A comprehensive structural MRI analysis revealing the neural architecture behind working memory fragility, cognitive compensation, and why some brains need external scaffolding to thrive.
My right cerebellum is doing disproportionate work, particularly a region called Lobule VI (+21% larger than the left). This enlarged structure projects through a normal thalamic relay to my left motor cortex, which is measurably thicker than the right (its biological age, on re-verification, matches the right side).
The left hemisphere reads "older" than the right on the structure-age model across nearly every paired structure, with the language regions furthest out front — but re-verification showed those same regions are also bigger and thicker than their mirrors, so the best reading is a brain built unusually lateralized, not one wearing down. Meanwhile, midline structures run smaller than average (brainstem, cerebellar vermis) while processing concentrates in the hemispheres.
The good news: overall brain age is normal (43.71 vs. my actual age of 44 — and an independent second model agrees), critical structures like the prefrontal cortex and basal ganglia are entirely typical, and a professional neuroradiology review found nothing to treat. The system is healthy; it's just built asymmetric — which, if you know me, explains a lot.
Two things happened since this page was written. First, an independent neuroradiology second opinion (board-certified, Swiss FMH/EDiNR) reviewed the underlying scan: the T1 is structurally unremarkable — no lesion, no atrophy, no malformation — and the one asymmetry flag that most worried me (a thalamic micro-structure at −72%) is visually symmetric and almost certainly a software measurement artifact, exactly what my own pipeline's QC notes suspected. Second, a full re-verification of every number on this page against the raw pipeline exports confirmed most of them — and overturned one framing.
The framing that changed: "wear" is probably the wrong word. The left-hemisphere language regions that read as "biologically older" are also thicker and larger than their right-side mirrors — the opposite of what aging or wear does to tissue. Leftward enlargement of language cortex is the most replicated normal asymmetry in the human brain, and the age model — trained on typical aging patterns — appears to be reading that normal lateralization as "older." The per-structure ages are also pipeline-relative (the model highlights anything ±5 years, and published cross-scanner comparisons show brain-age models disagree by 5–22 years between setups), so the absolute year-counts on this page are best read as relative rankings, not calendar time.
What survived verification unchanged: the right-cerebellar dominance (Lobule VI +21%), the midline reduction cluster (brainstem, vermis, 3rd ventricle — all outside the pipeline's own normative ranges), the uniformly left-older subcortex, and the classic left-larger language-area anatomy (planum temporale, frontal operculum), which reproduces across two independent atlases. The story is less "worn down by work" and more "unusually lateralized by construction" — which fits the working-memory architecture this page exists to explain.
Next step, per the radiologist: a dedicated clinical MRI with the sequences this research scan didn't include (FLAIR, DWI, SWI) — the ones that can actually answer the old-head-injury question (a helmetless ATV wreck my senior year of high school with a few minutes of memory loss, never evaluated at the time — now properly documented, and worth one real look).
Imagine one side of my "coordination center" (cerebellum) grew bigger to help out more. It's like having one really strong arm that does most of the lifting. It works, but that arm gets tired.
The "relay station" (thalamus) that passes messages between brain parts? Totally normal. Its pieces are working fine — one microscopic piece triggers a measurement glitch, and a radiologist confirmed it's the software, not the brain. The traffic jam isn't on the highway; it's at the destination.
The parts that handle language are unusually big and thick on my left side. Most brains lean that way a little; mine leans hard. A computer model reads that as "older" — but it's more like a tailor mistaking a custom fit for wear. That's how this brain was built.
Instead of using the middle parts of my brain equally, my brain shifted work to the sides, especially the left. It's like favoring one leg when you walk. You get there, but differently.
The "thinking" part (prefrontal cortex)? Normal size and thickness. The "habit" parts (basal ganglia)? All normal. The overall age of my brain? Perfect. The foundation is solid.
This isn't a "broken brain" story — and after a radiologist and a re-check of every number, it isn't a "worn out" story either. It's a custom build: unusual proportions, healthy parts, tuned for words and patterns. It runs the way it was built to run.
The most robust finding, replicated across three independent analyses.
| Structure | Finding | Normal Range | Status |
|---|---|---|---|
| Total Cerebellum Asymmetry | +7.70% R>L | -2.86% to +4.47% | Outside Normal |
| Lobule VI Asymmetry | +21.23% R>L | -15.72% to +14.87% | Outside Normal |
| Cerebellar Gray Matter | +8.37% R>L | -1.70% to +5.97% | Outside Normal |
| Cerebellar White Matter | 1.40-1.76% | 1.56-2.89% | Below Normal |
Every subcortical structure reads left-older on the structure-age model — 58 of 63 paired structures overall. Per the August 2026 update, read the year-figures as model-relative rankings, not calendar time.
| Region | Left Age | Right Age | Difference |
|---|---|---|---|
| Temporal Lobe (avg) | 53.1 years | 46.9 years | L is 6.2 yrs older |
| Insular/Opercular (avg) | 52.4 years | 45.2 years | L is 7.2 yrs older |
| Subcortical (avg) | 48.7 years | 45.2 years | L is 3.5 yrs older |
| Right Occipital Pole | 54.95 years | 58.73 years | +14.7 yrs above chrono |
Critical structures that are entirely typical, and often more informative than the differences.
| Structure | Finding | Why It Matters |
|---|---|---|
| Prefrontal Cortex (Volume & Thickness) | Normal | Working memory issues aren't from obvious PFC deficits |
| Thalamic Nuclei (12 of 13) | Normal | The relay station is NOT the bottleneck. (The 13th, the ~20-voxel mammillothalamic sliver, flags at −72% — professionally confirmed as segmentation artifact, Aug 2026) |
| Basal Ganglia (all structures) | Normal | This is a cerebellar story, not a basal ganglia story |
| Total Hippocampus | Normal | Long-term memory architecture is intact |
| Overall Brain Age | 43.71 years | Global brain health is good (chrono: 44) |
The brain appears organized away from midline processing toward lateralized processing.
| Structure | Finding | Normal Range | Status |
|---|---|---|---|
| Brainstem | 1.12-1.23% | 1.15-1.64% | Below Normal |
| Cerebellar Vermis VI-VII | 0.12-0.15% | 0.15-0.23% | Below Normal |
| 3rd Ventricle | 0.127-0.137% | 0.006-0.123% | Above Normal |
This isn't a story of pathology or deficit. It's a story of a brain that organized itself differently, lateralizing processing, developing one cerebellar hemisphere more than the other, and adapting to its own architecture over 44 years.
The structures that read "older" on the age model are the same ones built biggest and thickest — the language and pattern machinery. That's not failure, and it turns out it isn't wear either; it's the anatomy of the specialization that does the compensating.
The prefrontal cortex is structurally normal (both volume and thickness). So why the working memory challenges? The answer may lie in circuit dynamics rather than local tissue deficits.
When the cerebellum (a "timing" and "prediction" structure) is asymmetric and the receiving cortex shows signs of strain, the coordination between structures may be where the challenge lives.
The left temporal lobe, home to language processing, is where the age model reads furthest ahead (7-10 pipeline-years) — and where the tissue is also largest and thickest relative to its mirror. For someone with constant internal verbal processing, recursive thinking, and decades of intense linguistic computation, the anatomy fits the life.
The build is where the work is.
If the brain's internal coordination circuits are working at capacity, external tools become essential, not optional. The 700+ browser tabs, the detailed documentation systems, the AI partnerships for real-time processing... these aren't crutches.
They're appropriate infrastructure for a cognitive architecture that needs external buffering because the internal coordination systems are already running at full capacity.
Detailed parcellation of all cerebellar lobules showing rightward asymmetry pattern.
| Lobule | Function | Asymmetry | Normal Range |
|---|---|---|---|
| Lobule VI | Motor planning, cognition | +21.23% R>L | -15.72% to +14.87% |
| Crus I | Executive function | +7.80% R>L | -12.30% to +17.83% |
| Crus II | Cognitive processing | +13.22% R>L | -16.67% to +23.91% |
| Lobule VIIIA | Sensorimotor | +12.39% R>L | -27.30% to +19.95% |
| Lobule VIIIB | Sensorimotor | +12.93% R>L | -28.82% to +26.98% |
| Lobule IX | Vestibular | +4.92% R>L | -9.69% to +15.90% |
Total hippocampus normal, but CA2-CA3 subfield shows rightward asymmetry.
| Subfield | Function | Asymmetry | Status |
|---|---|---|---|
| CA1 | Memory encoding/retrieval | -3.31% L>R | Normal |
| CA2-CA3 | Social memory, pattern separation | +22.27% R>L | Outside normal |
| CA4-DG | Pattern separation | +8.35% R>L | Normal |
| Subiculum | Output to cortex | -3.68% L>R | Normal |
12 of 13 thalamic nuclei within normal bounds. Motor relay nuclei show subtle leftward asymmetry consistent with receiving input from enlarged right cerebellum. The exception — the mammillothalamic nuclei, a ~0.02 cm³ sliver — flags at −72% asymmetry, which a 2026 professional neuroradiology read confirmed as segmentation artifact, not anatomy (the structures are visually symmetric).
| Nucleus | Function | Volume % | Status |
|---|---|---|---|
| VLPN | Motor relay (cerebellum) | 0.121% | Normal |
| VLAN | Motor relay (cerebellum) | 0.014% | Normal |
| Pulvinar | Visual attention | 0.174% | Normal |
| Mediodorsal | Executive function | 0.093% | Normal |
| VAN | Motor planning | 0.038% | Normal |
| Centromedian | Arousal, attention | 0.018% | Normal |
| LGN | Visual relay | 0.010% | Normal |
| MGN | Auditory relay | 0.011% | Normal |
Biological age estimates for brain regions. Left hemisphere consistently older.
| Structure | Right Age | Left Age | Chronological: 44 |
|---|---|---|---|
| Planum Temporale | 47.12 | 54.46 | +10.5 yrs (L) |
| Sup. Temporal Gyrus | 46.97 | 54.16 | +10.2 yrs (L) |
| Frontal Operculum | 43.24 | 52.41 | +8.4 yrs (L) |
| Hippocampus | 45.90 | 50.21 | +6.2 yrs (L) |
| Putamen | 45.13 | 49.52 | +5.5 yrs (L) |
| Thalamus | 45.01 | 47.84 | +3.8 yrs (L) |
| Occipital Pole | 58.73 | 54.95 | +14.7 yrs (R) |
Cortical thickness across regions. Prefrontal cortex is normal, a critical negative finding.
| Region | Thickness | Status |
|---|---|---|
| Frontal Lobe (total) | 2.37 mm | Normal |
| Middle Frontal Gyrus | 2.28 mm | Normal |
| Superior Frontal Gyrus | 1.92 mm | Normal |
| Precentral Gyrus | 1.68 mm | Normal |
| Precentral Gyrus (L vs R) | 1.78 vs 1.58 mm | Left 11.8% thicker |
| Temporal Lobe | 3.15 mm | Normal |
| Parietal Lobe | 1.90 mm | Normal |
| Occipital Lobe | 2.26 mm | Normal |
Data quality metrics and analysis pipeline details.
| Analysis | SNR / Quality | Confidence |
|---|---|---|
| vol2Brain | SNR 41.15 | HIGH |
| AssemblyNet | SNR 41.26, QC: A | HIGH |
| BrainStructureAges | QC: A | HIGH |
| DeepThalamus | Scale 0.96 | HIGH |
| CERES | SNR 32.23 | HIGH |
| HIPS | Scale 0.96 | HIGH |
| pBrain | SNR 17.50 | LOW (interpret with caution) |
Interactive pages with complete measurements from each volBrain analysis pipeline. Click any available report to explore the full dataset.
Structural MRI acquired as part of a UT Austin research study in May 2024. Analysis performed using volBrain's cloud-based neuroimaging platform.
| Image Source | UT Austin Research Study, May 2024 |
| Resolution | 0.8mm isotropic (T1-weighted + T2-weighted sequences) |
| Quality Grade | Research-grade, superior to standard clinical 1mm resolution |
| Analysis Platform | volBrain (7 specialized neuroimaging pipelines) |
| Normative Comparison | Age and sex-matched reference data with 95% confidence intervals |
| Cross-Validation | Key findings replicated across 2-3 analyses (several share one segmentation family; a 2026 cross-atlas check reproduced the classic language-area asymmetries but not every secondary finding — see the August 2026 update above) |
| Professional Review | Independent neuroradiology second opinion (FMH/EDiNR, Aug 2026): T1 structurally unremarkable; flagged micro-structure asymmetry confirmed as measurement artifact |
"Understanding my own brain architecture isn't just personal curiosity. It's proof of concept for what's possible when AI helps humans understand themselves deeply."
This is why I'm building AIs & Shine. Because if structural MRI data, AI-assisted analysis, and careful synthesis can reveal this level of insight about one brain, imagine what "Life Models" could do for millions of people who experience their minds differently.
Human. Deeply seen.
That's not just a tagline. It's what this page represents.