[FRIAM] The epiphenomenality relation

[uǝlƃ ☤>$]

Variables are ... well, "things that vary". So in the language surrounding iteration, I'm not saying "variable X occurs before Y". I'm saying X and Y take on values *before* an iterate. And they take on values *after* an iterate. Then ΔX and ΔY may be non-zero. I.e. x1, x2 ∈ X and y1, y2 ∈ Y and the iteration looks like Iter(x1,y1) → <x2,y2>.

In this context, X is not a cause of Y. Iter() is a cause of the variation in X and Y.

What Frank said was that the variation in X might be *predictive* of the variation in Y. So, even if you don't know the values y1 and y2, you can "get a feel for" y1 and y2 by looking at x1 and x2.

Re: "latent variables" - Iter() might be defined in terms of 3 variables, X, Y, and Z. And we might have access to X and Y, but not Z. (I.e. we know the values x1, x2, y1, and y2. But we don't know the values z1 or z2.) It's possible that X be predictive of Y whether or NOT X and Y depend on Z. But if they do depend on Z, then we might be able to go beyond merely "predictive of" and say something about causality ... e.g. we might be able to say something like Z causes both X and Y, which would then explain why X and Y correlate.

I hope that helps.

On 11/30/21 4:10 PM, thompnickson2 at gmail.com wrote:
> My problem is, of course, that if variable X occurs before Y and is predictive of it, then it is a cause, by definition.  I am groping for an understanding of a “latent” variable.  I promise I am not arguing, here. 

-- 
"Better to be slapped with the truth than kissed with a lie."
☤>$ uǝlƃ