Understanding the intricate web of connections within the human brain remains one of the greatest challenges in science. This complex network of billions of neurons, communicating through brief electrical pulses, forms the foundation of every thought, memory, and action. A fundamental goal for neuroscientists is to map this connectivity, to determine which neurons influence others and how information flows through these circuits.
Scientists typically monitor this neural communication by recording the electrical pulses, or “spikes,” that neurons generate. When plotted over time, these signals form a pattern known as a spike train. However, these spike trains are not neat and rhythmic; they are often chaotic, bursty, and irregular. This unpredictability makes it exceptionally difficult to determine cause and effect. Is a spike in Neuron A causing a subsequent spike in Neuron B, or are they both responding to a third, unobserved neuron, or is it just a coincidence?
Traditional methods for detecting causality in data often struggle with the unique nature of neural signals. Many techniques require data that is sampled at regular time intervals or assume that the relationships between components are simple and linear. The brain, with its complex and nonlinear dynamics, rarely fits these assumptions. This has left a significant gap in the tools available to researchers trying to build an accurate functional map of the brain directly from observed activity.
In a new study published in the journal Physical Review E, a team of researchers from Japan has introduced a novel technique designed to overcome these hurdles. The method provides a powerful new way to identify causal relationships directly from the irregular spike trains of neurons. The research was led by Assistant Professor Kazuya Sawada from the Tokyo University of Science, in collaboration with Professor Tohru Ikeguchi, also from the Tokyo University of Science, and Associate Professor Yutaka Shimada from Saitama University.
Their approach is an inventive adaptation of an existing framework for analyzing complex systems, known as convergent cross mapping (CCM). The core principle of CCM is that if one component in a system (Neuron A) causally influences another (Neuron B), then the behavior of Neuron B must contain some signature or “shadow” of Neuron A’s activity. By analyzing the data from Neuron B alone, one should be able to make increasingly accurate predictions about Neuron A’s past behavior. If such predictions are not possible, then a causal link is unlikely.
The challenge was that conventional CCM cannot be directly applied to the irregularly timed events of a spike train. The research team implemented two key modifications to solve this. First, instead of looking at the spike times themselves, they focused on the time intervals between consecutive spikes. This sequence of interspike intervals (ISIs) transforms the irregular timing of events into a more continuous series of data points that reflects the neuron’s dynamic state.
Second, they devised a new procedure to establish a temporal correspondence between the ISI data from different neurons. Since the spikes do not happen at the same time, a system was needed to match a point in time in one neuron’s activity stream with the closest corresponding point in another’s. This alignment is essential for making the cross-system predictions that lie at the heart of the method.
With these modifications, the team created a new way to assess causality. The method calculates how accurately one can predict a given neuron’s ISI sequence using the data from another neuron. If the prediction accuracy improves as more data becomes available, it signals a genuine causal connection.
“The method proposed in our paper differs from previous ones in that it can be directly applied to spike sequences and identify causal relationships in data generated by complex, nonlinear systems that cannot be represented by simple rules,” Sawada highlights. This allows for the estimation of neuronal connectivity from easily observable spike train recordings.
To verify that their new technique worked as intended, the scientists applied it to a well-established mathematical model of neurons. In this simulated environment, they could program the exact connections between neurons and then see if their method could correctly identify them. They tested scenarios with bidirectional connections, unidirectional connections, and no connections at all. The proposed approach successfully detected the correct causal structure in each case. It also performed well even when a degree of random noise was added to the system, a feature that mimics the inherent unpredictability of real biological environments.
The development of this tool opens up new avenues for exploring the brain’s functional architecture with greater precision. It could allow researchers to construct more detailed maps of how information is processed and routed through neural circuits.
“The connections between brain neurons are not yet fully understood, and causality detection methods can be used to estimate not only structural and anatomical connections but also effective connections,” explains Sawada. “If we could clarify the nature of such effective connections within the brain, it would contribute to a better understanding of disorders and mental illnesses caused by neuronal connections, potentially paving the way for new therapies.” The approach could offer new insights into conditions like epilepsy, which involves abnormal synchronized firing, or schizophrenia and bipolar disorder, which are thought to involve imbalances in neural circuitry.
Sawada noted that the current study focused on small networks of two or three neurons. A primary goal for future research is to expand the method’s capability to analyze much larger and more complex networks, bringing it closer to the scale of real brain circuits.
The impact of this work may extend beyond neuroscience. Data composed of irregularly timed events, known as point processes, appear in many other fields. The principles behind this new method could inspire similar analytical techniques for determining causal links in disciplines like finance, seismology, and logistics, where understanding the drivers of complex events is of immense importance.
The study, “Detecting causality based on state space reconstruction from interspike intervals for neural spike trains,” was published July 28, 2025.
