Cognition without Cortex


As the tittle states, there are birds that show very impressive cognitive abilities, but don’t have a cortex. What do you think about this? Here is a paper and some images that explain a little bit more about this.


Cortex? OK - calling it that is a mammal thing.

Pallium? Lots of critters have one. What is a Pallium?

Note that the swellings of the neural tube into lobes has a bump that is called the telencephalon. That develops in different ways in different branches of the animal kingdom but birds definitely have one.

You may notice that the avian pallium has many of the same structures as the mamian counterpart. It is shaped and layered differently but I would not say that birds “don’t have a cortex” because we don’t call it that.

@gmirey may have some additional insight on this.

I would like to add that the efforts of the Numenta group has been a laser-like focus on the cortical column and it’s computation. What would be different if the group had started with the avian model? The two models are very different but achieve much the same high level functions.

This thread reinforces my assertion that the higher level functions arises from the connections of processing units and that the minor details of the local computation is not as important.

I see that the best way forward is to work back and forth between what computational functions that are provided by the low level structures and what high level functions emerge. Each line of research will inform the other on what is computational-level and what is implementation-level details. I see that focus on just one level will make it hard to work out the differences between these levels.


I’ve asked the same question about the validity of cortical column models being where intelligence is truly seated. Because as you have pointed out, birds don’t have everything that mammals have but still show signs of higher intelligence.

Turns out though, they have something similar to cortical columns that have developed independently of the neo cortex. Which I think is even better news because it means that the learning principle is a universal standard (or at least neural standard) that nature will always try to tend towards as the go to answer for intelligence.

There are a couple of papers on it that I have read but this one should give you a nice direction for your own personal studying. Neocortical Evolution.

The avian versions are less structured and less layered than mammalian cortical columns but they seem like they talk and communicate in a similar fashion. There are some papers that suggest that maybe the cortical neurons layering the way they do is mostly just a metabolic efficiency based adaptation rather than an important structural one.

Either way, it’s definitely interesting and whatever the reasoning, it gives a better insight into what is actually important in learning systems.


its all about reducing noise till you get to something useful,


Haha, I’ve posted the same point some years ago [on this forum].

I think we need to figure out the general algorithm or principle that causes intelligence.

By the way, check this out. This is also very important to figuring this out:
( What Bodies Think About: Bioelectric Computation Outside the Nervous System - NeurIPS 2018)

What happens in layers 5a and 6a?

Don’t be fooled by the size of the bird’s brain:

“Bird Brains:
They’re often disregarded for being small. It turns out, however, that bird brains pack surprisingly high numbers of neurons. REALLY high…”

Herculano Houzel home


Our ancestors have had the kind of cortex we have now for well over a million years, but discovered ‘human’ artifacts seem to go back no further than ~60,000 years ago suggesting that our brains have only incrementally supported more intelligence among humans than birds or other beasts.

Our manual dexterity is still primitive compared to octopus dexterity and they usually live for less than 2 years.

I think the idea of cortical layers being significant for memory formation and recall may be a stretch - all that is needed is a way to connect activated neurons into an engram that can be triggered in whole when something similar occurs in part. This will enable association.

sequence memory is the similar but has the character of fading signals associated with arising signals as a natural transition between engrams.


Another possibility is that bootstrapping into a functional language is not automatic.


there is very low likelihood that an operating system evolved waiting for a functional language: we know intelligence emerged, stone tool making emerged, use of fire emerged, spoken language emerged, collaborative hunting and gathering emerged, social dominance and agriculture emerged, mining and smiting emerged, industrialization emerged.

with AI, we do not mean to wait for all those things to emerge, but the model of our intelligence is still worth refining so that one day we can do something like booting up a human mind equivalent from some kind of holographic data store.


The different between chimps and humans is very small.

If the key difference (whatever that may be) was a chance mutation it may have taken time for that to become reinforced enough to build language. Remember, at least two humans had to have this mutation to talk to each other and it was likely that the first mutations were not as good as the current skills.


the origin of language is fascinating with several great theories
but I do believe that many animals have verbal communication and can think and have visual recognition intelligence, spatial creative intelligence, and some even have a kind of math even with tiny brains, and no specific cerebrum to speak of.
Sapiens has one of the more complex brains, great hands, upright stance, and poor insulation so the drive to shelter combined with the ability to collaborate on building projects (predating agriculture and animal husbandry) exceeds the abilities of wolves for instance.
Who knows, if wolves had fingers, we could be barking and howling up a storm.


For a million years? I don’t think so. Our species didn’t even exist a million years ago. The brains of our ancestor species were very different (in this context), that’s why all of them died out but our species survived.
In fact, even our species underwent an evolution of the brain around 50k years ago (I don’t remember the exact date). Some people say that sparked all of modern human innovation and the ability for humans to have languages, make complex tools, etc. There are many theories as to why this happened. Some people say it’s the diet, some say it’s the “Stoned Ape Theory,” and so on. : ) What’s more is that some people argue even in the recent millennia our brain has also advanced (mainly through a better diet iirc).


The brain cases for our ancestors and their ancestors before them, have been preserved going back farther than the 60kyrs for which we are aware of cave art and residue of human social activity. Those brain cases seem to accommodate similar shaped brains, most likely capable of most of what we now do although not yet exercised in literacy (as we know it) until after 6 thousand years.

Prior to social cultural life our ancestors had hunter gatherer challenges which engaged their brains extensively leading to food herd extinctions and migrations over the globe.

I think it is a mistake to assume that the brain evolved significantly in the last 5k years, one cannot even begin to rule out the effects population density resulting from breeding to the capacity of agricultural sustenance, and the results of playtime combined with sheer numbers of “our kind” interacting.

These were most likely not new phenotypic expressions of genetic mutations.

In any case consciousness is not human specific. it works with the kinds of data it gets, and we go from there.


@rhyolight just posted a link to papers slightly relevant to the current focus of this thread: