KJKelly Jakubowski

Kelly Jakubowski

*This is an algorithmically generated author record based on publicly available information.

Current Organizations / Institutes

The organizations/institutes from the scholar's latest article

Previous Organizations / Institutes

The organizations / institutes from the scholar's other articles in the past ten years


Kelly Jakubowski has a total of 313 co-authors. Here are the top co-authors by number of shared publications:


Latest Publications

  • Open Access

    Music illuminates sociocultural dimensions of autobiographical memory

    by Cara Turnbull, Benjamin M. Kubit, Kelly Jakubowski, Elizabeth H. Margulis

    Researchers have long theorized that autobiographical memory relies on sociocultural structures, such as life scripts and event schemas. The potency of music as a marker of social and personal identity, combined with its ubiquity in everyday life, allows us to connect these everyday scripts to larger patterns of socially and culturally shaped behavior. We collected free-response descriptions of autobiographical memories evoked by short clips of familiar songs from university undergraduates (n = 233). The songs spanned three popular genres and aligned with two significant periods of listeners’ lives. We analyzed the similarities and differences in the content of memories for particular songs using natural language processing. Memories cued by songs from the same category displayed striking similarities, suggesting that particular types of music cue memories that are similar between listeners but that vary according to characteristics such as music genre and release date. Two control conditions, in which a new set of participants reported memories evoked by key words from the lyrics (n = 121) and short narrative summaries of the lyrics (n = 129), indicate that these results cannot be explained by within-category similarities in the songs’ lyrical content. Instead, we suggest that the intersubjective structure of music-evoked autobiographical memories reflects socioculturally-defined scripts tied to song characteristics such as genre and release date. Thus, while memories may feel idiosyncratic and personal, there are also substantial similarities in life experiences between individuals with broadly similar backgrounds, and these shared experiences can be spontaneously called to mind through the seemingly innocuous act of turning on the radio.

  • Open Access

    Sampling Thoughts During Music Listening: A Systematic Comparison of Probe-caught, Self-caught and Retrospective Thought Reporting Methods

    by Nao Toyama, Kelly Jakubowski

    Music listening experiences dynamically unfold over time. However, as previous research has relied predominantly on short music excerpts and retrospective post-listening self-reports, how qualitative thought content unfolds during extended listening remains underexplored. Crucially, addressing this requires effective real-time sampling, yet direct empirical comparisons of such methods––i.e., experimenter-led versus participant-led approaches––and their impact on data quality and listener experience are remain lacking. To address these gaps, the present study systematically compared three experience-sampling methodologies (probe-caught, self-caught, and a retrospective control) using participants’ own smartphones during an 18-minute video of a classical music performance. While the control group yielded significantly fewer thoughts, both experimental groups (probe- and self-caught) reported a comparable total number of thoughts. The probe-caught method exhibited greater inter-participant stability and generated significantly more detailed qualitative descriptions. Importantly, participants’ subjective perceived interference on their listening experience was low and did not differ across groups, confirming that real-time thought reporting did not disrupt the natural listening experience compared with traditional retrospective reporting. By validating the efficiency, stability, and minimal invasiveness of the probe-caught method, this study provides an empirical foundation for future research investigating real-time subjective experiences across diverse listening environments.

  • Open Access

    Music influences the content of spontaneous thought

    by Wei Wu, Elizabeth H. Margulis, Kelly Jakubowski

  • Open Access

    Music influences the content of spontaneous thought

    by Wei Wu, Elizabeth Hellmuth Margulis, Kelly Jakubowski

    Spontaneous thought is a form of cognition that is relatively unconstrained and free flowing. Recent accounts have raised the possibility that the generation and content of spontaneous thought can be influenced by even weak perceptual constraints. We leveraged this perspective to develop an approach using music, a ubiquitous stimulus that permeates our everyday environments, as a window into the mechanisms that shape spontaneous thought content. Adults spanning four age groups listened to music from ten different genres and completed the open-ended task of writing descriptions of any thoughts that arose. Natural language processing analysis was applied to obtain computational similarity (word2vec) of the semantic meaning of thought descriptions. We found that thoughts sustained while hearing music from the same genre shared greater semantic similarity than those sustained while hearing music from different genres, and participants from the same age group experienced more similar thoughts than those from different age groups. Properties that varied naturally with genre, including the music’s auditory features, expressed emotions, and contextual associations, all contributed significantly to explaining the greater semantic similarity of thoughts within genres. These results demonstrate that spontaneous thought can be jointly shaped by perceptual and experiential constraints, opening new avenues for understanding when and why particular thoughts come to mind, with implications for creativity and mental health.

  • Open Access

    Basslines, Bodies, and Beehive Haircuts: What do we think about when we listen to music?

    by Hazel A. van der Walle, Elizabeth H. Margulis, Kelly Jakubowski

    Music influences diverse mental content—fictional narratives, autobiographical recall, visual imagery, multimodal imaginings—with wide-reaching implications for everyday cognition, creativity, and wellbeing. Prior studies often examine thought types in isolation, limit musical diversity, or reduce experiential complexity through quantitative ratings and automated text-analytic methods such as natural language processing, flattening contextual richness. To address this, we introduce Music-Influenced Mental Content (MIMC) as a unified conceptual framework capturing the full heterogeneity of mental contents shaped by music. We analysed free-response MIMC descriptions from 1,786 UK and US participants (aged 18–75), each hearing 10 of 120 instrumental clips spanning 10 genres (Ambient, Classical, Electronic, Film, Funk, Jazz, Metal, Pop, 60s-pop, 80s-pop). Mixed-method qualitative and network analyses of 14,091 MIMC descriptions reveal that, rather than a collection of isolated content types, MIMC is best understood as an interconnected system. Organised across three themes—Narrational, Referential, and Experiential—this study provides one empirically grounded instantiation of the MIMC framework within a Western listener sample and stimulus set.

  • Open Access

    Mapping the Abstraction of Song Representations from Perception to Autobiographical Memory

    by Riesa Cassano-Coleman, Kelly Jakubowski, Elise A. Piazza

    Humans abstract away from sensory features to form invariant representations of external stimuli, but relatively little is understood about this process in naturalistic audition. Popular music is a common and complex auditory stimulus that is often intertwined with long-term memory and life narratives, offering a unique opportunity to study the mapping from perception to higher-order cognition. We took advantage of a rich, naturally occurring comparison–popular songs versus cover versions, which share a core “song identity” but vary in low-level features–to test the influence of auditory and emotional features on holistic song perception and music-evoked autobiographical memory. In Study 1, we found that a range of auditory features (most critically, tempo and acoustic modulation energy) influenced holistic perceptual representations of songs, as measured by similarity judgments. In Study 2, we found that different levels of song processing drove different aspects of music-evoked autobiographical memories: perceptual features influenced the likelihood of a memory's retrieval, whereas reinstatement of memory details (i.e., vividness) was more strongly linked to core song identity. Taken together, these results provide a new understanding of how complex features of real-world sounds are hierarchically integrated from perception to long-term autobiographical memory.

  • Open Access

    Framed Listening: Can accompanying contextual cues shape Music-Influenced Mental Content?

    by Hazel Aileen van der Walle, Elizabeth Hellmuth Margulis, Kelly Jakubowski

    When we listen to music, what we think about is rarely shaped by sound alone. This study tested whether contextual framing cues—phrases locating music within a familiar social or cultural situation—systematically align listeners’ music-influenced mental content (MIMC). Using a between-participants design, 160 participants heard 16 unfamiliar instrumental clips (Electronic, Jazz, Metal, and Eighties-pop) each accompanied by one of four context cues (bar, concert, movie, video game). Semantic and lexical similarity of participants’ MIMCs described after each clip were quantified using complementary natural language processing pipelines: sentence-embedding model, MiniLM (semantic; 1,919,820 individual-level pairwise comparisons), and TF-IDF (lexical; 2,016 aggregated document-level comparisons). Listeners who received the same context cue produced more semantically similar MIMC than those given both a different clip and context cue—a finding replicated at the lexical level and confirmed in linear mixed models controlling for genre, clip, and individual response tendencies. Music remained the stronger driver of convergence: MIMC were significantly more similar with same-clip pairs than same-context pairs. Both factors contributed additively, with combined clip-and-context alignment producing the greatest convergence overall. Genre moderated this balance: Metal consistently showed the greatest context framing susceptibility across both pipelines and was the only genre where contextual framing exceeded music as a driver of within-genre similarity; Jazz showed the smallest framing susceptibility. Believability of the clip–context pairing significantly predicted lexical (TF-IDF) but not semantic (MiniLM) similarity, suggesting framing operated through associative priming rather than conscious plausibility judgement. These findings provide an empirical test of source sensitivity in MIMC, demonstrating that even newly introduced contextual frames are sufficient to meaningfully constrain the distributed subjectivity of MIMC in music listening experiences.

  • Open Access

    Audio features impact the semantic content of music-evoked thoughts

    by Wei Wu, Tuomas Eerola, Elizabeth Hellmuth Margulis, Kelly Jakubowski

    Music permeates our everyday environments and can elicit a wide range of thoughts for the listener. To understand the mechanisms through which music steers ongoing thoughts, previous studies have investigated the effects of different music genres or excerpts on the content of music-evoked thoughts. However, 'genre' is a very broad and potentially ambiguous category (e.g., “pop music”) and 'excerpt' is too narrow to allow generalization of results. The current study explored how music shapes thoughts with a mid-level approach based on features derived from the audio. We collected open-text descriptions of music-evoked thoughts from approximately 1800 participants and extracted 27 audio features across seven acoustic domains from our music stimuli (120 instrumental excerpts from a range of genres). By applying natural language processing methods to the open-text data, we measured the general semantic similarity for the thoughts (via word2vec) and decomposed thought contents into 16 semantic topics (via Latent Dirichlet Allocation). Then we used regression models to map the audio profile of the music to the semantic profile of the thoughts. We found that the audio features effectively explained both variance in the general structure of the semantic similarity space and in the topic-specific contents of the thoughts. Our results demonstrate that auditory features serve as a window into how music impacts ongoing thoughts, providing a key method for future targeted investigations of how music can be used in everyday and clinical settings to shape listeners’ mental experience.

  • Open Access

    Exploring the Influence of Music’s Semantic Associations on Autobiographical Memories

    by Fruzsina Lilla Szücs, Kelly Jakubowski, Tuomas Eerola

    While the effect of music on autobiographical memories has been examined in the past, it has not yet been studied how semantic associations with music cues influence the content and phenomenology of autobiographical memories. In a repeated-measures design experiment with 53 participants, we tested how 14 musical cues depicting semantic associations, in either a more concrete (e.g., “birds”) or abstract (e.g., “French café ambience”) manner, influenced the recall of autobiographical memories and the phenomenological characteristics of these. Some music excerpts evoked a high proportion of memories containing our expected semantic association (e.g. “Loss” and “Latin-American ambience”). Others were less successful in evoking the expected semantic association, but nevertheless evoked memories with a high degree of shared content (e.g., “Calmness”). There were no statistically significant differences found in the phenomenological characteristics of the memories evoked by more concrete versus abstract music cues. These results indicate that music can convey shared semantic meaning across listeners that shapes the content of associated memories, although this capacity varies substantially across different musical cues. Findings from this study have implications for how music may be used as a memory cue in comparison to other cue types in both clinical and everyday settings.

  • Open Access

    Basslines, Bodies, and Beehive Haircuts: What do we think about when we listen to music?

    by Hazel Aileen van der Walle, Elizabeth Hellmuth Margulis, Kelly Jakubowski

    Music influences distinct and specific mental content, including fictional narratives, autobiographical recall, and visual mental imagery. These imaginative responses are relevant in specialised contexts like therapy and play a vital role in everyday cognition, creativity, memory, attention, and well‑being. However, prior studies are typically constrained to one isolated thought type, use limited musical diversity, or interpret intricate imaginings through quantitative ratings—flattening the complexity of these experiences. Computational qualitative analyses can process large datasets, but fail to capture contextual richness and nuance available through human-driven qualitative analysis. What is missing is an open, manual qualitative examination of the fuller breadth of music-influenced mental content (MIMC)—an approach capable of preserving the depth and human subtleties that quantitative, predefined, isolated, and computational approaches overlook. We analysed participants’ free‑response MIMC descriptions, not restricting probing to isolated types or predefined thought-type categories. Music stimuli—120 instrumental clips from the MUSIFEAST-17 stimulus set—spanned 10 stylistically diverse genres: Ambient, Classical, Electronic, Film, Funk, Jazz, Metal, Pop, 60s-pop, 80s-pop. A total of 1,786 UK and US participants aged 18–75 years listened to 10 clips (one from each genre), reported whether they experienced imaginative responses, and provided open-text descriptions. Rigorous, mixed-method qualitative analysis was conducted on 14,091 MIMC descriptions, involving multiple coding passes, multi-coder reliability checks, thematic analysis, and co-occurrence network analysis. The resulting thematic framework functions as a typology of music-influenced “thoughtscapes”, revealing multi-faceted MIMC unfolding across narrational, referential, and experiential dimensions, offering a structured account of a fuller spectrum of listeners’ imaginative experiences.