Poster (After Call) FZJ-2026-05213

http://join2-wiki.gsi.de/foswiki/pub/Main/Artwork/join2_logo100x88.png
A Quality Measure for Evaluation of Spatio-temporal Patterns in Parallel Spike Trains

 ;  ;

2026

Bernstein Conference 2026, BC2026, FrankfurtFrankfurt, Germany, 28 Sep 2026 - 1 Oct 20262026-09-282026-10-01 ()

Abstract: Spiking neural networks that perform on spike-time-based coding and computation schemes havebeen attracting the interest of researchers [1, 2], as they are expected to enable faster, sparser, andmore energy-efficient information processing than the conventional, non-spiking neural network ar-chitectures. In computational neuroscience, the synfire chain [3, 4] has been proposed as a modelfor brain processing based on spike timing. Since this model predicts stably propagating packetsof synchronized spikes, the search for timed spiking activity in the form of synchronous activity orspatio-temporal pattern (STP) in biological spike train data was initiated. However, previous reportson STPs in biological neural networks have been limited (see e.g. [5, 6, 7] among others), due toa lack of a standard method to detect and evaluate such signatures. We have proposed a measurefor evaluating the quality of STPs in terms of the unlikeliness of their occurrences based on multipleparameters of the STPs, such as their size (the number of neurons involved), duration (the maximumlag between the spikes in a pattern), and the firing rates of the involved neurons. For stationary spiketrain data, a formula for its computation was analytically derived, and, in more general cases of non-stationary spike trains, it can be estimated from stochastic simulations or surrogate samplings of theoriginal data [8]. In this presentation, we introduce this quality measure and report its application todatasets of 100 spike trains simultaneously recorded from the macaque motor cortex while the animalperformed a reach-to-grasp arm movement task [9]. We find that the STPs of the highest quality inthe data are composed of only a small subset of the recorded neurons. Furthermore, these patternsoften occur at the same timing overlapping each other, with spikes shared among different patterns.We quantify this observation by examining the overlap matrix, which represents the degree of overlapbetween every pair of patterns, and infer which patterns form a cluster of densely overlapping patternsand which neurons compose those patterns. Based on these results, we discuss the mechanisms under-lying this phenomenon and their implications for spike-time-based computation, which may provideinsights into the information processing in spiking neural networks, both biological and artificial.[1]J. Göltz et al., Nat. Mach. Intell., 3(9), 823-835, 2021.[2]J. Oberste-Frielinghaus et al., arXiv, arXiv:2508.12975, 2025.[3]M. Abeles, Local Cortical Circuits: An Electrophysiological Study, 1982.[4]M. Abeles, Corticonics: Neural Circuits of the Cerebral Cortex, 1991.[5]Y. Prut et al., J. Neurophysiol., 79(6), 2857-2874, 1998.[6]Y. Ikegaya et al., Science, 304(5670), 559-564, 2004.[7]W. Xie et al., Nature, 635(8040), 935-942, 2024.[8]G. Palm et al., Biol. Cybern., accepted, 2026.[9]T. Brochier et al., Sci. Data., 5(1), 180055, 2018.


Contributing Institute(s):
  1. Computational and Systems Neuroscience (IAS-6)
Research Program(s):
  1. 5231 - Neuroscientific Foundations (POF4-523) (POF4-523)
  2. Algorithms of Adaptive Behavior and their Neuronal Implementation in Health and Disease (iBehave-20220812) (iBehave-20220812)
  3. GRK 2416 - GRK 2416 MultiSenses-MultiScales: Neue Ansätze zur Aufklärung neuronaler multisensorischer Integration (368482240) (368482240)
  4. DFG project G:(GEPRIS)561027837 - Einbettung neuronaler Ensembles und deren Signaturen (964/25) (561027837) (561027837)

Appears in the scientific report 2026
Click to display QR Code for this record

The record appears in these collections:
Document types > Presentations > Poster
Institute Collections > IAS > IAS-6
Workflow collections > Public records
Publications database

 Record created 2026-10-26, last modified 2026-11-09



Rate this document:

Rate this document:
1
2
3
 
(Not yet reviewed)