How tiny wasps cope with being smaller than amoebas • Hacker News Board

Re: How tiny wasps cope with being smaller than amoebas

#41

Post by agravier » Fri, Jul 06, 2012, 6:43 PM UTC

Xcelerate wrote: post #38

cs702 wrote: post #31

Wow: with only 7,400 neurons (compared to 340,000 for the common housefly and 850,000 for honeybees), this wasp can somehow fly, search for food, find the right places to lay its eggs, etc. That ridiculously tiny neural network is one freakingly efficient computing device!

7,400 seems very capable of being simulated. In fact, that's a small enough number that I'll bet you could map those connections out by hand with some type of microscope. To someone more knowledgeable in this subset of biology, is this possible?

It is possible, but the neurons are actually complex computing units that take a plethora of signals into account: subtle temporal behavior (relative timings of post and pre-synaptic activations), complex chemistry in the cell as well as the in the synaptic cleft, and many more less understood things. Secondly, as you can imagine, the connectivity is far from random. Like in larger nervous systems (or even more than in large pools or neurons), these computing units take precise roles, part due to the developmental process, part due to later stage learning. But in the end, the machinery is precise, and simulating it requires understanding all of it.

I believe that this understanding should not come though imagery alone, but also through studies of the development of these organisms' brains. Like we are now forming artificial neural maps through "machine" learning techniques that are accelerated versions of their biological counterparts, I believe that we should develop developmental algorithms into large-scale simulations. The goals being to first understand how it works, second to see if the models are able to replicate the end result, a good indication that they are useful models.

(I do research in computational neuroscience.)

Re: How tiny wasps cope with being smaller than amoebas

#42

Post by ChuckMcM » Fri, Jul 06, 2012, 6:49 PM UTC

cs702 wrote: post #31

Wow: with only 7,400 neurons (compared to 340,000 for the common housefly and 850,000 for honeybees), this wasp can somehow fly, search for food, find the right places to lay its eggs, etc. That ridiculously tiny neural network is one freakingly efficient computing device!

I was going to say the same thing, coding for such behaviors has to either be pretty simple, or there is more going on here than just neurons.

Re: How tiny wasps cope with being smaller than amoebas

#43

Post by grannyg00se » Fri, Jul 06, 2012, 7:07 PM UTC

cs702 wrote: post #31

Wow: with only 7,400 neurons (compared to 340,000 for the common housefly and 850,000 for honeybees), this wasp can somehow fly, search for food, find the right places to lay its eggs, etc. That ridiculously tiny neural network is one freakingly efficient computing device!

I'd love to see this flight in action. I would expect that at their size, they are pretty much limited to going where the primary air current brings them. With perhaps a tiny amount of individual influence.

Re: How tiny wasps cope with being smaller than amoebas

#44

Post by Someone » Fri, Jul 06, 2012, 7:22 PM UTC

cs702 wrote: post #31

Wow: with only 7,400 neurons (compared to 340,000 for the common housefly and 850,000 for honeybees), this wasp can somehow fly, search for food, find the right places to lay its eggs, etc. That ridiculously tiny neural network is one freakingly efficient computing device!

"this wasp can somehow [...] find the right places to lay its eggs"

I bet this wasp is playing the "you do not need eyes to hit bullseye if you throw a million darts" game. It is sufficien t if some of these wasps, almost by accident, happen to lay their eggs in the right place.

Re: How tiny wasps cope with being smaller than amoebas

#45

Post by Apocryphon » Fri, Jul 06, 2012, 8:07 PM UTC

Is that wasp truly really small? Or is that amoeba just really big?

Re: How tiny wasps cope with being smaller than amoebas

#46

Post by excuse-me » Fri, Jul 06, 2012, 8:09 PM UTC

raganwald wrote: post #20

excuse-me wrote: post #13

Earlier quoted context omitted.

Well as long as you make it clear that they were created that size because God likes a challenge - rather than any sort of adaptation to their environment - that should be OK

Hmmm, downvotes. Hey Fred! Get in here, and bring the spare sensor filaments. I think we have some sarcasm detectors on the fritz. https://en.wikipedia.org/wiki/Poes_law

Doesn't HN let you see a poster's other posts/comments? My background should be pretty obvious!

ps How many christian fundies post on HN anyway?

Re: How tiny wasps cope with being smaller than amoebas

#47

Post by yoklov » Fri, Jul 06, 2012, 8:28 PM UTC

ChuckMcM wrote: post #42

cs702 wrote: post #31

Wow: with only 7,400 neurons (compared to 340,000 for the common housefly and 850,000 for honeybees), this wasp can somehow fly, search for food, find the right places to lay its eggs, etc. That ridiculously tiny neural network is one freakingly efficient computing device!

I was going to say the same thing, coding for such behaviors has to either be pretty simple, or there is more going on here than just neurons.

While I cannot say for certain that neurons are not the only thing responsible for these behaviors, I'm fairly confident in saying that we don't know nearly enough about neurons to assume that 7,400 equates to "pretty simple".

Re: How tiny wasps cope with being smaller than amoebas

#48

Post by Splines » Fri, Jul 06, 2012, 8:29 PM UTC

zb wrote: post #34

raldi wrote: post #10

What are the major technical barriers before we can identify the input and output channels to this insect's brain and start iterating through all possible input values, recording the corresponding output values? And once we can do that, could we use that data to fly a virtual insect around a virtual world?

http://www.nytimes.com/2011/06/21/science/21brain.html?pagew... This article suggests that it's computationally infeasible even for an organism with only 302 neurons, all of which have already been completely mapped.

Oddly enough, there's another story on the front page that address this guy:

http://news.ycombinator.com/item?id=4208454

Re: How tiny wasps cope with being smaller than amoebas

#49

Post by dllthomas » Fri, Jul 06, 2012, 8:56 PM UTC

yoklov wrote: post #47

ChuckMcM wrote: post #42

Earlier quoted context omitted.

I was going to say the same thing, coding for such behaviors has to either be pretty simple, or there is more going on here than just neurons.

While I cannot say for certain that neurons are not the only thing responsible for these behaviors, I'm fairly confident in saying that we don't know nearly enough about neurons to assume that 7,400 equates to "pretty simple".

It's probably "pretty simple" compared to 340,000.

That said, I wonder whether some of the complexity was effectively pushed down into complicating behavior of the individual neurons. If there were strong evolutionary pressures pushing down the number of neurons (as seems likely, for so extreme a reduction) it'd be interesting to see what kinds of hacks were included to get it there.

Re: How tiny wasps cope with being smaller than amoebas

#50

Post by vibrunazo » Fri, Jul 06, 2012, 9:32 PM UTC

raldi wrote: post #10

What are the major technical barriers before we can identify the input and output channels to this insect's brain and start iterating through all possible input values, recording the corresponding output values? And once we can do that, could we use that data to fly a virtual insect around a virtual world?

Honest newbie question: Why would we even want to do that? Exclusively for academic medical purposes? (ie learning how our body works)

It seems to me, that since evolution is highly imperfect, that trying to mimic living beings, might not be the best idea. While we do mimic them a lot, mostly it seems to be for learning until we can do better. But in this case. Since it's so expensive and not viable to mimic them properly. Shouldn't we just try to come up with our own algorithms instead of trying to copy nature's historically bloated algorithms?

That seems to be a lesson we learned from machine learning. I remember when neural networks were first being talked about seriously a decades or so ago. The goal seemed to be to try to mimic neurons just for mimicing neurons sake. Many critics would say we should probably try other learning algorithms that were obviously more efficient. And nowadays we barely see actual neural networks being used seriously in commercial production simply because we have better algorithms that are not trying to emulate nature, just because.

Wouldn't trying to emulate a virtual insect be repeating the same mistake? If all we want is to design a virtual robot that can look for food and control its wings. I sure as hell don't need thousand something emulated neurons to build that.