousal were predicted with reasonable accuracy (r = 0.67 and r = 0.81, respectively), while classification models achieved an accuracy of 0.87 as measured with the Matthews correlation coefficient. Across modelling approaches, linear and tree-based methods generally outperformed neural networks in regression tasks, whereas neural networks and support vector machines (SVMs) showed the highest performance in classification tasks. We highlight key recommendations for future MER research, emphasizing the need for greater transparency, feature validation, and standardized reporting to improve comparability across studies.


  • Open Access

    Measuring Emotional Experiences with Music: Content Validity Assessment for Episode Model Constructs

    by Connor Kirts, Suvi Saarikallio, Cameron J. Anderson, Scott Bannister, Julian Céspedes-Guevara, Gladys J. Heng, Noah Henry, Kelly Jakubowski, Friederike Koehler, Amanda E. Krause, et al.

    This study focuses on the first stage of instrument development, content validity, and provides guidance on what steps should be taken when designing and evaluating content through the development of an instrument to reflect a recent theory concerning emotional episodes. To establish this instrument, we (1) operationalized the theory, identifying 25 sub-constructs for topics such as listening attention, meaning generation, preferences, familiarity, reward, and functions attributed to the use of music to regulate affectual states; (2) proposed a set of items (N = 495) to represent the relevant constructs operationalized here from the Episode Model; (3) analyzed and reduced the item pool with natural language processing (NLP); (4) assessed whether items were indeed reflective of their assumed construct and would garner appropriate responses using feedback supplied by subject matter experts; (5) refined the item pool based on expert feedback; (6) reassessed the revised items which resulted in a reduced item sample (N = 168). Through this collaborative content validity process, experts supported the theoretical positioning of the Episode Model through their agreement with the operationalized constructs. Expert insight shaped the implementation of theory from loosely associated items into a more tightly interrelated set of items, comprising fewer and more distinct constructs. We conclude by discussing the purpose of these content validity processes and outline the next stages of instrument development to construct a robust instrument which contextualizes emotional episodes experienced with music.


  • Open Access

    Analysis from multiple perspectives (AMP): Applying decision hygiene to analysis of musical structure

    by Max Delle Grazie, Cameron J Anderson, Jonathan De Souza, Michael Schutz

    Music analysis is a complex and subjective task requiring a considerable degree of judgment on questions often lacking verifiable answers. In many cases, this subjectivity leads to seemingly intractable disagreements. Although disagreements can offer useful insight, whether they represent genuine differences in perspective is not always clear. To contribute to this challenging aspect of music analysis, here we introduce a procedure inspired by recommendations for improving decision making in other domains lacking verifiable answers, such as judicial sentencing. Our approach involves a 3-phase procedure, combining independent analyses with information sharing and re-evaluation among five graduate-level music analysts. We show that this procedure reduces self-identified errors/oversights in music analysis while preserving meaningful differences in perspective. As a proof of concept, we apply this procedure to 381 excerpts from 16 historic sets of preludes to assess relative mode , a complex musical property alluded to in previous scholarship but never formally explored in theoretical applications. This procedure (yielding complementary qualitative and quantitative data) demonstrates how a new, group-based approach to music analysis can offer insights unavailable from more traditional single-scholar approaches.

  • Open Access

    Beyond the notes: Clarifying the role of expressivity in conveying musical emotion

    by Cameron J Anderson, Jamie Ling, Michael Schutz

    Understanding how performance expression affects perceived emotion requires separating the effects of notated music from its interpretation by performers. Previous studies suggest that compositional cues (e.g., the pitches of a melody) primarily convey valence (negative–positive emotional quality), whereas performance cues (e.g., performance timing, intensity) convey arousal (low–high emotional intensity). However, these conclusions largely follow from simple single-line stimuli that lack the complexity of real-world music. To explore compositional and performance contributions to emotion in more complex works, we conducted experiments comparing participants’ ( N = 120) valence and arousal ratings of 48 recorded excerpts from a Grammy-winning pianist against parallel deadpan versions lacking emotionally expressive aspects. By comparing differences in ratings of stimuli presented in expressive and deadpan conditions, we corroborate past findings highlighting performance contributions to perceived emotion, while also providing novel insight into the relative importance of analyzed cues. Our findings reveal that removing expressive aspects (i.e., the deadpan condition) significantly affects arousal ratings of 21 excerpts, but valence ratings of only 4. Additionally, we highlight how cues differ in importance between expressive and deadpan conditions through a novel analytical approach employing elastic nets. Our analyses shed new light on how performance expression affects emotions communicated across complex musical works with different levels of compositional cues.


  • Open Access

    Measuring Emotional Experiences to Music – Content Validity Assessment for Episode Model Constructs

    by Connor Kirts, Suvi Saarikallio, Cameron J Anderson, Scott Bannister, Julian Cespedes-Guevara, Gladys Jiamin Heng, Noah Henry, Kelly Jakubowski, Friederike Koehler, Amanda E Krause, et al.

    This study focuses on content validity procedures in music psychology. We set out to develop an instrument to measure a recent theory concerning emotional episodes (Eerola et al., 2024) related to music and their associated constructs. To establish this instrument, we: (1) operationalised the theory; (2) proposed and adapted a set of items to represent the relevant constructs; (3) analysed and reduced the item pool; (4) assessed the content validity of the items using subject matter experts; and (5) refined and reassessed the item pool based on expert feedback. In Step 1, we identified 25 sub-constructs. In Step 2, these were represented by an initial pool of 495 items, of which 290 were adapted from existing instruments. Step 3 involved eliminating 159 semantically similar items. In Step 4, twelve subject matter experts in music and emotion evaluated a sample of 366 items for relevance, providing suggestions for reclassification and identifying missing content. Expert agreement was used to determine item retention. Based on their feedback, a revised pool of 302 items underwent a second round of expert review. This process resulted in a set of 168 relevant items and 115 of the most representative items encompassing the emotional episodes, the descriptive schemes, and the contextual constructs across 23 sub-constructs. Together, relevant and representative scores indicate the quality of the content but future psychometric processes will need to test construct validity. We conclude by discussing the content validity of the constructs and frame a modular scheme that implements the underlying theoretical framework.

  • Open Access

    Breaking with common practice: Exploring modernist musical emotion

    by Max Delle Grazie, Cameron J Anderson, Michael Schutz

    Experimental research on musical emotion has identified clear links between specific aspects of musical structure and emotional responses. However, growing recognition of changes in the affective meaning of specific cues over time raises intriguing questions about the degree to which these links hold across historical eras. In particular, the traditional focus on compositional principles from common-practice period music (ca. 1600–1900) might not capture how emotion is perceived in later compositions. Here we explore perceived emotion ratings in a set of 24 preludes by Dmitri Shostakovich (Op. 34), comparing the effects of cues in his preludes vs. those by Bach and Chopin. We find that prosodic cues (i.e., pitch height, timing) play a stronger role than mode in these pieces. Because music theorists widely recognize Shostakovich’s music as tonal, this result reflects not his abandonment of mode, but rather his decision to use it differently than his predecessors. This provides an important perspective complementing a growing body of research using score-based analyses to explore historical changes in the “meaning” of specific cues. Our findings illustrate how modern compositions can provide novel insight into cues’ historically changing roles in emotional communication.


  • Open Access

    A Meta-Analysis of Music Emotion Recognition Studies

    by Tuomas Eerola, Cameron J Anderson

    This meta-analysis examines music emotion recognition (MER) models published between 2014 and 2024, focusing on predictions of valence, arousal, and categorical emotions. A total of 553 studies were identified, of which 96 full-text articles were assessed, resulting in a final review of 34 studies. These studies reported 204 models, including 86 for emotion classification and 204 for regression. Using the best-performing model from each study, we found that valence and arousal were predicted with reasonable accuracy (r = 0.67 and r = 0.81, respectively), while classification models achieved an accuracy of 0.87 as measured with Matthews correlation coefficient. Across modelling approaches, linear and tree-based methods generally outperformed neural networks in regression tasks, whereas neural networks and support vector machines (SVMs) showed highest performance in classification tasks. We highlight key recommendations for future MER research, emphasizing the need for greater transparency, feature validation, and standardized reporting to improve comparability across studies.


  • Open Access

    Understanding Feature Importance in Musical Works: Unpacking Predictive Contributions to Cluster Analyses

    by Cameron J. Anderson, Michael Schutz

    Cluster analysis provides insight into musical patterns in composition, performance, and perception. Despite its wide adoption in music research, understanding how specific features affect clustering solutions remains challenging. For example, features such as mode (i.e., major/minor), timing, signal amplitude, and pitch are often intercorrelated, making it difficult to understand their specific role within different clusters. To demonstrate how accumulated local effects (ALEs) can help with this challenge, here we analyze 48 excerpts from complete sets of preludes by Bach and Chopin, showing how specific features contribute to two- and three-cluster analyses. These exploratory analyses reveal that ALEs can identify salient or subtle data patterns from cluster analyses by tracking how changes in features affect cluster membership. We explore these insights in visualizations quantifying feature importance and an interactive companion application ( https://maplelab.net/feature-importance/ ) featuring the analyzed audio. Following a demonstration of this method, we suggest how it can be applied to explore topics of interest to researchers in music information retrieval, empirical musicology, and music cognition alike.


  • Open Access

    Exploring historic changes in musical communication: Deconstructing emotional cues in preludes by Bach and Chopin

    by Cameron J. Anderson, Michael Schutz

    A growing body of research analyzing musical scores suggests mode’s relationship with other expressive cues has changed over time. However, to the best of our knowledge, the perceptual implications of these changes have not been formally assessed. Here, we explore how compositional choices of 17th- and 19th-century composers (J. S. Bach and F. Chopin, respectively) differentially affect emotional communication. This novel exploration builds on our team’s previous techniques using commonality analysis to decompose intercorrelated cues in unaltered excerpts of influential compositions. In doing so, we offer an important naturalistic complement to traditional experimental work—often involving tightly controlled stimuli constructed to avoid the intercorrelations inherent to naturalistic music. Our data indicate intriguing changes in cues’ effects between Bach and Chopin, consistent with score-based research suggesting mode’s “meaning” changed across historical eras. For example, mode’s unique effect accounts for the most variance in valence ratings of Chopin’s preludes, whereas its shared use with attack rate plays a more prominent role in Bach’s. We discuss the implications of these findings as part of our field’s ongoing effort to understand the complexity of musical communication—addressing issues only visible when moving beyond stimuli created for scientific, rather than artistic, goals.