Flexible, abstract rhythm perception in bumblebees
Data files
Feb 18, 2026 version files 89.56 KB
-
Data_S1.csv
10.23 KB
-
Data_S2.csv
68.87 KB
-
Data_S3.csv
7.13 KB
-
README.md
3.33 KB
Abstract
Flexible, abstract rhythm perception underpins human music, dance, and speech, but has only been demonstrated in a few select birds and mammals. Here, we show that bumblebees also form robust abstract rhythm representations. Free‑flying bees learned to discriminate two arbitrary repeating flashing light sequences, balanced to preclude the use of any local cues. Bees successfully recognized these learned rhythmic patterns at novel, faster, and slower tempi. Bees trained on vibrational patterns transferred their learning to equivalent flashing light patterns, demonstrating cross-modal rhythm perception. These findings reveal that even an insect brain can encode and generalize arbitrary complex temporal patterns, suggesting that abstract rhythm perception can emerge from relatively simple neural architectures and pointing to deep evolutionary roots for a domain‐general rhythm cognition across animals.
Dataset DOI: 10.5061/dryad.b8gtht7qt
Description of the data and file structure
File: Data_S1.csv
Description: Data S1 contains all the raw training data used for the statistical analyses and visualization in the main text for Figure 1.
- ‘beeID’ entries are the unique identification numbers for each individual bee used in each experiment;
- ‘colony’ entries are the unique identification numbers for each colony;
- ‘setting’ entries indicate which pattern was rewarding where 1 means that pattern A contained sugar and pattern B contained quinine and 2 means the opposite contingency.
- ‘exp’ entries are the experiments (rate = regular patterns in experiment 1; pattern = irregular pattern in experiment 2; tempo 1, 2, and 3 = irregular patterns at varying tempos in experiment 3, where the tempo listed is the one the bee was tested on;
- ‘trial’ entries indicate each choice/landing by the bee during the test;
- 'choice' entries indicate whether the bee landed on the correct pattern (1) or incorrect pattern (0) during the test.
File: Data_S2.csv
Description: Data S2 contains all the raw training data used for the statistical analyses and visualization in the main text for Figure 2.
- ‘beeID’ entries are the unique identification numbers for each individual bee used in each experiment;
- ‘colony’ entries are the unique identification numbers for each colony;
- ‘setting’ entries indicate the pattern-reward-side contingency, where 1 means reward on right for pattern A and reward on left for pattern B while 2 means reward on left for pattern A and reward on right for pattern B;
- ‘testseq’ entries indicate the order in which the two patterns appeared in the two tests (1 means pattern A was tested first and 2 means pattern B was tested first);
- ‘trial’ entries indicate the trial number during the training phase or whether it was testA or testB;
- ‘pattern’ entries indicate which pattern was displayed during the training trial or test;
- ‘choice’ entries indicate whether the bee chose the correct (1) or incorrect (0) side arm.
File: Data_S3.csv
Description: Data S3 contains all raw behavioral records from the irregular pattern test (experiment 2), used for the statistical analyses and visualization in Figure S2.
- 'beeID' entries are the unique identification numbers for each individual bee;
- 'colony' entries are the unique identification numbers for each colony;
- 'setting' entries indicate which pattern was rewarding where 1 means that pattern A contained sugar and pattern B contained quinine and 2 means the opposite contingency;
- 'behavior' entries specify the response type, including hovering followed by landing on the rewarded conditioned stimulus ('hover_land_CS+'), hovering followed by landing on the aversive quinine-associated conditioned stimulus ('hover_land_CS−'), hovering followed by rejection of the rewarded conditioned stimulus ('hover_reject_CS+'), and hovering followed by rejection of the aversive quinine-associated conditioned stimulus ('hover_reject_CS−').
- 'duration' entries report the hovering time (in seconds) preceding the bee's landing or rejection decision.