DOE PAGES title logo U.S. Department of Energy
Office of Scientific and Technical Information

Title: Role of the site of synaptic competition and the balance of learning forces for Hebbian encoding of probabilistic Markov sequences

Abstract

The majority of distinct sensory and motor events occur as temporally ordered sequences with rich probabilistic structure. Sequences can be characterized by the probability of transitioning from the current state to upcoming states (forward probability), as well as the probability of having transitioned to the current state from previous states (backward probability). Despite the prevalence of probabilistic sequencing of both sensory and motor events, the Hebbian mechanisms that mold synapses to reflect the statistics of experienced probabilistic sequences are not well understood. Here, we show through analytic calculations and numerical simulations that Hebbian plasticity (correlation, covariance, and STDP) with pre-synaptic competition can develop synaptic weights equal to the conditional forward transition probabilities present in the input sequence. In contrast, post-synaptic competition can develop synaptic weights proportional to the conditional backward probabilities of the same input sequence. We demonstrate that to stably reflect the conditional probability of a neuron’s inputs and outputs, local Hebbian plasticity requires balance between competitive learning forces that promote synaptic differentiation and homogenizing learning forces that promote synaptic stabilization. The balance between these forces dictates a prior over the distribution of learned synaptic weights, strongly influencing both the rate at which structure emerges and the entropy ofmore » the final distribution of synaptic weights. Together, these results demonstrate a simple correspondence between the biophysical organization of neurons, the site of synaptic competition, and the temporal flow of information encoded in synaptic weights by Hebbian plasticity while highlighting the utility of balancing learning forces to accurately encode probability distributions, and prior expectations over such probability distributions.« less

Authors:
 [1];  [2];  [3]
  1. Lawrence Berkeley National Lab. (LBNL), Berkeley, CA (United States). Life Sciences and Computational Research Divisions
  2. Stanford Univ., CA (United States). Dept. of Applied Physics
  3. Univ. of California, San Francisco, CA (United States). Dept. of Physiology. Center for Integrative Neuroscience; Howard Hughes Medical Inst., Chevy Chase, MD (United States)
Publication Date:
Research Org.:
Lawrence Berkeley National Laboratory (LBNL), Berkeley, CA (United States)
Sponsoring Org.:
USDOE
OSTI Identifier:
1628198
Grant/Contract Number:  
AC02-05CH11231
Resource Type:
Accepted Manuscript
Journal Name:
Frontiers in Computational Neuroscience
Additional Journal Information:
Journal Volume: 9; Journal ID: ISSN 1662-5188
Publisher:
Frontiers Research Foundation
Country of Publication:
United States
Language:
English
Subject:
97 MATHEMATICS AND COMPUTING; 59 BASIC BIOLOGICAL SCIENCES; Mathematical & Computational Biology; Neurosciences & Neurology; Hebbian plasticity; pre/post-synaptic; probability; sequences; birdsong

Citation Formats

Bouchard, Kristofer E., Ganguli, Surya, and Brainard, Michael S. Role of the site of synaptic competition and the balance of learning forces for Hebbian encoding of probabilistic Markov sequences. United States: N. p., 2015. Web. doi:10.3389/fncom.2015.00092.
Bouchard, Kristofer E., Ganguli, Surya, & Brainard, Michael S. Role of the site of synaptic competition and the balance of learning forces for Hebbian encoding of probabilistic Markov sequences. United States. https://doi.org/10.3389/fncom.2015.00092
Bouchard, Kristofer E., Ganguli, Surya, and Brainard, Michael S. Tue . "Role of the site of synaptic competition and the balance of learning forces for Hebbian encoding of probabilistic Markov sequences". United States. https://doi.org/10.3389/fncom.2015.00092. https://www.osti.gov/servlets/purl/1628198.
@article{osti_1628198,
title = {Role of the site of synaptic competition and the balance of learning forces for Hebbian encoding of probabilistic Markov sequences},
author = {Bouchard, Kristofer E. and Ganguli, Surya and Brainard, Michael S.},
abstractNote = {The majority of distinct sensory and motor events occur as temporally ordered sequences with rich probabilistic structure. Sequences can be characterized by the probability of transitioning from the current state to upcoming states (forward probability), as well as the probability of having transitioned to the current state from previous states (backward probability). Despite the prevalence of probabilistic sequencing of both sensory and motor events, the Hebbian mechanisms that mold synapses to reflect the statistics of experienced probabilistic sequences are not well understood. Here, we show through analytic calculations and numerical simulations that Hebbian plasticity (correlation, covariance, and STDP) with pre-synaptic competition can develop synaptic weights equal to the conditional forward transition probabilities present in the input sequence. In contrast, post-synaptic competition can develop synaptic weights proportional to the conditional backward probabilities of the same input sequence. We demonstrate that to stably reflect the conditional probability of a neuron’s inputs and outputs, local Hebbian plasticity requires balance between competitive learning forces that promote synaptic differentiation and homogenizing learning forces that promote synaptic stabilization. The balance between these forces dictates a prior over the distribution of learned synaptic weights, strongly influencing both the rate at which structure emerges and the entropy of the final distribution of synaptic weights. Together, these results demonstrate a simple correspondence between the biophysical organization of neurons, the site of synaptic competition, and the temporal flow of information encoded in synaptic weights by Hebbian plasticity while highlighting the utility of balancing learning forces to accurately encode probability distributions, and prior expectations over such probability distributions.},
doi = {10.3389/fncom.2015.00092},
journal = {Frontiers in Computational Neuroscience},
number = ,
volume = 9,
place = {United States},
year = {Tue Jul 21 00:00:00 EDT 2015},
month = {Tue Jul 21 00:00:00 EDT 2015}
}

Journal Article:
Free Publicly Available Full Text
Publisher's Version of Record

Figures / Tables:

FIGURE 1 FIGURE 1: Production and learning of sequences and their probabilistic characterizations. (A) (top) Example spectrogram (power at frequency vs. time) of the song from one Bengalese finch. Songs are composed of categorical vocalizations (syllables), organized into probabilistic sequences. (bottom) Forward probability transition diagram for the song displayed above. Here, eachmore » node corresponds to a syllable, and edge width corresponds to conditional probability of transitioning to a syllable from a given syllable. Dots at the end of each edge denote the forward direction of song during production. (B) Example spectrograms of the song of one Bengalese finch recorded at a juvenile age (55 days of age, top) before exposure to tutor song, the song from the same bird after 10 days of tutoring, (65 days of age, middle), and the song of the tutor (bottom). To ease visual demonstration of learning, we have chosen a bird trained with a linear sequence of syllables. (C) Backward and forward probabilities have different semantics. For a given state of the system (e.g., “x”), the backward probability describes the distribution of previous states [PB(si, sj) = P(si(t−1)|sj(t))]. In contrast, the forward probability describes the distribution of upcoming states [PF(si, sj) = P(sj(t+1)|si(t))]. Generally speaking, for a given transition, the backward and forward probabilities will not be equal [e.g., PB(xc) = 1, PF(xc) = 0.4].« less

Save / Share:

Works referenced in this record:

Equilibrium Properties of Temporally Asymmetric Hebbian Plasticity
journal, January 2001


Learning Patterns and Pattern Sequences by Self-Organizing Nets of Threshold Elements
journal, November 1972


Learning Input Correlations through Nonlinear Temporally Asymmetric Hebbian Plasticity
journal, May 2003


Intrinsic Stabilization of Output Rates by Spike-Based Hebbian Learning
journal, December 2001


Spike-timing-dependent plasticity: common themes and divergent vistas
journal, December 2002


Signal-Driven Computations in Speech Processing
journal, August 2002


Complex Sequencing Rules of Birdsong Can be Explained by Simple Hidden Markov Processes
journal, September 2011


Control of Spoken Vowel Acoustics and the Influence of Phonetic Context in Human Speech Sensorimotor Cortex
journal, September 2014


Target Cell-Dependent Normalization of Transmitter Release at Neocortical Synapses
journal, May 2005


Corticonics
book, January 2011


Neuronal populations and single cells representing learned auditory objects
journal, August 2003

  • Gentner, Timothy Q.; Margoliash, Daniel
  • Nature, Vol. 424, Issue 6949
  • DOI: 10.1038/nature01731

Local Dendritic Activity Sets Release Probability at Hippocampal Synapses
journal, August 2008


The Persistence of Long-Term Memory
journal, September 2004


Conservation of total synaptic weight through balanced synaptic depression and potentiation
journal, April 2003


Synaptic plasticity at hippocampal mossy fibre synapses
journal, November 2005

  • Nicoll, Roger A.; Schmitz, Dietmar
  • Nature Reviews Neuroscience, Vol. 6, Issue 11
  • DOI: 10.1038/nrn1786

Functional Significance of Long-Term Potentiation for Sequence Learning and Prediction
journal, January 1996


Rapid Hebbian axonal remodeling mediated by visual stimulation
journal, May 2014


Mechanisms Underlying the Rapid Induction and Sustained Expression of Synaptic Homeostasis
journal, November 2006


Modeling Synaptic Plasticity in Conjunction with the Timing of Pre- and Postsynaptic Action Potentials
journal, February 2000


Activity-dependent scaling of quantal amplitude in neocortical neurons
journal, February 1998

  • Turrigiano, Gina G.; Leslie, Kenneth R.; Desai, Niraj S.
  • Nature, Vol. 391, Issue 6670
  • DOI: 10.1038/36103

Competitive Hebbian learning through spike-timing-dependent synaptic plasticity
journal, September 2000

  • Song, Sen; Miller, Kenneth D.; Abbott, L. F.
  • Nature Neuroscience, Vol. 3, Issue 9
  • DOI: 10.1038/78829

The Adaptive Self-organization of Serial Order in Behavior: Speech, Language, and Motor Control
book, January 1986


Role of Homeostasis in Learning Sparse Representations
journal, July 2010


Learning and production of movement sequences: Behavioral, neurophysiological, and modeling perspectives
journal, November 2004


Ocular dominance column development: analysis and simulation
journal, August 1989


Synaptic computation underlying probabilistic inference
journal, December 2009

  • Soltani, Alireza; Wang, Xiao-Jing
  • Nature Neuroscience, Vol. 13, Issue 1
  • DOI: 10.1038/nn.2450

Some Networks That Can Learn, Remember, and Reproduce any Number of Complicated Space-Time Patterns, I
journal, January 1969


What can we learn from synaptic weight distributions?
journal, December 2007


Increases in phonotactic probability facilitate spoken nonword repetition
journal, February 2005


A Compact Statistical Model of the Song Syntax in Bengalese Finch
journal, March 2011


Statistical Learning by 8-Month-Old Infants
journal, December 1996


Probabilistic reasoning by neurons
journal, June 2007


What Can a Neuron Learn with Spike-Timing-Dependent Plasticity?
journal, November 2005

  • Legenstein, Robert; Naeger, Christian; Maass, Wolfgang
  • Neural Computation, Vol. 17, Issue 11
  • DOI: 10.1162/0899766054796888

Compressed Sensing, Sparsity, and Dimensionality in Neuronal Information Processing and Data Analysis
journal, July 2012


The Neural Basis of Temporal Processing
journal, July 2004


Intrinsic Stability of Temporally Shifted Spike-Timing Dependent Plasticity
journal, November 2010


Canonical Microcircuits for Predictive Coding
journal, November 2012


Emergence of simple-cell receptive field properties by learning a sparse code for natural images
journal, June 1996

  • Olshausen, Bruno A.; Field, David J.
  • Nature, Vol. 381, Issue 6583
  • DOI: 10.1038/381607a0

Heterosynaptic changes accompany long-term but not short-term potentiation of the perforant path in the anaesthetized rat.
journal, June 1985


The Role of Constraints in Hebbian Learning
journal, January 1994


Spike Timing-Dependent Plasticity of Neural Circuits
journal, September 2004


The Latency and Duration of Rapid Movement Sequences: Comparisons of Speech and Typewriting
book, January 1978


Spike-timing-dependent plasticity: common themes and divergent vistas
journal, December 2002


Increases in phonotactic probability facilitate spoken nonword repetition
journal, February 2005


Spike Timing-Dependent Plasticity of Neural Circuits
journal, September 2004


The Persistence of Long-Term Memory
journal, September 2004


Mechanisms Underlying the Rapid Induction and Sustained Expression of Synaptic Homeostasis
journal, November 2006


Local Dendritic Activity Sets Release Probability at Hippocampal Synapses
journal, August 2008


What can we learn from synaptic weight distributions?
journal, December 2007


Synaptic Economics: Competition and Cooperation in Synaptic Plasticity
journal, September 1996


Cortical Development and Remapping through Spike Timing-Dependent Plasticity
journal, October 2001


Determinants of Spike Timing-Dependent Synaptic Plasticity
journal, December 2001


Rate, Timing, and Cooperativity Jointly Determine Cortical Synaptic Plasticity
journal, December 2001


Activity-dependent scaling of quantal amplitude in neocortical neurons
journal, February 1998

  • Turrigiano, Gina G.; Leslie, Kenneth R.; Desai, Niraj S.
  • Nature, Vol. 391, Issue 6670
  • DOI: 10.1038/36103

Emergence of simple-cell receptive field properties by learning a sparse code for natural images
journal, June 1996

  • Olshausen, Bruno A.; Field, David J.
  • Nature, Vol. 381, Issue 6583
  • DOI: 10.1038/381607a0

Competitive Hebbian learning through spike-timing-dependent synaptic plasticity
journal, September 2000

  • Song, Sen; Miller, Kenneth D.; Abbott, L. F.
  • Nature Neuroscience, Vol. 3, Issue 9
  • DOI: 10.1038/78829

Conservation of total synaptic weight through balanced synaptic depression and potentiation
journal, April 2003


Neuronal populations and single cells representing learned auditory objects
journal, August 2003

  • Gentner, Timothy Q.; Margoliash, Daniel
  • Nature, Vol. 424, Issue 6949
  • DOI: 10.1038/nature01731

Probabilistic reasoning by neurons
journal, June 2007


Synaptic computation underlying probabilistic inference
journal, December 2009

  • Soltani, Alireza; Wang, Xiao-Jing
  • Nature Neuroscience, Vol. 13, Issue 1
  • DOI: 10.1038/nn.2450

Synaptic plasticity at hippocampal mossy fibre synapses
journal, November 2005

  • Nicoll, Roger A.; Schmitz, Dietmar
  • Nature Reviews Neuroscience, Vol. 6, Issue 11
  • DOI: 10.1038/nrn1786

Functional Significance of Long-Term Potentiation for Sequence Learning and Prediction
journal, January 1996


Heterosynaptic changes accompany long-term but not short-term potentiation of the perforant path in the anaesthetized rat.
journal, June 1985


Signal-Driven Computations in Speech Processing
journal, August 2002


Rapid Hebbian axonal remodeling mediated by visual stimulation
journal, May 2014


Ocular dominance column development: analysis and simulation
journal, August 1989


Compressed Sensing, Sparsity, and Dimensionality in Neuronal Information Processing and Data Analysis
journal, July 2012


The Neural Basis of Temporal Processing
journal, July 2004


Modeling Synaptic Plasticity in Conjunction with the Timing of Pre- and Postsynaptic Action Potentials
journal, February 2000


Intrinsic Stabilization of Output Rates by Spike-Based Hebbian Learning
journal, December 2001


What Can a Neuron Learn with Spike-Timing-Dependent Plasticity?
journal, November 2005

  • Legenstein, Robert; Naeger, Christian; Maass, Wolfgang
  • Neural Computation, Vol. 17, Issue 11
  • DOI: 10.1162/0899766054796888

The Role of Constraints in Hebbian Learning
journal, January 1994


Role of Homeostasis in Learning Sparse Representations
journal, July 2010


Intrinsic Stability of Temporally Shifted Spike-Timing Dependent Plasticity
journal, November 2010


Some Networks That Can Learn, Remember, and Reproduce any Number of Complicated Space-Time Patterns, I
journal, January 1969


Synaptic Basis for Whisker Deprivation-Induced Synaptic Depression in Rat Somatosensory Cortex
journal, April 2006


Two Coincidence Detectors for Spike Timing-Dependent Plasticity in Somatosensory Cortex
journal, April 2006


The time course and amplitude of EPSPs evoked at synapses between pairs of CA3/CA1 neurons in the hippocampal slice
journal, March 1990


Synaptic transmission between individual pyramidal neurons of the rat visual cortex in vitro
journal, January 1991


Variable Sequencing Is Actively Maintained in a Well Learned Motor Skill
journal, October 2012


Stable Hebbian Learning from Spike Timing-Dependent Plasticity
journal, December 2000


Neural Encoding and Integration of Learned Probabilistic Sequences in Avian Sensory-Motor Circuitry
journal, November 2013


Temporal Structure in Zebra Finch Song: Implications for Motor Coding
journal, January 2006


Matching Recall and Storage in Sequence Learning with Spiking Neural Networks
journal, June 2013


Dynamic Encoding of Speech Sequence Probability in Human Temporal Cortex
journal, May 2015


Complex sequencing rules of birdsong can be explained by simple hidden Markov processes
text, January 2010


A compact statistical model of the song syntax in Bengalese finch
text, January 2010


Works referencing / citing this record:

Timing during transitions in Bengalese finch song: implications for motor sequencing
journal, September 2017

  • Troyer, Todd W.; Brainard, Michael S.; Bouchard, Kristofer E.
  • Journal of Neurophysiology, Vol. 118, Issue 3
  • DOI: 10.1152/jn.00296.2017

Advantages of comparative studies in songbirds to understand the neural basis of sensorimotor integration
journal, August 2017

  • Murphy, Karagh; James, Logan S.; Sakata, Jon T.
  • Journal of Neurophysiology, Vol. 118, Issue 2
  • DOI: 10.1152/jn.00623.2016

Auditory-induced neural dynamics in sensory-motor circuitry predict learned temporal and sequential statistics of birdsong
journal, August 2016

  • Bouchard, Kristofer E.; Brainard, Michael S.
  • Proceedings of the National Academy of Sciences, Vol. 113, Issue 34
  • DOI: 10.1073/pnas.1606725113