All you need is LUV

This commentary addresses three issues (learning, unification, and validation) brought up by Rey's bold proposal. First, it suggests integrating another type of learning to his unification: chunking, defined in a broader sense than in his paper and encompassing not just verbal and sequential information, but also visual information. Second, it proposes that a multilevel explanatory approach - in particular including cognitive and neuronal levels - might offer a more effective strategy than Rey's reductionist approach. Finally, it outlines several challenges raised by developing unified computational models. Whether Rey's unified radical associationism will succeed, or whether a multilevel explanatory approach is more promising, is a fascinating question important not only for psychology and neuroscience, by also for epistemology.