Hacker News new | past | comments | ask | show | jobs | submit login
Our Mind-Boggling Sense of Smell (nautil.us)
28 points by prostoalex 2 days ago | hide | past | favorite | 11 comments

> it takes only two synapses for information to travel from the air, through your nose, and to the core cortex

I was expecting it would be just a low level feature detector since smell is like like a bag-of-words format (unordered) while vision is at least 2D (space) and hearing 1D (time).

One fascinating thing about smell is that you can smell only on inhalations. On exhalations, the sensors shut down. Hard to figure out what the evolutionary benefit of that was. But so it is for humans, and I believe for most mammals.

I suppose it makes sense, otherwise the smells from our breath and most recent meals digestion would mix with and confuse our sense of the smells surrounding us.

I'm definitely glad we have this feature first thing in the morning before brushing my teeth!

Dogs noses have those slits in the side to move exhaled air out to the side without disturbing what's in front of the big nostrils.

Fun to watch this work when my dog is ratting for voles in the yard.

  > And where do you put indole, an unpleasant smelling
  > crystalline organic compound found in coal tar and
  > feces?
Dilute indole actually smells quite pleasant and is widely used in perfumery.

Civet is also a perfume note, but I'll still take your word for indole.

Makes a great case that smell is very different than our “favorite” sense, vision. I heard her interview on Sean Carroll’s podcast, was super interesting.


Smell is the primary sense and the sooner that Computer Vision embraces that fact the better it will be for any system that needs any form of context sensitivity or relevance detection.

It's all well and good to have awesome algorithms for detecting X, Y, or Z in a stream of stimuli but it's smell that tells us which things to anticipate.

There is no "primary sense". It's primary only in terms of detecting molecules, but the presence or absence of molecules is just one environmental dimension.

If the sense of smell is pattern recognition, would training it make us better at recognizing patterns?

Feynman trained smell, but I don't recall if he mentions any practical payoff.

Bonus clip: https://www.youtube.com/watch?v=MvhwpUiQXdY

Guidelines | FAQ | Support | API | Security | Lists | Bookmarklet | Legal | Apply to YC | Contact