Denyse O’Leary has put up a fascinating post about a cockatoo (a Tanimbar corella, Cacatua goffiniana, like the one pictured above) that can crack locks unassisted, without having to be offered a reward at each step along the way. This raises the interesting question of whether the bird’s behavior can properly be described as intelligent. Here’s a brief excerpt from the report by Sandrine Ceurstemont (3 July 2013) in New Scientist magazine:
Alex Kacelnik and colleagues at the University of Oxford set a number of cockatoos a challenge: pick a lock to access a nut visible behind a transparent door. The birds had to remove a pin, followed by a screw and a bolt, before turning a wheel to release a latch (see video, above).
Five birds were successful after some guidance, or with practice, but one of the cockatoos – called Pipin – broke in unassisted in under 2 hours. It was also the only bird to remove the screw with its foot instead of its beak.
To test whether the birds had simply memorised a sequence of tasks, or whether they had a physical understanding of each device, the team altered the set-up by breaking, removing or re-ordering some of the locks, as shown in the video.
This did not stump the birds, suggesting that they are aware of how objects act on each other, says Kacelnik. It also shows that the parrots do not need to be rewarded every step of the way to solve a problem.
At first blush, this experiment appears to address the traditional objections. The bird wasn’t being rewarded; it learned from its experience; it used what it had learned in a creative fashion on subsequent occasions. Surely that deserves to be called intelligent behavior, right? Not so fast.
I posted the following critical remarks in a comment on Denyse’s post, but in response to a suggestion from KeithS, I have put them on a new post. Readers are welcome to weigh in with their own opinions.
Here are my reasons for skepticism:
(a) The vast majority of the birds required guidance. “Five birds were successful after some guidance, or with practice, but one of the cockatoos – called Pipin – broke in unassisted in under 2 hours.”
(b) The one that didn’t require guidance still took two hours to finish the job;
(c) Trial and error alone may not be sufficient to explain what the cockatoo did, but how about a combination of three unintelligent mechanisms, none of which require insight on the bird’s part: (i) trial and error beak movements in a novel situation; (ii) memory of a repertoire of techniques that worked in previous similar situations; and (iii) instant feedback via touch of objects giving way in response to being pushed in a certain way? I’d bet that those three techniques would suffice to explain what the bird did, given that it took two hours to do the job;
(d) There was no control testing of human beings solving a similar task, to see how long they took, and whether they used a more efficient procedure for opening the locks;
(e) Could a robot be designed that could do the same thing, without training? I’ll bet it could;
(f) Normally we require a person who has accomplished an intelligent feat to explain how he/she did it, and why he/she did it that way. If the person can’t do that, then we usually don’t refer to that person’s feat as intelligent, but merely intuitive. For instance: one trick I can do is fast mental calculations, but I can always go back and explain how I did them and why I did them that way. I’m not super-fast; someone with a really good auditory memory could calculate a lot faster than I can. However, I’ve read that there are some super-fast people who just “hear the answer” inside their heads but who can’t say where it comes from. Should we call their calculations intelligent? I think not;
(g) More generally, can we speak of true intelligence in the absence of language? It might seem so: lots of good mathematicians engage in picture thinking and hate to “show workings.” But even they still need to use some words to explain what they’re getting at, and I’m not persuaded that they could do their work without any language at all;
(h) In response to KeithS’s comment about the mind being immaterial: well, it has to be, whether materialists like it or not. A material system is by definition incapable of generating a purely formal concept like “true,” “false,” “reality,” “prime” or even “probable.” Why? Because material properties such as speed, temperature etc. are in a different category from formal concepts. It’s a category mistake to explain the latter in terms of the former: it’s like trying to explain humor in terms of color. You wouldn’t even try to do that: you can see immediately that it won’t work: the enterprise is doomed from the start. Ditto for the attempt to explain formal concepts in terms of material ones. But material objects are by definition entities, all of whose properties (and behavior) can be characterized in physicalistic, spatio-temporal terms. So whatever we are, we’re not material objects. At least some of our properties transcend the material. Do animals also transcend the material to some degree? I’d be happy to allow that they might, but I wouldn’t appeal to feats such as these to prove my case. I’d be more impressed with the fact that they are capable of genuine (non-opportunistic) friendship, as many pet-owners will attest. That suggests that they have a primitive concept of a significant “other,” which resists explanation in simple materialistic terms. Unlike us, though, pets can’t talk about what they mean by this concept. And therein lies one vital difference between us.
What do you think? Over to you.