When talking about apples "breeding true" from seed, several conditions must be met. First of all, the apple must be self fertile meaning a different variety is not required for fertilization and subsequent fruit production. Most apples do not meet this qualification but some do. The second condition is that the self fertile apple must be have been fertilized by one of its own kind. This typically happens when a grower plants thousands of apples in a given plot so that the likelihood is that the bees will pollinate the apples from surrounding trees (all of which are the same). For example, a grower might have hundreds, if not thousands, of Dolgo crab apples (reportedly growing "true from seed") together with the mind toward selling them as "true from seed" Dolgo seedlings. Same with Kerr applecrab or Antonovka apple. When these very same Dolgo apples, however, are crossed with an apple other than one of their own kind, very different offspring results - there is nothing "true" about that particular match. As an example, consider the Dolgo crabapple. Supposedly it "breeds true" when crossed with itself and growers advertise, then sell these seedlings as "Dolgo seedlings." Usually, it is very economical to buy them as compared to a grafted apple. They are much, much cheaper than purchasing a grafted apple - example $7.00 versus $31.00 per tree in 2012 pricing. Typically the grower will plant thousands of Dolgo seeds together in one spot, sell the trees and/or and harvest the seeds for the next batch of seedlings the grower intends to plant. Now, this very same Dolgo when crossed with a different apple will produce a wildly different offspring. Dolgo and Haralson, for example, produced the Kerr applecrab. Kerr, BTW, inherited Dolgo's self fertility Dolgo has been crossed with a whole host of other apples and produced unique (sometimes very popular) offspring which are very different from the "breeds true" Dolgo seedling.
Even the so-called "Dolgo seedlings" are not really all that "true." at all - similar might be a better word. I purchased 10 Dolgo seedlings from Lawyer nursery (a very good wholesale supplier, BTW) just to satisfy my own curiosity about how "similar" they were. Several years later, I have trees which are not all that much like my "true" i.e. grafted Dolgo. They are similar, for sure, but even this sample of 10 trees are all sufficiently different that I can easily see differences in them and tell each one apart from the others. If you want exactly the same apple as you picked or ate from xyz tree, then you need to graft that tree. If you want a "close enough for government work" apple AND IF that particular apple is self-fertile and also grown in proximity to other similar seedlings, then, you might get something fairly close in terms of the seedling. One thing which is very important to remember is that the bees which pollinate these apples cannot read books. They don't know the "rules" of the game and they are not interested in what you want. If they are carrying pollen from a tree 1/2 a mile away on their little leg hairs and, then, they just so happen to land on one of these mass-planed plots, the offspring of that particular match is going to be very different from what was expected. Unless the bees promise to cooperate, one will never reliably get "breeds true" apples from seed. One the other hand, variety is the spice of life and surprises are nice, too!