A 2010 study (via Whole Health Source) compared a Paleo diet with a Mediterranean diet. For twelve weeks, twenty-nine volunteers could eat as much as they wanted, whenever they wanted. Half ate Paleo (“lean meat, fish, fruit, vegetables, root vegetables, eggs, and nuts”), half Mediterranean (“whole grains, low-fat dairy products, vegetables, fruit, fish, and oils and margarines”).
The main result was that the Paleo food was “more satiating per calorie”. The Paleo eaters ate less but no more often than Mediterranean eaters. The paper does not report weight loss. (See an earlier report of the same experiment for that.)
I suspect the Paleo diet was less familiar than the Mediterranean diet (of course I can’t be sure from the descriptions). My theory of weight control says familiarity matters: less-familiar food pushes your set point lower than familiar food because its smell-calorie associations are weaker. The smells of less-familiar food is less associated with calories than the smells of more familiar food. With a lower set point, you will need less food to feel full.
If familiarity matters, this causes big problems for clinical studies. It means that short-term results (e.g., after 6 months) may be quite different than long-term results (e.g., after 2 years) — and most clinical trials last about six months. Short term, says my theory, any new food will cause weight loss. Indeed, a wide range of diets that cause dieters to eat new foods, such as the cabbage soup diet, cause short-term weight loss. Over the long term, however, the new foods become familiar and, according to my theory, the set point goes back up as the new smell-calorie associations are learned. Indeed, on most diets there is great long-term weight regain. If familiarity matters, we need data sets lasting a long time (e.g., nine or ten years) to understand weight control. Such data sets allow enough time for the chosen diets to become familiar so that (a) the diets being compared are equal in familiarity and (b) we can see their long-term effects — which may easily be different from their short-term effects.
Thanks to Steve Hansen.