Just for the sake of the argument, I took this box of cereals as an example. Sure, it does have an explicit suggestion of portion size in small print: 30 grams. These 30 grams will give you 133 kcal. Ok. But there is a much more salient part providing a strong implicit cue. There's a huge bowl filled to the rim and it is there right in front of you. If we only look at the visible cereals in that bowl (and the suggested cereals that are cut off at the left), this is already about double of those 30 grams, thus equalling more than 250 kcal. If you imagine how many cereals are beneath the visible surface of that bowl, you'll easily have a quadruple portion, having over 500 kcal. If you really want to limit yourself to 30 grams, you can only have 35 of these cereals in your bowl.
Now, this issue is not limited to these cereals, neither is it a Kellogg's problem. Take a good look around. Fruit juice boxes often depict a large glass. Nutella and other choco pastes show a copious amount spread on a sandwich, and cookie packages often show more than just one cookie.
There are, of course, good reasons to present these foods and beverages as such. It is certainly more appealing in most cases to use more copious presentations. The hidden cost, however, is that it implicitly sets portion standards we might anchor our own choices on.
UPDATE AUGUST 2015:
We did already publish one study on this topic. You'll find it here: http://dx.doi.org/10.1016/j.appet.2015.07.003
Neyens, E., Aerts, G., & Smits, T. (2015). The impact of image size manipulation and sugar content on children's cereal consumption. Appetite, 95(December), 152-157.
Abstract: Previous studies have demonstrated that portion sizes and food energy-density influence children's eating behavior. However, the potential effects of front-of-pack image-sizes of serving suggestions and sugar content have not been tested. Using a mixed experimental design among young children, this study examines the effects of image-size manipulation and sugar content on cereal and milk consumption. Children poured and consumed significantly more cereal and drank significantly more milk when exposed to a larger sized image of serving suggestion as compared to a smaller image-size. Sugar content showed no main effects. Nevertheless, cereal consumption only differed significantly between small and large image-sizes when sugar content was low. An advantage of this study was the mundane setting in which the data were collected: a school's dining room instead of an artificial lab. Future studies should include a control condition, with children eating by themselves to reflect an even more natural context.UPDATE 2 February 2017
Both the anecdotal evidence from this blog (that cereal packages suggest a larger serving in their depicted food than stated in the nutrition table) and the evidence that this matters (it affects consumers) have now been conceptually replicated. Tal, Niemann & Wansink published in BMC Public Health reporting a systematically large depicted serving suggestion on cereal packages (Study 1) and its effects on adult consumers (Study 2).