What drives our aesthetic attraction to birds?

The study investigates which traits make some bird species more visually attractive to humans. Birds hold cultural, educational, and recreational value, and birdwatching is a global pastime. Beyond emotional or historical associations, humans respond to direct visual stimuli such as color, shape, size, and ornaments, alongside non-visual cues like sounds and behavior. Prior work suggests aesthetically appealing species can face higher threat from trade but also receive more conservation attention; however, existing studies have been geographically or taxonomically limited. The authors propose that human aesthetic preferences may be rooted in evolutionary and experiential principles: vivid colors (especially blue and red) evoke positive or arousing responses; people prefer fresh/vivid over dull colors; multicolored and extravagant forms (e.g., unusual shapes, long ornaments) draw attention; and rarity of features can increase preference (negative frequency-dependent selection). The research goal is to provide a global, trait-based understanding of visual features driving human attraction to birds and to examine links with non-visual factors such as range size and threat status.

The paper situates its research within several strands of literature: (1) empirical findings that aesthetically appealing or colorful species may be disproportionately targeted by wildlife trade and receive more research attention; (2) psychological and ecological theories on color preference and affect, including associations of blue (sky, clean water) and red (arousing, high-calorie foods) with emotional responses; (3) evidence that humans generally prefer vivid and multicolored objects and are attracted to extravagant, uncommon shapes; (4) the concept of negative frequency-dependent selection in human preferences for rare phenotypes; and (5) prior bird-focused studies indicating preferences for blue and green hues and for smaller species, though these studies were limited to certain taxa or regions. The authors highlight the need for a comprehensive, global analysis using direct measures of aesthetic preference across nearly all bird species.

Data source and attractiveness scoring: The authors used an internet-based application (iratebirds.app), available in 21 languages, to collect human ratings of birds’ visual attractiveness from photographs sourced from the Macaulay Library (Cornell Lab of Ornithology). Users rated attractiveness on a 1–10 scale. To minimize photo-quality bias, they prioritized higher-quality photos (library’s 1–5 quality score) and used multiple photos per species (mean ≈ 5, range 1–15). Over 6000 respondents from 78 countries provided about 400,000 ratings for ~11,000 bird taxonomic units. A regression model estimated species-level attractiveness while accounting for photo quality and random effects (photo/identity, species, genus, family, order). Sex-specific scores were obtained for dichromatic species; sex was coded as male, female, or unknown. A subset analysis including user demographics (country, age, birdwatching activity, environmental awareness, nature-related attitudes) showed high concordance with the main estimates (r = 0.92), indicating limited bias from these factors. The final database contained 11,187 attractiveness estimates at the species/sex level (7060 species with a single estimate; 2963 male and 116 female separate estimates for dichromatic species).

Predictors (visual): Traits included body mass; ornaments (crest scored on a continuous scale from absent to long; beak and tail length relative to body mass in mm); coloration represented by six largely uncorrelated color variables (black, white, yellow, blue, red, green); and a color elaboration measure quantifying departure from the global average brown-grey coloration. Color diversity (number of color loci) and dull colors (grey, brown, orange, purple, rufous) were tested in separate models due to collinearity.

Predictors (non-visual): Range size (extent of resident breeding range); IUCN threat status reclassified into Threatened (VU, EN, CR), Non-threatened (LC, NT), and Unknown (DD, NE); trophic level (carnivores, omnivores, herbivores); migratory status (migratory vs resident); and latitude of range center.

Data processing and modeling: Continuous predictors (including log-transformed body mass and range size) were standardized (mean 0, SD 1). Collinearity checks (Pearson’s |r| ≤ 0.5 among retained predictors) guided variable inclusion. The response was the attractiveness score rescaled to (0,1) by dividing by 10 and modeled with a beta distribution using glmmTMB (R 4.1.0). Random intercepts: Order (41 levels), Family (247), Genus (2248), and SpeciesID (8852) to handle phylogenetic structure and repeated sex-specific measures. Main aesthetic model (Eq. 1) included sex, body mass, crest, relative beak and tail length, color elaboration, and the six color variables. A second model (Eq. 2) added non-visual traits (trophic level, IUCN, range size, latitude, migration). Model diagnostics and selection employed the performance package. Robustness checks included sex-specific datasets (male, female, unknown) and models partitioning light vs dark hues for certain colors. Due to collinearity, separate models assessed color diversity and dull color sets.

• Across 8852 species analyzed globally, several traits associated with higher visual attractiveness emerged:
  1. Coloration: Strong preferences for elaborated/more vivid colors (departure from brown-grey) and for blue and red hues; lower attractiveness associated with black and dull colors (brown/grey). A separate model indicated multicolored plumage (higher color diversity) is most attractive.
  2. Ornaments: Longer tails were positively associated with attractiveness; longer crests showed a positive but weaker association; relative beak length had minimal effect.
  3. Size: Smaller species were more attractive than larger species.
  4. Non-visual: Larger range size correlated with higher attractiveness, possibly reflecting familiarity.
• Effect size estimates from the beta GLMM (N = 9649; mean ± 95% CI): • Sex (Female): 0.00 [−0.06, 0.07] • Sex (Male): 0.06 [0.00, 0.07] • Body mass (log-scaled): −0.18 [−0.29, −0.06] • Crest (score): 0.02 [−0.02, 0.06] • Relative beak size: 0.01 [−0.02, 0.07] • Relative tail size: 0.14 [0.09, 0.20] • Color variables (scaled):
  • Black: −0.08 [−0.14, −0.02]
  • White: −0.02 [−0.07, 0.02]
  • Yellow: −0.09 [−0.20, 0.02]
  • Blue: 0.16 [0.10, 0.22]
  • Red: 0.14 [0.10, 0.19]
  • Green: 0.04 [−0.06, 0.16] • Color elaboration (degree of departure from brown-grey): strongest positive association with attractiveness among predictors.
• Robustness: Patterns held across subsamples restricted to male (N = 2531), female (N = 1012), and unknown-sex/largely monomorphic species (N = 6070), though effects were clearer for males (likely due to sample size/trait variance). Results for light vs dark hues were consistent with main findings.
• Threat status: Contrary to expectations, no evidence that threatened species are more attractive than non-threatened ones in this global analysis.
• Validation: Attractiveness estimates accounting for user demographics strongly correlated with main estimates (r = 0.92), indicating limited demographic confounding.

The findings directly address the central question by identifying visual traits that systematically increase human attraction to birds: vivid and elaborated colors (notably blue and red), multicolored plumage, small body size, and extravagant ornaments (especially long tails). Preference against black and dull colors further refines this profile. The positive association with range size suggests familiarity may enhance perceived attractiveness. These patterns align with theories of ecological relevance and negative frequency-dependent selection, wherein rare or unusual phenotypes elicit stronger preferences. Conservation implications are twofold: (1) leveraging aesthetic appeal in conservation marketing and public engagement can enhance willingness to support conservation; (2) aesthetic desirability can increase demand in wildlife trade, creating risks for attractive species. The absence of a global link between threat status and attractiveness indicates complex, context-dependent relationships (e.g., trade focus, taxa studied) and underscores the need to consider additional modalities (e.g., song) and sociocultural factors. Overall, quantifying aesthetics at a global scale provides actionable insights to harness birds’ cultural ecosystem services while monitoring potential trade-related threats.

This study provides a global, trait-based synthesis of human visual preferences for birds, showing that vivid, elaborated colors (especially blue and red), multicolored plumage, small body size, and extravagant ornaments (notably long tails) drive higher attractiveness, while dull/black coloration and larger body size reduce it. Non-visual traits such as larger geographic range also correlate with attractiveness, possibly via familiarity effects. The work advances beyond previous regional or taxon-limited studies by using direct, large-scale human ratings across nearly all bird species and linking them to standardized trait data via hierarchical models. Future research should experimentally test the roles of color combinations and harmony, cultural and demographic influences, familiarity, and non-visual cues (e.g., acoustic signals) in shaping preferences; integrate multisensory assessments; and further parse how aesthetic appeal intersects with conservation status and trade dynamics. These insights can inform conservation marketing, educational outreach, and strategies to mitigate biases that overlook less attractive but conservation-needy species.

• Cultural and demographic representation: Respondents were predominantly from Western countries; although demographic-adjusted estimates correlated strongly with main scores (r = 0.92), global cultural variation may not be fully captured.
• Visual-only focus: Attractiveness was assessed from photographs, excluding auditory cues (e.g., song) and behavioral context that may influence preferences and trade.
• Photo-based biases: Despite controlling for photo quality and using multiple images, residual effects of image selection/pose/background may remain.
• Trait collinearity and modeling constraints: Some color variables were excluded from the main model due to collinearity; dull colors and color diversity were analyzed separately, which may limit direct comparability across models.
• Sex-specific data imbalance: Effects were clearer for males, potentially due to smaller sample size and lower trait variability in female datasets.
• Familiarity not directly measured: Range size was used as a proxy; the study did not directly quantify individual familiarity with species.
• Threat status inference: No global association was detected; differences from trade-focused studies and taxonomic scope (all birds vs. passerines) limit cross-study comparability.