Efficient Coding Predicts Synaptic Conductance

This paper demonstrates that a parameter-free biophysical model, derived from Shannon's information theory and the principle of minimal energy, accurately predicts how synaptic information efficiency (bits per Joule) decreases as conductance deviates from its natural value, confirming that synapses operate at signal-to-noise ratios that maximize energy efficiency.

The Gist

Here is an explanation of the paper using simple language, analogies, and metaphors.

The Big Idea: The Brain's "Energy Bill"

Imagine your brain is a massive, high-tech city. Every time a thought happens, tiny messengers (neurons) send signals to each other across bridges called synapses.

Sending these messages costs energy (like electricity). The big question scientists have been asking is: How does the brain get the most "bang for its buck"? In other words, how does it send the maximum amount of information (bits) while spending the minimum amount of energy (Joules)?

This paper argues that evolution has tuned our synapses to be the ultimate energy-efficient couriers. They operate at a "sweet spot" where they send the most data for the least cost.

---

The Problem: Why Not Just Turn Up the Volume?

In a previous study (Harris et al., 2015), scientists tried to mess with the synapses. They artificially changed the "conductance" (think of this as the width of the bridge or the volume knob on a radio).

• The Natural State: When the bridge was its natural width, the signal was clear, and the energy cost was perfect.
• The Experiment: When they made the bridge wider or narrower (forced it away from its natural state), the efficiency dropped. The brain wasted energy, and the message got "fuzzier."

The Mystery: We knew efficiency dropped when the settings were wrong, but we didn't know exactly why or how fast it dropped. It was like knowing a car gets terrible gas mileage if you drive too fast or too slow, but not knowing the mathematical formula for that bad mileage.

---

The New Discovery: The "Noise vs. Energy" Trade-off

A recent study (Malkin et al., 2026) found a rule: Energy limits how quiet the background noise can be.

Imagine you are trying to hear a whisper in a crowded room.

• The Noise: The chatter of the crowd is like "synaptic noise" (random electrical static).
• The Energy: To hear the whisper clearly, you need to spend energy to block out the noise.

Malkin's team found that synapses are already doing the best job possible: they are using their limited energy budget to make the background noise as quiet as physically possible. This is called the Minimal Energy Boundary.

But here is the catch: Just being quiet (low noise) doesn't guarantee you are sending the most information. You could be whispering a secret very quietly, but if you are whispering a boring story, you aren't being efficient.

---

The Solution: Stone's "Perfect Formula"

James Stone (the author of this paper) took that "Minimal Energy Boundary" rule and combined it with Shannon's Information Theory (the math behind how we send data over the internet).

He built a model that predicts exactly how efficiency changes when you tweak the synapse.

The Analogy: The Perfect Radio Station
Imagine a radio station trying to broadcast music to a city.

1. The Signal: The music.
2. The Noise: Static on the line.
3. The Power: The electricity used by the transmitter.

Stone's model says:

• The brain's synapses are tuned to a specific "volume" (conductance) where the ratio of Music to Static is perfect.
• If you turn the volume up too high (too much conductance), you waste energy on unnecessary power, and the efficiency drops.
• If you turn it down too low, the static drowns out the music, and you waste energy trying to fix the garbled message.

The Magic Number:
Stone derived a specific mathematical curve (Equation 21) that predicts the efficiency drop.

• The Cool Part: This formula has zero "free parameters."
  • What does that mean? Usually, when scientists fit a curve to data, they tweak knobs and dials until the line matches the dots.
  • Stone's approach: He didn't tweak anything. He built the formula entirely from the laws of physics (how electricity and chemicals work in the brain).
  • The Result: When he plugged in the raw physics numbers, the curve landed perfectly on the experimental data from the Harris study. It was a "blind prediction" that turned out to be right.

---

Why This Matters

This paper confirms a beautiful idea about evolution: The brain is an engineer's dream.

It suggests that over millions of years, our brains have been fine-tuned to operate at the absolute limit of efficiency.

• We don't just minimize noise (as Malkin found).
• We also maximize the information per dollar (Stone found).

The Takeaway:
Your brain isn't just a messy biological machine; it is a highly optimized information processor. Every time a synapse fires, it is doing so at the exact "sweet spot" where it sends the most bits of information for the least amount of energy. If you try to force it to work differently, it immediately becomes wasteful and inefficient, just like a car engine running at the wrong RPM.

In short: The brain is the ultimate energy saver, and math proves it.

Technical Summary

Here is a detailed technical summary of the paper "Efficient Coding Predicts Synaptic Conductance" by James V. Stone.

1. Problem Statement

The paper addresses the mechanism behind synaptic efficiency, defined as the amount of information (bits) transmitted per unit of energy (Joule).

• Background: Previous work by Harris et al. (2015) demonstrated that synapses transmit the maximum number of bits per Joule at their "natural" conductance values. When conductance is artificially deviated from this natural value, efficiency drops rapidly. However, the mathematical form of this efficiency drop remained unexplained.
• Recent Context: Malkin et al. (2026) established a "minimal energy boundary," showing that synaptic noise variance is minimized for a given energy budget ($E$) according to the relationship $\sigma^{-2} = kE^5$.
• The Gap: While the minimal energy boundary ensures low noise, it is a necessary but not sufficient condition for maximizing information efficiency. The paper seeks to derive a theoretical model that explains exactly how efficiency falls as conductance deviates from its natural value, using the minimal energy boundary as a foundation.

2. Methodology

The author derives a parametric efficiency function by integrating Shannon's Information Theory with the biophysics of synaptic transmission.

• Biophysical Definitions:
  • Synaptic conductance ($\mu$) is defined as $\mu = npq$ (where $n$ is active zones, $p$ is release probability, $q$ is vesicle size).
  • Noise variance ($\sigma^2$) is derived from the binomial nature of vesicle release.
  • Malkin et al.'s minimal energy boundary is adopted: $\sigma^{-2} = kE^5$, linking precision to energy.
• Information Theoretic Derivation:
  • Mutual Information ($I$): Expressed using Shannon's formula: $I = \log_2(1 + \text{SNR})^{1/2}$, where SNR is the signal-to-noise ratio.
  • Energy-Conductance Relationship: Using Ohm's law, energy ($E$) is expressed as proportional to conductance ($G$): $E = \beta G$.
  • Substitution: By substituting the energy-conductance relationship into the minimal energy boundary equation, the SNR is expressed as a function of conductance: $\text{SNR} \propto G^{2.5}$.
  • Efficiency Function ($\varepsilon$): Defined as $\varepsilon = I/E$. Substituting the derived expressions yields the core model equation:
    $$\varepsilon(G) = \frac{\log(1 + cG^{2.5})}{\beta G}$$
    Where $c$ is a constant determined by the condition that efficiency is maximized at the natural conductance ($G=1$).
• Validation: The derived function (Equation 23) was fitted to experimental data from Harris et al. (2015). The model was tested with a fixed theoretical exponent ($\alpha = 2.5$) and compared against a best-fit numerical search ($\hat{\alpha}$).

3. Key Contributions

• Derivation of a Parameter-Free Model: The paper presents a closed-form mathematical function for synaptic efficiency that contains no free parameters. The shape of the curve is derived entirely from the biophysics of the synapse and the minimal energy boundary condition.
• Unification of Noise and Efficiency: It demonstrates that the minimal energy boundary (minimizing noise) combined with Shannon's theory leads to a specific signal-to-noise ratio that maximizes information efficiency.
• Theoretical Justification for Empirical Data: It provides the first theoretical explanation for the specific rate at which efficiency declines when synaptic conductance deviates from its natural value, a phenomenon previously only described empirically.

4. Results

• Model Fit: The theoretical model (Equation 21 with $\alpha = 2.5$) was fitted to the experimental data from Harris et al. (2015).
  • Goodness-of-Fit: The fit was statistically significant ($p = 0.0013$, $R^2 = 0.746$).
  • Parameter Comparison: A numerical search for the best fit yielded $\hat{\alpha} = 2.451$, which is extremely close to the theoretically derived value of $\alpha = 2.5$.
• Visual Confirmation: As shown in Figure 2b of the paper, the theoretical curve (dashed line, $\alpha=2.5$) is almost identical to the best-fit curve (solid line, $\hat{\alpha}=2.451$) and accurately tracks the experimental data points across a wide range of conductances.
• Peak Efficiency: The model correctly predicts that maximum efficiency occurs at $G=1$ (the natural conductance), consistent with experimental observations.

5. Significance

• Evolutionary Principle: The results strongly support the hypothesis that neuronal systems in the brain have evolved to operate at the limit of thermodynamic and information-theoretic efficiency (maximizing bits per Joule).
• Predictive Power: The model proves that the "minimal energy boundary" is not just about noise reduction; it dictates the specific operating point of synapses for optimal information transfer.
• Constraint on Biological Systems: The finding that the exponent $\alpha$ is fixed at 2.5 implies that if experimental data were to deviate significantly from this value, it would suggest that synapses do not operate at the minimal energy boundary. The close match confirms that biological synapses are tightly constrained by these physical laws.
• Theoretical Rigor: By eliminating free parameters, the study moves beyond curve-fitting to provide a fundamental biophysical explanation for synaptic coding efficiency.