Complex traits such as voluntary locomotion are thought to require both motivation and functional ability. However, for historical and practical reasons comparatively less work has been done to elucidate the neurobiology underlying variation in locomotion, despite the fact that locomotion is necessary for vertebrate survival, is highly heritable and therefore can be under intense selective pressure, and biomedically speaking can ameliorate many metabolic diseases (U.S. Department of Health and Human Services (2008) Physical Activity Guidelines for Americans. U.S. Department of Health and Human Services, Washington, DC.). Past work has supported the concept that reward circuitry within the brain, particularly the nucleus accumbens and the neurotransmitter dopamine, participate in the control of various forms of locomotion (e.g. open field, exploratory, spontaneous, or wheel-running behaviour) (Garland et al. 2011). Consistent with the idea that reward-related neurobiological differences underlie voluntary activity, in an article published in this issue of The Journal of Physiology Roberts et al. (2014) demonstrate an interesting relationship between the maturation of neurons within the nucleus accumbens and locomotion in rats bred for high voluntary running (HVR) or low voluntary running (LVR). The investigators first performed RNA sequencing which unexpectedly identified cell-cycle gene expression differences between the rat lines, including synaptogenesis-promoting genes, and cell adhesion molecules. Indeed, upon further investigation, the number of immature and mature medium spiny neurons (MSNs) in the nucleus accumbens of LVR rats was reduced at least 4–fold compared to the HVR animals, indicating MSN development is inherently lower in the LVR line. Remarkably, however, voluntary running reverses these trends such that LVR animals showed an increase while HVR showed a decrease in MSN number. This is the first report suggesting that voluntary running affects neurogenesis in the nucleus accumbens and furthermore, can do so differentially with genetic background. Moreover, this demonstration of neuronal plasticity occurs within only 6 days. The fact that morphological differences in classical reward-based neurons relate to total distance run, and the extraordinary plasticity inherent in these neurons, raises many more interesting questions. For instance, we do not yet know if changes in the MSN population are secondary to, or required for, changes in locomotion to occur. Furthermore, it remains unclear why wheel running would lower MSN content in HVR animals while promoting it in the LVR line, and by what mechanism this occurs. One important point that bears on the interpretation is the fact that when RNA sequencing was performed, a cohort of wild-type Wistar rats were also included as controls. The results reveal that the HVR and wild-type lines segregate together for half of the differentially expressed genes, while the LVR and wild-type lines segregate for the other half. This segregation pattern suggests that both linetypes have responded to opposing directional selection through manipulation of the same gene network, namely, neuronal cell-adhesion molecules. It is feasible, and perhaps likely, that selection for low or high running could fundamentally require an entirely different suite of genes. But in this study, both linetypes modified similar gene clusters compared to wild-type rats, which serves as a priori evidence that neuronal development within the nucleus accumbens underlies variation in running or vice versa. Unfortunately, wild-type rats were not included in the MSN studies, so conclusions cannot be drawn as to which linetype more closely reflects the basal state, or how MSNs in wild-type rats respond to running. But given the large MSN changes observed in HVR and LVR rats and their interaction with running, it would be interesting to know how the wild-type progenitors respond. Both past and current work provide strong support for the importance of reward circuitry in locomotory behaviour, but the current study suggests a more complex picture involving neuronal development and plasticity rather than isolated dopamine communication per se, especially considering that dopamine content within the nucleus accumbens of LVR and HVR lines did not differ before, during, or after running. Larger questions also emerge such as (1) what aspect(s) of exercise modulate neurotrophic factors, (2) are the changes in the nucleus accumbens responsible for the correlated exercise-induced changes in anxiety, stress, or learning behaviour, and (3) do the modifications to MSNs ultimately influence other brain regions, such as the mediobasal hypothalamus, to affect body weight, food intake and the hormonal milieu which are also known to change with exercise in both humans and rodents? Whatever the answers may be, it seems likely that the use of selective experiments will continue to be a useful tool in illuminating the neurobiological basis for activity.