6 ms·

So there's an experiment I'm going to run. I've already got permission from the ethics committee. And by ethics committee I mean the friend of mine who owns the

So there's an experiment I'm going to run. I've already got permission from the ethics committee. And by ethics committee I mean the friend of mine who owns the birds. I'm going to attempt to train parrots to train each other to talk. Here's the premise. We make switches that the birds know how to press and understand that pressing causes actions. Easy enough. Then we hook the switches up to some sort of reward dispenser. Press button, get sunflower seed or almond. So far pretty simple. Then, we put the birds in a game-theoretic situation like the prisoners dilemma. If both birds press the left button, they both get the high reward (almond). If they both press the right button, they both get the low reward (sunflower seed). If one presses left and the other presses right, the left presser gets nothing and the right presser gets the highest reward (almond AND seed, oh wow). Our first question is if the birds will converge upon the Nash Equilibrium for this game. There's a second, even more interesting question we want to answer. The birds are capable of talking to each other, but they don't have a general purpose language. So far as we can tell, their vocabulary consists of 2 squawks, 3 squawks, 5 squawks, and rarely 6 squawks. The possible signals the birds might send each other are arbitrary and meaningless in the context of this experiment, but they can signal each other. Thus the question becomes, will the birds invent a meaning for their squawks in order to cooperate at this game? Can the evolution of language be explained by simply trying to maximize payoffs in the presence of other birds trying to maximize payoffs? Our third question is if the birds ever start lying to each other? If so, can I get the birds to manage my Diplomacy games for me? To the extent that the whales clearly have a structured language, I suggest a way to decode it. Ask not what the generic patterns of whale sounds might be, but instead ask what payoffs the whales need to negotiate over. What positive sum behavior do the whales engage in that would be impossible without signaling?

hetspookjee 4y ago
You can observe these behaviours in the wild, and especially among corvids alike. Sure it’s less convenient than using animals in your own enclosure, and perhaps less alluding as they’re not “your” animals. But I reckon for anyone, besides yourself and your friend, involved it would be beneficial if you’d conduct experiments like these through observation of wild behaviour instead. Also have you looked at the literature? Surely these experiments have been conducted several times already by many students in all kinds of flavours.
jciochon 4y ago
How do you account for confounding variables? For example, unseen or alternative methods of communication e.g. body language like bobbing, or clicking their beaks? Or even the tonality of the squawking? Whistling? It seems like you might still get meaningful results for some of your questions, but also that the behavior of the birds could be far more complex than foreseen, and maybe even missed entirely.
- IIAOPSW 4y ago
  >How do you account for confounding variables? For example, unseen or alternative methods of communication e.g. body language like bobbing, or clicking their beaks? Or even the tonality of the squawking? Whistling? All perfectly acceptable. The relevant feature is merely the ability for the birds to put arbitrary signals in some communications channel, without those symbols having any a priori semantics. We would certainly know if they were winning more often than they should. It might take a while to figure out what non-squawk method they were using, but it wouldn't go over our heads.
eurasiantiger 4y ago
Do not use food as a reward, it will not work because the animal will just use the buttons to dispense food, not to communicate. Social communication is rewarding and reinforcing in itself.
zasdffaa 4y ago
> ...whales clearly have a structured language Well, that's new to me. References?
- wolverine876 4y ago
  The first part of the OP describes it.
  - zasdffaa 4y ago
    Structure... but 'language'? Not the same thing.
    - wolverine876 4y ago
      I see your point! Edit: From the OP: Early in the book, he [British documentary filmmaker Tom Mustill] goes to visit Payne [researcher Roger Payne, who discovered or first studied whale songs] who’s now eighty-seven. Why, he asks, do humpbacks sing? And what do their songs mean? Payne says he can’t say: “I would desperately love to know.”
      - ASalazarMX 4y ago
        > Why, he asks, do humpbacks sing? And what do their songs mean? Payne says he can’t say: “I would desperately love to know.” "Any good looking females around? I am robust and can protect a family!" "DAE hate the noisy floaters that eat all the fish?" "Oh salty water, you are more salty than usual this day, oh salty water" "I'm tired of barnacles, mom says it's nothing, but they're nasty and they're ruining my life!" "Stop singing nonsense Waldorf, others want to sing too! If I was a few hundred miles closer I would slap you! Shut up shut up shut up!" Joking aside, the songs probably aren't very intellectual. If they were smarter, whalers and fisheries wouldn't be so successful.
daqhris 4y ago
I'd like to be informed of the outcome of this experiment. Mail, Twitter, Subscription... whatever you choose. The last paragraph got me hooked. You asked the right hard question. Tell me more if you've started working on it.
msrenee 4y ago
Can I ask what species of parrot?
- IIAOPSW 4y ago
  The type that can't talk people words but loves to yell "SQUAWK". Rainbow colored, about 100 grams. (I don't actually know the breed, they're not my birds).
  - msrenee 4y ago
    Rainbow Lorikeet maybe? There's such a range of intelligence and behavior between parrot species. It matters a lot what species you're working with.
motohagiography 4y ago
Arguably, a game that relied on increasingly complex arithmetic could become viral among birds. Geometry (lines and circles) would be next. When we school horses, we use circles and straight lines because those things don't exist in nature, but horses can learn them with instruction and repetition, and then we use those figures as a foundation or baseline to refine subtle deviations from the figures so that the horse can learn to interpret our intent from our physical cues in the context of a baseline figure. What dressage is really is the construction of a physical language with another being who has a completely different brain and existential experience, from the first principles of geometry. People go on about horse dancing and spiritual woo, but almost two decades in, this is what I have been able to derive from it. Master riders will tell you it's a science as much as an art, and it usually takes many years to sink in why they would say that, but like constructing a language with birds to teach each other, while they may not articulate it, with guidance, animals can absolutely experience and percieve geometry, and based on that shared apprehension, we can build on it to relate in very sophisticated ways that come to resemble language. The idea isn't crazy. I practice it more days than not. :)
- IIAOPSW 4y ago
  I suppose its only fair that crows should understand geometry in terms of "how the man walks".
  - motohagiography 4y ago
    I'd say if you could get a crow to fly in a circle, and then in a straight line, that is sufficient to establish your intentions of: right and left, bigger and smaller, longer and shorter, forward and sideways, faster and slower, more and less, listening and freely, and then combinations of those. Given that's 16 instructions in all, in 8 groups of a-or-b, a language that formed programs of combinations of each of them (without memory or state) would be pretty rich. Once you add abstraction to that, (something we're not sure about in animals), we're into some very complex behaviour and instructions in a way of relating. Communication with animals is very achievable in this model. What's not clear is how to express higher order abstractions in the model once you have those basics built up from geometry. Assistance dogs are "trained," to do tasks for their owners like opening doors, assisting with dressing, finding objects, etc. These are abstractions over the above instructions, and combined with them, we're well into relating to another being anyway.
forgotmypw17 4y ago
Crows do it. >A five-year study of crows living near Seattle in Washington State show the birds can remember a "dangerous human" and are able to share their knowledge of the learned danger with their offspring and other crows. https://www.cbc.ca/news/science/crows-share-intelligence-about-enemies-1.1014915 https://www.cbc.ca/news/science/crows-share-intelligence-abo... Lasting recognition of threatening people by wild American crows https://docdro.id/8UxS2eA https://docdro.id/8UxS2eA

hetspookjee 4y ago

You can observe these behaviours in the wild, and especially among corvids alike. Sure it’s less convenient than using animals in your own enclosure, and perhaps less alluding as they’re not “your” animals. But I reckon for anyone, besides yourself and your friend, involved it would be beneficial if you’d conduct experiments like these through observation of wild behaviour instead. Also have you looked at the literature? Surely these experiments have been conducted several times already by many students in all kinds of flavours.

jciochon 4y ago

How do you account for confounding variables? For example, unseen or alternative methods of communication e.g. body language like bobbing, or clicking their beaks? Or even the tonality of the squawking? Whistling? It seems like you might still get meaningful results for some of your questions, but also that the behavior of the birds could be far more complex than foreseen, and maybe even missed entirely.

IIAOPSW 4y ago
>How do you account for confounding variables? For example, unseen or alternative methods of communication e.g. body language like bobbing, or clicking their beaks? Or even the tonality of the squawking? Whistling? All perfectly acceptable. The relevant feature is merely the ability for the birds to put arbitrary signals in some communications channel, without those symbols having any a priori semantics. We would certainly know if they were winning more often than they should. It might take a while to figure out what non-squawk method they were using, but it wouldn't go over our heads.

eurasiantiger 4y ago

Do not use food as a reward, it will not work because the animal will just use the buttons to dispense food, not to communicate. Social communication is rewarding and reinforcing in itself.

zasdffaa 4y ago

> ...whales clearly have a structured language Well, that's new to me. References?

wolverine876 4y ago
The first part of the OP describes it.
- zasdffaa 4y ago
  Structure... but 'language'? Not the same thing.
  - wolverine876 4y ago
    I see your point! Edit: From the OP: Early in the book, he [British documentary filmmaker Tom Mustill] goes to visit Payne [researcher Roger Payne, who discovered or first studied whale songs] who’s now eighty-seven. Why, he asks, do humpbacks sing? And what do their songs mean? Payne says he can’t say: “I would desperately love to know.”
    - ASalazarMX 4y ago
      > Why, he asks, do humpbacks sing? And what do their songs mean? Payne says he can’t say: “I would desperately love to know.” "Any good looking females around? I am robust and can protect a family!" "DAE hate the noisy floaters that eat all the fish?" "Oh salty water, you are more salty than usual this day, oh salty water" "I'm tired of barnacles, mom says it's nothing, but they're nasty and they're ruining my life!" "Stop singing nonsense Waldorf, others want to sing too! If I was a few hundred miles closer I would slap you! Shut up shut up shut up!" Joking aside, the songs probably aren't very intellectual. If they were smarter, whalers and fisheries wouldn't be so successful.

daqhris 4y ago

I'd like to be informed of the outcome of this experiment. Mail, Twitter, Subscription... whatever you choose. The last paragraph got me hooked. You asked the right hard question. Tell me more if you've started working on it.

msrenee 4y ago

Can I ask what species of parrot?

IIAOPSW 4y ago
The type that can't talk people words but loves to yell "SQUAWK". Rainbow colored, about 100 grams. (I don't actually know the breed, they're not my birds).
- msrenee 4y ago
  Rainbow Lorikeet maybe? There's such a range of intelligence and behavior between parrot species. It matters a lot what species you're working with.

motohagiography 4y ago

Arguably, a game that relied on increasingly complex arithmetic could become viral among birds. Geometry (lines and circles) would be next. When we school horses, we use circles and straight lines because those things don't exist in nature, but horses can learn them with instruction and repetition, and then we use those figures as a foundation or baseline to refine subtle deviations from the figures so that the horse can learn to interpret our intent from our physical cues in the context of a baseline figure. What dressage is really is the construction of a physical language with another being who has a completely different brain and existential experience, from the first principles of geometry. People go on about horse dancing and spiritual woo, but almost two decades in, this is what I have been able to derive from it. Master riders will tell you it's a science as much as an art, and it usually takes many years to sink in why they would say that, but like constructing a language with birds to teach each other, while they may not articulate it, with guidance, animals can absolutely experience and percieve geometry, and based on that shared apprehension, we can build on it to relate in very sophisticated ways that come to resemble language. The idea isn't crazy. I practice it more days than not. :)

IIAOPSW 4y ago
I suppose its only fair that crows should understand geometry in terms of "how the man walks".
- motohagiography 4y ago
  I'd say if you could get a crow to fly in a circle, and then in a straight line, that is sufficient to establish your intentions of: right and left, bigger and smaller, longer and shorter, forward and sideways, faster and slower, more and less, listening and freely, and then combinations of those. Given that's 16 instructions in all, in 8 groups of a-or-b, a language that formed programs of combinations of each of them (without memory or state) would be pretty rich. Once you add abstraction to that, (something we're not sure about in animals), we're into some very complex behaviour and instructions in a way of relating. Communication with animals is very achievable in this model. What's not clear is how to express higher order abstractions in the model once you have those basics built up from geometry. Assistance dogs are "trained," to do tasks for their owners like opening doors, assisting with dressing, finding objects, etc. These are abstractions over the above instructions, and combined with them, we're well into relating to another being anyway.

forgotmypw17 4y ago

Crows do it. >A five-year study of crows living near Seattle in Washington State show the birds can remember a "dangerous human" and are able to share their knowledge of the learned danger with their offspring and other crows. https://www.cbc.ca/news/science/crows-share-intelligence-about-enemies-1.1014915 https://www.cbc.ca/news/science/crows-share-intelligence-abo... Lasting recognition of threatening people by wild American crows https://docdro.id/8UxS2eA https://docdro.id/8UxS2eA