Why Brains Break and Where: Physics of Neurodegeneration

Why These Neurons? Why This Patient? Why Now?

Selective vulnerability in neurodegeneration has puzzled neurologists for decades. At the Alzheimer’s Association International Conference of 2025, we describe some computational neuroscience network models that attempt to address these questions with a hypothesis called Thermodynamic-Informational Entropic Relationships (TIER). We turned to physics—specifically thermodynamic entropy—to understand why systems fail. Engineers have long known: push anything far enough, fast enough, or hard enough, and it breaks. In physics, this breaking down over time has a name: thermodynamic entropy.

What Is Entropy, Really?

When physicists say “entropy,” doctors often think heat or metabolism. But thermodynamic entropy is simpler: it’s why things fall apart over time.

Build a sandcastle. Without maintenance, waves and wind scatter it. Why? Because there are millions of ways for sand to be scattered, but only a few ways to be a castle. Disorder is probable. Order is improbable. This isn’t philosophy—it’s physics. And brains can’t escape physics.

Information as Physical Work

Learning takes work. Physical work. Learning– even perceiving– can mean moving proteins, building receptors, reshaping synapses. It’s construction work at molecular scale. That work is information processing.
Claude Shannon, father of the information theory that pioneered the computer age, showed that information equals surprise. A surprising input demands bigger synaptic changes. Bigger changes mean more physical work. More work– more often, faster, over more distance—means more wear and tear—more entropy.

Here’s where distance matters. Work = Force × Distance. Some neurons span 15 centimeters—from your parietal lobe to frontal cortex. HUmans are kind of special that way as our cortex expanded in evolution. But there’s a cost. Molecular motors haul cargo down these highways thousands of times daily. For decades. That’s vastly more work than short, local connections.

The Pyramid Falls From the Top

As the brain expanded, it did so systematically. Primary sensory neurons talk to unitmodal sensory neurons with talk to polymodal systems integrating all information together. That’s how a barking sound matches with an ambiguous figure to assure you that it is, in fact, a dog. The information ascends pyramid-like heirarchies of informational abstraction and integration. But… imagine a human pyramid where the guy on top must integrate every wobble from everyone below. Who falls first? The guy desperately trying to balance all that incoming information.

Your brain is that pyramid—a hierarchy where information flows upward from simple to complex. Primary sensory areas form the stable base. They process basic features: edges, tones, touches. The middle layers combine these into objects and meanings. But those “polymodal” regions at the very top? They integrate everything—sight with sound with memory with meaning.

Here’s the problem with being on top: Every level below sends information up. The peak processes not just its own work, but the accumulated complexity of everything beneath. Now add two more challenges. First, our evolved cortical expansion means these top regions connect across vast distances—up to 15 centimeters of axon per connection. Second, engineers have long known that complex systems are inherently fragile. More parts mean more potential failure points. The brain’s highest regions combine maximum complexity with maximum distance with maximum integration demands.

The Elephant on a Needle

Look at our brain from the side, a large mass balanced on a tiny brainstem. The brain has done something both awesome and absurd, the equivalent of balancing an elephant on a needle. This huge and hierarchically expanded cortical pyramid is the elephant. Evolution expanded our cortex to 86 billion neurons, creating unprecedented cognitive abilities. But what’s the needle?

Ancient support structures: The locus coeruleus contains approximately 50,000 neurons in young adults (only 3,000 remain by age 80) maintaining norepinephrine supply across your entire brain. The cholinergic basal forebrain—maybe 200,000 neurons in the nucleus basalis of Meynert—modulates billions of cortical neurons. That’s like all of Germany’s farmland being watered by a single village’s wells.

This isn’t poor design. It’s evolution’s calculated bargain: these tiny structures can maintain your brilliant hierarchical cortex through reproductive years by working themselves to death. Evolution got what it wanted—your genes passed on. But the bill comes due.

The Siphon Effect

As those top cortical regions—already working hardest—age and become less efficient, they demand more from their support systems. The locus coeruleus, trying to prop up an increasingly wobbly pyramid with just thousands of cells, burns out first—not because it started the problem, but because it tried to fix it.

When these exhausted cells die, they’re so overworked they’re packed with toxic tau protein. They spread this pathology through their vast networks. But here’s the crucial detail: the tau only takes root in tissue that’s already stressed from its own hierarchical work burden. The top of the pyramid, already wobbling, gets poisoned by the very structures that tried to stabilize it.

The cortex that overworked its helpers becomes victim to their toxic remains. It’s a vicious cycle—the failing systems make each other worse, spreading failure as they go.

You Can’t Stop Sensing

The unique vulnerability of certain brain regions comes in part from what it can’t stop doing, the constancy of its work:

-You can choose not to move—motor cortex rests

-You can choose not to speak—language areas pause

-You can’t choose not to sense—integration never stops

The “default mode” identified as the network most prone to Alzheimer’s pathology isn’t exactly resting. It’s defaulting to nearly constant work from birth to death. Every second, your sensory systems process information, push it up the hierarchy, integrate it at the top. No breaks. No vacations.

Beyond Alzheimer’s: Why Different Diseases Hit Different Targets

This framework explains patterns across all neurodegeneration, not just Alzheimer’s. Each disease reflects a different combination of the same physical forces:

Parkinson’s Disease hits different neurons because the work stress is different. Alpha-synuclein accumulates where synapses fire rapidly, not where axons are long. The basal ganglia process continuous sensory-motor integration but through short, local connections. High frequency, short distance—different equation, different protein, same physics. Still sensory integration, still constant work, but synaptic cycling stress rather than transport stress.

Frontotemporal Dementia could be rarer because it has fewer risk factors. Yes, frontal cortex receives sensory input from sensory polymodal regions, BUT—and this is crucial—the basal ganglia brake those signals before they travel back down long projections. Less frequent transmission equals lower risk equals rarer disease. Von Economo neurons? They’re long and sparse, but they process emotional valence, not millisecond sensory changes. Partial vulnerability predicts partial prevalence.

ALS shows what happens with extreme distance but reduced frequency. Motor neurons span up to a meter—maximum distance. But they fire only during voluntary movement, not constantly like sensory systems. The twist? The anterior horn isn’t just motor output—it integrates sensory feedback for reflexes, creating unexpected work burden. Still less prevalent than Alzheimer’s because it lacks the triple threat of hierarchy plus distance plus constant input.

The Cerebellar Exception proves the rule. Despite intense activity, the cerebellum resists degeneration. Why? No hierarchy—just parallel processing. Massive redundancy. Different architecture, different physics, different outcome. Though notably, the vermis, which receives direct sensory input, shows the most age-related loss—exactly what the framework predicts.

The pattern is clear: Prevalence drops as risk factors disappear. Alzheimer’s (all risks) → Parkinson’s (high frequency, no hierarchy) → FTD (long distance, gated output) → ALS (extreme distance, intermittent). Physics predicts epidemiology.

Predictions From Physics

This framework makes specific, testable predictions:

Why don’t mice get Alzheimer’s? Their neurons span millimeters, not centimeters. Less distance equals less work equals less wear. Physics protects them. You need genetic manipulation to give mice Alzheimer’s because their natural architecture resists it.

Motion sickness as early warning? If you’re prone to motion sickness, your brain struggles integrating visual and vestibular signals. That’s inefficient processing—extra work reconciling conflicting inputs. Decades of this inefficiency might compound into earlier failure. Not tested yet, but maybe worth exploring.

When amyloid drugs might actually work: In rare genetic Alzheimer’s, amyloid overproduction could be the primary thermodynamic insult. Remove it early, you remove the main driver. But in normal aging? Amyloid might be more like fever in infection—a response to underlying problems, not the cause itself.

Working With the Physics

Understanding these principles transforms how we might intervene:

Timing is everything: Education builds neural infrastructure when you’re young—like exercise building muscle. The same intensive learning might break aging systems—like forcing arthritic joints to sprint. There may be an inflection point where too much help becomes stressful and potentially harmful.
Support the supporters before they fall: SNRIs might reduce Alzheimer’s risk by helping the locus coeruleus maintain function before the siphon effect begins. The data’s striking: 27% reduction overall, but 50% with long-term use. Early, sustained support matters more than late rescue attempts.

Synchronize to save energy: 40Hz stimulation might work by synchronizing neural oscillations, reducing the work of communication. Think of soldiers breaking step on a bridge—same result, less structural stress.

The inflammation connection: Chronic, low-grade inflammation over decades might increase the work of neural communication. This could explain why lifelong NSAID use shows protection while short-term trials fail. You can’t undo forty years of extra work in six months.

The cerebellar wild card: Your cerebellum times neural operations with incredible precision. Lose that timing, and cortical work might skyrocket. Could cerebellar training be protective? Physics suggests yes, though it’s not yet very well tested.

The Bottom Line

This framework proposes that neurodegeneration follows fundamental physics. Your brain’s greatest achievement—its hierarchical integration of complex information across vast distances—creates its greatest vulnerability. The elephant of expanded cortical processing balances precariously on the needle of ancient support structures.

Evolution built a brain to last 50 years, not 90. The same systems that make us brilliantly human—our ability to synthesize, connect, and understand—burn themselves out in service of consciousness. But understanding the physics might help us extend the warranty. We can’t stop entropy any more than we can stop gravity. But we might learn to work with it rather than against it.

“What shines must burn”—Viktor Frankl understood this truth about human nature. Now we might understand why, and perhaps how to tend the flame more carefully.

Dr. Peter Pressman studies brain aging at Oregon Health & Science University. This framework represents ongoing research into fundamental principles of neurodegeneration.