Neural Noise

Time delays in computational models of neuronal and synaptic dynamics

2 min read

  • Time delays in neuronal and synaptic models matter for realistic dynamics and control (Frontiers).1
  • Supports neural data analysis and model-based decoding; tier-2.1 1

Weekly enrichment (2026-07-20)

  • The piece is an opinion article by Mojtaba Madadi Asl (Institute for Research in Fundamental Sciences and PIAIS, Tehran) in Frontiers in Computational Neuroscience, vol. 19, published 10 November 2025 (received 6 September, accepted 27 October 2025), doi:10.3389/fncom.2025.1700144.23
  • Core thesis: dendritic and axonal delays should be modeled as separate parameters rather than lumped; their sum (τ = τd + τa) is the total transmission delay that shapes neuronal dynamics, while their difference (ξ = τd − τa) sets the time lag perceived at the synapse and governs spike-timing-dependent plasticity (STDP).23
  • Reported delay magnitudes: dendritic delays are typically sub-millisecond to a few ms, whereas axonal delays range from a few ms in thalamo-cortical and cortico-tectal projections up to tens of ms in cortico-cortical connections.2
  • STDP sign depends on the delay balance: when τd > τa the pre-before-post order is preserved (potentiation, but with larger |Δt′|); when τd < τa the back-propagated postsynaptic spike reaches the synapse first and flips the outcome to depression; when τd = τa the delay-free case (Δt′ = Δt) is recovered.2
  • In reciprocally coupled plastic networks, delay configurations yield distinct emergent motifs — strong bidirectional loops when dendritic delays exceed axonal ones, versus loosely connected (both synapses depressed) or unidirectional structures otherwise (Madadi Asl, Valizadeh & Tass, Sci. Rep. 7:39682, 2017).24
  • Clinical link: demyelination (e.g., in multiple sclerosis) lengthens axonal delays and lowers conduction velocity, disrupting reliable transmission and impairing oscillatory responses; the framework also motivates timing-based therapeutic stimulation (coordinated reset, time-shifted stimulation) to desynchronize pathological oscillations in Parkinson’s disease models.2
  • Empirical anchor for delay values: a modelling atlas built from 780 epilepsy patients (F-TRACT; 34,354 stimulations, 774,445 cortico-cortical evoked potentials) estimated a median cortico-cortical axonal conduction delay of 10.2 ms in subjects older than 15 years, with only 16% of delays exceeding 20 ms and a median conduction velocity of 3.9 m/s.5
  • Illustrative model: a neural-mass MEG/EEG model reproduced an inter-areal delay of ~18 ms that decomposes into ~10 ms axonal propagation plus ~8 ms synaptic delay, concretely showing the “sum of delays” concept the opinion article formalizes.6

Footnotes

  1. https://news.google.com/rss/articles/CBMipwFBVV95cUxPTkQxZnk5LU04eDFFSTVxMDVpem1yd2NLeUxfVTdEUWhHbDY4NVVpWkRjaHZmNVlnNEw3bTZfbnpmWXJVQ2NCQ3RUakZwanJEcGIzYjU5QWlGdG85T1haQTFnLTRiWUN4aDFyMWJoUGlIbnY4MlYyeUExaWVoUnB5a2pwLTJjUXBITGYxdUhUcklhU0dJb01td0VUUWoxLUFWLXo3MUZrdw?oc=5 ↩ ↩2 ↩3

  2. https://www.frontiersin.org/journals/computational-neuroscience/articles/10.3389/fncom.2025.1700144/full ↩ ↩2 ↩3 ↩4 ↩5 ↩6

  3. https://doi.org/10.3389/fncom.2025.1700144 ↩ ↩2

  4. https://doi.org/10.1038/srep39682 ↩

  5. https://pmc.ncbi.nlm.nih.gov/articles/PMC9166555/ ↩

  6. https://www.fil.ion.ucl.ac.uk/~karl/A%20neural%20mass%20model%20for%20MEG.pdf ↩

Graph View