Leggett--Garg Tests in Neural Dynamics: Probing Non-Diffusive Stochastic Structure in Single Neurons

This paper proposes an experimental framework using Leggett-Garg inequalities to distinguish between standard diffusive and non-diffusive persistent stochastic models in single-neuron dynamics, suggesting that observed violations would indicate non-Markovian temporal memory and contextual structure without requiring microscopic quantum coherence.

The Gist

The Big Idea: Is the Brain Just a "Random Walk"?

Imagine a neuron (a brain cell) as a tiny messenger trying to send a signal. For a long time, scientists have thought of this messenger's movement like a drunk person stumbling through a crowd.

• The Old View (Diffusive): The messenger moves randomly, bumping into things, with no real direction. If you stop and look at where they are, then look again a moment later, their position changes in a smooth, predictable way that just slowly fades away. This is called "diffusion."
• The New Proposal (Persistent): The author, Partha Ghose, suggests the messenger might actually be more like a runner with a strong memory. If the runner decides to go left, they keep going left for a while before suddenly switching to right. They have "persistence." They don't just stumble; they have momentum and a finite speed.

The paper asks: Can we tell the difference between the "drunk stumbler" and the "persistent runner" just by watching their timing?

The Test: The "Leggett-Garg" Check

To answer this, the paper proposes a specific test called a Leggett-Garg inequality.

Think of this test like checking if a story makes sense.

1. The Setup: Imagine you are watching a light switch that is either ON (+1) or OFF (-1).
2. The Rule: If the light follows a simple, predictable path (like the "drunk stumbler"), the relationship between its state at time A, time B, and time C must follow a strict mathematical limit. It's like saying, "If I walk from my house to the store, and then to the park, my total distance can't be more than the sum of the two trips."
3. The Violation: If the light behaves like the "persistent runner," it might create a pattern where the math breaks. The relationship between the three times becomes "wobbly" or oscillatory (like a wave going up and down).

The Paper's Claim:

• If the neuron acts like a simple diffuser (Wiener noise), it will never break this rule.
• If the neuron acts like a persistent runner (Kac process), it can break this rule because its movement has a "wave-like" memory.

Why This Matters (And What It Doesn't Mean)

This is the most important part to get right: The author is NOT saying the brain is "quantum" in the sci-fi sense.

• What people often think: "Quantum" means tiny particles acting weirdly, like electrons being in two places at once.
• What this paper says: We are looking for "quantum-like" math. The "persistent runner" model creates a mathematical pattern that looks exactly like the patterns found in quantum physics (specifically, the Dirac equation).

The Analogy:
Imagine a drum.

• If you hit it randomly, the sound dies out smoothly (Diffusion).
• If you hit it rhythmically, the sound creates a complex, vibrating wave (Persistence).
• The paper says: "If we hear that complex wave in the brain, it proves the brain isn't just stumbling randomly. It has a 'memory' and a 'rhythm'."

The author calls this "contextual temporal structure." In plain English: The brain's past actions influence its future actions in a way that isn't just simple randomness.

How to Do the Experiment

The paper outlines a simple, practical way to test this in a real lab:

1. Record: Use a needle to listen to a single neuron's electrical activity (membrane potential).
2. Simplify: Turn that complex signal into a simple "Yes/No" list.
   • Did a spike happen? Yes (+1).
   • Did no spike happen? No (-1).
3. Compare: Look at the signal at three different times (Time 1, Time 2, Time 3).
4. Calculate: Do the math to see if the "Leggett-Garg" limit is broken.

The Catch (The "Clumsiness" Loophole):
In physics, measuring something usually changes it (like checking a tire pressure lets air out). The paper admits we can't measure the brain without touching it. However, they suggest a workaround: Record the brain continuously without stopping to poke it, and then analyze the data later. This way, the "poking" doesn't mess up the specific timing we are trying to measure.

The Conclusion

If this experiment shows that the Leggett-Garg limit is broken, it means:

1. The "Drunk Stumbler" model is wrong. The neuron isn't just diffusing randomly.
2. The "Persistent Runner" model is likely right. The neuron has internal memory, moves at a finite speed, and creates wave-like correlations.
3. It's not magic. This doesn't prove the brain is a quantum computer. It just proves that the brain's noise is more structured and "rememberful" than we thought, and that this structure happens to use the same math as quantum mechanics.

In short: The paper proposes a way to prove that neurons have a "rhythm" and "memory" that makes them more complex than simple random walkers, using a mathematical test usually reserved for quantum particles.

Technical Summary

Technical Summary: Leggett–Garg Tests in Neural Dynamics

Problem Statement
The paper addresses the challenge of distinguishing between two competing classes of stochastic models for single-neuron dynamics:

1. Diffusive Models: Based on Wiener noise and the cable equation, these are Markovian, trajectory-based processes that yield monotonic decay of temporal correlations.
2. Persistent Stochastic Models: Based on Kac-type finite-velocity processes, these possess intrinsic memory and finite propagation speed, leading to the Telegrapher's equation.

While "quantum-like" features in neural dynamics (such as coherence and interference) have been proposed, they often admit classical analogues. The author argues that violations of Bell-type inequalities do not necessarily imply microscopic quantum nonlocality but rather the impossibility of describing correlations via a single underlying trajectory-based or non-contextual probabilistic model. The central problem is to determine if neural dynamics exhibits non-diffusive temporal correlations incompatible with a purely trajectory-based diffusive description, without invoking claims of microscopic quantum coherence in the brain.

Methodology
The paper proposes an experimental program utilizing the Leggett–Garg Inequality (LGI) as a temporal analogue of Bell-type constraints. The methodology involves:

• Defining the Observable: A coarse-grained dichotomic observable $Q(t) \in \{+1, -1\}$ is defined for a single neuron. This can be based on spike occurrence within a time window, the sign of the membrane potential relative to a threshold, or the internal state of a persistent random walk.
• Theoretical Framework: The LGI is formulated as $K = C_{12} + C_{23} - C_{13} \leq 1$, where $C_{ij} = \langle Q(t_i)Q(t_j) \rangle$. This inequality holds under the assumptions of Macrorealism (MR) (definite properties exist independent of observation) and Non-Invasive Measurability (NIM) (measurements do not disturb the system).
• Analytic Comparison:
  • Diffusive Case: The paper demonstrates that standard diffusive dynamics (Ornstein–Uhlenbeck process) leads to exponentially decaying correlations ($C_{ij} \propto e^{-\gamma|t_i-t_j|}$), which strictly satisfy the LGI ($K \leq 1$).
  • Persistent Case: The paper analyzes Kac-type processes where the internal state $s(t)$ flips with a Poisson rate $\lambda$. This leads to the Telegrapher's equation. Through analytic continuation, the Telegrapher's equation is shown to map to a Dirac-like envelope equation, yielding oscillatory temporal correlations of the form $C_{ij} \sim \cos(\omega(t_i - t_j))e^{-\gamma|t_i-t_j|}$.
• Experimental Protocol: The proposed procedure involves continuous intracellular recording of membrane potential $V(t)$, offline construction of the dichotomic variable $Q(t)$, and calculation of correlations from a single time series or ensemble averages. The paper acknowledges that strict NIM is physically impossible but argues that "clumsiness loopholes" can be managed by ensuring measurement-induced perturbations are negligible compared to intrinsic fluctuations or statistically controlled.

Key Contributions and Results

1. Theoretical Distinction: The paper establishes that purely diffusive dynamics cannot violate the LGI, whereas persistent stochastic dynamics with finite propagation speed can.
2. Mechanism for Violation: It is shown that the wave-like structure arising from persistence and finite velocity in Kac processes generates oscillatory correlations. Specifically, for damped oscillatory correlations, the Leggett–Garg combination $K$ can exceed 1 (e.g., $K = 3/2$ for specific phase conditions), thereby violating the inequality.
3. Mathematical Connection: The work demonstrates a formal mathematical link between the Telegrapher's equation (governing persistent stochastic transport) and the Dirac equation via analytic continuation. This connection suggests that the "quantum-like" structure in neural dynamics arises from the effective envelope equation rather than microscopic quantum mechanics.
4. Conservative Interpretation: The paper provides a framework to interpret LGI violations not as evidence of microscopic quantum coherence in the brain, but as evidence against a simple trajectory-based diffusive description. A violation implies the presence of memory, persistence, and contextual temporal structure.

Significance and Claims
The paper claims that Leggett–Garg tests offer a "possible experimental probe" for contextual and non-Markovian structure in neural dynamics. The significance lies in providing a rigorous method to distinguish between standard diffusive models and persistent stochastic models (Kac-type) without requiring the controversial assumption of microscopic quantum coherence in the brain.

The author emphasizes that a violation of the LGI in this context would indicate:

• The breakdown of a purely diffusive, trajectory-based description.
• The presence of memory and persistence in neural signal propagation.
• Compatibility with Kac-type stochastic dynamics that mathematically mimic quantum temporal correlations.

The paper concludes that these tests open a novel experimental window into the underlying mechanisms of neural signal propagation, specifically targeting the non-diffusive, persistent stochastic nature of the dynamics, while maintaining a conservative stance that these phenomena are effective descriptions of neural-level dynamics rather than evidence of fundamental quantum physics in the brain.