Abstract
By recording multiple cells simultaneously, electrophysiologists have found evidence for repeating spatiotemporal spike patterns, which can carry information. How this information is extracted by downstream neurons is unclear. In this theoretical paper, we investigate to what extent a single cell could detect a given spike pattern and what the optimal parameters to do so are, in particular the membrane time constant $\tau$. Using a leaky integrate-and-fire (LIF) neuron with instantaneous synapses and homogeneous Poisson input, we were able to compute this optimum analytically. Our results indicate that a relatively small $\tau$ (at most a few tens of ms) is usually optimal, even when the pattern is much longer. This is somewhat counter intuitive as the resulting detector ignores most of the pattern, due to its fast memory decay. Next, we wondered if spike-timing-dependent plasticity (STDP) could enable a neuron to reach the theoretical optimum. We simulated a LIF neuron equipped with additive spike-timing-dependent potentiation and homeostatic rate-based depression, and repeatedly exposed it to a given input spike pattern. As in previous studies, the LIF progressively became selective to the repeating pattern with no supervision, even when the pattern was embedded in Poisson activity. Here we show that, using certain STDP parameters, the resulting pattern detector can be optimal. Taken together, these results may explain how humans can learn repeating visual or auditory sequences. Long sequences could be recognized thanks to coincidence detectors working at a much shorter timescale. This is consistent with the fact that recognition is still possible if a sound sequence is compressed, played backward, or scrambled using 10ms bins. Coincidence detection is a simple yet powerful mechanism, which could be the main function of neurons in the brain.
Introduction
In sensory systems, specific stimuli evoke reproducible, repeating sequences of spikes across multiple neurons, known as "spatiotemporal spike patterns." While electrophysiologists have observed these patterns, it remains unclear how downstream neurons actually "read" or extract this information.
Do neurons need complex, multi-step integration to recognize a long pattern? Or can a single neuron, acting as a simple coincidence detector just one synapse away, do the job? This paper addresses two primary questions: First, what are the absolute theoretical optimal parameters—specifically the membrane time constant ($\tau$)—for a single Leaky Integrate-and-Fire (LIF) neuron to detect a pattern? Second, can a biologically plausible, unsupervised learning rule like Spike-Timing-Dependent Plasticity (STDP) allow a neuron to naturally find this optimum?
Methods
The research combines theoretical derivations with computational simulations. The study relies on synthetically generated Poisson spike trains, using "frozen noise" for repeating patterns and random Poisson processes for background noise.
Analytical Derivation
A LIF neuron was modeled receiving inputs from $N$ afferent neurons firing according to Poisson processes. Figure 3 illustrates the setup: a repeating "frozen noise" pattern (cyan) is embedded in random background activity. The goal is to maximize the Signal-to-Noise Ratio (SNR) of the membrane potential $V$, defined by the difference between the peak voltage during the pattern ($V_{max}$) and the average noise level ($\bar{V}_{noise}$).
This SNR was mathematically optimized to determine the ideal membrane time constant ($\tau$), the optimal subpattern duration ($\Delta t$), and the optimal wiring strategy. The derivation also accounted for spike jitter ($T$). As shown in Figure 4, the expected signal shape can be modeled as a trapezoid, where the peak height $h$ depends on the relationship between the detection window $\Delta t$ and the jitter $T$.
Spiking Simulations
To test biological plausibility, a LIF neuron was simulated with $10^4$ afferents. Synapses featured additive STDP for potentiation and a homeostatic rate-based term for depression. The neuron was repeatedly exposed to a 100ms "frozen noise" pattern with 3.2ms jitter, embedded in random background Poisson noise, without any supervisory signal.
Key Insights and Results
- Short memory is a feature: The optimal membrane time constant ($\tau$) is surprisingly short (a few tens of milliseconds), even for patterns hundreds of milliseconds long. It is mathematically more efficient to detect a brief, highly synchronous "signature" subpattern than to integrate the entire sequence.
- Theoretical Optimum: For a test case with jitter $T=3.2$ ms and firing rate $f=3.2$ Hz, the optimal parameters were $\tau = 18$ ms and $\Delta t = 23$ ms, yielding a massive SNR of ~80.
- STDP as an Unsupervised Optimizer: Without labels, STDP naturally prunes and strengthens synapses, turning a naive neuron into a near-optimal coincidence detector. In simulations, selectivity emerged after only ~12 presentations (about 5 seconds of simulation time), with full convergence after ~500 presentations.
- Sparse activity boosts signal: Lower background firing rates yield a higher optimal SNR, providing a theoretical justification for the low average firing rates observed in the brain (< 1 Hz).
Discussion
This study demonstrates that a single-cell readout can be highly effective at detecting spatiotemporal patterns by focusing on short "signatures." However, a single neuron cannot detect an entire long pattern in its entirety.
Future work suggests the brain likely employs populations of these coincidence detectors. Downstream neurons could combine outputs from multiple subpattern detectors using axonal delay lines or spatial conversion chains to recognize extended sequences. While the study used a simplified homeostatic depression rule, the robustness of the coincidence detection mechanism suggests it could be a fundamental function of neurons in the brain.
Summary updated