Unlike reptiles, who are born with species-specific morphology and behavior that hardly changes as they grow into adults, mammals are born with a class-specific neonatal phase that renders the morphology and behavior of each species different from the physiology and behavior of their species-specific adulthood. Mammals must undergo a transformation phase, called youth, between the neonate and the adult. This youthful metamorphic and meta-behavioral phase is necessary while the mammal is remodeling from sucking milk to munching grass or hunting meat. During its youthful phase a mammal is not simply growing linearly into its adult form and behavior but is a hybrid trying out different patterns of neonatal and adult micro-motor pieces of behavior that change almost daily in frequency as pieces of neonatal behavior inactivate and adult behavioral motor pieces begin to activate. The offsets of neonatal behavior and the onsets of species-specific adult behavior drive the youthful mammals into varied combinations of behavior sequences, that, since they are hybrids of two very different systems, don't look functional and are often unusual enough to amuse an observer. Such random hybrid combinations of micromotor pieces of behavior are commonly called play. But, then, play is not a unitary drive as many researchers have assumed but the consequence of youths hybridizing between neonatal and adult behaviors that are waning and waxing in intensity and frequency at each successive stage of youthful ontogeny. During the youthful period, any random combination of motor sequences that are useful for any reason can be remembered, repeated, and learned, so the youthful phase of mammalian ontogeny has potential for learning. A fixed neonatal system of physiological adaptations and complex behaviors is built into the neonate and might be modified for the worse by learning at this stage of its ontogeny. At a mammal's species-specific adult stage it is presumably well adapted to a stable niche. So, again, at this phase of ontogeny learning might be unnecessary. However, during its youthful transition phase a mammal is shifting between two systems of physiology and behavior designed to fit two very different stable environments. Thus its abilities and motives are in constant flux and trial and error learning is perhaps its best means of adjusting to almost daily changes in its own physiology and behavior. The environment of many adult mammals has been changing fast, during the past two and a half million years of successive glaciations and melting with new niches opening and old niches disappearing or degrading. Learning provides a way of coping with swiftly changing environments, and the fastest and most efficient means of providing adult mammals some more learning ability, and a somewhat less fixed and invariable species-specific behavior system would have been a selection for extending the youthful learning phase of mammalian ontogeny into adulthood. An extension of the youthful phase into the adult phase would likely have also loosened the typical species-specific sequences that would have been, in ancestral adults, a fixed system of invariable sequences that fit them to a formerly stable environment. A number of mammalian species and even some orders, such as Primates, probably Cetaceans, and including many Carnivores have, by an evolutionary process of neoteny, retained into adulthood some of their youthful phase of ontogeny. Such an adaptation provided these more neotenic adults a more open system of learning that could modify remnants of ancestral species-specific behavior to better fit adults of that species to a swiftly changing environment. We hypothesize an interesting phenomenon in mammalian phylogeny: that a number of mammalian species have been modified over the past several million years by selection for the metamorphic phase of ontogeny extended into the adult phase. Mammalian phylogeny appears to be recapitulating a reversal in the usual direction of ontogeny.