90% certainty that this is bs because I’m waiting for a flight and I’m sleep deprive, but:
For most people there’s not a very clear way or incentive to have a meta model of themselves in a certain situation.
By meta model, I mean one that is modeling “high level generators of action”.
So, say that I know Dave:
Likes peanut-butter-jelly on thin cracker
Dislikes peanut-butter-jelly in sandwiches
Likes butter fingers candy
A completely non-meta model of Dave would be:
If I give Dave a butter fingers candy box as a gift, he will enjoy it
Another non-meta model of Dave would be:
If I give Dave a box of Reese’s as a gift, he will enjoy it, since I thing they are kind of a combination between peantu-butter-jelly and butter fingers
A meta model of Dave would be:
Based on the 3 items above, I can deduce Dave likes things which are sweet, fatty, smooth with a touch of bitter (let’s assume peanut butter has some bitter to it) and crunchy but he doesn’t like them being too starchy (hence why he dislikes sandwiches).
So, if I give Dave a cup of sweet milk ice cream with bits of crunchy dark chocolate on top as a gift, he will love it.
Now, I’m not saying this meta-model is a good one (and Dave is imaginary, so we’ll never know). But my point is, it seems highly useful for us to have very good meta-models of other people, since that’s how we can predict their actions in extreme situations, surprise them, impress them, make them laugh… etc
On the other hand, we don’t need to construct meta-models of ourselves, because we can just query our “high level generators of action” directly, we can think “Does a cup of milk ice cream with crunchy dark chocolate on top sound tasty ?” and our high level generators of action will strive to give us an estimate which will usually seem “good enough to us”.
So in some way, it’s easier for us to get meta models of other people, out of simple necessity and we might have better meta models of other people than we have of our own self… not because we couldn’t construct a better one, but because there’s no need for it. Or at least, based on the fallacy of knowing your own mind, there’s no need for it.
I’d agree that this is useful to think on, but I tend to use “meta model” to mean “a model of how to build and apply models across distinct people”, and your example of abstracting Dave’s preferences is just another model for him, not all that meta.
I might suggest you call it an “abstract model” or an “explainable model”. In fact, if they make the same predictions, they’re equally powerful, but one is more compressible and easier to transmit (and examine in your head).
Hmh, I actually did not think of that one all-important bit. Yeap, what I described as a “meta model for Dave’s mind” is indeed a “meta model for human minds” or at least a “meta model for American minds” in which I plugged in some Dave-specific observations.
I’ll have to re-work this at some point with this in mind, unless there’s already something much better on the subject out there.
But again, I’ll excuse this with having been so tried when I wrote this that I didn’t even remember I did until your comment reminded me about it.
90% certainty that this is bs because I’m waiting for a flight and I’m sleep deprive, but:
For most people there’s not a very clear way or incentive to have a meta model of themselves in a certain situation.
By meta model, I mean one that is modeling “high level generators of action”.
So, say that I know Dave:
Likes peanut-butter-jelly on thin cracker
Dislikes peanut-butter-jelly in sandwiches
Likes butter fingers candy
A completely non-meta model of Dave would be:
If I give Dave a butter fingers candy box as a gift, he will enjoy it
Another non-meta model of Dave would be:
If I give Dave a box of Reese’s as a gift, he will enjoy it, since I thing they are kind of a combination between peantu-butter-jelly and butter fingers
A meta model of Dave would be:
Based on the 3 items above, I can deduce Dave likes things which are sweet, fatty, smooth with a touch of bitter (let’s assume peanut butter has some bitter to it) and crunchy but he doesn’t like them being too starchy (hence why he dislikes sandwiches).
So, if I give Dave a cup of sweet milk ice cream with bits of crunchy dark chocolate on top as a gift, he will love it.
Now, I’m not saying this meta-model is a good one (and Dave is imaginary, so we’ll never know). But my point is, it seems highly useful for us to have very good meta-models of other people, since that’s how we can predict their actions in extreme situations, surprise them, impress them, make them laugh… etc
On the other hand, we don’t need to construct meta-models of ourselves, because we can just query our “high level generators of action” directly, we can think “Does a cup of milk ice cream with crunchy dark chocolate on top sound tasty ?” and our high level generators of action will strive to give us an estimate which will usually seem “good enough to us”.
So in some way, it’s easier for us to get meta models of other people, out of simple necessity and we might have better meta models of other people than we have of our own self… not because we couldn’t construct a better one, but because there’s no need for it. Or at least, based on the fallacy of knowing your own mind, there’s no need for it.
I’d agree that this is useful to think on, but I tend to use “meta model” to mean “a model of how to build and apply models across distinct people”, and your example of abstracting Dave’s preferences is just another model for him, not all that meta.
I might suggest you call it an “abstract model” or an “explainable model”. In fact, if they make the same predictions, they’re equally powerful, but one is more compressible and easier to transmit (and examine in your head).
Hmh, I actually did not think of that one all-important bit. Yeap, what I described as a “meta model for Dave’s mind” is indeed a “meta model for human minds” or at least a “meta model for American minds” in which I plugged in some Dave-specific observations.
I’ll have to re-work this at some point with this in mind, unless there’s already something much better on the subject out there.
But again, I’ll excuse this with having been so tried when I wrote this that I didn’t even remember I did until your comment reminded me about it.