# counterfactual correlation

A particular kind of conditional in which the speaker associates

some possible situation (which is known to not actually be true)

with something that could be an outcome of it.

In lojban this is probably best rendered using

mu'ei (see also possible worlds and mu'ei), and certainly cannot be

rendered with ganai ... gi ..., and here's why:

(~p v q) (ganai P gi Q) (p -> q)

implies any of the following:

(~~q v ~p) (~q -> ~p)

(~(p & A) v q) where A is true and isn't ~p.

~(p & ~q) DeMorgan's law

and a bunch of other stuff.

It's easy to find examples of "if" in english (counterfactuals

claims among them) which do *not* obey the above, and thus cannot

possibly correspond to claims with TFTT as the truth function.

Examples:

If I had a million dollars then I would quit my job. (p -> q ?)

does *not* imply

If I don't quit my job then I don't have a million dollars. (~q -> ~p)

And,

John needs oxygen to live. (We need to state this to be able to add it below, according to the rules)

If John were born on Mercury then he wouldn't need oxygen to live. (p -> q ?)

does *not* imply

If John were born on Mercury and he needs oxygen to live then he [p & A) -> q)

wouldn't need oxygen to live.

(Feel free to furnish this page with additional (or better) examples).

The true intent behind "million dollars" is "My becoming a millionaire would greatly increase the chances of my quitting my job.", since we don't want to claim that there is no other condition under which I'd quit, etc. Play with brivla containing the rafsi for cunso, or consider an experimental NU that abstracts the probability of the event occurring. --la xod

This is exactly what i mean (mi'e maikl.). It is necessary to deconstruct all such statements before translating them into Lojban. After awhile, the desire to phrase one's actual utterances in that way would disappear; & argument would cease...