I'm having trouble in understanding function evaluation in Haskell.

I came across the following line, which is somewhat cryptic to me.

(liftM . (+)) 1 [2]  

I thought that to evalutate two composed functions,

I should apply right function to get a result and then

apply left function with the result.

e.g. f.g x y = f (g x y) = f z = result

So I guessed that Haskell evaluated above expression

as follows:

(liftM . (+)) 1 [2]   --->

  ((liftM . (+)) 1) [2]  --->       (A)

  (liftM (+1)) [2] -->

  [3]

Why did Haskell, however, not try to fully evaluate addition, 

like following?

(liftM . (+)) 1 [2]   --->

  liftM ((+) 1 [2])   --->

  error

Does (f . g) x y z equal ((((f . g) x) y) z)  in haskell?

Any guide will be appreciated.

Chul-Woong