It's funny that you say it's more Groundhog Day than Sherlock Holmes when your soy milk example, to me, is pretty much as Holmesian as it gets. I've never intentionally tried the kind of deduction that you describe, but at least for me and my circle of friends, I can see it easily falling apart just a few steps in. Plenty of stores carry soy milk nowadays, not just vegetarian ones. I can't possibly hope to know all the restaurants where my friends live. Most of my friends who cook at home get by with dull knives and ancient cookware. And not only do I have no idea which stores sell kitchen supplies, but it's not always obvious whether someone does their shopping in person or on Amazon. This is especially the case for hobbies and activities I'm less familiar with, such as yoga.
I suppose if I had meticulous knowledge of my area and practiced a lot, I could get my guesses above 50%. But it seems like something I'd have to put a lot of effort in, not a skill I would just acquire casually.
Do you find that this comes easily to you?
(I'm not saying that figuring things out about people is impossible, of course. It makes sense to divide up the probability space and increase your odds. Again with your coffee example, you could probably make a reasonable educated guess that the person is vegetarian, or even that they cook at home. But where they shop? Where they eat? There's so many choices these days, and people like trying new things. Even if I had a hunch, I certainly couldn't outright tell them with specificity about all their consumer habits, like you described with your coworker. The probability of me being right would just be way too low.)
I appreciate you posting here, it's pretty interesting to read.
> Holmesian as it gets
Yeah, it wasn't a great example, just sprung to mind. But I think you got the basic gist of how it works. I think the difference is that Holmes deduces his conclusions by following clues, but Groundhog Day aggregates information to find patterns...which can then be used to make deductive guesses. I guess it's more of an inductive approach at the beginning.
> Do you find that this comes easily to you?
If I don't put effort into turning it off, which I do most of the time these days because I associate doing it with bad feelings, it's just part of the normal process my brain goes through when I'm getting to know somebody. So about as much effort as having a bunch of conversations and remembering a few things. I have a terrible memory so I have to spend some effort reinforcing a few key facts. But I can then usually derive the rest with reasonable accuracy without having to remember it all.
Why remember things when I can just figure them out each time?
It's actually part of the classic "gifted curse" I personally deal with. I never really bothered to learn good memorization techniques because I can usually figure things out from the few kernels I do remember. But when topics get too complex, or I'm just outright missing data, the jenga pile falls over and I'm rather lost.
I'm also slightly dyscalculic and dyslexic (took a few tests and everything for formal and second diagnoses), which provides their own sets of challenges -- this is surprisingly common among the very gifted, but we usually "solve" it ourselves internally and thus the research on it is pitifully poor. I didn't personally find out about it conclusively till I was well into my mid-twenties, though some early childhood tests raised the possibility.
There's a specific reason I'm not inserting IQ into what I'm talking about, I agree that it's not a good focal point for a discussion about intelligence.
And I'm not saying I have perfect prediction of human behavior. Just that, with many people, if I spend enough time observing them, I can start to predict what they'll do or say with better than chance accuracy. The best analogy I can give is not Sherlock Holmes, but Groundhog day. What people miss in that movie is that the main character (Phil Connors played by Bill Murray) is spending his time mapping out the behavior patterns of the girl of his dreams (Rita by Andie MacDowell) to see if he can find the behavior pathways that will lead to her falling in love with him (or sleeping with him depending on your take on the story). Many people follow routines and habits, and if you observe them long enough it's like spending the same "day" over and over with them. You can start to mentally note and map out their behavior and then start making predictions about that behavior.
Most people do it all the time, we track what foods or films or whatever our friends like and dislike, and then make recommendations to them, or buy them presents based on that. We know them, so we think we can predict their behavior on a like/dislike vector. But what if you observed other things? Say you find a friend has an extreme allergy to milk and milk products. You can guess then that, if they drink coffee, they'll use a milk substitute, maybe soy. If they use soy, and you know what stores in their area carry soy milk and which don't and the ones that do are only the vegetarian stores, then you might be able to make a probabilistic guess that they're vegetarian. If they're vegetarian, you know there aren't many restaurants that are vegetarian in their area, so they probably cook at home, meaning they have a decent kitchen. So you can guess they have decent cookware, guessing at their income level you can guess if it's $100 an item cookware of $1000 an item cookware. There's only so many vendors at each price point for that, and only so many sell in my area etc., and on and on to yoga studio and car choice and neighborhood.
You can probabilistically guess all kinds of things from a cup of coffee. A little more observation, a comment here and there, a couple questions from you to direct your guesses and you can improve your chances and before you know it you can run down the list of consumer choices many individuals make which defines their space.
Even singular biases or opinions can tell you worlds about a person. It's kind of like stereotyping, but it's a little more informed than that. You can guess, for example, a person with strong opinions on say gun rights is likely aligned with several other related categories. Differences in those aligned categories might tell you other things, and let you know if they're conservative, liberal, anarchist or something else. Even a single comment can inform your predictive model about a person and set boundaries on what they would or wouldn't do.
You also have to know things. I couldn't do this with somebody from a country or area I don't know anything about. I don't know the politics, neighborhoods, social signals, in and out group dynamics and on and on.
Another example, there's an old phrase, "everybody skis in Aspen", meaning that knowing a person is from Aspen implies they do snow sports, but knowing that doesn't inform you of much other than they probably own a pair of skis or a snowboard and there's a fair chance they come from money.
But knowing a person you meet in someplace like Tampa, is an avid skier (an outlier) tells you all kinds of things about them. That they travel, that they probably don't go to anywhere on the East Coast (the snow sucks), that they probably have a passport (to go to the Canadian Rockies or the Alps), and gives you an idea of their disposable income they set aside for trips and skiing, which then can inform you of their total income and on and on and on.
People are like chaotic systems, and every piece of information you know restricts the space they can operate in. Within limits you can infer many many things. And it's not superhuman.