🧵 "The Smell of Inhibition. A Code in the Nose?"

ICYMI, this week's SFI Seminar by Fractal Faculty Stuart Firestein (@Columbia) on "what started out ass a very simple-seeming problem [re: #olfaction] and turned out to be very complicated":

"Everything we know about the world comes through these little holes in our head and the skin covering our body, processed through tissue specialized to interpret it."

"The thing to notice about [sight and hearing] is that they're [processing] fairly low-dimensional stimuli."
"Even a simple smell is composed of a VARIETY of molecules, and these are high-dimensional from a chemical point of view. And it's also a somewhat discontinuous stimulus. How do we get from this bunch of molecules to this unitary perception of something like a rose?"
"You could make the argument that humans have the most discriminating palate of any animal on the planet — including dogs. If you have a dog, you know it'll eat nearly anything."

Stuart Firestein on the olfactory system as a model for understanding sensory processing:
"There are only three kinds of neurons in this system. Like many sensory neurons in other systems, they're these kind of ciliated structures...as far as we know there are no interactions between these cells until they get back to the olfactory bulb and synapse."
"Normally a chloride channel would be inhibitory, but olfactory neurons do this cute gthing where they actually pump chloride into a relatively high intercellular concentration, so these channels are actually chloride defluxes. It's a built-in amplifier, as it were."
"We wanted to ask a kind of a silly but fundamental question: What actually is an odor?"

"There are some very odd bits of business that go on with these odor molecules...for some of them, the only difference is that they rotate polarized light in one direction or another."
"These two odor molecules differ only by one carbon atom, and yet Hexyl acetate smells like bananas and Heptyl acetate smells like pears. On the other hand, here are two molecules that don't look ANYTHING LIKE each other, and yet smell IDENTICALLY like sandalwood."

• • •

Missing some Tweet in this thread? You can try to force a refresh
 

Keep Current with Santa Fe Institute

Santa Fe Institute Profile picture

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

Read all threads

This Thread may be Removed Anytime!

PDF

Twitter may remove this content at anytime! Save it as PDF for later use!

Try unrolling a thread yourself!

how to unroll video
  1. Follow @ThreadReaderApp to mention us!

  2. From a Twitter thread mention us with a keyword "unroll"
@threadreaderapp unroll

Practice here first or read more on our help page!

More from @sfiscience

Mar 10
🧵 "#Possibility Architectures: Exploring Human #Communication with Generative #AI"

Today's SFI Seminar with ExFac Simon DeDeo @LaboratoryMinds (@CarnegieMellon), streaming now:
"A key feature of this is talk is that we make sense of what each other are saying IN PART by what they say, but ALSO by what we expect of them."
"Language transmits info against a background of expectations – syntactic, semantic, and this larger cultural spectrum. It's not just the choices of make but [how] we set ourselves up to make later choices."

@LaboratoryMinds re: work led by @clairebergey:
Read 15 tweets
Mar 9
"I think what really drives [the popularity of the #multiverse in #scifi] is regret... There's a line in @allatoncemovie where #MichelleYeoh is told she's the worst version of herself."

"I don't think we should resist melting brains. I think we should just bite the bullet."
"When you measure the spin of an electron, or the position...what happened to all of the other things you could have seen? Everett's idea is that they're all real. They all become real in that measurement."
- SFI Fractal Faculty @seanmcarroll at @guardian
theguardian.com/science/audio/…
"At the level of the equations there is zero ambiguity, but the metaphors break down. The two universes it splits into aren't as big as the original universe. The thickness of the two new universes adds up to the thickness of the original universe."
Read 4 tweets
Dec 14, 2022
"Compositionality in Vector Space Models of Meaning"

Today's SFI Seminar by @marthaflinders, streaming:


Follow this 🧵 for highlights!
"Scientists gather here
Santa Fe Institute, oh so near
Inquiring minds seek truth"

#haiku about SFI c/o @marthaflinders & #ChatGPT

...but still, #AI fails at simple tasks:
"One way to represent the kind of #compositionality we want to do is with this kind of breakdown...eventually a kind of representation of a sentence. On the other hand, vector space models of #meaning or set-theoretical models put into a space have been very successful..."
Read 14 tweets
Dec 13, 2022
"Humans are prone to giving machines ambiguous or mistaken instructions, and we want them to do what we mean, not what we say. To solve this problem we must find ways to align AI with human preferences, goals & values."
- @MelMitchell1 at @QuantaMagazine:
quantamagazine.org/what-does-it-m…
“All that is needed to assure catastrophe is a highly competent machine combined with humans who have an imperfect ability to specify human preferences completely and correctly.”

- Stuart Russell (@UCBerkeley) as quoted by @MelMitchell1 in her latest @QuantaMagazine article
"It’s a familiar trope in #ScienceFiction — humanity threatened by out-of-control machines who have misinterpreted human desires. Now a not-insubstantial segment of the #AI research community is concerned about this kind of scenario playing out in real life."
- @MelMitchell1
Read 6 tweets
Dec 12, 2022
"Training Machines to Learn the Way Humans Do: an Alternative to #Backpropagation"

Today's SFI Seminar by Sanjukta Krishnagopal
(@UCBerkeley & @UCLA)

Starting now — follow this 🧵 for highlights:
Image
"When we learn something new, we look for relationships with things we know already."

"I don't just forget Calculus because I learned something else."

"We automatically know what a 'cat-dog' would look like, if it were to exist."

"We learn by training on very few examples." Image
1, 2) "[#MachineLearning] is fundamentally different from the way humans learn things."

3) Re: #FeedForward #NeuralNetworks

"You choose some loss function...maybe I'm learning the wrong weights. So I define some goal and then I want to learn these weights, these thetas." ImageImageImage
Read 13 tweets
Nov 30, 2022
🧵 "#Criticality: A Balance Between #Robustness and #Adaptability"

Today's SFI Seminar by @cgershen @cgg_mx:
Image
Regarding systems at the #EdgeOfChaos
Identifying #PhaseTransition dynamics
Citing @WagnerEvolution on #Robustness
Contrasting it with #Chaos (fragile, hard to predict, etc.)
And then #Criticality, somewhere in the middle... ImageImageImageImage
With #criticality, due to #PowerLaw(-like) distributions:

"Most of the changes will be small, but for some, you can have large changes."

See also #fractals

Key history point: Stuart Kauffman's #random #Boolean #networks (1969!)

Updated by @cgershen in 2004 ImageImageImageImage
Read 12 tweets

Did Thread Reader help you today?

Support us! We are indie developers!


This site is made by just two indie developers on a laptop doing marketing, support and development! Read more about the story.

Become a Premium Member ($3/month or $30/year) and get exclusive features!

Become Premium

Don't want to be a Premium member but still want to support us?

Make a small donation by buying us coffee ($5) or help with server cost ($10)

Donate via Paypal

Or Donate anonymously using crypto!

Ethereum

0xfe58350B80634f60Fa6Dc149a72b4DFbc17D341E copy

Bitcoin

3ATGMxNzCUFzxpMCHL5sWSt4DVtS8UqXpi copy

Thank you for your support!

Follow Us on Twitter!

:(