I think it's probably based off the empirical observation that in studies of free-living individuals, moderate increases in exercise output simply don't result in weight loss.
The thinking is that our bodies are very good at recognizing a calorie deficit and respond very effectively by increasing spontaneous intake to balance the calorie budget. It often only takes very small increases in intake -- so low they can easily go unnoticed (or that people deny they could have happened, again why metabolic ward studies are so vital) -- to obliterate the calorie expenditure of moderate exercise.
The thinking is that our bodies are very good at recognizing a calorie deficit and respond very effectively by increasing spontaneous intake to balance the calorie budget. It often only takes very small increases in intake -- so low they can easily go unnoticed (or that people deny they could have happened, again why metabolic ward studies are so vital) -- to obliterate the calorie expenditure of moderate exercise.