I'm waiting to get the full-text of this article, "Why Has the BMI Gone Up So Drastically in the Last 35 Years?", that looks promising. From the abstract:
"The additives of fats and sugars in combination, not separately, best predicted increases in BMI accounting for 97% of the variance in the linear regression analyses."
97% is obviously very high!
"In line with the major physiological theories emphasizing palatability as the addictive stimulus in models of incentives and addiction, fats and sugars in combination rather than calories per se or particular food groups accounted for the increases in BMI."
