3D flightpaths reveal the development of spatial memory in wild hummingbirds
Pritchard, D. J.; Hurly, T. A.; Photopoulou, T.; Healy, S. D.
Show abstract
Many animals learn to relocate important places and reflect this spatial knowledge in their behaviour. Traditionally evidence for learning is examined experimentally by studying spatial memory. However, tools developed for analysing tracking data from widely ranging animals allow a more holistic analysis of behaviour. Here we use the two together in novel combination of experimental and modelling approaches to analyse how patterns of hummingbird movements change as birds learn to find a reward in a location indicated by a pair of landmarks. Using hidden Markov models (HMMs) we identified two movement states which we interpret as Search and Travel and compared these to experimental behavioural measures of spatial memory. When birds had a single training trial to learn a flowers location, both the behavioural measures and HMMs showed that hummingbirds relied on landmarks to guide search. Hummingbirds focussed hovering around the rewarded location and were more likely to be in the Search state, and more likely to switch from Travel to Search, when closer to the rewarded location, but only when the landmarks were present. When birds had had 12 additional training trials, however, the HMMs and behavioural measures showed differences in how reliant birds were on landmarks. While behaviours like hovering were still strongly affected by removing landmarks, the likelihood of being in or entered the Search state was the same regardless of whether the landmarks were present or removed. These results suggests that hummingbirds rapidly learn to use nearby landmarks to structure where they search, but as birds gain experience the role of these landmarks changes. While familiar local landmarks were still essential for precise search, experienced birds were able to use alternative cues to guide broad-scale transitions between behaviour. HMMs and traditional behavioural measures each capture a different aspect of this learning, with neither approach alone accurately described the role of landmarks in spatial learning.
Matching journals
The top 5 journals account for 50% of the predicted probability mass.
Similar papers in this journal
- Great tits show serial reversal learning in the perseverance phase but not in the new learning phase 95%
- How do backward walking ants (Cataglyphis velox) cope with navigational uncertainty? 95%
- Wild raccoons are more innovative but not bolder than another ecological generalist, the Virginia opossum, on a pull-string task 93%
Similar papers in this journal
- Fear of the new? Geckos hesitate to attack novel prey, feed near objects and enter a novel space 94%
- Cleaner wrasse failed in early testing stages of both visual and spatial Working Memory paradigms 93%
- Common marmosets (Callithrix jacchus) excel in a one-trial spatial memory test, yet perform poorly in a classical memory task 93%
Similar papers in this journal
Similar papers in this journal
"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.