Research on animal cognition is currently struggling with two main problems. Firstly, mainstream views on reasoning in philosophy of mind and epistemology are unable to explain a growing body of empirical results in animal cognition research. The second problem concerns the interpretation of the same empirical results within comparative psychology, namely, whether they are best explained by appeal to (‘lower’) associative or (‘higher’) cognitive mechanisms, where neither side is providing convincing explanations. This problem is amplified by a crisis of the dichotomy itself between the ‘associative’ and the ‘cognitive’. In response to these two problems, I develop an affect-based construal of cognition, which broadens the conception of cognition to also include affective processes. I introduce the concept of affective representations that exhibit both ‘associative’ and ‘cognitive’ characteristics. This concept allows for new kinds of explanations of comparative research results, and indicates a future direction to explore human and nonhuman cognition.