Hacker News new | past | comments | ask | show | jobs | submit login
Crows can perform as well as 7- to 10-year-olds on cause-and-effect tasks (ucsb.edu)
179 points by evo_9 on July 25, 2014 | hide | past | favorite | 72 comments



Another researcher thinks effects like this are due to language in humans. (Maybe not directly linked, but somewhat related)

Basically the experiment was to have a rectangular room with 4 white walls, and then another box with one blue wall.

Researchers put some food in the room to the left of the blue wall, showed it to a rat, then spun the rat around, and the rat got it right 50% of the time. (IE random chance.)

Researchers did the same experiment with kids, and kids under 6 failed every time. Though after age 6, kids start using spacial language and can say the phrase "left of the blue wall", which implies to the researcher we use language to connect different disparate parts of the brain.

More on this during this radiolab episode: http://www.radiolab.org/story/91725-words/


>which implies to the researcher we use language to connect different disparate parts of the brain

I've always felt this was the case in my own mind. Language is the serialization format for abstract concepts.

This is why I'm very interested in constructed languages. I think it might help human cognition if we had a more efficient serialization format for internal concept transport.


> Language is the serialization format for abstract concepts.

Woa nice! Good analogy. Will steal.


Well, it is 'a sentence is a formatted thought' from a different angle.


it's a definition, not an analogy.

language is just our best tool for sharing feelings, concepts, etc. funny to think it's (probably) ultimately not the best tool for that. :)


Not that I disagree with your point, but I don't think spoken/written language is our best tool for sharing feelings. I'd say body-language is better for that. Or certain forms of art (such as visual art or music). All forms of communication, obviously.

Language probably is our best way of communicating abstract concepts, however (now, are feelings abstract or concrete? ;-).


if your definition of "feelings" stops at "happy" and "sad", sure. but feelings quickly become too complicated to express without language.

example: i want to go out to dinner tonight, but i feel like we went out to dinner too many times this week.

example: i am happy that the Yankees won, but i am unhappy with Derek Jeter's performance.

communicating these feelings clearly will, without better technology, require words.

more importantly, interpreting body language is simply prone to error. assuming two honest parties, it is much harder to misinterpret words.


That's not quite how I meant to use the word feeling above; I'd characterise what you describe as sentiment. I suppose I might have used the term emotion rather than feeling.


you're right, feelings seems like a mischaracterization on my part. in that case i agree body language is can be a better tool for communicating feeling than verbal language.

not to take this off the deep end, but i'll try to clarify my point.

language (body or verbal) is just our current best attempt at getting your brain to feel what my brain feels (feelings, sentiments, concepts, anything). my hunch they are both terribly insufficient compared to some technology we may be able to develop in the future.

that is, why serialize the data at all? a literal data transfer seems more reliable. for now, it's because we have no other choice.


> Language is the serialization format for abstract concepts.

For you maybe, but some people work with images only and most people use a mix of language and images. Autistic people tend to work exclusively with images, or exclusively with words. Some also use music or numbers. You don't need to speak any audible language to invent or understand things. Your own internal language is enough. You'll just be so isolated that you're unlikely to be understood by the masses, that's all.


> > Language is the serialization format for abstract concepts.

> For you maybe, but some people work with images only and most people use a mix of language and images.

There must be some variation across the population in this regard (https://news.ycombinator.com/item?id=5256057). Though I can think in images, concepts, words, sounds, smells, and tastes, I've said before on HN that my thoughts don't seem "official" until they've been serialized as a stream of words (https://news.ycombinator.com/item?id=5256126).


Whether or not autistic people use imagery is irrelevant, he said "language is the serialization," not "for everybody the only way they think of abstract concepts is via serialization with language."


> This is why I'm very interested in constructed languages

xu do kakne bacru la jbobau


No, but it's cool how I was able to find a mechanical parser (not something based on machine learning) that translated that for me.

I might learn it in the future.


> Researchers did the same experiment with kids, > and kids under 6 failed every time.

How can this be??? If they didn't have the ability to keep track of which way was which, shouldn't they have failed 50% of the time, like the rats?


I think this should be read as "kids under 6 failed [to perform better than chance] every time"


Correct. My bad. They acted just like the rats.


I heard during AI Class (6.034 w/Patrick Winston) that the same experiment was repeated with adults but they had to say some pre-determined words the whole time, and they regressed back to the level of pre-6 year olds.

Speech uses the same part of the brain as thought apparently, so making someone say words prevents them from thinking during that time.


they had to say some pre-determined words the whole time, and they regressed back to the level of pre-6 year olds

Only slightly related, but this is why I listen to audiobooks every night to fall to sleep. If I'm listening to someone else, I almost literally can't "think" so my brain chatter is basically eradicated and I fall asleep. I know it's not the same for everyone, but I basically turn into a mindless automaton if I'm actively listening to someone else as I'm an extremely verbal thinker.


This must be why I find it so difficult to talk to people about important things. If you explain something to me verbally, I will not get it. I will hear you, and I will listen intently, and I will have no idea what you just said.

Not that I wouldn't know what you said, but I won't have a deep understanding, I won't be able to use what you said to get work done. This is especially bad with phone conversations.

But if you write it down, then we're talking. I can digest written stuff extremely well.


Interesting, I often wonder in these kinds of cases whether that implies that Kitty, the best crow, has the equivalent of an internal language or symbolic logic that they're using to frame these ideas? Also, if provided with an easy way to interact with human language, would they be able to develop / learn concepts like grammatical structure, or always be limited to simple relationships like "Red = Food"? Would their reasoning ability improve if provided with such a language framework?


[just your first question] I don't think it necessarily implies an internal language, as crows, not being mammals, have a different brain structure. Specifically, they lack a neocortex. They instead have an independently evolved higher brain structure called a hyperpallium. Corvids (including crows) have the largest of these, amongst birds. Kinda, the apes of the bird world.

So it's possible the relevant processing centers are not separate in them, as they are in mammals, and do not need an internal language to bridge them. Therefore, while it's possible they do have the same problem and solution as us, this experiment doesn't imply it.

I wonder if there's a way to test which it is? e.g. something analogous the interfering with this internal language in humans that a commentor mentioned.


There is a culture in africa for which the concept of relative direction doesn't exist. How would they perform at this test?


In a linguistics course in college, this tribe was used as an example of the Sapir-Whorf hypothesis. Supposedly, members who were taught English and spoke it well still had trouble with relative direction because their original language (and culture) didn't have the concept. I always liked the example even if I thought it was dubious. Very cool to hear that someone else can corroborate that and make it at least a shared anecdote.


I believe the cultures in question know which way is North at all times, so this would be super easy for them, because the absolute position of an object that doesn't move never changes.


What kind of magical thinking leads people to accept these anthropological legends unquestioningly? 'there's a tribe who have no concept of left and right but always know which way is North' 'There's a tribe which has no concept of numbers greater than four but understands logarithms'...


What makes you think people accept it unquestioningly? There's plenty of evidence: http://pages.ucsd.edu/~jhaviland/Publications/ETHOSw.Diags.p...


You're actually demonstrating my point. That study is about an actualy Australian Aboriginal group, but the above commenters talk about nameless tribes in Africa. This is a common pattern in folk-anthropology where genuine cultural findings from field studies get overinterpreted, confused with other studies, elaborated and mythologized to the point where members of mysterious hypothetical African tribes have superhuman abilities to always know which way is North.


Reread parent. The person is spun around so they can't know absolute directions.


When they are first shown the room, they would think to themselves "the food is north of the blue wall". After being spun around they might not know which way is north, but they'll figure it out as soon as they spot the blue wall.


But what would happen if the researcher spun around the room as well?


Then their sense of "north" would become decoupled from true geographical north, but would still suffice for navigation within the current reference frame.


Just last night I watched a really, really great documentary on crow intelligence called "A Murder of Crows". I highly, highly recommend it. https://www.youtube.com/watch?v=s472GjbLKQ4



Never saw that, thx. His leap of faith worked out well for them. Another good bird IQ documentary, Nova's Inside Animal Mind: Birds... https://www.youtube.com/watch?v=-spBaywak7M

An even better read, IMO, but I am partial to the Corvus corax... https://www.goodreads.com/book/show/254704.Mind_of_the_Raven


Stoffel the honey badger is pretty innovative as well, (especially considering he was never trained): http://www.wimp.com/badgerhoudini/


Two reasons to think birds excel at spatial reasoning:

1. Migration. 2. Nest-building.

If you gave the same pile of materials to a robin and to me, I think the robin might do a better job of building a nest with it. I base this conjecture upon robin nests I've seen around my house.


Most humans don't experience much of the z axis without tools.

Brain-machine interface with weak AI (translation) could drastically alter the intelligence debate.


what in the world are you talking about? "most" humans exist in 3 dimensions (x,y,z), all the time, with or without tools.


I'm talking about spatial intelligence and not the difference between Aristotelian and Boolean logic. Obviously I have no idea what it feels like to stack shorter than a pancake (i.e., existing in 2 dimensions). Pedantry aside, the difference between existence and experience seems obvious.

I've never seen a human fly and well over 99% of my work is performed in the x and y axis. When communicating directions, I'm rarely concerned about z. Simple graphing tools (e.g., paper and pen) can be used to illustrate my point further.


Crows and magpies are the smartest birds on earth. European magpies pass the mirror test, something very rare animals and only children over 18 months do: http://www.newscientist.com/article/dn14552-mirror-test-show...


Crows are impressive animals. The more we learn about animal intelligence, the more I wonder what exactly it is that we do with all this extra brain matter we lug around.

http://www.fromquarkstoquasars.com/crows-could-be-the-key-to...


Crows can do amazing things; a documentary shows crows dropping nuts on roads, waiting for passing cars to crack them open.

But the problem then is to get onto the road to eat the content of the nut without being run over.

Well, crows learned to drop the nuts at pedestrian crossings, and wait for read lights stopping all cars, to go collect the reward. Problem solved.


I live in a neighborhood that has lots of crows and lots of walnut trees and not much car traffic. The neighborhood crows have figured out that dropping the walnuts in the intersections increases the chances of them getting crushed.

I'm running my own little social experiment: whenever I'm walking in the 'hood and I see a crow drop a walnut, I walk over and crush it. So far the crows have not started dropping nuts near me when they see me, but I have hopes :-)

Related, crow vending machine: http://www.ted.com/talks/joshua_klein_on_the_intelligence_of...


I have observed crows doing this with chestnuts on a crossing. And also a crow burying a chestnut, noticing that it was observed by another crow, then pretending that it buried the nut in another spot and vigorously defending the fake location from the other crow.


But can they beat chickens at tic-tac-toe?

http://www.nytimes.com/video/nyregion/100000001595360/tick-t...


as soon as crows start performing experiments on humans, I might share your wonder.


Looking forward to the Crow Computing revolution.


Enabled by crowed funding


You, guys, just crowsed the line.


This is kind of off-topic but birds can be pretty advanced in surprising ways. In particular, birdsongs can be pretty complex and some species have evolved specific cognitive functions for understanding, replicating, and producing birdsongs. It is conceivable that features evolved in relation to some tasks may also enhance behavior in other cognitive tasks. I would not be surprised if quite a few species besides crows perform just as well for the task mentioned in the article, and others.

Here is a neat review relating to songbirds:

http://languagelog.ldc.upenn.edu/nll/?p=3261


I sometime wonder if wild crows and ravens can understand a lot of human language, but have a cultural ban on revealing this too often due to how much trouble it tends to cause.


Gary Larson did a 'Far Side' cartoon where cows were standing, enjoying conversations in the field until one of them yelled 'Car!', where they all went silent and dropped to all fours.


This echoes to a statement in the "A Murder Of Crows" documentary linked in this thread, which goes like "They are smart, but we are the smartest". If there is an intelligence vastly superior than ours, what are the chances that we find out it exists? Can an "inferior" intelligence identify an other one as "superior"? Put in another way: if a superior AI emerges from one of our experiments, will we be able to understand its output?


If we ask it to dumb it down sure. Also often it's a lot harder to find a conclusion then to understand the conclusion and how to arrive at the same conclusion. Unless it's so complex our minds cannot comprehend it without years of study on 1 particular problem. But that is already occurring in silence today not ?


I'd say that if it was in its interest to communicate something, an intelligent being would be better equipped to do so than an unintelligent being.


In Tolkein's universe, the ravens of Erebor could understand and speak Westron. As with the rest of his works, Tolkein drew from Norse mythology, in which Odin's messenger ravens brought him news from around the world.


In Tolkien's universe, there are trees that understand and speak Westron...


well, to take the other perspective: little kids are not smarter than little animals.

it's not like the typical 7-10 year old kid is a fountain of smartness. supervision is still needed, it is not able to sustain its existence by itself, would starve or get killed through its own inexperience - which makes those little animals superior.


This is certainly true of the average 7-10 year old kid raised in suburbia by ordinary parents and educated and public school, but childhood and helplessness have been vastly prolonged artificially.

Maybe not a 10 year old, but a 12 year old in the Industrial Revolution would have been earning enough to cover his costs and contribute to the family. A 14 year old could also be doing just fine as a mother. It's not as if teenagers' poor life-skills are functions of their age, we just choose not to let them have the requisite experience until they are "old enough." Children are technically capable of much more than we give them credit for, we just would rather they spend time gaining the education to be knowledge workers instead of replaceable cogs in factories.

You couldn't just throw an average modern 10-year-old on the streets and expect him to be fine, but it's historical fact that it is possible for children not to be so helpless.


A huge fraction of those historical self-sufficient kids also died young for preventable reasons.


But is that because they were treated like adults, or because healthcare was primitive and unevenly distributed?


Well, a huge fraction of animals also die young for preventable reasons.


Survival can be a result of intelligence, but good survival abilities does not imply intelligence in itself. Tons of things survive without much intelligence at all, such as bacteria, or jellyfish.

In short, survival skills are not a good metric for intelligence, you can have one without the other.


There is a similar experiment showing pigeons outperforming Harvard students at an even simpler cognitive task.

One response to this would be to conclude that it's not like the typical Harvard student is a fountain of smartness; it is not able to sustain its existence by itself, would starve or get killed through its own experience (true enough; the average Harvard student wouldn't survive a month lost alone in the jungle) - which makes little animals superior.

A more sensible response would be to notice that humans actually are smarter than birds (I'm prepared to revisit that conclusion if any crows or pigeons can post reasoned arguments to the contrary), but that you can prove anything is superior to anything else if you cherry-pick the test carefully enough.


Unless they're your own kids, in which case they're really intelligent!

It goes without saying that a 7 yr old human can do a much wider variety of intellectual tasks an adult crow, however.


Excellent PBS Nature episode on the subject too: http://www.pbs.org/wnet/nature/episodes/a-murder-of-crows/fu...


We'll go to amazing lengths to modify these tests so that crows can perform at a 7 year old level, and we all accept that as fair, but when someone talks about modifying tests so that kids of a certain class/race background can score highly everyone thinks you're cheating and violating the sanctity of the test.

Imagine if Google made an alternate set of tests for low income applicants that let them in at a 200% higher rate and announced "low income applicants perform at 22 year old level in Google applications with modified test".


These smart crows make chimps seem like mere monkeys.


In Islam we are told to love kids till 7 by holding them close to us. Just conversation and love. Tell them to pray when they are 10 and use harsh language if necessary after 10. I know i am talking to a highly intelligent audience in this forum, most of which is atheist. I love you as a fellow human and would request you to know about Islam through Quran not through the Media or by looking at Muslims(coz most of us are doing it wrong)


> In Islam we are told to love kids till 7 by holding them close to us. Just conversation and love.

The Jesuits put this a little more sinisterly: "Give me the child until he is seven and I’ll give you the man".


You may have been expecting to post this in another thread. Otherwise, it's off-topic.




Join us for AI Startup School this June 16-17 in San Francisco!

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact

Search: