Guidelines
Suggest edit

Do not tune perceptually uniform multi-hue palettes to chase retrieval performance

For retrieve tasks, avoid switching among perceptually uniform multi-hue sequential color scales on quantitative color encodings to prevent over-interpreting negligible performance differences and address unnecessary palette tuning for viewers comparing values against a legend.

  • purpose:refine
  • basis:empirical
  • task:retrieve
  • data:quantitative
  • quality:fidelity
  • lever:encoding
  • reading-mode:lookup
  • polish:palette

advice

Multi-hue palette tuning

Do not switch among perceptually uniform multi-hue sequential ramps to chase retrieval speed or accuracy. For example, viridis, plasma, and magma did not produce a significant performance difference in the tested quantitative retrieval task.

reason

Why this tuning does not pay off

Within the tested perceptually uniform multi-hue family, the alternatives produced very similar retrieval behavior. That means family choice mattered more than fine-grained switching inside the family.

Mechanism: When readers are already using an ordered perceptually uniform multi-hue ramp, changing to another member of the same family does not substantially alter how they recover relative value distance from color.

Evidence: The review records a weak accuracy ordering among viridis, plasma, and magma and a tie for time, with no significant pairs; the original study likewise found no significant differences among these multi-hue palettes for either response time or error in the retrieval task (Zeng & Battle, 2023; Liu & Heer, 2018).

context

Use when this is the decision

  • User Goal: Improve retrieval performance on an existing perceptually uniform multi-hue quantitative color scale.
  • Task: Compare relative closeness of encoded values using a legend.
  • Data: Scalar quantitative values encoded with a perceptually uniform multi-hue sequential palette.
  • Chart Setting: The planned revision is only a swap among the tested multi-hue ramps.
  • Audience: Readers decoding magnitude from color.
  • Success Criterion: A measurable gain in speed or accuracy.

exceptions

When not to use this rule

Break it when: You are deciding between a perceptually uniform multi-hue palette and a different palette family. Why: This rule only covers within-family swaps among the tested multi-hue sequential ramps.

costs

What this costs

Sacrifice: You give up palette iteration as a primary performance fix within the tested multi-hue family. Risk: A real retrieval problem can persist if the issue lies outside this family-level choice. Mitigation: If performance is still weak, compare the current multi-hue ramp against a different palette family instead of another within-family variant.

mistakes

Common failure around this change

Mistake: Replacing one perceptually uniform multi-hue ramp with another after readers struggle, without changing anything else. Why it fails: The tested multi-hue ramps were statistically indistinguishable on the retrieval task.

check

How to test the tuning decision

Failure Sign: Retrieval performance does not improve after a swap between multi-hue variants. Quick Check: A/B test the current multi-hue ramp against another multi-hue ramp on the same chart and retrieval question. Stronger Test: Record time and error for both versions; if the difference stays negligible, stop tuning within the family.

fix

What to change

  • Stop rotating among perceptually uniform multi-hue variants as the main retrieval-performance intervention.
  • Keep one multi-hue ramp and compare it against a different palette family if retrieval remains weak.
  • Evaluate any further palette change with the same legend-based retrieval question.

References

Liu, Y., & Heer, J. (2018). Somewhere Over the Rainbow: An Empirical Assessment of Quantitative Colormaps. Proceedings of the 2018 CHI Conference on Human Factors in Computing Systems, 1–12. https://doi.org/10.1145/3173574.3174172
Zeng, Z., & Battle, L. (2023). A Review and Collation of Graphical Perception Knowledge for Visualization Recommendation. Proceedings of the 2023 CHI Conference on Human Factors in Computing Systems, 1–16. https://doi.org/10.1145/3544548.3581349