One of the many things on my mind recently is the application of all this (some would say little bit of) specialized knowledge to broader frameworks of understanding, and applying them to other fields. In theory, this is one of the ways knowledge should spread and evolve; and in an attempt to find the utility of theory in practice, I conducted an experiment (linked just in case…).
I brewed a wonderful Panther Coffee (best coffee in Miami, and I don’t get paid for these endorsements…), Columbia, Finca El Ventilador, brewed in a Chemex (unbleached paper filter, rinsed), and tested the liquor in 3 different cups.
The coffee aroma has a fruity body and a caramel base, an easy flowing mouth feel, and nothing but other wonderful joyous notes. On to the experiment!
Cup one is a standard home coffee mug, made in Malaysia. I expected it to be the worst of the lot. On the first round the coffee from it displayed seemingly nice highlighted acidity tending towards a nearly-sour tart berry note. I was surprised that that cup would highlight anything.
Cup two is a higher quality “specialty coffee” cup made in Japan. After cup one, the coffee tasted downright flat in it; again, not expected. What is going on here?
This is an antique English cup, probably from the inter-war years. Wasn’t sure what to expect from this guy as it was my first time ever drinking from it. What I didn’t expect was for the coffee to have a beautiful round and mellow body tending towards dark dry fruit (figs, dates, etc). This wasn’t even in the same ball park as the flavor profile I was getting from cup 1.
And now, a quick note on perception based preferences: Many things go into ones perception of flavor, and the decisions (revealed preferences) we make on a daily basis. Taste is not as simple as [physical sensation] -> [perception], it needs to pass through the filter of [cognition]. Now our model looks like this: [physical sensation] -> [cognition] -> [perception]; cognition, the filter of thought, emotions, and all other stimuli at this very moment, will characterize the way you think about any product you are tasting. That means product “A” will not always yield “Taste 3”.
So how is it possible to run tests on flavor?
Please humor me and let me skip many of the details and explain how the test I was ruining should have worked. I believe this will be a more enlightening way of showing you the power of flavor profiling (not capitalized, I’m not talking about the singular method by the same name).
I fully expected the antique cup (cup 3) to be best, followed by the Japanese cup (cup 2), and then cup 1. That was my expectation – the null hypothesis. Yet, we know that expectation will influence taste and that influence can deceive us. So their are 2 possible outcomes, I will accept my stated non-parametric ranking and fail to reject the null (I would not have proven anything) OR I would find these cups fall in a different order. That 2nd possibility is the key; for that to happen, it would mean that the change in flavor was so great, that it overcame my pre-conceived notion of cup quality.
Should I have failed to rejected the null, I would have had to use other tests to see if I was just missing subtlety. I fully expected not to reject the null. But I did. I preferred, after 2 rounds of sipping, cup 1 to all the other cups; but cup 1 also had a different flavor profile than the 2 other cups.
I had a very hard time believing that mass produced cup 1 truly was better, but what was I to do? Enter the glass cup. Glass is not neutral the way people mean (inert), but it is more neutral than the majority of ceramics.
I used the glass cup to test what the “unaffected flavor profile” would be; that is unaffected by interaction with the ceramic materials. The flavor profile did not match cup 1. Now we’re on to something!
I sat there sipping from cup 1 when it hit me; or rather when I couldn’t help but notice that that enjoyable acidity from the small sips during the initial rounds built to an overwhelmingly un-enjoyably acridity. The cup was acidifying (or at least reacting with) the coffee!
I sat back, waited ~15ish minutes, and went back to the cupping counter to re-try the other 2 cups. The antique won handily, and I failed to reject the null hypothesis. More tests will be necessary to determined if it is truly better now.
I did learn that I need to throw out the set of cup 1’s! (probably poison).
Problems with this design
- Only one person (me) doing the tasting – lack of data
- Not blind – I knew which cup was which the whole time
- same coffee throughout the test – lack of data and repeatability, no generality
- Tasted the cups in the same order each time – knock on effects, halo effects, carry over effects, masking effects, you get the idea…
Lots more work to be done!
Now, you might ask, what spurred me to do this?
I’m getting on a plane in about an hour to Panama heading for the coffee region – my work is classified (for the moment), but you dear readers, I promise, will be the first to know.
All the Best,