The current thesis set out to investigate whether the behavioural process of derived stimulus relating could facilitate a better understanding, prediction and influence of human likes and dislikes than that afforded by direct contingency accounts alone. Across a series of six experiments and in the absence of co-occurrence, reinforcement or instruction, stimuli spontaneously acquired evaluative functions by participating in derived coordination, opposition and comparative relations. By exerting fine-grained contextual control over how Pokémon characters, fictitious brand products or potential prizes were related to one another, we systematically manipulated the direction and magnitude of evaluative responding. These relational effects were evident when a range of direct and indirect (IAT, IRAP and Affective Priming) tasks were employed. When taken together, our work suggests that a sophisticated experimental analysis of evaluative responding cannot focus solely on simple stimulus pairings – at least where verbally trained humans are concerned. Rather, a comprehensive understanding of human likes and dislikes requires a shift in current research practices, with direct and derived stimulus relations explored in tandem.