# Synchrony-Division Neural Multiplexing: An Encoding Model

^{1}

^{2}

^{3}

^{4}

^{5}

^{6}

^{*}

## Abstract

**:**

## 1. Introduction

## 2. Results

#### 2.1. Different Temporal Filters Map Distinct Features of a Mixed Stimulus

#### 2.2. Low-Dimensional Feature Space of the Neural Response Can Be Characterized by the STAs of Synchronous and Asynchronous Spikes

#### 2.3. Different Nonlinear Functions Are Associated with Synchronous and Asynchronous Spikes

#### 2.4. An Augmented LNL Cascade Model for Synchrony-Division Multiplexing

## 3. Discussion

#### 3.1. Subspace Feature Extractors: iSTAC vs. STC

#### 3.2. Choice of Static Nonlinearity in the LNL Model

#### 3.3. Generalized Linear Model (GLM) for Augmented LNL

## 4. Materials and Methods

#### 4.1. Stimulated Mixed Input

#### 4.2. Simulated Neural Ensemble and Its Response to the Mixed Input

#### 4.3. Generalized Linear Model (GLM) Details

#### 4.4. STA and STC Estimators

#### 4.5. iSTAC Estimator

## Author Contributions

## Funding

## Institutional Review Board Statement

## Informed Consent Statement

## Data Availability Statement

## Conflicts of Interest

## Appendix A

## Appendix B

## References

- Li, Y.G.; Winters, J.H.; Sollenberger, N.R. MIMO-OFDM for wireless communications: Signal detection with enhanced channel estimation. IEEE Trans. Commun.
**2002**, 50, 1471–1477. [Google Scholar] [CrossRef][Green Version] - Laughlin, S.B.; Sejnowski, T.J.J.S. Communication in neuronal networks. Science
**2003**, 301, 1870–1874. [Google Scholar] [CrossRef] [PubMed][Green Version] - Johnson, D.H. The relationship between spike rate and synchrony in responses of auditory-nerve fibers to single tones. J. Acoust. Soc. Am.
**1980**, 68, 1115–1122. [Google Scholar] [CrossRef] [PubMed] - Harvey, M.A.; Saal, H.P.; Dammann, J.F., III; Bensmaia, S.J. Multiplexing stimulus information through rate and temporal codes in primate somatosensory cortex. PLoS Biol.
**2013**, 11, e1001558. [Google Scholar] [CrossRef][Green Version] - Lankarany, M.; Al-Basha, D.; Ratté, S.; Prescott, S.A. Differentially synchronized spiking enables multiplexed neural coding. Proc. Natl. Acad. Sci. USA
**2019**, 116, 10097–10102. [Google Scholar] [CrossRef][Green Version] - Sullivan, W.; Konishi, M. Segregation of stimulus phase and intensity coding in the cochlear nucleus of the barn owl. J. Neurosci.
**1984**, 4, 1787–1799. [Google Scholar] [CrossRef][Green Version] - Saal, H.P.; Harvey, M.A.; Bensmaia, S.J. Rate and timing of cortical responses driven by separate sensory channels. eLife
**2015**, 4, e10450. [Google Scholar] [CrossRef] - Saal, H.P.; Bensmaia, S.J. Touch is a team effort: Interplay of submodalities in cutaneous sensibility. Trends Neurosci.
**2014**, 37, 689–697. [Google Scholar] [CrossRef] - Paninski, L. Maximum likelihood estimation of cascade point-process neural encoding models. Netw. Comput. Neural Syst.
**2004**, 15, 243–262. [Google Scholar] [CrossRef] - Latimer, K.W.; Barbera, D.; Sokoletsky, M.; Awwad, B.; Katz, Y.; Nelken, I.; Lampl, I.; Fairhall, A.L.; Priebe, N.J. Multiple timescales account for adaptive responses across sensory cortices. J. Neurosci.
**2019**, 39, 10019–10033. [Google Scholar] [CrossRef] - Churchland, M.M.; Byron, M.Y.; Sahani, M.; Shenoy, K.V. Techniques for extracting single-trial activity patterns from large-scale neural recordings. Curr. Opin. Neurobiol.
**2007**, 17, 609–618. [Google Scholar] [CrossRef] [PubMed][Green Version] - Paninski, L.; Pillow, J.; Lewi, J. Statistical models for neural encoding, decoding, and optimal stimulus design. Prog. Brain Res.
**2007**, 165, 493–507. [Google Scholar] - Rezaei, M.R.; Popovic, M.R.; Lankarany, M. A Time-Varying Information Measure for Tracking Dynamics of Neural Codes in a Neural Ensemble. Entropy
**2020**, 22, 880. [Google Scholar] [CrossRef] [PubMed] - Prescott, S.A.; Ratté, S.; De Koninck, Y.; Sejnowski, T.J. Pyramidal neurons switch from integrators in vitro to resonators under in vivo-like conditions. J. Neurophysiol.
**2008**, 100, 3030–3042. [Google Scholar] [CrossRef] [PubMed][Green Version] - Paninski, L. Convergence properties of three spike-triggered analysis techniques. Netw. Comput. Neural Syst.
**2003**, 14, 437–464. [Google Scholar] [CrossRef][Green Version] - Schwartz, O.; Pillow, J.W.; Rust, N.C.; Simoncelli, E.P. Spike-triggered neural characterization. J. Vis.
**2006**, 6, 13. [Google Scholar] [CrossRef][Green Version] - Pillow, J.W.; Simoncelli, E.P. Dimensionality reduction in neural models: An information-theoretic generalization of spike-triggered average and covariance analysis. J. Vis.
**2006**, 6, 9. [Google Scholar] [CrossRef][Green Version] - Lankarany, M.; Prescott, S.A. Multiplexed coding through synchronous and asynchronous spiking. BMC Neurosci.
**2015**, 16, P198. [Google Scholar] [CrossRef][Green Version] - Bojak, I.; Nowotny, T. 27th Annual Computational Neuroscience Meeting (CNS* 2018): Part Two. BMC Neurosci.
**2018**, 19, 65. [Google Scholar] - Ramachandran, P.; Zoph, B.; Le, Q.V. Searching for activation functions. arXiv
**2017**, arXiv:1710.05941. [Google Scholar] - Pillow, J.W.; Paninski, L.; Uzzell, V.J.; Simoncelli, E.P.; Chichilnisky, E. Prediction and decoding of retinal ganglion cell responses with a probabilistic spiking model. J. Neurosci.
**2005**, 25, 11003–11013. [Google Scholar] [CrossRef] [PubMed][Green Version] - Pillow, J.W.; Shlens, J.; Paninski, L.; Sher, A.; Litke, A.M.; Chichilnisky, E.; Simoncelli, E.P. Spatio-temporal correlations and visual signalling in a complete neuronal population. Nature
**2008**, 454, 995–999. [Google Scholar] [CrossRef] [PubMed][Green Version] - Paninski, L. Estimation of entropy and mutual information. Neural Comput.
**2003**, 15, 1191–1253. [Google Scholar] [CrossRef][Green Version] - Sharpee, T.; Rust, N.C.; Bialek, W. Analyzing neural responses to natural signals: Maximally informative dimensions. Neural Comput.
**2004**, 16, 223–250. [Google Scholar] [CrossRef][Green Version] - Moskovitz, T.H.; Roy, N.A.; Pillow, J.W. A comparison of deep learning and linear-nonlinear cascade approaches to neural encoding. BioRxiv
**2018**, 463422. [Google Scholar] [CrossRef] - Rezaei, M.R.; Gillespie, A.K.; Guidera, J.A.; Nazari, B.; Sadri, S.; Frank, L.M.; Eden, U.T.; Yousefi, A. A Comparison Study of Point-Process Filter and Deep Learning Performance in Estimating Rat Position Using an Ensemble of Place Cells. In Proceedings of the 2018 40th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC), Honolulu, HI, USA, 18–21 July 2018; pp. 4732–4735. [Google Scholar]
- Morris, C.; Lecar, H. Voltage oscillations in the barnacle giant muscle fiber. Biophys. J.
**1981**, 35, 193–213. [Google Scholar] [CrossRef][Green Version] - Ratté, S.; Hong, S.; De Schutter, E.; Prescott, S.A. Impact of neuronal properties on network coding: Roles of spike initiation dynamics and robust synchrony transfer. Neuron
**2013**, 78, 758–772. [Google Scholar] [CrossRef][Green Version] - Pirschel, F.; Kretzberg, J. Multiplexed population coding of stimulus properties by leech mechanosensory cells. J. Neurosci.
**2016**, 36, 3636–3647. [Google Scholar] [CrossRef][Green Version] - Destexhe, A.; Rudolph, M.; Paré, D. The high-conductance state of neocortical neurons in vivo. Nat. Rev. Neurosci.
**2003**, 4, 739–751. [Google Scholar] [CrossRef] - Prescott, S.A.; De Koninck, Y.; Sejnowski, T.J. Biophysical basis for three distinct dynamical mechanisms of action potential initiation. PLoS Comput. Biol.
**2008**, 4, e1000198. [Google Scholar] [CrossRef][Green Version]

**Figure 1.**Slow and fast features of a mixed signal can be inferred from responses of a homogeneous ensemble of neurons using the iSTAC method. (

**A**) Slow and fast signals comprising a mixed signal. (Bottom) Sample raster plot of 30 model neurons receiving the common mixed signal (and independent noise). Spikes evoked by the fast and slow signals cannot be distinguished visually. (

**B**) The iSTAC method was applied to spike-triggered mixed signal and eigenvalues and eigenvectors were obtained (see Methods). (

**Left**) The eigenvalues of the iSTAC matrix reveal two significant components of the population code. (

**Right**) The projection of spike-triggered mixed signal onto the main eigenvectors of the iSTAC matrix. Two clusters can be visually distinguished. (

**C**) The 1st and 2nd eigenvectors of the iSTAC matrix, V1 and V2, respectively, are shown against the spike-triggered average (STA). V1 resembles the STA filter reflecting slowly-varying changes in the signal. Unlike V1, V2 resembles a high-pass filter (differentiator) that reflects fast features of the mixed signal.

**Figure 2.**Synchronous and asynchronous spikes represent slow and fast features of the mixed signal, respectively. (

**A**) Synchronous (red) and asynchronous (blue) spikes are distinguished by setting a threshold on the instantaneous firing rate calculated by a narrow kernel (see Methods). Synchronous spikes evoked by the fast signals can be distinguished visually. (

**B**) The projection of spike-triggered mixed signal onto the STA

_{Sync}and STA

_{Async}. Two (visually) distinguishable clusters belong to asynchronous spikes representing the slow feature of the signal (blue dots) and synchronous spikes representing the fast features (red circles). (

**C**) The spike-triggered average of synchronous (red) and asynchronous (blue) spikes, namely, STA

_{Sync}and STA

_{Async}, respectively, is shown against the STA of all spikes (similar to Figure 1C).

**Figure 3.**Block diagram of decoding the mixed signal from spikes using (

**A**) the STA filter (light brown), (

**B**) a weighted sum of the 1st and 2nd eigenvectors of the iSTAC method (green), and (

**C**) a weighted sum of filtered asynchronous spikes (by STA

_{Async}) and filtered synchronous spikes (by STA

_{Sync}) (purple). Original mixed signal (black) is overlaid with reconstructed signal (color) in the plots. As can be seen in these figures, the reconstructed signal based on STA

_{Sync}and STA

_{Async}—similar to that obtained by eigenvectors of iSTAC method—can capture both slow and fast components of the signal accurately. “*” indicates the spiking signals.

**Figure 4.**Static nonlinearities underlying asynchronous spikes. (

**A**) Block diagram of LNL model for asynchronous spikes. (

**B**) Static nonlinearity calculated for asynchronous spikes is obtained by mapping the output of filtered stimulus to the instantaneous rate of asynchronous spikes (calculated with a wide kernel, σ = 25 msec). Static nonlinearity calculated based on 1st eigenvectors of the iSTAC method, ${v}_{1}$, (

**Left**) and STA

_{Async}(

**Right**). The solid black shows fitted rectifiers. (

**C**) The PSTHs constructed using the fitted nonlinearities based on ${v}_{1}$ were drawn against the PSTH of asynchronous spikes. (

**D**) The PSTHs constructed using the fitted nonlinearities based on STA

_{Async}were drawn against the PSTH of asynchronous spikes.

**Figure 5.**Static nonlinearities underlying synchronous spikes. (

**A**) Block diagram of LNL model for synchronous spikes. (

**B**) Static nonlinearity calculated for the synchronous spikes is obtained by mapping the output of filtered stimulus to the instantaneous synchronous events (calculated with a narrow kernel, σ = 1 msec). Static nonlinearity calculated based on 2nd eigenvectors of the iSTAC method, ${v}_{2}$, (

**Left**) and STA

_{Sync}(

**Right**). The solid black shows fitted sigmoid functions. (

**C**) The PSTHs constructed using the fitted nonlinearities based on ${v}_{2}$ were drawn against the PSTH of synchronous spikes. (

**D**) The PSTHs constructed using the fitted nonlinearities based on STA

_{Sync}were drawn against the PSTH of synchronous spikes.

**Figure 6.**Two-stream LNL model, referred to as augmented LNL model, enables co-existence of temporal and rate codes. (

**A**) Block diagram of the augmented LNL model for combining rate of asynchronous and synchronous spike events. (

**B**) The PSTHs estimated using a conventional Poisson GLM (red) are shown against the original PSTH (calculated with a 1 msec Gaussian kernel). (

**C**) The PSTHs estimated using the segmented LNL using temporal filters of iSTAC method. (

**D**) The PSTHs estimated using LNL using the segmented LNL using STA

_{Async}and STA

_{Sync}.

**Table 1.**Mean absolute error (MAE) and root mean squared error (RMSE) performance measure comparison between Poisson GLM and augmented LNL model on test data.

(MAE) | $\mathbf{R}\mathbf{M}\mathbf{S}\mathbf{E}$ | |||||
---|---|---|---|---|---|---|

Sync | Async | Mixed | Sync | Async | Mixed | |

Poisson GLM | 0.245 | 0.223 | 0.228 | 0.338 | 0.313 | 0.326 |

Augmented LNL ($\mathit{S}\mathit{T}{\mathit{A}}_{{V}_{1},}\mathit{S}\mathit{T}{\mathit{A}}_{{V}_{2}}$) | 0.101 | 0.104 | 0.102 | 0.137 | 0.134 | 0.135 |

Augmented LNL ($\mathit{S}\mathit{T}{\mathit{A}}_{\mathit{A}\mathit{S}\mathit{y}\mathit{n}\mathit{c}},\mathit{S}\mathit{T}{\mathit{A}}_{\mathit{S}\mathit{y}\mathit{n}\mathit{c}}$) | 0.103 | 0.107 | 0.106 | 0.140 | 0.151 | 0.149 |

Disclaimer/Publisher’s Note: The statements, opinions and data contained in all publications are solely those of the individual author(s) and contributor(s) and not of MDPI and/or the editor(s). MDPI and/or the editor(s) disclaim responsibility for any injury to people or property resulting from any ideas, methods, instructions or products referred to in the content. |

© 2023 by the authors. Licensee MDPI, Basel, Switzerland. This article is an open access article distributed under the terms and conditions of the Creative Commons Attribution (CC BY) license (https://creativecommons.org/licenses/by/4.0/).

## Share and Cite

**MDPI and ACS Style**

Rezaei, M.R.; Saadati Fard, R.; Popovic, M.R.; Prescott, S.A.; Lankarany, M. Synchrony-Division Neural Multiplexing: An Encoding Model. *Entropy* **2023**, *25*, 589.
https://doi.org/10.3390/e25040589

**AMA Style**

Rezaei MR, Saadati Fard R, Popovic MR, Prescott SA, Lankarany M. Synchrony-Division Neural Multiplexing: An Encoding Model. *Entropy*. 2023; 25(4):589.
https://doi.org/10.3390/e25040589

**Chicago/Turabian Style**

Rezaei, Mohammad R., Reza Saadati Fard, Milos R. Popovic, Steven A. Prescott, and Milad Lankarany. 2023. "Synchrony-Division Neural Multiplexing: An Encoding Model" *Entropy* 25, no. 4: 589.
https://doi.org/10.3390/e25040589