Can it be that adding food makes people believe they are eating less?

A recent study by Brian Wansink and Pierre Chandon report that this can indeed be the case (this version of the study was done with John Tierney of the NYT)

Half of the people were shown pictures of a meal consisting of an Applebee’s Oriental Chicken Salad and a 20-ounce cup of regular Pepsi and they were asked to estimate the amount of calories in the entire meal. The other participants were shown the same salad and drink plus two Fortt’s crackers prominently labeled “Trans Fat Free.” The crackers added 100 calories to the meal, but given that they were “diet” how will their presence influence the estimated amount of calories in the entire meal?

The first group estimated that the meal contained 1,011 calories, which was a little high. The meal actually contained 934 calories — 714 from the salad and 220 from the drink.  But, the second group  estimated the total amount of calories to go down.  Now the average estimate for the whole meal was only 835 calories — 199 calories less than the actual calorie count, and 176 calories less than the average estimate by the other group for the same meal without crackers.

The original study was interpreted as a halo effect of items labeled as diet.  I suspect that this is correct, but I think that it is also possible that people have a hard time computing totals and that instead they compute averages – which makes the estimation when the crackers are present to be lower.