Using the same paradigm, we are able to confirm and extend our previous finding of transient training-induced gray matter changes in the adult human brain. Our results show that dynamic alterations in gray matter structure can occur very rapidly within a time range of a single week (Figue 2).
This time-course favours fast adjusting neuronal systems, such as spine and synapse turnover 
as the underlying factor for gray matter increase, rather than such slow evolving mechanisms as neuronal or glial cell genesis 
. It is important to mention, that gray matter does not necessarily mean that we are measuring neurons or even cells as such. It is possible that other factors could subtly alter voxel values resulting in tissue being “misclassified” as gray matter. In general, an increase in gray matter could be due to an increase in cell size, neural or glial cell genesis, spine density or even changes in blood flow or interstitial fluid 
. A strong argument against the assumption that MRI signal changes capture cortical neurogenesis comes from a recent post mortem
study measuring the integration of (14)C, generated by nuclear bomb tests during the Cold War with DNA. This was used to establish the age of neurons in the major areas of the human cerebral neocortex and provided evidence, that neocortical neurogenesis may be restricted to the developmental period 
. However, the contra argument is supported by the assumption that newly generated cells can migrate to distant anatomical sites 
. Finally, a recent study by Pereira et al. demonstrating in vivo
correlates of exercise-induced neurogenesis in the hippocampus confirms the theoretical possibility of angiogenesis underlying plasticity processes 
. Further work is needed to clarify whether vascular changes due to increased cerebral blood volume and/or cerebral blood flow may have additional effects to the observed changes 
Independent of the precise histological nature of these structural alterations, our results support structural forms of neuroplasticity to be important in processing the information in dynamic networks according to novel informational demands 
. Interestingly, neither performance (minutes endurance juggling) nor exercise (hours per day) was able to predict structural changes in the occipito-temporal cortex.
Importantly, the ability to initially learn a three-ball cascade juggling task is correlated with an increase in gray matter, whereas further improvement of the skill over time due to training does not seem to alter brain structure. Animal experiments suggest that learning is associated with synaptogenesis and glial hypertrophy, whereas a simple increase in motor activity is “only” related to angiogenesis 
As a general pattern, the increase in gray matter in all regions () is only detectable during constant training of the visual-motor skill and recedes when exercise is stopped, although the participants were still able to juggle. We suggest that the qualitative change (i.e. learning of a new task) is more critical for the brain to change its structure than simple training of this task once learned; however, when we detect such a change in brain structure, it may well be a combination of both. In the process of learning, it is a normative characteristic of the nervous system to change to be able to encode and appropriately implement new knowledge 
. Further studies need to address the question whether the skill as such or whether exercising this skill is more important for functional and structural adaptations of the brain.
In addition to the gray matter change in the temporal area of the visual cortex, we found a change of brain gray matter which followed the same time pattern (increase during exercise and receding when exercise stopped) bilaterally in the frontal and temporal lobes and the cingulate cortex. Because this finding was not reported in our previous study 
and because this finding did not survive the correction for multiple comparisons, these data have to be viewed with caution and may be unspecific. One possible reason why we detected these areas in the present study as compared to the study in 2004 may be the higher sensitivity of a higher field strength (3 Tesla vs. 1.5 Tesla) and/or of a larger group size in the present study (n
20 vs. n
12 in the former study) and an improved estimation of the mean and variance due to a higher number of repeated measures.
One of the unsolved obstacles of voxel based morphometry is the fact, that MR morphometry studies done at different research centers are almost impossible to compare due to scanner- and site- specific properties 
. Therefore, multicenter studies are currently only feasible with significant limitations. The present study is the first to include data from two different cohorts scanned on two different scanners and even different field strengths.
It is an intriguing question why our brains do not expand over time, if we assume that that there is an increase in gray matter that is sustained with learning and/or practicing a skill. The most intuitive answer is, that the alterations are not sustained but that, once the learning process is over and the functional networks sufficient for the new task, the gray-matter changes reverse to their original size. However, given that such changes may last for at least 3 months without further exercising 
, we suggest that these regionally restricted changes are rather sublte and will not change the net-size or weight of the brain. It has also to be pointed out, that an increase in gray matter volume (i.e. a change of the classification of individual voxels from white to gray matter) will prompt an inverse effect (i.e. regionally loss in white matter volume) in adjacent white matter. The major future challenge is to understand the behavioural consequences and cellular mechanisms underlying training-induced neuroanatomic plasticity.