, 18 tweets, 16 min read
Very proud of my paper with @AlexhTaylor out today in @NatureComms.

We show that kea use probabilities to predict sampling events, and can integrate social & physical information into their predictions.

Full paper: nature.com/articles/s4146…

{thread}
@AlexhTaylor @NatureComms Imagine that I place my hand in a jar containing mostly blue candies and a few yellow candies, and then showed you a closed fist. Which candy do you think I must have taken from the jar? {1/16}
@AlexhTaylor @NatureComms If you guessed “blue”, you are correct! You used the relative frequencies of blue to yellow candies in the jar (i.e. the ratio of the two types of candy) to make your guess. {2/16}
@AlexhTaylor @NatureComms Humans are very good at making these probabilistic judgements and can make them from a very early age: sciencedirect.com/science/articl… {3/16}
@AlexhTaylor @NatureComms Importantly, even infants can use the relative frequencies of objects in a jar to make these judgements. They don’t use simpler strategies such as relying on the absolute number of rewarding or unrewarding objects. {4/16}
@AlexhTaylor @NatureComms We first sought out to test whether kea, a parrot species endemic to New Zealand, could also use probabilities in this way. We started off by teaching kea that black tokens could be exchanged for a food reward, and orange tokens were worthless. {5/16}
@AlexhTaylor @NatureComms We then presented kea with three tasks, where each time they observed as a human experimenter sampled a hidden token from one of two jars. They picked the hand that they thought was most likely to contain the prized black token. {6/16}
@AlexhTaylor @NatureComms Just like infants, kea also used the relative frequencies of rewarding objects (black tokens) and unrewarding objects (orange tokens) to make their choices. This shows that they actually use probabilities to make these judgements. {7/16}
@AlexhTaylor @NatureComms We then decided to test whether kea might be able to combine probabilistic information with other types of knowledge. We first showed kea that we could not sample a black token through a physical barrier. {8/16}
@AlexhTaylor @NatureComms Kea then had to pick between two jars, both of which had the same numbers of black and orange tokens throughout. However, both had a solid barrier down the middle, with tokens unevenly distributed above and below it. {9/16}
@AlexhTaylor @NatureComms Kea readily took this physical barrier into account, considering only the accessible parts of each jar when making their judgements. {10/16}
@AlexhTaylor @NatureComms We then tested whether kea could integrate social knowledge into their predictions, like chimpanzees can: sciencedirect.com/science/articl… {11/16}
@AlexhTaylor @NatureComms Kea first watched as two human experimenters picked out a black token from a jar. One of them was biased, looking carefully into the jar and picking out black tokens when they were in the minority. {12/16}
@AlexhTaylor @NatureComms The unbiased experimenter looked up, randomly picking a black token when they were in the majority. {13/16}
@AlexhTaylor @NatureComms At test, both experimenters picked from jars with 50% black tokens and 50% orange tokens: given chance alone, they should be equally likely to sample a black token. However, the kea remembered the biased experimenter’s preference and hedged their bets. {14/16}
@AlexhTaylor @NatureComms Domain-general intelligence requires an ability to combine different types of information, and was originally thought to have evolved once, in the great apes. We now know that at least one bird species are also capable of this form of knowledge integration. {15/16}
@AlexhTaylor @NatureComms TLDR; kea can make predictions based on probabilities, and adjust their predictions based on physical and social information. They combine information to adjust their predictions in a flexible way, suggesting that they have domain-general intelligence, as humans do. {end}
Missing some Tweet in this thread? You can try to force a refresh.

# Enjoying this thread?

Keep Current with Amalia Bastos

Stay in touch and get notified when new unrolls are available from this author!

This Thread may be Removed Anytime!

Twitter may remove this content at anytime, convert it as a PDF, save and print for later use!

# Try unrolling a thread yourself!

1) Follow Thread Reader App on Twitter so you can easily mention us!

2) Go to a Twitter thread (series of Tweets by the same owner) and mention us with a keyword "unroll" `@threadreaderapp unroll`

You can practice here first or read more on our help page!