The Power of Analogy and Artificial Intelligence

A “parquet deformation,” from Douglas Hofstadter’s Metamagical Themas, 1985

A la Douglas Hofstadter, the metamathematical wizard, the “man who will break your spell check,” the power of analogy gives artificial intelligence the edge it needs to squeeze us out of the equation. A new artificial cognition design technique known as ---  “structure-mapping engine (SME), the new model is capable of analogical problem solving, including capturing the way humans spontaneously use analogies between situations to solve moral dilemmas.” (phys.org 2016)

Fluidity, intuition, morality – these are things that are hard to come by in computers these days. But one of the things that allows us think in this way is the power of analogy.

And not only is the “language of smell” an entirely metaphorical concept – both in the way it uses other senses to generate its descriptions (sweet, sharp), and in the way the smells are described (orange is like lemon, but not like butter; peach is somewhere between fruity and butter) – but more importantly, this phenomenon, the language of smell, is a powerful analogy for the primitive mind. Teach a robot to smell, and we’re out of here.

Notes:

Structure-mapping engine enables computers to reason and learn like humans, including solving moral dilemmas

phys.org, June 2016

The Ever Impending Electronic Nose

CMOS Sensor, image source

The electronic nose has been coming forever. We already have artificial noses in the form of mass spectrometers. The new kind use integrated circuits, the same CMOS chips that are in cellphones. They aren’t tied to a thinking, feeling human, but at least the initial step of identification is happening.

This article in particular is touting the use of an artificial nose to analyze breath samples, citing that “breaths contain gases from the stomach and that come out of blood when it comes into contact with air in the lungs. The breath test is a blood test without taking blood samples. Breath contains information about practically every part of a human body.”

This is good stuff, and explains why your breath starts to smell funny when you’re hungry, or why diabetic-breath smells like acetone, but I must keep going and repeat the sales pitch of this scientist:

"If you think about the industry around sensors that emulate our senses, it's huge," said Dr. O, also a professor in the Erik Jonsson School of Engineering and Computer Science and holder of the Texas Instruments Distinguished University Chair. "Imaging applications, hearing devices, touch sensors—what we are talking about here is developing a device that imitates another one of our sensing modalities and making it affordable and widely available. The possible use of the electronic nose is almost limitless. Think about how we use smell in our daily lives."

But yes, let’s think for a minute about how we use smell in our daily lives. Smell is so below the radar that we don’t consciously register most of its ongoings. So when we take this CMOS sensor to a brain, what exactly would we like it to do? Of course we can’t engineer the olfactory bulb itself, or the subsequent limbic system (in concert with our memory) that ultimately creates our experience of smell. But if we were to isolate it, and use it for specific things, like breath analysis, then what else would it do?

Gas leaks, obviously. Maybe it could alert my roommate to change the litter box? Or tell a parent that their teenager was drinking last night (that’s a simple breathalyzer built into the air system of the house; watch out kids). You would think we’d already have one to smell the maple syrup smell coming from the burning transmission fluid in your engine. And what else? Can I smell the presidential candidates with my CMOS supernose? Who knows. Smell a house on Zillow? Let’s bring it to Japan and see what they do with it.

Post Script:

On thinking about this further, I must stress the difference between what I’ll call active smelling and passive smelling. What we do as humans is passive smelling. For the most part, we are not actively looking for a particular smell. (And, in fact, this is no way to smell, unless you’re a fragrance artist.) It’s more like the Tao of Perception – you must have a “soft awareness” where you are ready for any smell, but you’re not actively smelling for anything in particular. That’s just the way smell works. So this is passive smelling.

Active smelling would be a sensor fitted for one particular odor compound. Acetone, for diabetic-breath, for example. Or it may be fit for a bunch of things on top of that. But you could never fit the sensor for all potential smells. That’s not how the human epithelium works. We would need an artificial olfactory bulb for that, because the bulb turns our 450 receptors into the trillion potential smells available. And further, to attach that to meaning, we would need, again as mentioned above, an entire body, and more than that, a body that has lived from birth. A Frankenstein created at 20 years old would not be able to smell. This must be a baby Frankenstein we’re talking about here, zygote even.

In closing, we can’t have an electronic nose that is “open ended.” It can look for particular things on Zillow – mold? Wet paint? (who cares) Frito Feet? (that’s the smell of dog feet, which could indicate a pet lived in the house?). Thanks for listening.

Notes:

Electronic nose using CMOS integrated circuits analyzes breath

phys.org, June 2016

The Pumpkin Conundrum

aka What Do Pumpkins Actually Taste Like?

source: Maniac Pumpkin Carvers

It has come to the attention of the cultivorous behemoth that is Consumerism that using pumpkins in autumn is a really good idea. It’s in your cookies, your coffee, and for god’s sake it’s in your condoms. And in all its prestigitory splendor, the big C has even managed to remove the actual pumpkin from the formula, giving us a great example of the Lingua Anosmia at work: There is no pumpkin in pumpkin flavor.

It doesn’t take long for a hyper-mediated society to figure this stuff out, and so Starbucks has committed to using real pumpkins in their recipe. The change came about more as a response to the Natural Food movement than a recognition of the irony in this bogus beta-carotene bonanza.

So what’s the deal; where does this celebrity squash come from? It starts with pumpkin pie, which via whipped cream goes awesome with coffee. You might just like the taste, or you might like the taste because you like Thanksgiving in general. One thing is for sure, you don’t like it for the pumpkin. Anyway, coffeeshops are as much about their product as they are their placeness. The experience has to be as good as the coffee, and what better way to make people comfortable than to remind them of Thanksgiving. And so it begins, the pumpkin-flavored coffee. But as a flavor, what exactly is it?

The flavor that has revolutionized autumn’s bottom line is Pumpkin Spice, and not Pumpkin proper. This refers to the ingredients most often used to flavor pumpkin pie, and they are cinnamon, ginger, nutmeg, and clove. Chances are, when you hear “pumpkin flavor”, you can expect these four ingredients. As my friend the chef pointed out, apple pie and pumpkin pie both use the same four spices, only in different ratios. So not only is Pumpkin Flavor not Pumpkin, it could almost be Apple.

But what about the pumpkin sans spice? Let’s combine a couple sources, The Good Scents Company and Sigma Aldrich. (I thought it noteworthy that in their circa 2012 catalog, Sigma Aldrich doesn’t index pumpkin as an organoleptic descriptor for any of their chemicals.)

Looks like there are only two chemicals in relation to pumpkin. The first is sorbyl acetate, its more formal name is trans,trans-2,4-Hexadienyl acetate. Good Scents describes its properties as fresh green, oily, herbal, pumpkin, fresh parsley, soapy, metallic. Sigma Aldrich describes it as pineapple, sweet, and wine-like. If it all makes sense to you, please leave your input in the comments section below(!). All I can add is that Pineapple and Sweet are quite the ubiquitous organoleptics in the SA catalog.

Next is (E,E)-2,4-decadienal, which Good Scents describes as oily, cucumber, melon, citrus, nut, meat, fatty, chicken, aldehydic, green, fried, and potato. Sigma Aldrich indexes the chemical as one of the "other" aromas in the following categories: Meaty, Citrus, and Fatty. Many, if not all of their categories have a sublisting “other”.

This is where we find Pumpkin, in the Other category. After all, “other” is the most favored of the Lingua Anosmia.

Post-Script

Pumpkin pie filling tends not to be made of the quintessential pumpkin, but a similar variety like butternut squash.

Where do you stand in the pumpkin spice debate?

nj.com, Sep 2015

On Orange Juice and Rancid Butter and the Volatility of Aroma Compounds

Snippets from Hidden Scents – this one is about the volatility of aroma compounds:

Were they to be artificially isolated, or even left on their own, their state might be subject to change depending on the conditions. Normal amounts of oxygen in the air are enough to change a sensitive odor molecule into something very different; such reactions are a major consideration in food design and artificial flavoring. Terpenes, which comprise an important element of citrus-smell, oxidize over time, changing the relative proportion of the overall mixture, and moving from “citrus” to “turpentine.” Moving in the opposite direction, butyric acid, which smells like the rancid butter from which its name is rendered, can react with ethanol to become ethyl butyrate, which smells of pineapple. Ethyl butyrate is then used as a primary ingredient in artificial orange juice flavor. Only with careful consideration can an odor be named with any degree of accuracy. The name type is just as important: Is this a reference to a specific chemical within the orange profile, or to the entire smell-of-orange as it occurs in the pureness of its essential oil?

***

In practical terms, when your OJ goes bad, it’s not your imagination, it really does smell like turpentine. And when you ask for pineapple juice at the bar, and they pull it out in that big metal can that was opened yesterday, and you all of the sudden smell rancid butter, it’s not the bartender. And to use molecules as a basis for organizing smells is not as useful as it seems.

On Sweetness and Flowers

finedininglovers.com

"French scientists identified a gene that's far more active in a heavily scented kind of rose than in a type with little odor. This gene, which produces an enzyme, revealed the odor-producing process."

AKA "Biosynthesis of monoterpene scent compounds in roses"

Roses have not been designed over centuries of human endeavor to smell good, but to look good. Roses may have been initially valued for their heavenly scent, but over time, they have been selected for reasons of visual beauty.

A selection from Hidden Scents:

Crossing as often as it does with the Food Industry, Fragrance is often interpreted to the mind as taste. “Sweet” is frequently used as an olfactory descriptor. However, calling caramel-smell Sweet is not a chemical consequence of the molecule, but an effect of memory. If sucrose is paired with an odor, any odor, it will eventually “smell sweet”. The potential effect on odor perception that is cued by such multi-sensory information is thought to be the minimization of “perceptual dissonance between the dominant sense (vision) and the minor sense” (Wilson & Stevenson 2006).

Smells are not just Sweet; Castoreum is Umami, and Body Odor, Sour. Yet something about the word, the taste, and the smell of “Sweet” is linked together, and can be evidenced in the fact that the word “sweet” is the most commonly used sub-category heading in the Aldrich catalog (2013), which is a chemical clearinghouse for Flavor and Fragrance scientists. The biases and idiosyncrasies inherent in the Aldrich Catalog as a universal aroma reference are immediate and comprehensive. The catalog is a flavor tool, first of all, and as a commercially-driven pursuit flavor reigns supreme over fragrance alone. The fact that “Sweet” as a sub-category has the most centrality (it falls under Balsamic, Woody, Fruity and Minty) could relate to a confluence of the pervasive availability of sweet aromas, or the sweet-tooth of a certain society, or that there really is something sensually similar about the relative phenomena. Unresolved in that matter, we can scrape-together at least that the description of a smell will tend to be communicated via other, more intentionally-simulable channels of association – anything but the olfactory-specific semantic network of experience, because of course, such a thing is so halting and reluctant in its cooperation, if at all.

In parallel to this gustatory congruency of Sweetness, it is a curious thing to consider: are Florals such an extensive part of the perfumer’s repertoire because of their associated visual aesthetic beauty? David Howes responds in telling of his work with the Papa New Guinean Kwoma people (Howes 2002:76). He presents plastic cards impregnated with scent, and painted with a corresponding color (cinnamon is brown, coconut is white, etc.). When surveyed on the most-liked odor, respondents voted Rose in the vast majority. After having hidden the red-colored Rose scent in an envelope, however, the overwhelming preference disappeared (along with Howes’ search for a universal aroma preference, à la Berlin and Kay, 1969).

Notes:

Berlin B & Kay P (1969). Basic Color Terms: Their Universality and Evolution. Berkeley: University of California Press.

Howes D (2002). Nose-wise: Olfactory Metaphors in Mind. In: Olfaction, Taste, and Cognition, ed. C Rouby, B Schaal, D Dubois, R Gervais, & A Holley, pp. 67-81. Cambridge, U.K: Cambridge University Press.

Sigma-Aldrich (2013). Aldrich Chemistry 2012-2014: Handbook of Fine Chemicals. Sigma-Aldrich.

Wilson D A, Stevenson R J (2006). Learning to Smell: Olfactory Perception from Neurobiology to Behavior. Baltimore: Johns Hopkins University Press.

The Flesh of the Anthroposphere

BrainGate

The title here is in reference to the human flesh search engine, and only in theory, not in practice. The ‘flesh engine’ is used for public humiliation, but the idea of it is simply the use of a distributed process using humans as opposed to computers – not artificial intelligence exactly, but an artificially-mediated human intelligence. Nonetheless, this organizing agent, the whole of a particular population, produces a unique database of smell-descriptions that defies our contemporary notions of a “searchable database” to create the dirtiest of dirty data. 

Snippets from Hidden Scents; this one is in relation to the way People organize the names of smells, as compared to the science of chemistry or the fragrance industry:

In this compartment of olfactory classification, smell is organized by the human organism at large, the flesh of the anthroposphere, if you will. Whereas scientific taxonomies and fragrance industry inventories are themselves products of human endeavor – created for humans by humans – the (unconscious) categorization of odors by the general population is a separate system altogether.

I Could Have Been a Fragrance Millionaire Jerry

AKA Calvin Kleinfleld

Smell my arm!

Actually, he didn’t have a chance. By the time Kramer came up with his idea, it had already been in the works for a few years – in real life, that is. Calone’s patent protection ended in 1989, and by 1990 the first Oceanic scent was on the market.

Some folks might be familiar with the 1992 episode of Seinfeld – The Pick – where Kramer’s fragrance epiphany gets swiped by a Calvin Klein marketing executive.

Kramer originally has the idea for a cologne that makes you smell like “the beach” in a previous episode – The Pez Dispenser – upon returning from a midwinter’s Polar Bear plunge. He calls it The Beach and pitches it to a CK exec who dismisses Kramer for a nutjob.

“Do you think people are going to pay eighty dollars to smell like dead fish and seaweed? That’s why people take showers when they come home from the beach.”

Next episode, it turns out they stole Kramer’s idea. He whiffs it on a supermodel. “You smell like the beach. What's the name of that perfume you're wearing?”

“The Ocean, by Calvin Klein”

That’s what it’s called – The Ocean. In the episode, that is. In real life however, the scent Kramer is looking for is called Calone (not a misspelling). It is one of the most important aroma compounds ever discovered, and represents not only an entire decade of men’s fragrance, but the advent of the “unisex” perfume market.

Calone is a tradename for a chemical that smells like the ocean. It was discovered by a pharmaceutical company researching benzodiazepine derivatives for anti-depression meds. When the researchers took notice of the chemical’s aromatic quality, they handed it over to a fragrance company to refine, and to name, and thus we have Calone – its chemical recipe hidden from the public due to intellectual property rights.*

Calone can be found in nature; it is a secondary metabolite of seaweed pheromone. However, it exists in the world of fragrance only as a synthetic compound made in a lab. Calone stands out, in fact, because it is the first synthetic chemical to be used in its synthetic form before its natural form (and to this day, nobody is trying to extract seaweed sweat in lieu of chemical synthesis).

Again, this is not to say that Calone doesn’t exist in nature, only that the fragrance industry in all its history has never used it that way. In fact squared, Calone was so new, and so unique, and so groundbreaking of a discovery, that it was given its own category amongst only four others in the total of all fragrance families. This, the Oceanic/Ozonic/Marine class of fragrance aromas, appeared circa 1990, and is known for its use in the new style of androgynous, “watery” scents of that time. (Calvin Klein’s Eternity and Davidoff’s Cool Water, both 1988, are prime examples, as is Giorgio Armani’s Acqua Di Gio 1996. And now, I can’t resist mentioning the original Wu-Tang fragrance of 1998 was described as simply a "Cool Water scent.")

Synthetic aromatic molecules are jackpot for the company who discovers them, but also for those down-the-line who can take advantage. It means they can make the same scent for a fraction of the price of the natural equivalent. In this case, the synthetic discovery had no natural equivalent – it was an entirely new smell. Heads went Calone-crazy and flooded the market. A tsunami, if you will. Calone is such a distinct olfactory representation of an era that its scent can reliably be referred to as “almost any man's fragrance since 1995 that comes in a blue bottle.”

And so with that, we should find it ironic (how appropriately Seinfeldian) that Kramer’s million dollar idea was about to become the scent of the decade.

Post Script:

I’d like to list some of the terms used to describe Calone, beginning with what Kramer would have called it:

The Beach

The Ocean

Sea Breeze

Marine-like sea spray

Ozonic (though ozone is odorless)

Fresh

Aquatic

Mossy

Seaweed

Dead Fish (but only Seinfeldian Calvin Klein executives call it that)

“generic guy smell” (later 1990’s)

“that androgynous smell” (earlier 1990’s)

methylbenzodioxepinone

CAS No 28940-11-6

Watermelon Ketone

(it doesn’t sound right, but Calone is related to a chemical found in melons)

POST POST SCRIPT:

*Discovered in 1966 by chemists from the pharmaceutical company Pfizer, who passed on their discovery to the perfumers of Camilli Albert Laloue in Grasse - here the substance got its name - and yet it should again take 20 years (1989 ended the patent protection) for calone could begin its triumphal procession. With New West by Yves Tanguy the fragrance became famous. [source translated from German] Anyone who knows where to find the formal documentation of this, feel free to make contact.

As told by one perfume blogger via a lecture at the Osmotheque by Yves Tanguy (the perfumer, not the Surrealist), the first use of Calone was in New West for Aramis 1990, which was simply Cristalle de Chanel 1974 plus Calone. That’s it.