How Living Things Might Compute Using Natural Physics

Old computers versus living systems

Today's computers store information in fixed structures. Tiny charged spots sit still in memory cells. Gates flip between clear on and off states. This works well but costs much energy.

Living things do something different. Their parts move and sway together in continuous waves. The system settles into low-energy patterns naturally. The computation is the physical motion itself, not rules applied to still symbols. This creates a basic conflict. Engineers want clear readable states. Nature wants smooth flowing stability.

A bridge idea: letting physics do the work

The study uses a method called Biological Reservoir Computing, which means using living tissue or similar matter as a pool of natural motion to process signals. Instead of training every internal connection, researchers let a physical system move on its own. A network of linked pulsing units receives input signals. The natural motion of the system spreads these signals into complex patterns. A simple trained layer then reads the result at the end.

This matches an idea called criticality, which is a sweet spot between too rigid and too messy. The system balances between too much order and too much chaos. It needs some order to remember, some disorder to separate different inputs. The study tests whether real physical motion can serve this role.

The special role of fast brain rhythms

Brain waves around 40 cycles per second appear in many mental tasks. The study proposes these are not random noise. They show a system settling into a partly synchronized, low-energy state. This state actively maintains information through rhythmic motion, not through frozen bits.

The researchers tune how strongly the pulsing units pull on each other. Stronger linking drives the system toward this fast-rhythm zone. The idea is that this balance point resists random thermal disturbance while staying flexible enough to respond to new input.

What the study did

The team built a computer simulation of one hundred linked pulsing units. Each unit had a natural tendency to pulse near 40 cycles per second. The researchers tracked the system's energy as it moved. They also tested how well the system remembered past input signals.

They compared this biological-style system against a standard digital reservoir. The digital system used fixed update rules and fading memory. The biological system used energy minimization and persistent phase patterns.

What was found

The energy-conserving physical system showed superior dynamic stability. Its energy tracking revealed protected memory traces in low-energy traps. The system also achieved roughly three times better energy use per useful work than the digital baseline when measuring activity costs alone.

Fast brain rhythms did mark the shift to computationally useful states. Their emergence predicted when the system entered a stable synchronized regime.

However, a critical problem appeared. Standard linear reading methods failed completely. The memory capacity measured this way was essentially zero. The information existed in the timing relationships, but simple linear probes could not see it. The study calls this the Readout Gap, which means the stored wave information cannot be retrieved by ordinary simple reading tools.

What the study does not show

The efficiency comparison does not include the constant maintenance cost of keeping physical pulsing units ready. Living tissue always burns some energy just to stay alive. Inactive digital circuits do not. The true advantage depends heavily on the specific hardware.

The study does not model real biochemical brain processes. It uses a simplified physical proxy. It does not test semantic memory or meaningful knowledge, only short temporal retention of random signals.

The global linking assumption is a simplification. Real brains use sparse, patchy connection patterns. The results may not transfer directly.

The study does not solve the Readout Gap. It identifies the need for non-linear, timing-aware reading methods but does not build or test them.

Words used here

Hamiltonian dynamics

the natural motion of a physical system as it conserves total energy while trading between different forms

Fast brain rhythms

brain waves pulsing around 40 times per second, seen when groups of neurons coordinate their firing

Scale inseparability

the idea that in living systems, the rules and the physical stuff doing the rules cannot be pulled apart

Criticality

a balanced state at the edge between frozen order and random chaos, where complex behavior often emerges

Readout Gap

the problem that information stored in wave patterns cannot be retrieved by standard simple reading methods

Learn more in plain language

• Reservoir computingWikipedia (English, more technical)
• Neuromorphic computingWikipedia (English, more technical)
• Performance per wattWikipedia (English, more technical)
• Hamiltonian mechanicsWikipedia (English, more technical)