Classical conditioning allows relating the fundamental research in non-human animals with contingency assessment tasks in humans, given that animals judge the relationship between a conditioned stimulus and an unconditioned stimulus in a way akin to humans judge the relationship between a cue and an outcome. Classical conditioning has been traditionally explained by associative models, but these models have been demonstrated have some limitations. Signal Detection Theory (SDT) can be a more appropriate alternative. In the present experiment contingency assessment is analyzed in four groups of rats that were exposed to a tone that was always followed by food (100%), and a tone-click compound for which different contingencies of reinforcement were employed (100%, 66%, 33% and 0%). The general design was A+/ AX+. The group in which the compound was always reinforced (100%) showed augmentation and blocking. In contrast, second order conditioning and conditioned inhibition were observed when the compound contingency reinforcement was 0%. The results showed that these phenomena appeared in different moments of training for the mentioned groups, whereas groups with intermediate reinforcement (33% and 66%) showed intermediate results. Results are analyzed using associative learning methodologies and TDS techniques. Theoretical implications of applying SDT to associative learning are discussed.