Overview
- The peer-reviewed study by Kazuya Sawada and colleagues was published July 28, 2025, in Physical Review E and is now being highlighted in institutional coverage.
- The method extends convergent cross mapping by pairing ISI-based state-space reconstruction with a temporal correspondence procedure tailored for spike trains.
- Numerical experiments on a standard neuron model recovered bidirectional, unidirectional, and absent coupling with high fidelity.
- Performance held up under weak coupling and internal noise, indicating robustness for biologically realistic conditions.
- Tests so far covered two to three neurons, with the authors prioritizing scale-up to larger networks and noting potential relevance for neuroscience and other point-process fields such as finance, seismology, and logistics.