What shrank
Cerebellar grey matter came to 6.03 per cent of total intracranial volume in the 50 adults aged 20 to 35, 5.75 per cent in the 80 aged 55 to 70, and 5.49 per cent in the 30 aged over 80. Cerebellar white matter went from 2.39 per cent to 2.32 per cent and then 2.17 per cent across the same three groups. The MRI picture is exactly what an ageing brain is expected to look like.[1]
The cerebellum-specific measures did not follow. Grip-load force peak timing differed by 20.81 milliseconds in the young group, 23.25 in the older group and 40.60 in the older-old, and the Bayes factor against a group difference came to 0.153 and 0.095. Shoulder muscle anticipation timing gave 0.225 and 0.154, final hand adaptation in the reaching task 0.079 and 0.105, and speech adaptation 0.128 and 0.448. A Bayes factor well under one counts as evidence for the absence of a difference rather than a failure to find one.[1]
The two exceptions
Two cerebellum-specific measures did change. Coordination between smooth pursuit and saccades sat at -1.02 in the young group against -0.77 in the older group, with a Bayes factor of 106.8. The slope of reaction time against rotation angle rose from 2.62 milliseconds per degree in the young to 3.6 milliseconds per degree in the older-old, with a Bayes factor of 746.1. Both are strong evidence for a real difference, and both belong to tasks where the cerebellum shares the work with the cortex.[1]
Set against the general sensorimotor measures, the split is sharp. Choice reaction time went from 0.78 seconds in the young group to 0.97 in the older and 1.4 in the older-old; baseline reaching error moved from -0.55 degrees to -1.27 and -1.96; the grip force safety margin grew from 2.55 newtons to 3.39 and 4.87. What ages here is the general machinery of movement, while the specifically cerebellar computations mostly hold. The alternative reading is that the tasks recruit wider networks which may quietly take over. Steady performance would then not mean an unchanged cerebellum, and the authors keep that door open when they write that general sensorimotor performance is not independent of cerebellar processing.[1]
What a scan can carry
The denominator sets the limit. This is 50 people aged 20 to 35, 80 aged 55 to 70 and 30 aged over 80, each measured once, all well enough to complete eight demanding tasks. It describes three groups at one moment; it does not follow one person's ageing, and no patient group sits beside them. Someone whose cerebellum has visibly thinned on a scan gets no individual prediction out of this, and what the study offers is a caution against reading the volume as if it were the function.[1]
That is still worth carrying into a clinic. Thinning on a cerebellar scan is routinely read as a reason to expect trouble with timing, with anticipating a grip, with adapting a reach. In this sample those very abilities held up past 80 while general reaction and accuracy fell away, so the image and the observation at the bedside were answering different questions. The two exceptions say where to look instead: eye-movement coordination and mental rotation, the tasks the cerebellum runs together with the cortex.[1]