The Legendary Study That Embarrassed Wine Experts (2014)

Wine experts regularly identify wines when blinded but not misled. The "subtle flavors" are objectively there -- you can see them on a gas chromatograph. But the experts aren't there to be poor-quality GCs. They're there to identify wines that other people will think are interesting.

Some people find the language they use to describe it off-putting, but they're just trying to describe a sensory experience, using familiar language in unfamiliar ways. It's readily understood by their colleagues.

There's an objective difference between that and "power of suggestion" that's poorly conveyed by this article. I believe that's why the article is dated and lacks follow-ups. It doesn't say what people think it does, and researchers either didn't bother to look in more details or didn't think that what they found was as interesting.

I'll reiterate that these subtle differences don't have to be important to you to enjoy wine. People get attached to the high prices and snobbery of very high end wines. But that's not really what wine tasting is about, and it's absolutely not necessary to enjoy wine.

Obviously if you had a Barolo and a Chablis and asked people to pick out the white wine then people would easily manage.

"They may well be using LC-MS-MS time of flight on every sample now, for all I know, to identify and quantify every single organic molecule."

There seems no reason to believe this isn't possible. After all, insects (bees being the specific instance I'm aware of) can detect pheromones to phenomenally infinitesimal amounts, so it all hinges on the sensitivity of the receptors/sensors.

I recall hearing a talk on a radio science program several decades ago by a guy who invented one of the sensors for GC-MS. Can't remember all the specifics but I do recall his illustration, it seemed rather impressive to me at the time. That was if you gave him a thimble full of a chemical that wasn't in seawater and he placed the thimble in a body of water such as a large harbour, then he'd be able to detect traces of it a week later. He also referred to the year this resolution was possible which was 1957 (perhaps that was the patent date). Whether he was exaggerating or not I don't know.

If that was the state of the art 64 years ago then I'd not be surprised of the sensitivities you suggest. I'm also reminded that a few years back, I came across a book in a second-hand bookshop titled Pesticide Identification at the Residue Level American Chemical Society, Advances in Chemistry series 104 (1971). In Chapter 1 titled Possible Limits of Ultramicro Analysis, page 3 has a fascinating and very informative diagram (figure 1) titled Map of tracer cosmos. [Apologies if I'm stating the commonplace here, this isn't my day job.] It's a triangle-shaped diagram together with LHS and RHS axes. The LHS axis is labelled is Levels of Molecules and graduated from 10^0 at base of triangle to 10^21 at its apex, the RHS one is labelled Levels of Concentration and it is graduated from 10^-21 at the triangle base to 10^-3 (near but not at the apex). The x axis along the base of the triangle is labelled Maximum Number of Compounds at Each Level of Concentration, and it's graduated at 10^0 at the base's centre point (.i.e.: perpendicular from the apex) and the gradations extend to 10^10 on either side of that point (at the edges of the base of the triangle).

This is a wonderful diagram, I can't ever recall coming across it in all those years I studied chemistry. What's immediately obvious are myriads upon myriads of different chemicals one's likely to find at extremely low concentrations. This book also has an excellent section on GC-MS, detection of residues and related topics. The point of troubling you with all that is that the detection sensitivities shown in the book are now 50 years old and they seems remarkably good. For example, organophosphorus and organochlorine pesticides could easily be detected down to 0.05 microgram back then—mind you, that's still one hell of a lot of molecules whatever that number is (the compounds are given but I'm too lazy to apply Avogadro to find out).

I've often thought I like to own a GC-MS or LC-MS, I'd have a ball, at least until the novelty wore off. ;-)

"By the way, little if any of this is a "trade secret". The analyses are largely standardised across the industry (and validated using reference laboratories)…"

Yeah, makes sense these days with those analytical tools. That discussion is like the Coca Cola formula argument, Pepsi worked that out years ago, nevertheless they still want to differentiate.

"Smell is one of the most, if not the most, primitive senses in the body. While we can philosophise about whether individuals experience the world in the same way, and proving it one way or another is very difficult, if not impossible, a great deal of this basic sense is essentially "hardwired".

Right, qualia and the mind-body problem at their basics is 101 philosophy and they go back to Aristotle et al; at the discussion's 'high' end, the heavyweights are still slogging it out (and it's well over my head). In essence, comparing the differences in one's own 'sense data' [as per Russell, G.E. Moore terminology] and or against the sense data of others is nonsensical and must be equated out of any measurement discussion (there being no external reference to clamp to). As discussed, that's where training and repeated testing becomes essential if objective measurements/comparisons are needed.

"a researcher using genetically modified insects with fluorescently-labelled neurons, which lit up when activated but were also visible when inactive as well (IIRC they used two colours). they could visualise, in real-time, the olfactory receptors, the neural structure linking it to the brain, primarily the area associated with memory interesting here was that under the microscope."

You're right, that is fascinating stuff. One may not be able to measure a person's qualitative (perceptual) response but I can envisage a system that would allow one to quantitatively measure say the response to those synthetic musks to which I was referring earlier. Detecting that info would be useful for any number of reasons (as with Ishihara colour vision tests, it could be used for, say, a preliminary test for those who want to be perfumers).

"Each different olfactory receptor appeared to activate a different combination of neurons."

Apropos my comment immediately above, that presupposes that any analyser can distinguish between a learned response and an innate one. For instance, my example of turning myself off chocolate as a kid. If one couldn't differentiate that learned signal from my innate sense of chocolate smell then it would be pretty useless. Then, that perceptual analysis would no doubt be processed elsewhere in the brain (I think I've answered my own question here).

"And in biological systems, information transmission is often based on frequency encoding rather than simple on or off I wouldn't be surprised if it was the latter."

I'm not familiar enough with biological systems to make sensible comment about that except to say that if we definitely know there's a solid basis for the 'frequency encoding' argument, then could both be relevant if synapses process variable amounts of neurotransmitters (as they do)?

"I suspect our experiences will be largely the same bar genetic differences around selected receptors, kind of like red-green colourblindness. We will have a combination of innate and learned responses to given stimuli." "Maybe like red-green colourblindness you can compensate for it if other similar receptors are slightly stimulated? I don't know."

I've spent much time in electronic vision systems, television, image sensors, etc. and I can say that none of them can match the human eye in sensitivity and dynamic range let alone say a cat's eye, which is about five times as sensitive (it can detect a single photon). Modern electronic sensors are remarkably sensitive and we can get them to do great things in our smartphones. The trouble is that it takes a great deal of electronics to get them to do that. Essentially, image sensors have a linear transfer and they top out hard (saturate) at peak white (at the white clipping point a picket fence no loner has any detail in the whites), and at the bottom noise is a limiting factor and they can suffer a form of 'black clipping' somewhat akin to reciprocity failure in colour film) if not properly biased. Film emulsions [for a given aperture] and the human eye have better dynamic ranges, as their transfer curves are S-shaped and thus compress image data at both ends. With that longwinded lead in, it seems to me that most of our body's receptors have large dynamic ranges, essentially they use floating-point arithmetic to set dynamic range, levels etc. For example, I've been in situations where I've had to install monitoring cameras where the ambient light was highly variable, in fact up to 10^6 : 1. The human eye adapted AOK but the electronics was pretty R/S even after automatic aperture and automatic gain control compensation were wound to maximum effect.

Finally, I noticed you mentioned a matter that only the cognoscenti from specific parts of the planet understand—that of Marmite (long gone is anyone on the western side of the pond, either from boredom, or more likely from sheer disgust—and or perhaps because we've now switched to a foreign language. You obviously have some understanding of the subject or you wouldn't have mentioned it. There are many things I want to know about it, but I'll confine myself to the matter of its consistency. Can you enlighten me about that? How is it achieved? Presumably by some evaporative process but there has to be a lot more to it than that (i.e.: getting from yeast to that consistency would seem to involve many steps).