Efficient computation via sparse coding in electrosensory neural networks

doi:10.1016/j.conb.2011.05.016

Current Opinion in Neurobiology

Volume 21, Issue 5, October 2011, Pages 752-760

https://doi.org/10.1016/j.conb.2011.05.016 Get rights and content

The electric sense combines spatial aspects of vision and touch with temporal features of audition. Its accessible neural architecture shares similarities with mammalian sensory systems and allows for recordings from successive brain areas to test hypotheses about neural coding. Further, electrosensory stimuli encountered during prey capture, navigation, and communication, can be readily synthesized in the laboratory. These features enable analyses of the neural circuitry that reveal general principles of encoding and decoding, such as segregation of information into separate streams and neural response sparsification. A systems level understanding arises via linkage between cellular differentiation and network architecture, revealed by in vitro and in vivo analyses, while computational modeling reveals how single cell dynamics and connectivity shape the sparsification process.

Highlights

► The electric sense encodes a wide range of stimuli including navigational cues, communication signals and prey signals. ► Electroreceptors on the body respond linearly to this range, except at high frequencies where population synchrony occurs. ► The electrosensory hindbrain structure (ELL), with limited cell types, has been studied to reveal how a sparse code emerges. ► This code relies on linked cellular and architectural properties including segregated maps and cell types. ► Computational studies ground this code in neural dynamics, and support studies of deeper sparsification in the midbrain.

Introduction

Peripheral sensory neurons typically encode a wide range of environmental stimuli. Moving more centrally, this dense coding is progressively replaced by a narrowing of neural responses to a subset of features of the whole stimulus ensemble [1, 2, 3, 4, 5]. Such sparse coding makes for easier read-out by downstream neurons, and is more energy efficient [6]. The mechanisms underlying sparsification are still elusive, and their elucidation is a fundamental goal of systems neuroscience. In particular, there is a quest for links between cellular/molecular properties – revealed through in vitro recordings – and network architecture and responses – revealed through in vivo and anatomical studies. Ideally the process would be understood for a behaving animal. Computational modeling has revealed that sparse coding requires nonlinear mechanisms in both the biophysical and network properties of sensory circuitry.

This review looks at the successive transformations of sensory input in the electrosensory system of weakly electric fish. This system has enabled major advances in understanding how differences in anatomy and cellular physiology underlie sparse coding. This is because links between in vitro and in vivo subthreshold and spiking dynamics are possible in the awake animal responding to a wide range of behaviorally relevant sensory input. The review is organized as follows. We first introduce the electrosensory system and its natural stimuli. We then review how electroreceptors can respond differentially to all these stimuli. Next, we describe the first order electrosensory target (hindbrain) emphasizing how linked cellular and architectural differentiation initiates the routing of these inputs into separate output streams. Finally, we summarize recent work on how an even sparser and more efficient representation of the stimulus ensemble is generated in the midbrain. Throughout, we highlight the dynamical and computational features of neural function that provide the substrate for the sparse code.

Section snippets

The electrosensory system

Gymnotiform wave-type weakly electric fish produce an electric field by repetitively discharging a specialized electric organ located in their tail (Figure 1a). This electric field is commonly referred to as the electric organ discharge (EOD). The EOD is sensed everywhere on the body as a transdermal potential difference. Perturbations of the EOD, caused by nearby objects with conductivity different than that of the surrounding water (e.g. prey, conspecifics), modify the transdermal potential

Electrosensory stimuli

EOD AM's can arise in multiple behaviorally relevant situations. Prey objects give rise to spatially localized AMs that contain low temporal frequencies (Figure 1a) [11]. Self-generated AMs through body movements can induce spatially diffuse low frequency AMs that interfere with prey detection, but covering the entire body surface [10, 12]. Weakly electric fish are often found in groups of 2–4 individuals [13]; the interference between the EODs of two neighboring fish causes a sinusoidal AM

Electroreceptors and coding of natural sensory input

Electroreceptors in A. leptorhynchus (∼15,000 total) are phase locked to the EOD and discharge at rates of 100–500 spikes/s in a highly variable but patterned manner when the EOD is unperturbed [15]. The baseline discharges of individual electroreceptors are uncorrelated, allowing them to act as independent coding channels [16]. Patterning in the spontaneous discharge is characterized by a long interspike interval (ISI) following on average by a short ISI and vice versa [16, 17, 18] and is

The electrosensory lateral line lobe: efficient circuits for feature extraction and sparse coding

Electroreceptors project to the electrosensory lateral line lobe (ELL). Each afferent trifurcates and synapses onto pyramidal cells within three structurally similar parallel segments: centromedial (CMS), centrolateral (CLS) and lateral (LS) [29, 30, 31, 32, 33]. Ampullary electroreceptors project to the medial segment (MS) and are not considered here (Figure 2). All three segments thus receive identical input and each forms an entire topographic map of the animal's body surface. The output

Feature selectivity in the electrosensory midbrain (TS)

The separation of electrocommunication and electrolocation channels has been nearly completed in TS. Indeed, TS neurons generally respond to a much narrower range of spatiotemporal frequencies than ELL pyramidal cells [65, 66, 67]. Some are tuned to low frequencies, others to mid-range or even high frequencies (Figure 3a). In addition, some TS neurons respond selectively to communication stimuli [68] (Figure 3a).

Moreover, TS neurons also respond selectively to objects (Figure 3b). Indeed,

Summary

Through a combination of intrinsic and network properties, electrosensory circuitry progressively segregates inputs of varying temporal and spatial scales encountered in different behavioral contexts. This results in the creation of separate neural streams for electrolocation and electrocommunication. Experimental and computational work have revealed general principles of sparse coding that can be summarized as follows:

1-
Electroreceptors use different strategies to encode different temporal

Funding

This research was supported by CIHR (M.J.C., A.L., L.M.), NSERC (M.J.C., A.L.), CFI (M.J.C.), and CRC (M.J.C.). The funders had no role in either study design, writing of the manuscript, or the decision to submit the manuscript.

References and recommended reading

Papers of particular interest, published within the period of review, have been highlighted as:

• of special interest
•• of outstanding interest

Acknowledgements

We would like to dedicate this review to Joseph Bastian, a friend, colleague, mentor, and pioneer in research on the electrosensory system.

References (72)

B.A. Olshausen et al.
Sparse coding of sensory inputs
Curr Opin Neurobiol
(2004)
S.A. Stamper et al.
Species differences in group size and electrosensory interference in weakly electric fishes: implications for electrosensory processing
Behav Brain Res
(2010)
M.J. Chacron et al.
Feedback and feedforward control of frequency tuning to naturalistic stimuli
J Neurosci
(2005)
J.U. Ramcharitar et al.
Global electrosensory oscillations enhance directional responses of midbrain neurons in eigenmannia
J Neurophysiol
(2006)
M.J. Chacron et al.
Delayed excitatory and inhibitory feedback shape neural information transmission
Phys Rev E
(2005)
N.J. Cowan et al.
The critical role of locomotion mechanics in decoding sensory systems
J Neurosci
(2007)
J. Bastian
Electrolocation II. The effects of moving objects and other electrical stimuli on the activities of two categories of posterior lateral line lobe cells in Apteronotus albifrons
J Comp Physiol A
(1981)
H.B. Barlow
Single Units and sensation: a neuron doctrine for perceptual psychology
Perception
(1972)
E.T. Rolls et al.
Sparseness of the neuronal representation of stimuli in the primate temporal visual cortex
J Neurophysiol
(1995)
R.Q. Quiroga et al.
Invariant visual representation by single neurons in the human brain [see comment]
Nature
(2005)

J. Perez-Orive et al.

Oscillations and sparsening of odor representations in the mushroom body

Science

(2002)

D. Attwell et al.

An energy budget for signaling in the grey matter of the brain

J Cereb Blood Flow Metab

(2001)

L. Chen et al.

Modeling signal and background components of electrosensory scenes

J Comp Physiol A Neuroethol Sens Neural Behav Physiol

(2005)

M. Kelly et al.

Electric field interactions in pairs of electric fish: modeling and mimicking naturalistic input

Biol Cybern

(2008)

D. Babineau et al.

Spatial acuity and prey detection in weakly electric fish

PLoS Comput Biol

(2007)

J. Bastian

Plasticity in an electrosensory system. I. General features of a dynamic sensory filter

J Neurophysiol

(1996)

M.E. Nelson et al.

Prey capture in the weakly electric fish Apteronotus albifrons: sensory acquisition strategies and electrosensory consequences

J Exp Biol

(1999)

J. Bastian

Plasticity of feedback inputs in the apteronotid electrosensory system

J Exp Biol

(1999)

G.J. Hupe et al.

Electrocommunication signals in free swimming brown ghost knifefish, Apteronotus leptorhynchus

J Exp Biol

(2008)

D. Gussin et al.

Limits of linear rate coding of dynamic stimuli by electroreceptor afferents

J Neurophysiol

(2007)

M.J. Chacron et al.

Electroreceptor neuron dynamics shape information transmission

Nat Neurosci

(2005)

R. Ratnam et al.

Non-renewal statistics of electrosensory afferent spike trains: implications for the detection of weak sensory signals

J Neurosci

(2000)

M.J. Chacron et al.

Suprathreshold stochastic firing dynamics with memory in P-type electroreceptors

Phys Rev Lett

(2000)

M.J. Chacron et al.

Interspike interval correlations, memory, adaptation, and refractoriness in a leaky integrate-and-fire model with threshold fatigue

Neural Comput

(2003)

J. Benda et al.

Linear versus nonlinear signal transmission in neuron models with adaptation currents or dynamic thresholds

J Neurophysiol

(2010)

M.J. Chacron et al.

Negative interspike interval correlations increase the neuronal capacity for encoding time-varying stimuli

J Neurosci

(2001)

M.J. Chacron et al.

Experimental and Theoretical demonstration of noise shaping by interspike interval correlations

Proc SPIE

(2005)

O. Avila Akerberg et al.

Nonrenewal spike train statistics: causes and consequences on neural coding

Exp Brain Res

(2011)

A. Longtin et al.

Correlations and memory in neurodynamical systems

W. Nesse et al.

Biophysical information representation in temporally correlated spike trains

Proc Natl Acad Sci USA

(2010)

J. Benda et al.

Spike-frequency adaptation separates transient communication signals from background oscillations

J Neurosci

(2005)

J. Benda et al.

A synchronization-desynchronization code for natural communication signals

Neuron

(2006)

N.J. Berman et al.

a Inhibition evoked from primary afferents in the electrosensory lateral line lobe of the weakly electric fish (Apteronotus leptorhynchus)

J Neurophysiol

(1998)

N.J. Berman et al.

b Interaction of GABA_B-mediated inhibition with voltage-gated currents of pyramidal cells: computational mechanism of a sensory searchlight

J Neurophysiol

(1998)

N.J. Berman et al.

c Distal versus proximal inhibitory shaping of feedback excitation in the electrosensory lateral line lobe: implications for sensory filtering

J Neurophysiol

(1998)

N.J. Berman et al.

Neural architecture of the electrosensory lateral line lobe: adaptations for coincidence detection, a sensory searchlight and frequency-dependent adaptive filtering

J Exp Biol

(1999)

Cited by (86)

Stochastic resonance: The response to envelope modulation signal for neural networks with different topologies
2022, Physica A: Statistical Mechanics and its Applications
Citation Excerpt :
Green treefrog can use two frequency bands in long-distance vocal communication [29]. It experimentally showed that the electric fish would generate sinusoidal amplitude modulations beat pattern if the electric generating organs of the two electric fishes were disturbed [30]. Moreover, Chialvo et al. first found that moderate noise enables individual neuron to optimally response to mixed periodic signals with harmonic frequencies [31,32].
Neurons receive complex multiple signals from different regions in the biology system, and the neuron information carried by the multi-frequency signals can be detected by the neural system. Noise can induce stochastic resonance and promote the response of the neural network to a weak signal. The responses to weak envelope modulation signals for the neural networks with different topologies are investigated under electromagnetic induction and Gaussian white noise. We analyze and compare the stochastic resonance with tuning system parameters. The results show that in the three different networks, network systems exhibit double resonance with the variation of noise and beat frequency. Appropriate noise and electromagnetic induction can promote the generation of stochastic resonance and affect the response of envelope modulation signal in the neural network. The difference is that the three different networks differ in the noise intensity required to induce stochastic resonance and the effective range of the electrical coupling strength. Unlike the other two networks, the Fourier coefficient in scale-free network increases with the electromagnetic induction, and the number of neurons has no affect on stochastic resonance. Our results are conducive to understanding the mechanism of stochastic resonance in envelope modulation signals and electromagnetic induction.
A new patterns of self-organization activity of brain: Neural energy coding
2022, Information Sciences
According to the basic principles and methods of information theory, the operation way of neural coding is studied and analyzed by using the minimum mutual information and the maximum entropy principle. This paper describes how the principles of minimum mutual information and maximum entropy are used to evaluate the amount of information in neural responses. Its main contribution is as follows: (1) that the expression of neural information is closely related to the utilization of neural energy, and it is found that the highly evolved nervous system strictly follows the two basic principles of economy and efficiency in energy consumption and utilization; (2) In order to verify the relationship between neural information processing and energy utilization, this paper uses the concept of energy-efficiency ratio to measure the economy and high efficiency of the nervous system in term of energy utilization by using the maximum entropy principle; (3) The numerical results show that the energy consumed by the nervous system reflects not only the internal law of neural information conduction and processing, but also the self-organization structure of neural information coding. The results suggest that energy neural coding, a novel neural information processing method, can be used to understand how brain activity works. Such a coding pattern can not only be extended to research the large-scale neuroscience field, but also unify brain models at all levels by use of the energy theory. This will provide a scientific theoretical basis for the exploration of how the brain works and the computational principles of brain-like artificial intelligence.
Mixed selectivity coding of sensory and motor social signals in the thalamus of a weakly electric fish
2022, Current Biology
High-level neural activity often exhibits mixed selectivity to multivariate signals. How such representations arise and modulate natural behavior is poorly understood. We addressed this question in weakly electric fish, whose social behavior is relatively low dimensional and can be easily reproduced in the laboratory. We report that the preglomerular complex, a thalamic region exclusively connecting midbrain with pallium, implements a mixed selectivity strategy to encode interactions related to courtship and rivalry. We discuss how this code enables the pallial recurrent networks to control social behavior, including dominance in male-male competition and female mate selection. Notably, response latency analysis and computational modeling suggest that corollary discharge from premotor regions is implicated in flagging outgoing communications and thereby disambiguating self- versus non-self-generated signals. These findings provide new insights into the neural substrates of social behavior, multi-dimensional neural representation, and its role in perception and decision making.
A theory of capacity and sparse neural encoding
2021, Neural Networks
Motivated by biological considerations, we study sparse neural maps from an input layer to a target layer with sparse activity, and specifically the problem of storing $K$ input-target associations $(x, y)$ , or memories, when the target vectors $y$ are sparse. We mathematically prove that $K$ undergoes a phase transition and that in general, and somewhat paradoxically, sparsity in the target layers increases the storage capacity of the map. The target vectors can be chosen arbitrarily, including in random fashion, and the memories can be both encoded and decoded by networks trained using local learning rules, including the simple Hebb rule. These results are robust under a variety of statistical assumptions on the data. The proofs rely on elegant properties of random polytopes and sub-gaussian random vector variables. Open problems and connections to capacity theories and polynomial threshold maps are discussed.
How auditory selectivity for sound timing arises: The diverse roles of GABAergic inhibition in shaping the excitation to interval-selective midbrain neurons
2021, Progress in Neurobiology
Across sensory systems, temporal frequency information is progressively transformed along ascending central pathways. Despite considerable effort to elucidate the mechanistic basis of these transformations, they remain poorly understood. Here we used a novel constellation of approaches, including whole-cell recordings and focal pharmacological manipulation, in vivo, and new computational algorithms that identify conductances resulting from excitation, inhibition and active membrane properties, to elucidate the mechanisms underlying the selectivity of midbrain auditory neurons for long temporal intervals. Surprisingly, we found that stimulus-driven excitation can be increased and its selectivity decreased following attenuation of inhibition with gabazine or intracellular delivery of fluoride. We propose that this nonlinear interaction is due to shunting inhibition. The rate-dependence of this inhibition results in the illusion that excitation to a cell shows greater temporal selectivity than is actually the case. We also show that rate-dependent depression of excitation, an important component of long-interval selectivity, can be decreased after attenuating inhibition. These novel findings indicate that nonlinear shunting inhibition plays a key role in shaping the amplitude and interval selectivity of excitation. Our findings provide a major advance in understanding how the brain decodes intervals and may explain paradoxical temporal selectivity of excitation to midbrain neurons reported previously.
Lower Baseline Variability Gives Rise to Lower Detection Thresholds in Midbrain than Hindbrain Electrosensory Neurons
2020, Neuroscience
Understanding how the brain decodes sensory information to give rise to behaviour remains an important problem in systems neuroscience. Across various sensory modalities (e.g. auditory, visual), the time-varying contrast of natural stimuli has been shown to carry behaviourally relevant information. However, it is unclear how such information is actually decoded by the brain to evoke perception and behaviour. Here we investigated how midbrain electrosensory neurons respond to weak contrasts in the electrosensory system of the weakly electric fish Apteronotus leptorhynchus. We found that these neurons displayed lower detection thresholds than their afferent hindbrain electrosensory neurons. Further analysis revealed that the lower detection thresholds of midbrain neurons were not due to increased sensitivity to the stimulus. Rather, these were due to the fact that midbrain neurons displayed lower variability in their firing activities in the absence of stimulation, which is due to lower firing rates. Our results suggest that midbrain neurons play an active role towards enabling the detection of weak stimulus contrasts, which in turn leads to perception and behavioral responses.

View all citing articles on Scopus

View full text

Efficient computation via sparse coding in electrosensory neural networks

Highlights

Introduction

Section snippets

The electrosensory system

Electrosensory stimuli

Electroreceptors and coding of natural sensory input

The electrosensory lateral line lobe: efficient circuits for feature extraction and sparse coding

Feature selectivity in the electrosensory midbrain (TS)

Summary

Funding

References and recommended reading

Acknowledgements

Curr Opin Neurobiol

Behav Brain Res

J Neurosci

J Neurophysiol

Phys Rev E

J Neurosci

J Comp Physiol A

Single Units and sensation: a neuron doctrine for perceptual psychology

Perception

Sparseness of the neuronal representation of stimuli in the primate temporal visual cortex

J Neurophysiol

Invariant visual representation by single neurons in the human brain [see comment]

Nature

Oscillations and sparsening of odor representations in the mushroom body

Science

An energy budget for signaling in the grey matter of the brain

J Cereb Blood Flow Metab

Modeling signal and background components of electrosensory scenes

J Comp Physiol A Neuroethol Sens Neural Behav Physiol

Electric field interactions in pairs of electric fish: modeling and mimicking naturalistic input

Biol Cybern

Spatial acuity and prey detection in weakly electric fish

PLoS Comput Biol

Plasticity in an electrosensory system. I. General features of a dynamic sensory filter

J Neurophysiol

Prey capture in the weakly electric fish Apteronotus albifrons: sensory acquisition strategies and electrosensory consequences

J Exp Biol

Plasticity of feedback inputs in the apteronotid electrosensory system

J Exp Biol

Electrocommunication signals in free swimming brown ghost knifefish, Apteronotus leptorhynchus

J Exp Biol

Limits of linear rate coding of dynamic stimuli by electroreceptor afferents

J Neurophysiol

Electroreceptor neuron dynamics shape information transmission

Nat Neurosci

Non-renewal statistics of electrosensory afferent spike trains: implications for the detection of weak sensory signals

J Neurosci

Suprathreshold stochastic firing dynamics with memory in P-type electroreceptors

Phys Rev Lett

Interspike interval correlations, memory, adaptation, and refractoriness in a leaky integrate-and-fire model with threshold fatigue

Neural Comput

Linear versus nonlinear signal transmission in neuron models with adaptation currents or dynamic thresholds

J Neurophysiol

Negative interspike interval correlations increase the neuronal capacity for encoding time-varying stimuli

J Neurosci

Experimental and Theoretical demonstration of noise shaping by interspike interval correlations

Proc SPIE

Nonrenewal spike train statistics: causes and consequences on neural coding

Exp Brain Res

Correlations and memory in neurodynamical systems

Biophysical information representation in temporally correlated spike trains

Proc Natl Acad Sci USA

Spike-frequency adaptation separates transient communication signals from background oscillations

J Neurosci

A synchronization-desynchronization code for natural communication signals

Neuron

a Inhibition evoked from primary afferents in the electrosensory lateral line lobe of the weakly electric fish (Apteronotus leptorhynchus)

J Neurophysiol

b Interaction of GABAB-mediated inhibition with voltage-gated currents of pyramidal cells: computational mechanism of a sensory searchlight

J Neurophysiol

c Distal versus proximal inhibitory shaping of feedback excitation in the electrosensory lateral line lobe: implications for sensory filtering

J Neurophysiol

Neural architecture of the electrosensory lateral line lobe: adaptations for coincidence detection, a sensory searchlight and frequency-dependent adaptive filtering

J Exp Biol

b Interaction of GABA_B-mediated inhibition with voltage-gated currents of pyramidal cells: computational mechanism of a sensory searchlight