Live data from Hacker News

Bird brains (2023)

dhanishsemar.com

21–30 of 230 posts

Re: Bird brains (2023)

#22

Given parrots can talk, there must be a neuron count that activates language (assuming anatomy allows it), similar to LLM parameter count.

Parrots can't "talk". They just mimick noises they've heard before

So what you’re saying is that parrots are stochastic parrots.

Re: Bird brains (2023)

#23

Given parrots can talk, there must be a neuron count that activates language (assuming anatomy allows it), similar to LLM parameter count.

Given parrots eat their own poop (https://lafeber.com/pet-birds/questions/parrots-eating-poop/), there must be a neuron count/density that activates self-poop eating (assuming anatomy allows it), similar to LLM parameter count.

Re: Bird brains (2023)

#24
post #2

Adding to this a chart of neuron count [1] [1] - https://en.wikipedia.org/wiki/List_of_animals_by_number_of_n...

Interesting... I would have thought Octopi have more total neurons than dogs, given their problem-solving capabilities. Now I wonder if the decentralized organization / hub and spoke model octopi alone exhibit offers some advantage when it comes to problem-solving

Something interesting about the octopus is that it is independent and learning from the time it is tiny.

It continually learns from the real world, as more and more neurons accumulate.

This layered learning may be an advantage in terms of compact representations.

No doubt, the human fetus brain learns much earlier than birth, or even from emergence of first neurons. But it isn't learning from the environment directly, or making survival critical choices, from first neural emergence.

--

Another octopus advantage maybe that it has relatively independent "brains" behind each eye, and along each leg. The distribution of brain in a way that reflects its physical distribution, might offer optimizations too.

We know humans benefit from partially independent spinal cord activity. This is suggestive evidence that the distributed intelligence of an octopus may be an advantage.

--

For exhibited intelligence per time, no other creature including humans comes anywhere close. They even learn "theory of mind", i.e. the ability to model other creatures situational awareness, ability to perceive, and likely responses to different situations.

To learn all that, without any mentoring or social examples, in the order of a year, along with their exotic body plan and amazing sensory configurations, would make the octopus a wildly implausible science fiction invention, if we didn't actually happen to have them living successfully in astonishing numbers, and pervasively in essentially all ocean environments.

It may have been enormous luck for us, that they live in an environment where technological progression would be very challenging.

The octopus is a very strong candidate for "smarter than humans", as an individual. If we equalize age, it isn't even a contest. If we normalize for lifespan, but equalize for lack of social mentorship, I expect they win decisively again.

(We often forget how much of our survival and progress is predicated on not being individuals. We have a species intelligence that is much higher than our individual intelligence. Since we as individuals gain so much from what is passed to us, we imagine that we would naturally know countless basic things, that if we actually grew up with people who did not know those things, would be far out of reach. Having people around to teach us things, allowed us evolve to be mentally lazy! Shades of current tool/dependency issues. The octopus has never had a crutch.)

--

There is no credible estimate of how many octopus individuals inhabit our oceans. But the number is in the billions at a minimum. Including young, it may be tens of billions or more.

Re: Bird brains (2023)

#26
> Dr. Irene Pepperberg studied an African grey parrot named Alex for 30 years. Alex could identify objects, colours, shapes, and numbers. He understood abstract concepts like "same" and "different." His vocabulary exceeded 100 words. When he died in 2007, his last words to Pepperberg were reportedly "You be good. I love you. See you tomorrow." I don't care how you define intelligence -- that one's hard to brush off.

The author takes forgranted the claim of intelligence; and does not assess at all whether the researcher simply said those words to the parrot every night. (Why not? It sounds exactly like what a researcher would tell a parrot before turning off the lights.) A quick search on Wikipedia says the parrot was also found dead in the morning, not in the implied "parrot has last words" scenario.

Re: Bird brains (2023)

#27

Earlier quoted context omitted.

All living beings have been evolving for the same amount of time.

Sure, but the speed of change is also related to lifespan. The longer lives you have (technically how long it takes to start reproducing and how many offspring you have), the less time you have to adapt. This means that for a given unit of time, shorter reproduction cycles and more offspring results in faster adaptation which is what OP meant and what the unhelpful pedantry doesn’t describe.

I suspect the more significant difference here is the selection pressures. Take a good look at any part of a bird and you'll see millions of years of selection for reduced weight. The cost of weight is just so much greater when you're flying. Interesting too that bats tend to have lower neuron counts than say rodents. Did dinosaurs have a more weight efficient brain before flight, or were they forced to shrink before re-evolving that complexity in a smaller package?

Re: Bird brains (2023)

#28

Earlier quoted context omitted.

Interesting... I would have thought Octopi have more total neurons than dogs, given their problem-solving capabilities. Now I wonder if the decentralized organization / hub and spoke model octopi alone exhibit offers some advantage when it comes to problem-solving

The prevailing research is “more neurons = intelligence” And that doesn’t make any sense, unless there really is no configuration necessary octopi bucking that trend is an example we need

No its pretty well understood that brain size in it self doesn't signify intelligence, even if correcting for body size. Density, connectedness, and complexity are important. Modeling the information processing capacity of animal brains it is shown that smaller brain like those of octopi and corvides are highly capable despite a relative low neuron count compared to humans.

Re: Bird brains (2023)

#29

Given parrots can talk, there must be a neuron count that activates language (assuming anatomy allows it), similar to LLM parameter count.

Parrots can't "talk". They just mimick noises they've heard before

Many animals can communicate.

Parrots can't speak fluent English, which shouldn't be surprising. Last I checked, no human is fluent in Parrot or Dolphin.

Though, at least one parrot may have demonstrated an ability to understand language at more than a surface level.

Post reply on HN