As a model organism, zebrafish have much to offer neuroscientific research and they are increasingly being used in behavioral neuroscience, for example to study the genetics of learning and memory. As fish are often considered “less clever” than mammals, it is important to understand how they learn and to establish optimal testing conditions. In this study, we compared the efficacy of food reinforcement and social stimuli in supporting Pavlovian conditioning, Pavlovian-to-instrumental transfer, and acquisition of a two-alternative forced choice visual discrimination. Although equally effective in conditioning and in motivating discrimination learning, fish responded with shorter latencies when they were anticipating food but responded for a greater number of trials when anticipating the social stimulus. After learning, the reward was changed: food-reinforcement was replaced with the social stimulus and vice versa. Performance accuracy did not change, but response latency did: the group previously rewarded with food, but now rewarded with the social stimulus, showed a decrease in response vigor. This is a negative contrast effect, which is well established in rats, but was thought to be absent in fish because they lacked goal representation. Our results show that zebrafish, like rats, do have goal representations. Furthermore, we have shown that whereas food has greater incentive salience than social stimuli, fish become satiated rapidly, but motivation to seek social stimuli is sustained. We conclude that zebrafish are well motivated by a mixed economy of social stimuli and food.