Methods
This randomized cross-over study, including its participant recruit- ment, test food composition, and experimental design, has been described previously [13]. Briefly, participants repeatedly consumed novel foods that differed in energy density (LE or HE) and viscosity (liquid or semi-solid). Creaminess was not modulated. The four versions of the food rendered two conditions in which sensory characteristics were ‘congruent’ with satiety expectations (i.e. LE/liquid and HE/ semi-solid) and two with opposite ‘incongruent’ characteristics (LE/semi-solid and HE/liquid). The foods were highly novel with re- spect to their sensory characteristics, in order to limit the existence of learned associations with the energy content and to have thickness as the dominant sensory cue to rely on when guiding food intake. Foods were flavored with a ‘pandan rice’ aroma in combination with a green coloring (Koepoe Koepoe); with ‘spicy orange’ aroma (De Lange) and orange coloring (Chr. Hansen); with ‘rose apple’ aroma (Givaudan SA Corp., Vernier, Switzerland) and yellow coloring (Chr. Hansen); and ‘fenugreek’ aroma (Het Blauwe Huis) with carmine (pink) coloring (Chr. Hansen). Flavor–color combinations were ran- domly assigned to the four test products, under the condition that they were equally distributed. All participants received each of the fla- vor–color combinations once. Expected satiety was indeed higher for the semi-solid foods than for the liquid foods at baseline [13]