Ravens picked out a tool for a job they could not do until the next day, and the argument about what that means is still running. · published 10 October 2026
Ravens chose the right tool or bartering token for use up to 17 hours later, matching apes on tool tasks and beating chimpanzees at bartering - though whether that reflects planning for the future or well-trained association is still argued in print.
These are the working notes, unedited. Everything the video
states as fact comes from here, and nothing else may be stated as fact.
The sections on limits and on what the finding does not show are
part of the brief, not an afterthought — see
the method.
Why planning is a hard claim. Choosing something now for a need that does not exist yet is taken as evidence of representing a future situation. It is easier said than demonstrated, because an animal can look like it is planning while simply repeating whatever has paid off before. Food-caching birds plan for caching, but that is a specialised instinct in a species built for it. The interesting question is whether a bird can plan for something outside that specialisation.
The study. Can Kabadayi and Mathias Osvath, “Ravens parallel great apes in flexible planning for tool-use and bartering”, Science, volume 357, 2017, pages 202-204.
What they did. Ravens were taught to open a puzzle box with one particular stone, and separately to trade a particular token with a human for food. Neither task has anything to do with caching. The birds were then offered a tray holding the useful object alongside distractors, at a time when the box or the trading partner was absent, and the opportunity to use it came only later - in some conditions up to 17 hours later.
What they found. The ravens selected the right object and kept it. In one tool condition they chose the stone 86 per cent of the time when the box had been out of sight for an hour and returned 15 minutes after the choice. They matched tool-using apes on the tool tasks and beat orangutans, bonobos and chimpanzees on bartering. They also declined an immediate small reward in favour of the tool, which is self-control rather than planning but was measured alongside it.
The dispute, which is the interesting part. The paper was challenged almost immediately. A 2017 piece in Trends in Cognitive Sciences asked whether leaner explanations would do, and a later computational re-analysis argued the behaviour can be reproduced by associative learning - the birds having learned that the stone is valuable, without mentally representing tomorrow at all. The comparison with four-year-old children was called superficial. Kabadayi and Osvath replied in 2018, under the title “Contrary to the Gospel, Ravens Do Plan Flexibly”. No one has conceded.
Limits
Small numbers of birds, all hand-raised and experienced with these researchers and these tasks. The training that makes the test possible is also what gives the associative account its opening: a bird that has been rewarded for picking the stone many times has a reason to pick the stone that does not require any thought about the future. Whether that fully accounts for the 17-hour delays is precisely what is disputed.
What this does not show
Not a demonstration that ravens imagine tomorrow the way a person does, and not a debunking either. This is a live disagreement between competent people about what a well-run experiment shows, which is the normal condition of most interesting results and is usually tidied away in the retelling.