For the First Time, Research Reveals Crows Use Statistical Logic (arstechnica.com) 40
An anonymous reader quotes a report from Ars Technica: [R]esearchers from the University of Tubingen found for the first time that crows can perform statistical reasoning. These results can help scientists better understand the evolution of intelligence (and may give us a better appreciation of what's going on in our backyard). [...] Dr. Melissa Johnston, a Humboldt Fellow at the University of Tubingen, certainly appreciated the specialness of these creatures, as she and her colleagues have been studying these animals for several years. "In our lab, it has been shown that crows have sophisticated numerical competence, demonstrate abstract thinking, and show careful consideration during decision-making," she said. In her most recent experiment, Johnston and her team pushed these abilities to a new extreme, testing statistical reasoning.
To do this, Johnston and her team began by training two crows to peck at various images on touchscreens to earn food treats. From this simple routine of peck-then-treat, the researchers significantly raised the stakes. "We introduce the concept of probabilities, such as that not every peck to an image will result in a reward," Johnston elaborated. "This is where the crows learn the unique pairings between the image on the screen and the likelihood of obtaining a reward." The crows quickly learned to associate each of the images with a different reward probability. In the experiment, the two crows had to choose between two of these images, each corresponding to a different reward probability. "Crows were tasked with learning rather abstract quantities (i.e., not whole numbers), associating them with abstract symbols, and then applying that combination of information in a reward maximizing way," Johnston said. Over 10 days of training and 5,000 trials, the researchers found that the two crows continued to pick the higher probability of reward, showing their ability to use statistical inference.
Pushing the crows even further, Johnston and her team waited a whole month before testing the crows again. Even after a month without training, the crows remembered the reward probabilities and could pick the highest number every time. Johnston and her team were excited that the crows could apply statistical reasoning in almost any setting to ensure their reward. "Working with the birds every day is very rewarding! They are very responsive animals, so I enjoy spending time with them," added Johnston. The findings have been published in the journal Current Biology.
To do this, Johnston and her team began by training two crows to peck at various images on touchscreens to earn food treats. From this simple routine of peck-then-treat, the researchers significantly raised the stakes. "We introduce the concept of probabilities, such as that not every peck to an image will result in a reward," Johnston elaborated. "This is where the crows learn the unique pairings between the image on the screen and the likelihood of obtaining a reward." The crows quickly learned to associate each of the images with a different reward probability. In the experiment, the two crows had to choose between two of these images, each corresponding to a different reward probability. "Crows were tasked with learning rather abstract quantities (i.e., not whole numbers), associating them with abstract symbols, and then applying that combination of information in a reward maximizing way," Johnston said. Over 10 days of training and 5,000 trials, the researchers found that the two crows continued to pick the higher probability of reward, showing their ability to use statistical inference.
Pushing the crows even further, Johnston and her team waited a whole month before testing the crows again. Even after a month without training, the crows remembered the reward probabilities and could pick the highest number every time. Johnston and her team were excited that the crows could apply statistical reasoning in almost any setting to ensure their reward. "Working with the birds every day is very rewarding! They are very responsive animals, so I enjoy spending time with them," added Johnston. The findings have been published in the journal Current Biology.
Not exactly (Score:5, Funny)
It's really only 7.2 out of 10 crows based on a survey of 400 crows.
Re: (Score:2)
As some one who had an African Grey for 29 years (Score:1)
Re: (Score:1)
I had a budgie who just died a few months ago at the age of 14. (Now I have a new one.)
My previous bird always knew what time it was. To the minute.
He was also a rock and roll dancer. If the music had the right beat he would do The Twist or the Lambeth Walk, and he could do both as well as anyone I've ever seen. If he didn't feel like dancing you'd see him tapping a toe on his perch in time with the music.
fractional calculations (Score:3)
It sounds like they trained the crows to know that when you peck on image #1, you never get a reward, but pecking on #2 or #3 sometimes gets a reward. There is no evidence or logic in that to support the wild cliam that the crows are doing "statistical reasoning" and saying the crows are doing "calculations" based on "fractional numbers". Only that crows have apparently trained the scientists to do such calculations.
It is already well known that birds can do some counting, of the "1,2,3,many" variety. Parrots and Corvids being the smartest ones we're aware of. Like Pepperberg's parrot Alex [wikipedia.org].
Speaking of pecking at images, over at the lab where Alex the African Grey lived the last part of his life, they gave him a touch-screen (peck-screen) to a custom web browser which they called "Inter-pet Explorer". Just for (them to watch) him to surf the web.
Re:fractional calculations (Score:5, Informative)
Re: (Score:3, Insightful)
The crows are not crunching numbers, which was the claim. They're merely associating rewards with pictures. They remember which pictures work best. That's not "calculating". I wouldn't even call it "reasoning". It's no surprise that crows (or any animal I can think of) can remember a dozen or more things, and which of those things is usually good or bad.
Re:fractional calculations (Score:4, Informative)
Re: (Score:1)
Optimal play here is to pick Y if seen then wait until another Y or the 10th
Re:fractional calculations (Score:4)
First, there is no claim in the paper that crows are "crunching numbers". Second, the title here is actually quite accurate "uses statistical logic" as reasoning about outcomes from an understanding of probable outcomes is indeed "statistical logic". But you get a lot farther out on a limb with "I wouldn't even call it "reasoning"" as you are basing this rejection on a gross mischaracterization of what the researchers successfully demonstrated. No, it was not just picking "good and bad things".
Reasoning about probabilities is likely a highly evolved feature of natural neural systems generally to maximize the probability of survival.
Re: (Score:2)
Working with birds (Score:2)
"Working with the birds every day is very rewarding! They are very responsive animals, so I enjoy spending time with them"
Confirmation bias.
Can you say that, children? I knew you could!
Re: (Score:2)
Yeah. Animal behavioral researchers always get the best data from animals that are uncooperative and difficult to work with. Sure.
Crows (Score:4, Funny)
They freak me out. You can just tell by the way they look at you that they are working out methods of world domination. I encountered many of these creepy beasts while working on my grandfather's farm as a kid. They still haunt my dreams to this day ... ahhh! Great, now I'm going to be up all night.
Re:Crows (Score:5, Interesting)
Re: (Score:2)
They are trying to puncture the tires and cause mass destruction! LOL They have you fooled with their "skills" of using passing cars to eat walnuts. That "noisy war" you hear ... actually battle plans for when the domination finally occurs. They are all getting up to speed on the plans.
Re: (Score:2)
Re: (Score:2)
Common caw (Score:3)
They learned it in Common Caw.
Crows don't just have good memories (Score:2)
Music? (Score:3)
For the next experiment, maybe they should give some crow-sized guitars and drum sets to these counting crows and see if they make rock music.
I, for one⦠(Score:2)
Re: I, for one⦠(Score:2)
Crows are amazing, but statistics are not (Score:3)
Crows are amazing in all sorts of ways, like how you can train them to bring you items of interest (e.g. money) in exchange for reward or how some species have been observed tumbling down roofs to induce some kind of altered state. But "statistical logic" is kinda given just by the nature of neuronal networks, be it natural or artificial. It's the basis of its functionality and also the reason for all sorts of biases and malfunctions we develop, such as superstition and just learning correlations instead of causations.
It's nice to see some quantification of the granularity to which crows are able discern different probabilities, though.
On the other hand, the paper nicely demonstrates how long it takes to move away science from the very human-centric worldview of the darkages, where people would not even dare to attribute a modicum of intelligence to non-human lifeforms, even when we already do (or at least should) know such things for decades from basically petri dish experiements.
Re: (Score:2)
I don't much like scoffing at a paper I have only skimmed in an area I know next to nothing about, but I have questions before I'd pay any attention to this result.
Surely every predator learns to do this kind of "statistical reasoning"? Hunting in some areas gives better return than in others. Foraging animals similarly with other foods. Every house dog I've ever known had an instinctive knowledge of which adult to beg for scraps from at the table for combinations of far more than two people, too.
"In the ex
Re: (Score:2)
It does seem like the biases could be completely unconscious and still provide the same result.
Is that "reasoning"? (Score:2)
Re: (Score:2)
Yes, it's reasoning. It's not logic, but it's reasoning. Logic is the one that has to do with words and names and such things. (From the Greek logos. Reason is from a Latin word meaning [approx.] ratio.)
Re: (Score:2)
Must be nice.. (Score:2)
My dog keeps failing precalculus.
Re: (Score:1)
the inevitable outcome (Score:3)
no (Score:2)
Crows are very smart. All learning is statistical. That they do "statistical logic" is nonsense. Most humans don't, either.
Crows doing group statistical logic must be Murder (Score:1)
Of course, anything they do as a group is murder.
Re: (Score:2)
I think they can communicate with each other too.. (Score:2)