Four soundtracks promise very different experiences. But when concentration is measured—not marketed—what does a faster response actually mean?

This visual essay re-examines trial-level data from an online flanker-task experiment. Participants completed attention trials while listening to Work Flow, Deep Focus, Popular Hits or Office Noise. The aim is not to declare a winning playlist. It is to test how the apparent advantage changes when we look at individual participants, practice over time, mistakes and task difficulty.

Explore: Hover over points for details, click legend items to isolate groups, and use zoom and menu controls where provided. All five interactive graphics were created in R using plotly.

In this dataset: 196 participants; 14,112 trials; 4 listening conditions. Values below are exploratory descriptions, not causal-effect estimates.

01 / The first impression

The fastest-looking soundtrack is not a universal result

The distribution below is based on one median per participant, not thousands of repeated trials treated as independent people. Reaction times use only correctly answered go trials (congruent, incongruent, neutral), expressed in milliseconds. Lower is faster. Differences in distributions do not, by themselves, establish statistical significance.

Source: Orpella et al. (2025), PLOS ONE, Supporting Information S4 (flanker task). Each mark represents one participant.

Reading the chart: Look for overlap between participant distributions, not just the distance between their central lines. A group average does not describe every listener.

02 / The twist

Is the difference stable—or learned during the task?

Attention tasks involve practice, anticipation and fatigue. To inspect this pattern, each participant’s correct-go trials are divided into five progression bands, based on trial number. Each line plots the median of participant-level band medians: individuals receive similar weight, rather than allowing faster responders to dominate through more valid trials.

Source: Orpella et al. (2025), PLOS ONE, Supporting Information S4 (flanker task). Descriptive progression summaries, not within-person model estimates.

Look for movement: The question is whether groups move together over the task, or whether one soundtrack’s trajectory differs. Hover each phase to check its participant count.

03 / The catch

Faster does not necessarily mean better

A response can be fast and wrong. Each point below combines a participant’s median correct-go response time and go-trial accuracy. Audio condition is encoded by colour; the number of valid correct-go trials is shown on hover. This reveals whether a quick group also tends to preserve accuracy.

Source: Orpella et al. (2025), PLOS ONE, Supporting Information S4 (flanker task). Accuracy uses all go trials (including incorrect trials); speed uses valid correct-go trials.

Interpretation: The upper-left area represents both faster and more accurate performance. Colour overlap is a reminder not to infer a group-level speed–accuracy trade-off from one statistic alone.

04 / Hidden task difficulty

The same soundtrack, a different challenge

The flanker task includes congruent, incongruent, neutral and no-go trials. A no-go response is correctly withheld, so reaction-time comparisons on no-go trials would be misleading. This chart therefore shows accuracy across all four conditions, with the number of trials disclosed in every cell. A selectable mode switches to median correct-go reaction time for the three go conditions.

Source: Orpella et al. (2025), PLOS ONE, Supporting Information S4 (flanker task). RT tab excludes no-go and incorrect-response trials. RT colour scaling is independent of the accuracy tab.

What changes when you switch? A setting that looks favourable for accuracy might not look favourable for speed. The task condition matters to what counts as success.

05 / Beyond playlists

Perhaps the listener matters as much as the soundtrack

Can differences between people help explain the scatter? The final chart uses a dropdown to explore three recorded participant characteristics: musical reward sensitivity, musical training and stress. The vertical axis stays fixed as correct-go reaction time. These are exploratory associations and should not be interpreted as diagnoses or causal explanations.

Source: Orpella et al. (2025), PLOS ONE, Supporting Information S4 (flanker task). Participant characteristics appear as recorded in the supporting data.

The final thought: These measures reveal variation, not a definitive explanation of who benefits from music. A design tailored to real people requires more than ranking playlists.

So, does focus music work?

The apparent response-speed advantage for Work Flow is an intriguing first impression, not a universal rule. The original study reported improvement in response speed over time for Work Flow without a corresponding loss in accuracy. In these five views, the story becomes more complicated: listeners overlap substantially in speed, trajectories change across trial phases, fast responses do not guarantee correct ones, task conditions change the comparison, and participant characteristics vary. The result is a question worth investigating—not a playlist prescription. These exploratory graphics cannot establish the causal effect of a soundtrack on any individual.

Methods, exclusions and limitations

  • Unit of comparison: one participant per mark for Charts 1, 3 and 5. Chart 2 first calculates participant-level band summaries. Chart 4 provides cell-level accuracy and participant-level reaction-time summaries.
  • Reaction time: converted from seconds to milliseconds; only finite positive RT values on correctly answered go trials were retained. No arbitrary reaction-time trimming was applied. The number of contributing observations may therefore differ by chart.
  • Accuracy: Correct == 1 among all relevant trials; no-go trials are retained for the accuracy heatmap but never treated as go reaction times.
  • Exploratory design: medians are resistant to outliers, but these descriptive comparisons are not adjusted causal estimates. Participant traits are not randomized. Practice effects, sample characteristics and unmeasured factors may influence the patterns.
  • Context: this is a secondary visual exploration of one published online experiment, not new experimental evidence. The study was supported by a grant from the company supplying Work Flow tracks; some researchers disclosed advisory ties. These are relevant considerations for interpretation, not grounds to reject the data.
  • Accessibility: text accompanies colour, group names appear in hover labels, axes include units, and findings are explained in prose. Hover interactions work best with a mouse, keyboard navigation of Plotly widgets may vary by browser.

Data provenance and references

Every chart: Orpella, J., Bowling, D. L., Tomaino, C., & Ripollés, P. (2025). Effects of music advertised to support focus on mood and processing speed. PLOS ONE, 20(2), Article e0316047. https://doi.org/10.1371/journal.pone.0316047 . Dataset: Supporting Information spreadsheet pone.0316047.s004.xlsx (S4). Licence: CC BY 4.0.

GenAI reference: OpenAI. (2026). ChatGPT (GPT-6) [Large language model]. https://chatgpt.com/

Software: R (R Core Team), and the community R packages readxl, dplyr, tidyr, ggplot2, plotly, htmltools, scales, knitr and rmarkdown. See installed package versions in the knitting session for reproducibility.

GenAI acknowledgement

OpenAI ChatGPT (GPT-6, October 2026) assisted with the review of R code, visualization and design. All data, results and final outputs were reviewed by the author in accordance with RMIT academic integrity guidelines.

Reproducibility

The original Excel dataset and complete .Rmd source are supplied with this work. Source code can be made visible from the Code menu at the top of the rendered HTML. All visualisations are generated directly in R from the supplied data.