Learning when to learn: hummingbirds adjust their exploration behaviour to match the value of information
Bibliographic record
Abstract
Abstract Exploration is a key part of an animal’s ability to learn. The exploration-exploitation dilemma predicts that individuals should adjust their exploration behaviour according to changes in the value of information. Here, we test this prediction by tracking ruby-throated hummingbirds as they foraged repeatedly from a large array of artificial flowers, wherein 25% of the flowers contained a sucrose reward. Similar to real-world floral dynamics, the reward locations were consistent in the short-term, but varied from day to day. Thus, the value of information about the flower contents would be greatest at the beginning of a daily foraging session, and decay toward the end of each session. We tracked five individual hummingbirds in repeated foraging sessions, comprising more than 3,400 floral probes. We analyzed two metrics of their exploration behaviour: (1) the probability that a bird would shift from probing one flower to another, and (2) the Shannon information entropy of a sequence of flowers probed. We show that initially, the hummingbirds increased their exploration behaviour as time elapsed within a session. As they performed more sessions and learned the rules of the environment, the hummingbirds switched to explore more diverse choices at the beginning of a foraging session, when the value of information was high, and less diverse choices toward the end of a session. Our results suggest that foraging hummingbirds can learn when to learn, highlighting the importance of plasticity in exploration behaviour. Highlights Exploration is a necessary part of learning Foragers must balance sampling for information with the use of known rewards Hummingbirds learned to explore more when the value of new information was high Apparent mistakes may actually represent an information-seeking strategy
Fetched live from OpenAlex and de-inverted. Abstracts are not stored in this database: the inverted indexes are 8.6 GB of the frame’s 9.3 GB of text, and the host has 13 GB free.
How this classification was reachedexpand
Full frame machine prediction
Teacher imitationNot calibrated prevalence, not ground truth. Human validation pending. The Gemma side is a direct model label for every work in the frame, read from the title-only record. The Codex side is a classifier learned from the 10,348 direct Codex labels and calibrated to design-weighted sample rates; fields without enough sample support carry no Codex call. Candidate is the union of the two sides; consensus is their intersection. These outputs are machine_predicted_unvalidated and are not human labels.
Distilled classifier scores by category (both heads)
| Category | Codex | Gemma |
|---|---|---|
| Metaresearch | 0.000 | 0.001 |
| Meta-epidemiology (narrow) | 0.000 | 0.000 |
| Meta-epidemiology (broad) | 0.000 | 0.000 |
| Bibliometrics | 0.000 | 0.000 |
| Science and technology studies | 0.000 | 0.000 |
| Scholarly communication | 0.000 | 0.000 |
| Open science | 0.000 | 0.000 |
| Research integrity | 0.000 | 0.000 |
| Insufficient payload (model declined to judge) | 0.001 | 0.000 |
Machine scores (provisional)
The two teacher heads of the student model, read on this work. A score orders the frame for review; it never asserts a category, and the validation status ships verbatim with every row.
Baseline scores from an immature model (maturity gate not passed, 7 training rounds). Scores rank; they never assert a category.
score_only:v0-immature-baseline · verbatim from the scoring run: score_only means the number may rank works, and no category label ships from itClassification
machine, unvalidatedMachine predicted; a candidate call from one source (direct Gemma or distilled Codex), not a consensus.
How this classification was reached, model by model and score by score, is at the end of the page under "How this classification was reached".