By learning efficiency, do you mean the cognitive efficiency/cognitive proficiency clusters? On the WISC-IV, CPI can be computed from the WMI and PSI, if you have the supplementary tables (in sold-separately reference book). On the WJIII/IV, they are obtained when the relevant supplementary cognitive subtests are administered. CPI will be available in the standard kit on the WISC-V.
Fluid reasoning is not usually included in cognitive efficiency/proficiency composites. Working memory almost always is, as is processing speed. RAN (rapid automatic naming) sometimes is (draws on retrieval efficiency and ps).
Fluid reasoning is not particularly trainable. Speed is somewhat trainable, and memory somewhat less. The available evidence suggests that working memory training does work, a bit, but only in a very narrow sense, with limited to negligible transfer. Even the best studied wm training program, CogMed, has unimpressively tiny effect sizes, but a masterful marketing program.
Phonological processing, on the other hand, is quite trainable, especially if one catches students when they are young, before they have developed their own kludge-y bad habits for approximating decoding. There are not sufficient tests in the standard battery that routinely assess phonological processing at all, let alone in any depth. This should be changing with the release of the WISC-V and WJIV, which both have PP clusters. Or one can throw in the CTOPP-2 or PAL-II, which are not comprehensive, but have nice PP measures.
