# Bees capable of high level cognition?

**URL:** <https://discourse.numenta.org/t/bees-capable-of-high-level-cognition/10083>\
**Category:** Lounge\
**Created:** [September 9, 2022, 11:47pm UTC](https://discourse.numenta.org/t/bees-capable-of-high-level-cognition/10083 "2022-09-09T23:47:39Z")\
**Posts on this page:** 8\
**Page:** 1

<div class="post-metadata">

**Author:** ![JarvisGoBrr](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.numenta.org/jarvisgobrr/32/6147_2.png) [@JarvisGoBrr](https://discourse.numenta.org/u/JarvisGoBrr)\
**Post date:** [September 9, 2022, 11:47pm UTC](https://discourse.numenta.org/t/bees-capable-of-high-level-cognition/10083/1 "2022-09-09T23:47:39Z")

</div>

what od you think about this?

looks like its not just mammalians, birds and octopuses that are capable of high level cognition?

[![](https://img.youtube.com/vi/exsrX6qsKkA/hqdefault.jpg "Bumblebees learn to roll balls for reward") ](https://www.youtube.com/watch?v=exsrX6qsKkA)

---

<div class="post-metadata">

**Author:** ![azat\_n](https://avatars.discourse-cdn.com/v4/letter/a/9fc348/32.png) [@azat\_n](https://discourse.numenta.org/u/azat_n)\
**Post date:** [September 10, 2022, 2:47am UTC](https://discourse.numenta.org/t/bees-capable-of-high-level-cognition/10083/2 "2022-09-10T02:47:47Z")

</div>

Would you really call this high-level ? Even a robot with reinforcement could earn that…

---

<div class="post-metadata">

**Author:** ![JarvisGoBrr](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.numenta.org/jarvisgobrr/32/6147_2.png) [@JarvisGoBrr](https://discourse.numenta.org/u/JarvisGoBrr)\
**Post date:** [September 10, 2022, 11:48am UTC](https://discourse.numenta.org/t/bees-capable-of-high-level-cognition/10083/3 "2022-09-10T11:48:35Z")

</div>

I always get the impression people underestimate insects.

What those bees just did was to look at a scene, parse a high level description of the objects in it, infer what affordances objects have, infer that some objects are actually agents that can act on their own, watch those agents perform high lwvel actions and learn new cause-effect relationships, then infer that they themselves can perform the same actions to get the same effect.

finally they take all this high level knowledge and convert into a complex sequence of low level leg movements to actually push the balL

The impressive part is the abcense of trial and error, they just observe and actually understand whats going on.

To me it seems way beyond any RL algorithm I ever seen.

---

<div class="post-metadata">

**Author:** ![cezar\_t](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.numenta.org/cezar_t/32/4451_2.png) [@cezar\_t](https://discourse.numenta.org/u/cezar_t)\
**Post date:** [September 12, 2022, 9:57pm UTC](https://discourse.numenta.org/t/bees-capable-of-high-level-cognition/10083/4 "2022-09-12T21:57:28Z")

</div>

> [@azat\_n](#):
>
> Would you really call this high-level ? Even a robot with reinforcement could earn that…

Yeah in approx. 1M trials. Animals can’t afford that so they evolved very quick learning algorithms.  
That needs to be benchmarked not how close they get to perfection (accuracy)

---

<div class="post-metadata">

**Author:** ![JarvisGoBrr](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.numenta.org/jarvisgobrr/32/6147_2.png) [@JarvisGoBrr](https://discourse.numenta.org/u/JarvisGoBrr)\
**Post date:** [September 12, 2022, 10:09pm UTC](https://discourse.numenta.org/t/bees-capable-of-high-level-cognition/10083/5 "2022-09-12T22:09:32Z")

</div>

even if those bees needed 100 trials to learn this, I’d still be confident to say that they are probably smarter than my dog.

---

<div class="post-metadata">

**Author:** ![cezar\_t](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.numenta.org/cezar_t/32/4451_2.png) [@cezar\_t](https://discourse.numenta.org/u/cezar_t)\
**Post date:** [September 12, 2022, 10:14pm UTC](https://discourse.numenta.org/t/bees-capable-of-high-level-cognition/10083/6 "2022-09-12T22:14:28Z")

</div>

I doubt it needed 100 trials, I don’t think they have that much … patience.

---

<div class="post-metadata">

**Author:** ![cezar\_t](https://yyz2.discourse-cdn.com/flex030/user_avatar/discourse.numenta.org/cezar_t/32/4451_2.png) [@cezar\_t](https://discourse.numenta.org/u/cezar_t)\
**Post date:** [March 27, 2024, 1:51pm UTC](https://discourse.numenta.org/t/bees-capable-of-high-level-cognition/10083/7 "2024-03-27T13:51:58Z")

</div>

Here-s a follow-up study, bumblebees learning to unlock food by watching bees that know how to. The trick is too difficult to figure out by themselves, so the first ones had to be trained with intermediate step rewards.

The cultural assumption might be a bit of a stretch, nonetheless they aren’t as stupid as previously assumed.

> **[What’s this buzz about bees having culture? Inside a groundbreaking experiment...](https://aeon.co/videos/whats-this-buzz-about-bees-having-culture-inside-a-groundbreaking-experiment)**
>
> We’re not the only animals that can learn complex, problem-solving behaviours from each other – bumblebees can do it too

---

<div class="post-metadata">

**Author:** ![BrainVx](https://avatars.discourse-cdn.com/v4/letter/b/ed8c4c/32.png) [@BrainVx](https://discourse.numenta.org/u/BrainVx)\
**Post date:** [April 16, 2024, 12:06am UTC](https://discourse.numenta.org/t/bees-capable-of-high-level-cognition/10083/8 "2024-04-16T00:06:47Z")

</div>

If the bee dance communicates an external flight path location to a source of food, then in comparison simple mirroring of a smaller pattern that looks like a lot shorter dance would seem more easy. i.e. the initial stick example to follow.

It’s possibly more about the mirroring feedback process (communication) and less of self determining what works or where the food is. The shortest distance is more of the self determination.

Are bees just really good at communicating the results of lots of trial an error / searching.