I've actually been reading a lot on this recently, including some of Wrangham's work. The problem I have with it, and this is a problem I see in most anthropological work, is that researchers (including Wrangham and his colleagues) constantly overgeneralize the conclusions they draw from their work. This is aside from the fact that they make some pretty stunning logical leaps to reach their conclusions.
For instance, let's assume that Wrangham et al.'s findings of cooked food, burned bones and a hearth at a hominin site 2 million years ago are accurate. This says very little about human behavior, for several reasons.
1. Just because one finds a hearth does not mean it was used for cooking. Fire was certainly used for light, heat, or spiritual aspiration long before it was used to process food.
2. Just because one finds bits of cooked food at an archeological site doesn't mean the food was cooked intentionally. Someone may have accidentally dropped or intentionally disposed of it in a fire used for purposes other than cooking.
3. Just because one finds bits of cooked food at an archeological site doesn't mean the food was eaten. Someone may have discarded it after they realized it was cooked, figuring it was ruined.
4. Just because one finds burned bones at a hearth doesn't mean the fire's architect cooked and ate the meat that was on the bones. Modern people toss non-food items (sticks, stones, debris) and leftovers (popsickle sticks, glass, cups, cans) into camp fires all the time. I'm doubt this behavior is novel.
5. Even if it could be proven that someone cooked food and ate it at one archeological site 2 million years ago doesn't mean all people around the world also did it. (This is the fallacy of overgeneralizing that I mentioned above.) Cooking food may have emerged in one locality 2 million years ago and only spread beyond the local area several hundreds of thousands of years later. The habit of cooking food may also have emerged and disappeared several times over human history, in several places around the world, much like agriculture. For all we know, the habit of cooking food may have even driven its first adherents to extinction, only to emerge independently a million years later.
Because of how sparse the archeological record is and how challenging it is to objectively interpret it, I think the question of how long we've been cooking food is a red herring.
We will never know when the first morsel of food was cooked because the odds of it being preserved in a recognizable way are near zero. We will never know the geographic extent of the habit of cooking food because the archeological record will never be complete enough to paint an accurate picture. The geographic extent of the habit of cooking food certainly changed over time, it spread and contracted, it shifted. The archeological record will never be detailed enough to allow us to discern the range of the habit of cooking food with respect to time. We must simply admit that we don't and can't know the real history of the behavior of cooking food.
And why does it matter anyway?
It seems to me that the important question is whether we should be cooking our food today. If we fare better on a raw food diet today, then we should be eating raw food. What people were doing 100,000 or 1,000,000 years ago shouldn't matter. Maybe they did it to their detriment? Why should we repeat their mistake?
I have to say that I feel better on a predominantly raw food today than I did while eating mostly cooked food. So the amount of time that Homo sapiens has been cooking its food doesn't sound like an important piece of information to me.
