The effect of surprise on flavour-nutrient learning

Through exposure, flavours can come to predict energy composition and signal whether a food is likely to be filling and/or pleasant. While this ‘flavour-nutrient learning’ appears to be relatively robust in animals, research in humans has delivered mixed results. In this study we consider whether ‘being surprised’ by a food promotes human flavour-nutrient learning. In an initial session, participants (N = 92) provided a measure of the expected satiation of a novel high or low energy-dense breakfast (matched for sensory characteristics). Participants then consumed the meal and hunger and fullness was recorded over three hours. A measure of surprise was derived from the mismatch between expected satiation and actual satiety. One week later, participants completed each measure once again. Energy density did not affect changes in expected satiation. However, clear effects of surprise were evident – those who found the foods to be more filling than anticipated showed an upward shift in expected satiety (week 2 compared with baseline). They also selected smaller ‘ideal’ portion sizes. The reverse was the case in participants who were less full than expected. Albeit less clear, we also found evidence that surprise affected a measure of food reward. Being less full than initially expected was associated with decreased reward. These findings are interpreted as broadly consistent with predictions based on the Rescorla–Wagner model of associative learning.