Pith. sign in

REVIEW 3 major objections 4 minor 78 references

CatPAL: Task-Aware Learning for Categorical Palette Recommendation

T0 review · 3 major / 4 minor · reviewed 2026-08-11 · deepseek-v4-flash

Challenge this review Re-run · record.json Download PDF Read on arXiv ↗ Review JSON

Pith's one-line read Categorical palette effectiveness is task-dependent, and a model trained on seven scatterplot tasks can predict palette accuracy and drive task-aware recommendations.

desk verdict Genuinely useful empirical study of task-dependent palette perception, but the recommendation system's headline benchmark rests on an unvalidated color extrapolation; referee it, expecting heavy revision. read the letter →

arxiv 2608.09108 v1 pith:PGO6BELR submitted 2026-08-10 cs.HC

Chin Tseng , Arran Zeyu Wang , Yunqi Li , Danielle Albers Szafir This is my paper · ORCID
classification cs.HC
keywords categoricalpalettestask-awarerecommendationBradley-Terrymodelredundantencodingscatterplottasksperceptualaccuracycolorperceptionshape
verification ladder T0 review T1 audit T2 compute T3 formal

The pith

A machine-rendered reading of the paper's core claim, the machinery that carries it, and where it could break.

The reading

CatPAL argues that the perceptual quality of a categorical palette is not a fixed property of the colors or shapes alone: the same palette ranks differently across seven common scatterplot tasks, so recommendations must be conditioned on the user's analytical goal. To make this concrete, the authors ran a three-stage crowdsourced pipeline that estimates per-task perceptual strength for 39 colors and 39 shapes with Bradley–Terry models, adaptively samples the most uncertain and task-sensitive comparisons, and measures how much redundant color+shape encodings gain or lose relative to the better single channel. They then use these measurements to train logistic-regression scorers that predict task accuracy for any candidate palette, and embed the scorers in CatPAL, a tool that returns ranked color, shape, or redundant palettes under user constraints such as preferred colors and task weights. If the central claim is right, palette design tools should shift from maximizing generic discriminability to optimizing for the task at hand, and redundancy should be treated as a context-dependent design choice rather than a universal default.

What carries the argument

The machinery is a three-part empirical model. Bradley–Terry models assign each of 39 colors and 39 shapes a latent perceptual strength $\beta$ per task and a category-count slope, fitted on pairwise accuracy from 224 participants; adaptive sampling then targets pairs with high uncertainty or high cross-task variance. A redundant-gain metric, $\Delta G = \text{Acc}(\text{color+shape}) - \max(\text{Acc}(\text{color}), \text{Acc}(\text{shape}))$, maps when combining channels helps or hurts across tasks and $k$. A palette-level logistic regression combines mean, minimum, and standard deviation of BT strengths, pairwise accuracy, and low-level perceptual features (CIELAB lightness, chroma, $\Delta E$; shape open/closed/filled composition) to predict $P(\text{correct} \mid \text{palette}, \text{task}, k)$, and unseen user colors are handled by nearest-neighbor $\Delta E$ lookup into the 39-item pairwise accuracy matrix. This scorer is what the CatPAL recommendation engine uses to build and rank palettes.

What would settle it

Run a crowdsourced study in which participants perform two of the seven tasks (e.g., cluster identification and outlier detection) with CatPAL-recommended palettes made of colors outside the 39-color pool, and compare observed accuracy against CatPAL's predicted scores; a Spearman rank correlation near zero or a calibration error above 0.05 between predicted and observed accuracy would show the unseen-color imputation does not hold.

Watch

Extended reading notes

Core claim

The central discovery is that task and palette choice jointly determine perceptual accuracy, with measurable rank reversals between tasks: the Kendall tau between color-palette rankings averages only 0.12 across tasks, and pairs like tightest-cluster and spatial-localization actually reverse order ($\tau = -0.20$). Color beats shape overall (82.1% vs 77.4% accuracy), but the advantage ranges from 8.3 points for outlier detection to 0.4 points for closest-to-reference, so shape is a practical substitute in some tasks. Redundant color+shape encoding shows a small average interference ($\Delta G = -0.023$), driven by harm at $k=3$ and benefit at $k=9$, meaning redundancy helps only when the single channel is under enough load. On the model side, the trained scorer orders held-out palettes consistently with human accuracy (Spearman rho 0.49 for color, 0.58 for shape) and is well calibrated, and CatPAL's recommended palettes score above designer, tool, and LLM baselines in the model-based comparison.

Load-bearing premise

The load-bearing premise is that nearest-neighbor $\Delta E$ imputation transfers pairwise-accuracy knowledge from the 39 tested colors to arbitrary user colors (with BT strengths for unseen colors set to zero), because the headline benchmark scores and the tool's custom-color feature both depend on that extrapolation.

Editorial extensions

If this is right

  • Palette effectiveness should be reported per task: a palette optimized for cluster identification can be predictably worse for spatial localization, so task-agnostic rankings hide real trade-offs.
  • Redundant color+shape encoding should be recommended conditionally: it can reduce accuracy at $k=3$ and improve it at $k=9$, so a single 'use redundancy' rule is wrong.
  • Shape-only palettes are a viable alternative in correlation and proximity tasks where the color advantage is under 2 points, which matters for accessible design.
  • A palette scorer with calibration error around 0.02–0.03 can serve as a design-time oracle to rank candidate palettes before user testing.

Reading between the lines

Editorial extensions of the paper, not claims the author makes directly.

  • Editorial inference: the two perceptual clusters the paper describes (local element detection vs global structure perception) suggest that a new scatterplot task could be assigned to a palette regime by classifying its perceptual demands, without collecting new data.
  • Editorial inference: the $k$-dependent $\Delta G$ pattern implies a broader principle, 'redundancy pays off when single-channel accuracy drops below some threshold,' which could be tested for other channels such as size or texture.
  • Editorial inference: if $\Delta E$ nearest-neighbor imputation fails for novel colors, the tool's flexibility with user-chosen brand palettes is the first thing to break; a targeted human study on out-of-pool palettes would settle this.
Share X Bluesky LinkedIn Reddit HN

Editorial analysis

A structured set of objections, weighed in public.

Desk editor's note, referee report, and a circularity audit.

Referee Report

3 major / 4 minor

Summary. This paper describes a three-stage crowdsourced pipeline that models categorical palette effectiveness across seven scatterplot tasks and category counts k=3, 6, 9. The authors fit per-task Bradley–Terry strength models for 39 colors and 39 shapes, use adaptive sampling to expand coverage, and measure redundant gain (ΔG) for color+shape palettes. They then train per-channel logistic regression scorers that predict trial correctness from palette features and embed these in CatPAL, a recommendation tool that supports user-specified colors or shapes and returns ranked palettes. The paper reports low and sometimes negative Kendall τ values between task-specific palette rankings, arguing that palette effectiveness is task-dependent, and benchmarks CatPAL against designer palettes, tool-generated palettes, and LLM palettes using predicted accuracy.

Significance. The raw empirical finding—that task-specific rankings of color and shape palettes are weakly correlated across tasks—is internally consistent and is a useful contribution to categorical palette research. The three-stage adaptive design, the public release of data and code, and the integration of the model into a practical web tool are clear strengths. If the predictive model were validated on held-out data and if the unseen-color imputation were tested against human performance, CatPAL would be a valuable task-aware design tool. As it stands, the model's validity and the benchmark claims are conditional on in-sample evaluation and an untested extrapolation, so the central recommendation claim is not yet fully supported.

major comments (3)
  1. [§6.1, §6.3] The ranking-fidelity and calibration results in §6.3 are computed on the same Stage 1+2 trials used to fit the logistic regression in §6.1 (40,320 training instances per encoding). No cross-validation, held-out task, or participant-split is described, so the reported Spearman ρ = 0.49/0.58 and calibration errors (0.033/0.019) are in-sample consistency checks rather than predictive validation. This does not establish that the scorer generalizes to new palettes, and the claim that the model 'reliably recognizes which palettes produce higher task accuracy' is stronger than the evidence. I recommend reporting cross-validated or held-out metrics, for example by participant split or by leaving out entire tasks.
  2. [§6.2, Table 2] All Table 2 scores are produced by the CatPAL scorer itself through the unseen-color path: BT strengths are set to zero for colors outside the 39-item pool, and pairwise accuracy is assigned by nearest-pool ΔE lookup. CatPAL's 20 recommended palettes are generated by optimizing exactly this scoring function, so the comparison is a self-consistent simulation rather than an independent evaluation of recommendation quality. A biased or miscalibrated imputation function could inflate CatPAL's apparent advantage over baselines. The manuscript acknowledges a user study as future work in §6.3 and §7.3, but the headline benchmark should be explicitly labeled as model-predicted scores until the imputation is validated or the comparison is restricted to the experimental color pool.
  3. [§6.2] The unseen-color extrapolation is load-bearing for CatPAL's central feature and is not supported by human data. The paper itself reports that nearest-pool BT beta imputation fails catastrophically (R² = −0.86) and sets BT strengths to zero for unseen colors; the fallback—nearest-neighbor ΔE lookup of pairwise accuracy, with 0.5 for same-neighbor pairs—assumes that pairwise accuracy is a function of ΔE alone. This assumption is questionable because pairwise accuracy also depends on hue direction, lightness, and category context, and no validation against human judgments for arbitrary colors is provided. At minimum, the paper should include a holdout study in which human accuracy is measured for a sample of unseen-color palettes, or the tool should be constrained to recommend only colors from the validated 39-item pool.
minor comments (4)
  1. [§6.3] The benchmarking paragraph contains an incomplete parenthetical: 'against a validated scorer (e.g., the designer palettes likely include colors that the engine's random seeding approach does not consider)' trails off without completing the thought; please finish the sentence.
  2. [§3.1, §6.2] Section 3.1 describes the exhaustive space using k=2–10, while the experiments only collect k=3, 6, 9; Section 6.2 clarifies that k is internally mapped to the nearest supported value, but this mapping should be stated earlier to avoid apparent inconsistency.
  3. [§4.2] The sentence 'For color, Stage 2 completed the comparison graph' is redundant immediately after reporting that unobserved pairs dropped to 19 (2.6%); consider rewording for clarity.
  4. [§5.2] The claim that the quadrant factor did not reach significance (F(3,80)=0.15, p=.929) is informative, but the 2×2 quadrant design with pooled strong/weak sets may have limited power to detect interactions; the no-pooling lookup in §6.2 is a reasonable design choice, yet the interpretation should acknowledge this limitation.

Circularity Check

2 steps flagged · score 6.0 of 10

CatPAL's validation and benchmark are in-sample: the logistic-regression scorer is fitted on the same Stage 1+2 trials it is then said to predict, and Table 2 scores CatPAL and all baselines with that same fitted scorer.

  1. fitted input called prediction [Section 6.1 'Model Construction' and Section 6.3 'Palette ranking']
    "The training data are drawn from the combined Stage 1 + Stage 2 dataset (224 participants), resulting in 40,320 training instances per encoding. ... We evaluate this by examining the correlation between the predicted and observed accuracies for each task ×k combination aggregated across all palettes from the experiments using Spearman ρ. Across all 21 combinations (7 tasks × 3 k values), µρ is 0.49 for color and 0.58 for shape"

    The 'observed accuracies' used to compute Spearman ρ and the calibration bins are the very Stage 1+2 trial outcomes on which the logistic-regression scorer was fitted; no held-out split is reported. The reported ρ=0.49/0.58 and calibration errors therefore measure in-sample training fit, not predictive validity. The paper then calls the scorer 'validated' and uses it to generate the headline predicted-accuracy rankings, so the validation loop is closed by construction: the scorer is evaluated against its own training labels.

  2. fitted input called prediction [Section 6.3 'Benchmarking against reference palettes and tools']
    "We then applied the model to estimate expected accuracy for 20 palettes generated by CatPAL using random seeding. ... The predicted-accuracy advantage of CatPAL recommendations over a conventional baseline therefore reflects an estimated real-world accuracy advantage, conditional on the calibration established above. While future work should formally test the recommendation engine against conventional approaches in a user study, this approach provides preliminary evidence of CatPAL's overall recommendation approach against a validated scorer."

    CatPAL selects palettes by greedy search that maximizes the same logistic-regression score (Section 6.2), and Table 2 reports those predicted scores as evidence that CatPAL outperforms baselines. Thus the benchmark is the optimization objective evaluating its own optima: the 'advantage' is a property of the fitted surrogate, not a measured human-performance difference. The paper explicitly concedes that the user study is future work, confirming that the central effectiveness claim rests on an in-sample, scorer-in-the-loop evaluation rather than external validation.

full rationale

The task-dependence claim (H1) is supported by raw experimental data, including the Kendall τ matrix across per-task BT rankings and the task×encoding interaction, and is not circular. The BT strength estimation and redundant-gain measurement are legitimate empirical derivations. The circularity is confined to the model-validation/benchmarking chain. The logistic-regression scorer is trained on the combined Stage 1+2 trial outcomes (Section 6.1); the 'palette ranking' and 'calibration' checks in Section 6.3 compare that scorer's outputs to the same trial outcomes, so ρ≈0.49/0.58 and calibration errors of 0.033/0.019 are in-sample fit statistics, not predictive validation. CatPAL then optimizes that same scorer (Section 6.2), and Table 2 scores CatPAL and all baselines with it, so the reported advantage is the optimization objective evaluating its own recommendations. The paper explicitly flags the missing user study, confirming that the benchmark is conditional on an unvalidated surrogate. The unseen-color path (BT β set to zero; pairwise accuracy via nearest-pool ΔE lookup) is load-bearing for Table 2 and for user-specified colors, and the paper admits beta imputation fails (R²=−0.86 in Section 6.2); this is a substantive correctness risk, but it is an extrapolation rather than a by-construction circularity, so it does not by itself raise the score above 6. Self-citations to prior work by the same group supply anchor items and legacy uncertainty values, but the central task-dependence finding rests on this paper's own experiments, so no additional self-citation-based circularity is identified.

Assumptions & free parameters 8 free parameters · 6 assumptions · 1 invented entities

The central model rests on a chain of fitted and hand-chosen quantities: BT strengths fitted to the trial data are reused as features and also generate the pairwise-accuracy matrices that enter as features; the Stage 3 Delta-G lookup uses per-quadrant cells whose main effect was not significant; and the tool's support for arbitrary user colors depends on an unvalidated nearest-neighbor imputation. The only newly introduced entity with a name, Delta G, is a summary metric rather than a physical postulate.

free parameters (8)
  • task-specific log(k) coefficient alpha^t = not reported in main text
    Scale of the category-count effect per task in the BT model, estimated by gradient descent in Section 4.1.4.
  • Stage 2 adaptive-sampling weights (gap filling, task shift, legacy risk) = 0.4, 0.4, 0.2
    Hand-chosen weights that determine which color/shape pairs receive additional data; directly shape the BT estimates used downstream.
  • Strong/weak quadrant pools in Stage 3 = top 13 and bottom 13 of 39 per encoding
    Manual split of items into quality pools for the 2x2 redundant-gain design; a different threshold would change the Delta-G lookup cells.
  • Minimum perceptual distance filter in greedy search = Delta E >= 12
    Hand-chosen threshold used to filter candidate colors when constructing palettes in CatPAL.
  • Category-count mapping for user input k = 2-4 -> 3; 5-7 -> 6; 8-10 -> 9
    Discretizes user-requested k to the three studied values; the paper states this mapping without validation.
  • Unseen-color imputation parameters = beta=0; nearest-pool neighbor by Delta E; 0.5 for same-neighbor pairs
    Ad hoc rule for scoring colors outside the 39-item pool; the paper reports that nearest-pool beta imputation fails (R^2 = -0.86).
  • Logistic regression coefficients (color and shape scorers) = not listed
    Model weights fit to the combined Stage 1+2 trial data; this is the central prediction function used by CatPAL.
  • Bayesian smoothing strength for sparse pairwise accuracies = not specified
    Applied to cross-task mean pairwise accuracy to stabilize sparse pairs; the shrinkage parameter is not stated.
assumptions (6)
  • domain assumption Bradley-Terry additivity: logit(P(correct|i,j,k,t)) = beta_i^t + beta_j^t + alpha^t log(k)
    Assumes pairwise discriminability of a palette decomposes into additive item strengths plus a log-linear category-count effect. Invoked as the core model in Section 4.1.4; no diagnostic for item-pair interactions or non-log-linear k effects is reported.
  • domain assumption Accuracy is a valid proxy for palette effectiveness
    All modeling uses proportion of correct trial responses as the dependent measure; response time, confidence, and task strategy are not modeled (Section 4.1.4).
  • domain assumption The 39-color and 39-shape pools from prior work are representative of categorical encoding options
    Anchors and candidates are drawn entirely from pools defined in earlier CatPAW/Shape It Up papers [65,66]; the generality of findings to other colors and shapes is assumed, and the paper reports BT strength transfer to novel colors is poor (R^2 = -0.86).
  • domain assumption Crowdsourced MTurk judgments with 20-second time limits are reliable measures of perceptual accuracy
    All experiments recruit MTurk participants with 95% approval, exclude failures on engagement checks, and time-limit trials; no screening for color vision deficiency beyond self-report is described.
  • ad hoc to paper The 50 hand-tuned datasets per task and k avoid ceiling/floor effects and confounds
    Stimulus parameters (sigma values, convex-hull differences, scagnostics thresholds) were tuned through pilot studies, and the 50 datasets per condition are treated as sufficient; no formal difficulty calibration is reported.
  • ad hoc to paper Greedy search over 400 randomly sampled candidate colors per step with Delta E >= 12 finds near-optimal palettes
    The recommendation engine relies on this stochastic search for out-of-pool colors; the paper does not compare against exhaustive enumeration or report search stability.
invented entities (1)
  • redundant gain Delta G
    purpose: Quantifies whether encoding both color and shape improves accuracy beyond the better single channel, and serves as a lookup table in CatPAL's scoring of redundant palettes.
    Delta G is defined as Acc(color+shape) - max(Acc(color), Acc(shape)) and estimated entirely from the paper's own Stage 3 trials; it is not validated against independent data or external benchmarks.

how reviews work

0 comments
Cite this review

Pith. "Pith review of CatPAL: Task-Aware Learning for Categorical Palette Recommendation." pith.science (2026). https://pith.science/paper/PGO6BELR

@misc{pith2026260809108,
  author       = {Pith},
  title        = {Pith review of: CatPAL: Task-Aware Learning for Categorical Palette Recommendation},
  year         = {2026},
  howpublished = {\url{https://pith.science/paper/PGO6BELR}},
  note         = {Machine review of arXiv:2608.09108}
}
read the original abstract

Designing effective categorical palettes requires balancing a range of factors, including perceptual distinctiveness, category count, and task effectiveness. The effectiveness of categorical encodings can vary substantially depending on the target analytical tasks; however, existing recommendation tools largely ignore task context when evaluating palette quality, resulting in inconsistent performance across tasks. We synthesize findings from a series of multi-stage user studies into a unified model of task-based effectiveness for color encodings, shape encodings, and their redundant combination across category counts and seven common scatterplot tasks. Our results show that task and palette choice jointly influence perceptual accuracy: different color and shape palettes exhibit varying levels of robustness across tasks, indicating that palette effectiveness is task-dependent. We estimate task-specific perceptual strengths for 39 colors and 39 shapes using Bradley-Terry models, refined through adaptive sampling to target uncertain and task-sensitive comparisons. We further quantify cross-channel interactions using a redundant gain Delta G metric to model performance across color and shape pairings. We then train a predictive model that scores candidate palettes based on task, category count, and perceptual features. This model drives effective palette recommendations in CatPAL, a task-aware palette recommendation system grounded in empirical data responsive to user constraints. Our findings highlight the importance of selecting categorical palettes aligned with specific analytical tasks and demonstrate how task-aware modeling enables more reliable palette design. CatPAL translates empirical results into a practical tool that supports user-specified colors or shapes and returns ranked palette recommendations adaptable to a range of tasks.

Figures

Figures reproduced from arXiv: 2608.09108 by the authors.

Figure 1
Figure 1. CATPAL: A task-aware categorical palette recommendation system for scatterplots. Given a set of scatterplot tasks and a category number, CatPAL selects top-performing palettes that maximize perceptual accuracy. Different tasks lead to different palette performance, as color and shape encodings vary in effectiveness across tasks. Drawing on empirical data, CatPAL captures these task-dependent effects and recommends p… view at source ↗

Discussion (0). Continue with ORCID to comment.

Reference graph

Works this paper leans on

78 extracted references · 41 canonical work pages

  1. [1]

    Albers, M

    D. Albers, M. Correll, and M. Gleicher. Task-driven evaluation of aggre- gation in time series visualization. InProc. ACM Hum. Factors Comput. Syst. (CHI), pp. 551–560, 2014. doi:10.1145/2556288.25572002

    arXiv 2014

  2. [3]

    Bostock, V

    M. Bostock, V . Ogievetsky, and J. Heer. D³ data-driven documents.IEEE Trans. Vis. Comput. Graph., 17(12):2301–2309, 2011. doi: 10.1109/tvcg. 2011.1858

    arXiv 2011

  3. [4]

    H. C. Braun, K. Mukherjee, S. R. Gorelik, and K. B. Schloss. Affective color scales for colormap data visualizations.IEEE Trans. Vis. Comput. Graph., 2025. doi:10.31219/osf.io/p3bva_v11, 2, 9

    work page doi:10.31219/osf.io/p3bva_v11 2025

  4. [5]

    Brehmer and T

    M. Brehmer and T. Munzner. A multi-level typology of abstract visualiza- tion tasks.IEEE Trans. Vis. Comput. Graph., 19(12):2376–2385, 2013. doi:10.1109/TVCG.2013.1243

    work page arXiv 2013

  5. [6]

    Bujack, T

    R. Bujack, T. L. Turton, F. Samsel, C. Ware, D. H. Rogers, and J. Ahrens. The good, the bad, and the ugly: A theoretical framework for the as- sessment of continuous colormaps.IEEE Trans. Vis. Comput. Graph., 24(1):923–933, 2017. doi:10.1109/TVCG.2017.27439789

    arXiv 2017

  6. [7]

    Burlinson, K

    D. Burlinson, K. Subramanian, and P. Goolkasian. Open vs. closed shapes: New perceptual categories?IEEE Trans. Vis. Comput. Graph., 24(1):574– 583, 2017. doi:10.1109/TVCG.2017.27450862, 4

    arXiv 2017

  7. [8]

    Carpendale

    S. Carpendale. Evaluating information visualizations. InInformation visualization, pp. 19–45. Springer, 2008. doi: 10.1007/978-3-540-70956-5_2 4

    work page doi:10.1007/978-3-540-70956-5_2 2008

  8. [9]

    Cleveland and R

    W. Cleveland and R. McGill. Graphical perception: Theory, experimenta- tion, and application to the development of graphical methods.Journal of the American Statistical Association, 79(387):531–554, 1984. doi: 10. 1080/01621459.1984.104780804

    arXiv 1984

Show all 78 references
  1. [11]

    Dasgupta, J

    A. Dasgupta, J. Poco, B. Rogowitz, K. Han, E. Bertini, and C. T. Silva. The effect of color scales on climate scientists’ objective and subjective performance in spatial data analysis tasks.IEEE transactions on visual- ization and computer graphics, 26(3):1577–1591, 2018. doi:...

    2018

  2. [12]

    Demiralp, M

    Ç. Demiralp, M. S. Bernstein, and J. Heer. Learning perceptual kernels for visualization design.IEEE Trans. Vis. Comput. Graph., 20(12):1933–1942,

    1933

  3. [13]

    M. A. Elliott, C. Nothelfer, C. Xiong, and D. A. Szafir. A design space of vision science methods for visualization research.IEEE Trans. Vis. Comput. Graph., 2020. doi:10.1109/TVCG.2020.30294134

    2020

  4. [14]

    H.-P. Frey, C. Honey, and P. König. What’s color got to do with it? the influence of color on visual attention in different categories.Journal of vision, 8(14):6–6, 2008. doi:10.1167/8.14.64

    2008 doi

  5. [15]

    Gleicher, D

    M. Gleicher, D. Albers, R. Walker, I. Jusufi, C. D. Hansen, and J. C. Roberts. Visual comparison for information visualization.Information Visualization, 10(4):289–309, 2011. doi:10.1177/14738716114165494

    2011 doi

  6. [16]

    Gleicher, M

    M. Gleicher, M. Correll, C. Nothelfer, and S. Franconeri. Perception of average value in multiclass scatterplots.IEEE Trans. Vis. Comput. Graph., 19, 2013. doi:10.1109/TVCG.2013.1832, 3

    2013

  7. [17]

    Gogolou, T

    A. Gogolou, T. Tsandilas, T. Palpanas, and A. Bezerianos. Comparing sim- ilarity perception in time series visualizations.IEEE Trans. Vis. Comput. Graph., 25(1):523–533, 2018. doi:10.1109/TVCG.2018.28650774

    2018

  8. [18]

    R. L. Goldstone and A. T. Hendrickson. Categorical perception.Wiley Interdiscip. Rev.: Cogn. Sci., 1(1):69–78, 2010. doi:10.1002/wcs.262

    2010 doi

  9. [20]

    C. C. Gramazio, D. H. Laidlaw, and K. B. Schloss. Colorgorical: Creating discriminable and preferable color palettes for information visualization. IEEE Trans. Vis. Comput. Graph., 23(1):521–530, 2016. doi:10.1109/TVCG .2016.25989181, 2, 5, 8, 9

    2016

  10. [21]

    Haroz and D

    S. Haroz and D. Whitney. How capacity limits of attention influence information visualization effectiveness.IEEE Trans. Vis. Comput. Graph., 18(12), 2012. doi:10.1109/TVCG.2012.2332, 4

    2012

  11. [22]

    Harrison, F

    L. Harrison, F. Yang, S. Franconeri, and R. Chang. Ranking visualiza- tions of correlation using weber’s law.IEEE Trans. Vis. Comput. Graph., 20(12):1943–1952, 2014. doi:10.1109/TVCG.2014.23469793, 4

    1943

  12. [23]

    Harrower and C

    M. Harrower and C. A. Brewer. Colorbrewer. org: an online tool for selecting colour schemes for maps.Cartogr. J., 40(1):27–37, 2003. doi: 10.1179/0008704032350020422, 8

    2003 doi

  13. [24]

    C. G. Healey. Choosing effective colours for data visualization. InPro- ceedings of Seventh Annual IEEE Visualization’96, pp. 263–270. IEEE,

  14. [25]

    Heer and M

    J. Heer and M. Stone. Color naming models for color selection, image editing and palette design. InProc. ACM Hum. Factors Comput. Syst. (CHI), pp. 1007–1016, 2012. doi:10.1145/2207676.22085472

    2012

  15. [26]

    M.-H. Hong, Z. N. Sunberg, and D. A. Szafir. Cieran: Designing sequential colormaps via in-situ active preference learning. InProceedings of the 2024 CHI Conference on Human Factors in Computing Systems, pp. 1–15,

    2024

  16. [28]

    D. R. Hunter. Mm algorithms for generalized bradley-terry models.The annals of statistics, 32(1):384–406, 2004. doi:10.1214/aos/10791201412

    2004

  17. [29]

    Kay and J

    M. Kay and J. Heer. Beyond weber’s law: A second look at ranking visualizations of correlation.IEEE Trans. Vis. Comput. Graph., 22(1):469– 478, 2015. doi:10.1109/TVCG.2015.24676713

    2015

  18. [30]

    S. Lin, J. Fortuna, C. Kulkarni, M. Stone, and J. Heer. Selecting semantically-resonant colors for data visualization.Comput. Graph. Fo- rum, 32(3pt4):401–410, 2013. doi:10.1111/cgf.121272

    2013 doi

  19. [31]

    Liu and J

    Y . Liu and J. Heer. Somewhere over the rainbow: An empirical assessment of quantitative colormaps. InProc. ACM Hum. Factors Comput. Syst. (CHI), p. 598, 2018. doi:10.1145/3173574.31741722

    2018

  20. [33]

    G. Martin. iwanthue: Colors for data scientists. https://medialab. github.io/iwanthue/, 2015. Accessed: 2025-03-26. 2, 8

    2015

  21. [34]

    Mukherjee, B

    K. Mukherjee, B. Yin, B. E. Sherman, L. Lessard, and K. B. Schloss. Context matters: A theory of semantic discriminability for perceptual encoding systems.IEEE Trans. Vis. Comput. Graph., 28(1):697–706,

  22. [35]

    Nothelfer, M

    C. Nothelfer, M. Gleicher, and S. Franconeri. Redundant encoding strengthens segmentation and grouping in visual displays of data.Jour- nal of Experimental Psychology: Human Perception and Performance, 43(9):1667, 2017. doi:10.1037/xhp00003142, 6

    2017 doi

  23. [36]

    G. J. Quadri and P. Rosen. A survey of perception-based visualization studies by task.IEEE Trans. Vis. Comput. Graph., 28(12), 2021. doi: 10. 1109/tvcg.2021.30982403

    2021

  24. [37]

    Reda and D

    K. Reda and D. A. Szafir. Rainbows revisited: Modeling effective col- ormap design for graphical inference.IEEE Trans. Vis. Comput. Graph., 27(2):1032–1042, 2020. doi:10.1109/tvcg.2020.30304392

    2020

  25. [38]

    Rensink and G

    R. Rensink and G. Baldridge. The perception of correlation in scatterplots. Comput. Graph. Forum, 29(3):1203–1210, 2010. doi: 10.1111/j.1467-8659. 2009.01694.x3, 4

    2010 arXiv

  26. [39]

    Saket, A

    B. Saket, A. Endert, and Ç. Demiralp. Task-based effectiveness of basic visualizations.IEEE Trans. Vis. Comput. Graph., 25(7):2505–2512, 2018. doi:10.1109/tvcg.2018.28297502, 4, 9

    2018

  27. [40]

    Salvi, K

    A. Salvi, K. Lu, M. E. Papka, Y . Wang, and K. Reda. Color maker: a mixed- initiative approach to creating accessible color maps. InProceedings of the 2024 CHI Conference on Human Factors in Computing Systems, pp. 1–17, 2024. doi:10.1145/3613904.36422652, 9

    2024

  28. [41]

    Samsel, S

    F. Samsel, S. Klaassen, and D. H. Rogers. Colormoves: Real-time in- teractive colormap construction for scientific visualization.IEEE com- puter graphics and applications, 38(1):20–29, 2018. doi:10.1109/mcg.2018 .0114615253

    2018 doi

  29. [43]

    Satyanarayan, D

    A. Satyanarayan, D. Moritz, K. Wongsuphasawat, and J. Heer. Vega-Lite: A grammar of interactive graphics.IEEE Transactions on Visualization and Computer Graphics, 23(1):341–350, 2017. doi: 10.1109/tvcg.2016. 25990301

    2017 doi

  30. [44]

    K. B. Schloss. Color semantics in human cognition.Current Di- rections in Psychological Science, 33(1):58–67, 2024. doi: 10.1177/ 10 © 2026 IEEE. This is the author’s version of the article that has been published in IEEE Transactions on Visualization and Computer Graphics. Th...

    2024

  31. [45]

    K. B. Schloss, C. C. Gramazio, A. T. Silverman, M. L. Parker, and A. S. Wang. Mapping color to meaning in colormap data visualizations.IEEE Trans. Vis. Comput. Graph., 25(1):810–819, 2018. doi: 10.1109/TVCG.2018. 28651472

    2018 doi

  32. [46]

    K. B. Schloss, L. Lessard, C. S. Walmsley, and K. Foley. Color inference in visual communication: the meaning of colors in recycling.Cogn. Res.: Princ. Implic., 3(1):1–17, 2018. doi:10.1186/s41235-018-0090-y2

    2018 doi

  33. [47]

    K. B. Schloss and S. E. Palmer. Aesthetics of color combinations. In Human Vision and Electronic Imaging XV, vol. 7527, pp. 365–376. SPIE,

  34. [48]

    K. B. Schloss and S. E. Palmer. Aesthetic response to color combinations: preference, harmony, and similarity.Atten. Percept. Psychophys., 73:551– 571, 2011. doi:10.3758/s13414-010-0027-01

    2011 doi

  35. [49]

    Schulz, T

    H.-J. Schulz, T. Nocke, M. Heitzler, and H. Schumann. A design space of visualization tasks.IEEE Trans. Vis. Comput. Graph., 19(12):2366–2375,

  36. [50]

    Sedlmair, A

    M. Sedlmair, A. Tatu, T. Munzner, and M. Tory. A taxonomy of visual cluster separation factors.Comput. Graph. Forum, 31(3pt4):1335–1344,

  37. [51]

    Setlur and M

    V . Setlur and M. Stone. A linguistic approach to categorical color assign- ment for data visualization.IEEE Trans. Vis. Comput. Graph., 22, 2016. doi:10.1109/TVCG.2015.24674712

    2016

  38. [52]

    Sharma, W

    G. Sharma, W. Wu, and E. N. Dalal. The ciede2000 color-difference formula: Implementation notes, supplementary test data, and mathematical observations.Color Res. Appl., 30(1):21–30, 2005. doi: 10.1002/col.20070 1

    2005 doi

  39. [53]

    Shneiderman

    B. Shneiderman. The eyes have it: A task by data type taxonomy for information visualizations. InThe craft of information visualization, pp. 364–371. Elsevier, 2003. doi:10.1016/b978-155860915-0/50046-92, 3, 9

    2003 doi

  40. [54]

    H. S. Smallman and R. M. Boynton. Segregation of basic colors in an information display.Journal of the Optical Society of America A, 7(10):1985–1994, 1990. doi:10.1364/av.1989.fb12

    1985 doi

  41. [55]

    Smart, K

    S. Smart, K. Wu, and D. A. Szafir. Color crafting: Automating the construction of designer quality color ramps.IEEE Trans. Vis. Comput. Graph., 26(1):1215–1225, 2019. doi:10.1109/tvcg.2019.29342842

    2019

  42. [56]

    Smilkov, N

    D. Smilkov, N. Thorat, C. Nicholson, E. Reif, F. B. Viégas, and M. Wat- tenberg. Embedding projector: Interactive visualization and interpretation of embeddings.arXiv preprint arXiv:1611.05469, 2016. 1

    2016 arXiv

  43. [57]

    Statgraphics Technologies

    I. Statgraphics Technologies. Statgraphics19, 2022. 8

    2022

  44. [58]

    D. A. Szafir. Modeling color difference for visualization design.IEEE Trans. Vis. Comput. Graph., 24(1):392–401, 2018. doi: 10.1109/TVCG.2017. 27443592, 9

    2018 doi

  45. [59]

    D. A. Szafir, R. Borgo, M. Chen, D. J. Edwards, B. Fisher, and L. Padilla. Visualization Psychology. Springer Nature, 2023. 4

    2023

  46. [60]

    D. A. Szafir, S. Haroz, M. Gleicher, and S. Franconeri. Four types of ensemble coding in data visualizations.Journal of Vision, 16(5):11–11,

  47. [61]

    D. A. Szafir, M. Stone, and M. Gleicher. Adapting color difference for design. InColor and Imaging Conference, vol. 22, pp. 228–233. Society for Imaging Science and Technology, 2014. doi: 10.2352/cic.2014.22.1. art000402

    2014 doi

  48. [62]

    Tableau, 2022

    Tableau. Tableau, 2022. 2, 8

    2022

  49. [63]

    Tominski, G

    C. Tominski, G. Fuchs, and H. Schumann. Task-driven color coding. In 2008 12th International Conference Information Visualisation, pp. 373–

    2008

  50. [64]

    Tseng, G

    C. Tseng, G. J. Quadri, Z. Wang, and D. A. Szafir. Measuring categori- cal perception in color-coded scatterplots. InProc. ACM Hum. Factors Comput. Syst. (CHI), 2023. doi:10.1145/3544548.35814162, 7, 9

    2023

  51. [65]

    Tseng, A

    C. Tseng, A. Z. Wang, G. J. Quadri, and D. Albers Szafir. Shape it up: An empirically grounded approach for designing shape palettes.IEEE Trans. Vis. Comput. Graph. (Proc. IEEE VIS 2024), 2025. doi: 10.1109/tvcg.2024. 34563852, 4, 5, 7, 9

    2024 doi

  52. [66]

    Tseng, A

    C. Tseng, A. Z. Wang, G. J. Quadri, and D. Albers Szafir. Redundant is not redundant: Automating efficient categorical palettes design unifying color & shape encodings with CatPAW. InProceedings of the 2026 ACM CHI Conference on Human Factors in Computing Systems (CHI), 2026. ...

    2026

  53. [67]

    Tseng, A

    C. Tseng, A. Z. Wang, G. J. Quadri, and D. A. Szafir. Revisiting categorical color perception in scatterplots: Sequential, diverging, and categorical palettes. InProceedings of the 26th EG/VGTC Conference on Visualization (EuroVis), 2024. doi:10.2312/evs.202410732

    2024 doi

  54. [68]

    A. Z. Wang, G. J. Quadri, M. Zhu, C. Tseng, and D. Albers Szafir. Charac- terizing visualization perception with psychological phenomena: Uncover- ing the role of subitizing in data visualization.IEEE Trans. Vis. Comput. Graph. (Proc. IEEE VIS 2025), 2026. doi:10.1109/tvcg.202...

    2025

  55. [69]

    Y . Wang, Z. Wang, T. Liu, M. Correll, Z. Cheng, O. Deussen et al. Im- proving the robustness of scagnostics.IEEE Trans. Vis. Comput. Graph., 26(1):759–769, 2019. doi:10.1109/TVCG.2019.29347963, 4, 9

    2019

  56. [70]

    C. Ware. Color sequences for univariate maps: Theory, experiments and principles.IEEE Computer Graphics and Applications, 8(5):41–49, 1988. doi:10.1109/38.77602

    1988 doi

  57. [71]

    Ware.Information visualization: perception for design

    C. Ware.Information visualization: perception for design. Elsevier, 2012. 1, 4

    2012

  58. [72]

    C. Ware, T. L. Turton, R. Bujack, F. Samsel, P. Shrivastava, and D. H. Rogers. Measuring and modeling the feature detection threshold functions of colormaps.IEEE Trans. Vis. Comput. Graph., 25(9):2777–2790, 2018. doi:10.1109/tvcg.2018.28557422

    2018

  59. [73]

    Wilkinson, A

    L. Wilkinson, A. Anand, and R. Grossman. Graph-theoretic scagnostics. InIEEE Symp. Info. Vis. (INFOVIS), pp. 157–164. IEEE, Oct 2005. doi: 10.1109/INFVIS.2005.15321422, 3, 4, 9

    2005

  60. [74]

    F. Yang, L. Harrison, R. Rensink, S. Franconeri, and R. Chang. Correlation judgment and visualization features: A comparative study.IEEE Trans. Vis. Comput. Graph., 2018. doi:10.1109/tvcg.2018.28109184

    2018

  61. [75]

    Zhou and C

    L. Zhou and C. D. Hansen. A survey of colormaps in visualization.IEEE Trans. Vis. Comput. Graph., 22(8):2051–2069, 2015. doi: 10.1109/tvcg. 2015.24896491, 2

    2015

  62. [76]

    Zimnicki, C

    C. Zimnicki, C. Tseng, D. A. Szafir, and K. B. Schloss. Effects of data distribution and granularity on color semantics for colormap data visu- alizations. In2023 IEEE Visualization and Visual Analytics (VIS), pp. 11–15. IEEE, 2023. doi:10.1109/vis54172.2023.000111, 2 11

    2023

  63. [380]

    doi:10.1109/iv.2008.242, 3, 4

    IEEE, 2008. doi:10.1109/iv.2008.242, 3, 4

    2008 doi

  64. [1996]

    doi:10.1109/visual.1996.5681182

    1996

  65. [2010]

    doi:10.1117/12.8491112

    doi

  66. [2012]

    doi:10.1111/j.1467-8659.2012.03125.x3, 4

    2012

  67. [2013]

    doi:10.1109/tvcg.2013.1203

    2013

  68. [2016]

    doi:10.1167/16.5.119

    doi

  69. [2021]

    doi:10.1109/tvcg.2021.31147802, 9

    2021

  70. [2024]

    doi:10.1145/3613904.36429032

Pith tools

Reviewed August 11, 2026 · model on record in the stance chip above.