CoChat

Potential role of a ventral nerve cord central pattern generator in forward and backward locomotion in Caenorhabditis elegans

Open access

The DOI matches its Crossref record; 56 of the 56 references it cites were found in a registry or an index; 12 of the 26 sources its claims cite were checked for support.

Abstract

C. elegans locomotes in an undulatory fashion, generating thrust by propagating dorsoventral bends along its body. Although central pattern generators (CPGs) are typically involved in animal locomotion, their presence in C. elegans has been questioned, mainly because there has been no evident circuit that supports intrinsic network oscillations. With a fully reconstructed connectome, the question of whether it is possible to have a CPG in the ventral nerve cord (VNC) of C. elegans can be answered through computational models. We modeled a repeating neural unit based on segmentation analysis of the connectome. We then used an evolutionary algorithm to determine the unknown physiological parameters of each neuron so as to match the features of the neural traces of the worm during forward and backward locomotion. We performed 1,000 evolutionary runs and consistently found configurations of the neural circuit that produced oscillations matching the main characteristic observed in experimental recordings. In addition to providing an existence proof for the possibility of a CPG in the VNC, we suggest a series of testable hypotheses about its operation. More generally, we show the feasibility and fruitfulness of a methodology to study behavior based on a connectome, in the absence of complete neurophysiological details.

Overview

CoChat Summary

TL;DR

This paper uses an evolutionary computational model of a repeating ventral nerve cord neural unit in *C. elegans* to test whether intrinsic central pattern generator (CPG) activity could explain forward and backward locomotion. The authors report that 1,000 optimization runs yielded 110 solutions meeting all locomotion criteria, supporting the feasibility of a VNC CPG and generating testable hypotheses about its circuit organization.

Key findings

  • In 1,000 optimization runs, 110 solutions (11%) satisfied all three locomotion criteria for forward and backward movement.
  • For the oscillation criterion alone, 511 solutions worked for forward locomotion, 463 for backward locomotion, and 308 for both.
  • The best circuit matched the main experimental features: A- and B-class motorneurons oscillated, dorsal and ventral motorneurons were in antiphase, and B-class activity dominated forward locomotion while A-class activity dominated backward locomotion.
  • Across the successful ensemble, a dorsal core subcircuit (AS, DA, DB) appeared in 108 of 110 solutions; the connection AS → VD was necessary for dorsoventral coordination in 90% of solutions and sufficient in 78%.

Methods

  • Built a computational model of a repeating ventral nerve cord unit based on a connectome analysis, using motorneurons and synapses from the proposed unit; the model included 34 unknown parameters and imposed anterior-posterior symmetry on many connections.
  • Used a genetic algorithm with 1,000 individuals per run, 1,000 generations per run, and 1,000 runs with different random seeds; fitness was evaluated separately for simulated forward (AVB input) and backward (AVA input) locomotion.
  • Fitness combined three criteria: oscillation, dorsoventral phase, and A/B dominance, with post-search filtering to exclude damped oscillations by re-simulating candidates for 3,000 time units.
  • Analyzed the best evolved circuit, the solution ensemble, and ablation/substitution tests to identify minimal functional subcircuits and their correspondence to the connectome.

Limitations

  • The paper states that the model makes “a number of simplifying assumptions” and is “deliberately minimal,” including omitted chemical synapses, gap junctions, motorneurons, and additional electrophysiological detail.
  • The model does not include muscles or mechanical parts of the body, so locomotion was inferred from neural traces rather than full neuromechanics; this is stated in the paper.
  • The text makes evident that the results are existence proofs from simulation, not direct biological confirmation: the paper says the VNC CPG is feasible and offers testable hypotheses, but experimental validation is still needed.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1162/netn_a_00036

Cites

56 of 56 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Figures

Figure 1. Repeating neural unit in the VNC, adapted from Haspel & O’Donovan (2011). Arrows represent chemical synapses (→). Connections ending in lines represent electrical synapses (|—|). Black and blue connections are anterior-posterior symmetric; red connections are not. Purple connections correspond to intersegmental synapses. Gray connections represents neuromuscular junctions.
Figure 1. Repeating neural unit in the VNC, adapted from Haspel & O’Donovan (2011). Arrows represent chemical synapses (→). Connections ending in lines represent electrical synapses (|—|). Black and blue connections are anterior-posterior symmetric; red connections are not. Purple connections correspond to intersegmental synapses. Gray connections represents neuromuscular junctions.CC BY · PubMed Central
Figure 2. Best evolved circuit. (A) Full circuit architecture. Neurons and connections taken from VNC neural unit (see Figure 1). The type of connection is depicted with different symbols (see legend). Unknown parameters of the circuit such as the strength of the chemical synapses and sign (whether they are excitatory or inhibitory), the strength of the gap junctions, and the intrinsic parameters of the neurons were evolved. The width of the connection is proportional to the evolved strength. External input from command interneurons AVB to B-class motoneurons and AVA to A-class motoneurons are represented in blue and red, respectively. (B) Simulated neural traces of the evolved network for forward and backward locomotion. During forward locomotion, AVB input is on. During backward locomotion, AVA input is on.
Figure 2. Best evolved circuit. (A) Full circuit architecture. Neurons and connections taken from VNC neural unit (see Figure 1). The type of connection is depicted with different symbols (see legend). Unknown parameters of the circuit such as the strength of the chemical synapses and sign (whether they are excitatory or inhibitory), the strength of the gap junctions, and the intrinsic parameters of the neurons were evolved. The width of the connection is proportional to the evolved strength. External input from command interneurons AVB to B-class motoneurons and AVA to A-class motoneurons are represented in blue and red, respectively. (B) Simulated neural traces of the evolved network for forward and backward locomotion. During forward locomotion, AVB input is on. During backward locomotion, AVA input is on.CC BY · PubMed Central
Figure 3. Activity of model neuron resembles electrophysiological properties of C. elegans neurons. We simulated the neural model in two different configurations to reproduce electrophysio logical recordings in Mellem et al. (2008): neuron AVA (left column) and neuron RMD (right column), as examples of passive and active dynamics, respectively. Parameters for the passive AVA-like neuron were θ = −1.8; τ = 0.03; w = 0.6. Parameters for the active RMD-like neuron were θ = −3.4; τ = 0.05; w = 5.1. For each of the panels, the bottom trace represents the external input and the top trace represents the output of the neuron. Responses to a depolarizing current ramp from the passive neuron configuration (A) and the active response configuration (B) of the model. Responses to a current steps for the passive neuron configuration (C) and the active response configuration (D). The active configuration reproduces the long-lived plateau potentials in response to depolarizing current steps observed in C. elegans neurons (Mellem et al., 2008).
Figure 3. Activity of model neuron resembles electrophysiological properties of C. elegans neurons. We simulated the neural model in two different configurations to reproduce electrophysio logical recordings in Mellem et al. (2008): neuron AVA (left column) and neuron RMD (right column), as examples of passive and active dynamics, respectively. Parameters for the passive AVA-like neuron were θ = −1.8; τ = 0.03; w = 0.6. Parameters for the active RMD-like neuron were θ = −3.4; τ = 0.05; w = 5.1. For each of the panels, the bottom trace represents the external input and the top trace represents the output of the neuron. Responses to a depolarizing current ramp from the passive neuron configuration (A) and the active response configuration (B) of the model. Responses to a current steps for the passive neuron configuration (C) and the active response configuration (D). The active configuration reproduces the long-lived plateau potentials in response to depolarizing current steps observed in C. elegans neurons (Mellem et al., 2008).CC BY · PubMed Central
Figure 4. Performance of the ensemble on the individual locomotion criteria. (A) Oscillation criterion. Fraction of solutions that accomplish oscillations in one or more neurons in the network. B-class neurons were evaluated during forward locomotion and A-class during backward. Thirty percent of all solutions showed oscillations in all relevant motorneurons. (B) Phase criterion. Dorsoventral phase lag between B-class neurons for forward locomotion and A-class neurons during backward locomotion. The majority of solutions oscillate in dorsoventral antiphase. (C) Dominance criterion. Distribution of the differences in mean activity between B- and A-class neurons during forward and backward locomotion. In the majority of solutions, the A-class dominates during backward and the B-class dominates during forward.
Figure 4. Performance of the ensemble on the individual locomotion criteria. (A) Oscillation criterion. Fraction of solutions that accomplish oscillations in one or more neurons in the network. B-class neurons were evaluated during forward locomotion and A-class during backward. Thirty percent of all solutions showed oscillations in all relevant motorneurons. (B) Phase criterion. Dorsoventral phase lag between B-class neurons for forward locomotion and A-class neurons during backward locomotion. The majority of solutions oscillate in dorsoventral antiphase. (C) Dominance criterion. Distribution of the differences in mean activity between B- and A-class neurons during forward and backward locomotion. In the majority of solutions, the A-class dominates during backward and the B-class dominates during forward.CC BY · PubMed Central
Figure 5. Activity of best evolved circuit during realistic stimulation by AVA and AVB command interneurons. (A) Activity of command interneurons recorded in (Kawano et al., 2011). Traces were normalized. Background color indicates forward and backward dominance depending on relative activations of AVA and AVB. (B) Traces from VA and VB neurons in the simulated circuit when stimulated by the traces in (A).
Figure 5. Activity of best evolved circuit during realistic stimulation by AVA and AVB command interneurons. (A) Activity of command interneurons recorded in (Kawano et al., 2011). Traces were normalized. Background color indicates forward and backward dominance depending on relative activations of AVA and AVB. (B) Traces from VA and VB neurons in the simulated circuit when stimulated by the traces in (A).CC BY · PubMed Central
Figure 6. Functional analysis of the best evolved circuit. (A1) Dorsal core subcircuit. Minimal CPG comprises all three neurons that innervate dorsal muscles. Legend for the synaptic connections at the bottom of the figure. Connection width is proportional to synaptic strength. (A2) Output traces of the dorsal core subcircuit under forward and backward external inputs. Traces maintain shift in relative activation of B/A class neurons. Input from neurons AVB and AVA are shown with blue and red bars respectively. (B1) Minimal functional subcircuit. All features of the locomotion pattern are present when VD, VB, and VA are added to the dorsal core subcircuit. (B2) Simulated neural traces for the minimal functional subcircuit. Main output difference compared with the full network is the increase in amplitude in VA neuron during backward locomotion. (C1) Extended minimal circuit. (C2) Simulated neural traces for the extended minimal circuit.
Figure 6. Functional analysis of the best evolved circuit. (A1) Dorsal core subcircuit. Minimal CPG comprises all three neurons that innervate dorsal muscles. Legend for the synaptic connections at the bottom of the figure. Connection width is proportional to synaptic strength. (A2) Output traces of the dorsal core subcircuit under forward and backward external inputs. Traces maintain shift in relative activation of B/A class neurons. Input from neurons AVB and AVA are shown with blue and red bars respectively. (B1) Minimal functional subcircuit. All features of the locomotion pattern are present when VD, VB, and VA are added to the dorsal core subcircuit. (B2) Simulated neural traces for the minimal functional subcircuit. Main output difference compared with the full network is the increase in amplitude in VA neuron during backward locomotion. (C1) Extended minimal circuit. (C2) Simulated neural traces for the extended minimal circuit.CC BY · PubMed Central
Figure 7. Dorsal core operation. (A) The cyclic operation of the dorsal core subcircuit can be broken down into six stages. Each stage represents a “quasi-stable” configuration of the subcircuit. Active neurons are colored; inactive neurons are shown in white. Active connections are shown; inactive connections are not shown. Dorsal core subcircuit operates similarly during forward and backward locomotion. (B) Phase plot of dorsal core neurons during backward and forward simulations. Stages 1, 3, 4, and 6 are shown with disks, according to panel (A). Dorsal core neural traces for backward (C) and forward (D). Neural traces normalized. Vertical lines mark stages 1, 3, 4, and 6 in the oscillation cycle. Although the timing of stages is the different for forward and backward, the sequence is the same.
Figure 7. Dorsal core operation. (A) The cyclic operation of the dorsal core subcircuit can be broken down into six stages. Each stage represents a “quasi-stable” configuration of the subcircuit. Active neurons are colored; inactive neurons are shown in white. Active connections are shown; inactive connections are not shown. Dorsal core subcircuit operates similarly during forward and backward locomotion. (B) Phase plot of dorsal core neurons during backward and forward simulations. Stages 1, 3, 4, and 6 are shown with disks, according to panel (A). Dorsal core neural traces for backward (C) and forward (D). Neural traces normalized. Vertical lines mark stages 1, 3, 4, and 6 in the oscillation cycle. Although the timing of stages is the different for forward and backward, the sequence is the same.CC BY · PubMed Central
Figure 8. Sign patterns found in the ensemble for the dorsal core subcircuit. (A) Different patterns of signs of the connections in the dorsal core. Black arrows represent excitatory chemical synapses. Circle-ending connections represent inhibitory chemical synapses. (B) Solutions in each group show a characteristic oscillatory pattern. Traces correspond to average ±standard deviation of the normalized neuronal outputs. Two oscillation cycles are shown. (C) Proportion of the total solutions in the ensemble with each of the different dorsal core sign patterns. Sign patterns I and II represent over 75% of the solutions. (D) Fitness for the solutions according to their dorsal core sign pattern. Solutions with sign patterns I and II comprise the best performing solutions.
Figure 8. Sign patterns found in the ensemble for the dorsal core subcircuit. (A) Different patterns of signs of the connections in the dorsal core. Black arrows represent excitatory chemical synapses. Circle-ending connections represent inhibitory chemical synapses. (B) Solutions in each group show a characteristic oscillatory pattern. Traces correspond to average ±standard deviation of the normalized neuronal outputs. Two oscillation cycles are shown. (C) Proportion of the total solutions in the ensemble with each of the different dorsal core sign patterns. Sign patterns I and II represent over 75% of the solutions. (D) Fitness for the solutions according to their dorsal core sign pattern. Solutions with sign patterns I and II comprise the best performing solutions.CC BY · PubMed Central
Figure 9. Ablation experiments. Vertical histograms show the normalized fitness distribution among the 108 solutions that evolved a dorsal core oscillator while each of the connections is exchanged for a compensatory input. The dashed line marks a minimum level of fitness of 20%. This may seem low, but the fitness is multiplicative over eight different components, such that on average each different component of fitness has above 80% performance. The percentage of solutions that maintain a fitness above this level is indicated at the top of the respective column. The white disk corresponds to the best evolved circuit. The red triangles correspond to the two solutions that did not evolve a dorsal core oscillator. The connections are divided into the following: (A) connections present in the dorsal core subcircuit; (B) connections in the minimal functional subcircuit; (C) connections responsible for minor contributions.
Figure 9. Ablation experiments. Vertical histograms show the normalized fitness distribution among the 108 solutions that evolved a dorsal core oscillator while each of the connections is exchanged for a compensatory input. The dashed line marks a minimum level of fitness of 20%. This may seem low, but the fitness is multiplicative over eight different components, such that on average each different component of fitness has above 80% performance. The percentage of solutions that maintain a fitness above this level is indicated at the top of the respective column. The white disk corresponds to the best evolved circuit. The red triangles correspond to the two solutions that did not evolve a dorsal core oscillator. The connections are divided into the following: (A) connections present in the dorsal core subcircuit; (B) connections in the minimal functional subcircuit; (C) connections responsible for minor contributions.CC BY · PubMed Central
Figure 10. Two forms of the minimal functional circuits found in the ensemble. Common connections shown in gray. Unique connection shown in black. In both forms, the dorsal motorneurons AS, DA, and DB serve as the core pattern–generating subcircuit; dorsal oscillations propagate to ventral motorneurons in an antiphasic manner via the chemical synapse AS → VD, and from there they propagate to VB via the chemical synapse VD → VB. In one form of the solution, the information gets passed on to VA via a gap junction (A) and in others via a chemical synapse (B).
Figure 10. Two forms of the minimal functional circuits found in the ensemble. Common connections shown in gray. Unique connection shown in black. In both forms, the dorsal motorneurons AS, DA, and DB serve as the core pattern–generating subcircuit; dorsal oscillations propagate to ventral motorneurons in an antiphasic manner via the chemical synapse AS → VD, and from there they propagate to VB via the chemical synapse VD → VB. In one form of the solution, the information gets passed on to VA via a gap junction (A) and in others via a chemical synapse (B).CC BY · PubMed Central
Figure 11. Minimal functional subcircuit present in connectome. The circuit identified in the analysis exists in three places in the anterior section of the connectome of the C. elegans hermaphrodite (A) and twice in the anterior section of the male connectome (B). Each functional subcircuit is shown with a different color (e.g., blue, red, and green). The muscles innervated by these subcircuits are colored accordingly. Neurons and muscles are depicted along the VNC according to their sequential order (not exact body position). Light shaded connections correspond to synapses that were identified as important for the minimal functional subcircuit along the rest of the VNC but that do not form a complete subcircuit. For clarity, all other chemical synapses and gap junctions are not shown.
Figure 11. Minimal functional subcircuit present in connectome. The circuit identified in the analysis exists in three places in the anterior section of the connectome of the C. elegans hermaphrodite (A) and twice in the anterior section of the male connectome (B). Each functional subcircuit is shown with a different color (e.g., blue, red, and green). The muscles innervated by these subcircuits are colored accordingly. Neurons and muscles are depicted along the VNC according to their sequential order (not exact body position). Light shaded connections correspond to synapses that were identified as important for the minimal functional subcircuit along the rest of the VNC but that do not form a complete subcircuit. For clarity, all other chemical synapses and gap junctions are not shown.CC BY · PubMed Central

Verification

Claims the paper makes

9 Valid · 2 Not valid · 5 Not checked · 92 Own results and interpretations

66 more claims

Create a free account to see all 108 claims and to verify the 1 reference still open.

See all claims — free account
  • It may state two separate claims as one. 29
  • Probably not a substantial claim. 8
  • The quote only partly states this claim. 3

Create a free account to see these claims.

Citations

Cites

Research works printed but not linked: 5

  1. Fouad A. D., 2017, bioRxiv
  2. Gao S., 2017, bioRxiv
  3. Waterston R., 1988, The nematod C. elegans, 281
  4. 10.1098/rstb.1976.0086 · DOI · Paper page
  5. Xu T., 2017, bioRxiv

Cited by

See all 39

Metadata

Authors
Olivares, E. O., Izquierdo, E. J., & Beer, R. D.
Affiliations
Indiana University Bloomington (US)
Corresponding author
Izquierdo, Eduardo J.
Venue
Network Neuroscience
Publisher
MIT Press
Published
2018-09
Volume
2
Issue
3
Pages
323-343
ISSN
2472-1751 (ISSN-L)
OpenAlex
W2771714108
Cited by
38
References
56
Citations by year
2026: 3 · 2025: 8 · 2024: 3 · 2023: 1 · 2022: 4 · 2021: 5 · 2020: 5 · 2019: 3 · 2018: 6
FWCI
2.20
Type
journal article
Language
en
Open access
Licence
cc-by
Locations
  • Network Neuroscience (publishedVersion, cc-by)
  • Europe PMC (PubMed Central) (publishedVersion, cc-by)
  • PubMed Central (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • arXiv (Cornell University) (submittedVersion)
  • DOAJ (DOAJ: Directory of Open Access Journals) (submittedVersion, cc-by-sa)
  • IUScholarWorks Open (Indiana University) (submittedVersion)
  • arXiv (Cornell University)
Funders
Division of Information and Intelligent Systems (IIS-1524647)
Topics
  • Life Sciences › Neuroscience › Cognitive Neuroscience › Neural dynamics and brain function
  • Life Sciences › Neuroscience › Cellular and Molecular Neuroscience › Neurobiology and Insect Physiology Research
  • Physical Sciences › Engineering › Control and Systems Engineering › Robotic Locomotion and Control
Keywords
central pattern generator, Caenorhabditis elegans, ventral nerve cord, locomotion, undulatory locomotion, connectome, neural circuit modeling, evolutionary algorithms, neural oscillations, segmentation analysis

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

The claim verdicts this report counts are experimental: their accuracy is not yet measured.

A language model drew the claims from the paper on 2026-10-10, under version 3 of the claim definition (model deepseek-v4.1-flash, prompt extract-v3).

A second model labelled each claim on 2026-10-10 (model jev-1.13-20260917, definition revision gold-guideline-v0.2).

Cited sources are checked with a language model’s help (gpt-6-sol).

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review

Data sources

Bibliographic data: Semantic Scholar, OpenAlex, Crossref.

Contains information from the Semantic Scholar Academic Graph Datasets, made available under the ODC Attribution License (ODC-BY 1.0).

Semantic Scholar’s data is described in: Kinney, R., et al. (2023). The Semantic Scholar Open Data Platform. arXiv:2301.10140.