I had this discussion with a friend the other day. He argued that you could perhaps train the dog to identify a specific characteristic of the wine. He thought that a dog couldn’t take all of the information about a wine to pick out a good one. You could teach it for example to pick out a wine that has been aged in a particular wood.
Thoughts?
Mice have been trained to distinguish red wine from rose and a few can even discern different types of red wine.
But training a dog to pick out a “Good” wine seems like it would be difficult, simply because a dog’s olfactory is so much better than that of humans. Furthermore, mixtures of compounds can have qualitatively different smells than when they are smelled separately, by humans at least.
Statistics have some success in choosing wine though. Orley Ashenfelter established that a formula estimated on a few parameters based on the growing conditions of the wine did a pretty good job of predicting wine quality (as measured by prices).
Im not going to deny it isn’t possible but one would first have to decide what makes a good wine…I guess it would be hard to train a dog for all the different smells, aka chemicals etc…
But hey, then can smell drugs, all kinds of them so why not Wine. But I suppose it would be hard to distinguish …
Even if the dog couldn’t be trained to identify common characteristics across a spectrum of different wines considered “good”, couldn’t you train the dog to single out a particular wine that’s agreed to be good? If one, why not two? And from there it’s just a matter of memorization.
You might have more success with the idea of training a dog to detect bad wine. In so far as it could be trained to detect the set of well known wine flaws. Unfortunately this isn’t going to help you find a good wine, just find an undrinkable one and avoid it.
A good wine isn’t just about the presence of things - it is about the balance. And for an animal with a vastly superiour sense of smell, trying to work out how to teach it what to ignore as well as what the balance of the residual things it should recognise in an aesthetic mix that appeals to humans is going to be impossible. Poor dog is going to be thinking, “Well that last wine didn’t smell too bad at all to me, but this other one this idiot human thinks is so good, smells like cat’s piss. In fact it smells like that stinky 15 year old tom I chased up a tree last week. What the heck are these guys drinking?” Worse, how are you going to teach an animal whose idea of heaven is rolling around in something that has been festering in your houshold rubbish for the last two weeks, that a bottle of 2002 Grange is to be preferred to Jacob’s Creek, or tell eau d’ chat from Chateau.
Dogs are simple. They aren’t making value judgements, especially based on human systems. You can teach a dog to judge “presence or absence” of certain scents, but to rate, rank and evaluate complex chemical mixtures is going to be impossible.
I agree, you could pretty easily teach a dog to point out wines that have gone truly bad, but I’ve been drinking wine (and training dogs!) for 20+ years, and I don’t always agree with others on what is “good” or “eh”.
Pigeons learned to distinguish good art from bad. From that, I’m guessing that the answer is “yes”.
Though, I will note that if you test humans they learn to like and prefer what they usually drink be it cheap or expensive. Quality in this case appears to be subjective. You’d be teaching the dog to distinguish expensive wines, which while of a consistent makeup, is basically considered “better” because we’ve decided so.
You can train a dog to find the absence of a scent? That’s pretty cool. How do you do it - the same way you teach them the presence?
Ironically, wine prices are a poor predictor of quality. 
Expensive wines are only perceived as better when you know the price. In a taste test when you don’t know the price, $100 wines are often beaten by $20 wines.
'Scuse me while I work this one out.
It is all part of the same training. We usually ask a dog to indicate presence of a scent by exhibiting a trained behavior. A dog in an obedience ring will pick up the object (dumbell) that has the owner’s scent on it. A drug-sniffing dog will usually be trained to sit. A search and rescue dog will dig, or bark - depending on the training.
But, the opposite behavior to that behavior is “go on to next object”. The dog usually needs a bit of training to do that, but not too much - the reward comes when the scented object is found, so the dog wants to go on to a rewardable object.
There is no reason that the dog couldn’t be rewarded for indicating a “non-scent”. Come to think of it, USDAA dogs are rewarded for going though a whole lot of crates and stuff quickly - usually with praise. If they weren’t, they would “false indicate” objects to get rewards.
I haven’t trained for non-scent indication, but I see no reason why not - so you have just given me my next training challenge. I’ll have to start with Corbi, though - Gavi is already in tracking, and I don’t want to confuse her. Corbi has never worked scent, so we’ll be starting totally fresh.
“Good art” and “bad art” are pretty subjective.
I wonder what the pigeons would think of my masterpiece, “Untitled” (Pigeon blood, feathers and birdshot on canvas, 2008.)