Sensorimotor behavior from an untrained brain-and-nerve-cord connectome
summary
The gist
Connectomes map adult fly neurons and synapses, but behavioral consequences from this wiring are unknown.
In short
The study simulates a fly's nervous system using an adult brain map (connectome) to predict behavior from untrained wiring. It found that specific pathways are active when they receive enough input, and that learning avoidance requires both sensory input and a specific balance in the mushroom body output neurons.
Key concepts
- Connectome
- A comprehensive map of all neurons and their connections in an organism's brain. This study used this map to simulate how the fly's wiring might actually lead to movement and behavior, testing if the physical connections alone dictate function.
- Mushroom Body Output Neurons (MBONs)
- These are key neurons that process information about odors and drive behavioral responses like turning toward or away from a scent. The paper shows that the fly's ability to learn avoidance depends on a specific balance of activity among these neurons.
- Valence Readout
- This is a calculated measure of whether the fly is approaching or avoiding an odor, derived from summing the spike counts of all MBONs. This readout is crucial because it determines how the fly interprets its sensory input to make a behavioral decision about an odor.
Terminology used across episodes
This episode discusses
- Sensorimotor behavior from an untrained brain-and-nerve-cord connectome · Paper Radio
- Whole-Brain Connectomic Graph Model Enables Whole-Body Locomotion Control in Fruit Fly
The paper
Sensorimotor behavior from an untrained brain-and-nerve-cord connectome · Read on arXiv
John R. Minnick, Jesus Gonzalez-Ferrer, Jinghui Geng, Jason K. Eshraghian, Mohammed A. Mostajo-Radji, David Haussler, Mircea Teodorescu
Department of Electrical and Computer Engineering, University of California, Santa Cruz, CA, USA. · UC Santa Cruz Genomics Institute, University of California, Santa Cruz, CA, USA. · Department of Computer Science and Engineering, University of California, Santa Cruz, CA · Department of Biomolecular Engineering, University of California, Santa Cruz
Transcript
Introduction to the show: ident: Genomics Radio. Generated commentary on the latest computational biology and genomics papers.
Ines: We're looking at the paper: "Sensorimotor behavior from an untrained brain-and-nerve-cord connectome".
Title and authors: Marcus: They start by setting up the simulation and then they show us what happens when you look at the results of this connectome as is, without imposing any specific behaviors on the fly.
Ines: The main finding there is that even in this untrained network, a pathway that we already know has a behavioral role activates if it supplies two percent or more of its target's input.
Yuki: That suggests that even an untrained network has some built-in pathways ready to go, which makes sense since behavior isn't just something you learn from scratch.
Ines: But what they found is that inside the simulated body, this intact wiring achieves ninety-five percent coverage of the odour sources.
Marcus: That’s a big number for connectivity within the system itself, but they also found that rewiring that keeps the neuron connection counts and signs relatively stable only reaches eighteen percent.
Yuki: So it sounds like navigation isn't just about having all these connections present, but about how well those connections are organized for function.
Ines: Right. And leg commands almost entirely disappear without sensory feedback, which really shows that navigation needs both the measured wiring and some kind of sensory input to actually work.
The paper's summary: Marcus: They then move on to how this system can actually learn something, focusing on plasticity within one class of synapses and how that affects approach versus avoidance.
Ines: When they introduce plasticity in one synapse class, you see the fly switch from approaching an odour to avoiding it, and this switch depends entirely on whether the Kenyon-cell activity relies on that specific odour.
Yuki: That ties into how we think about associative learning; if the cell activity is tied to the stimulus, then changing that link changes the whole behavioral outcome.
Ines: The trained fly avoids a particular odour, and its closest approach drops from one point six millimeters without plasticity to about nineteen point six millimeters with it.
Marcus: That difference is huge—a factor of nearly a hundred—and they give us the probability of that shift happening, which is six point eight times ten to the power of negative eight, and Cliff's delta is one point zero zero.
Yuki: So it shows that this specific mechanism for switching from approach to avoidance is really quite sensitive and depends on that odour dependency.
Ines: They also look at how generalization happens between odours, which they find occurs through code overlap when the Kenyon cells depend on the odour.
The paper's improvements: Marcus: Now, the authors suggest several ways to improve this model and test their ideas further. One major improvement is being able to predict behavioral outcomes from untrained wiring by running a spiking network alone inside a simulated body using published brain model parameters.
Ines: That lets researchers see exactly what the system reproduces from known biology and where it actually fails when you just run the connectome as it is read, which is really valuable for understanding limits.
Yuki: That sounds like a way to systematically test hypotheses about wiring structure without having to fit unknown parameters against an external task first.
Marcus: Another improvement they propose is the ability to test pathway functionality by driving populations outside of the body and reading an identified target neuron or motor pool, checking if a pathway conducts when it supplies two percent or more of its target’s input.
Ines: That's a direct way to check that relationship, moving beyond just looking at what happens inside the body.
Yuki: And they also suggest assessing the necessity of sensory feedback by comparing performance inside versus outside, noting that navigation needs both the measured wiring and sensory feedback.
Conclusion: Ines: So to wrap up, this paper on "Sensorimotor behavior from an untrained brain-and-nerve-cord connectome" shows that reading a connectome at face value can be quite far from actual animal behavior, especially concerning the antennal lobe where cholinergic local neurons are involved and their projections pass through electrical synapses.
Marcus: The study emphasizes that scaling one synapse class down to zero point one five of its face value brings the Kenyon-cell code into the fly's range, and modeling it with a wrong sign leads to a reduction in active Kenyon cells from sixty eight percent down to forty two percent when verified signs are applied.
Yuki: It really highlights that the APL neuron keeps the code sparse in animals, but when you model it this way, the resulting activity level changes significantly depending on those input signs.
Ines: They also have three key experimental predictions: a pathway should drive its target if it supplies a few percent of its target's input synapses and fail at one-hundredth, steering signals depend on ongoing activity in the leg motor network because they lower that activity, and Kenyon cells active without an odour should be devalued by reinforcement not paired with the odour.
Marcus: The model demonstrates that a trained interface can drive a body regardless of wiring, suggesting navigation depends on the sensorimotor loop irrespective of command origin.
Yuki: It’s a reminder that the connection between sensing, acting, and learning is what really drives behavior in these systems.
More episodes
- 2610.40344-Morphological Features of Murine Pulmonary Arterial Networks for Diagnosing Vascular Disease
- Daily Summary for 2026-11-19
- 2609.10193-Are You Learning Biological Signal or Shortcuts? Auditing and Mitigating Bias in Protein-Protein Interaction Datasets
- 2610.38236-Generative discovery of chemical reaction networks
- 2507.11176-An Interpretable AI framework Quantifying Traditional Chinese Medicine Principles Towards Enhancing and Integrating with Modern Biomedicine
- The day in conversation for 2026-11-19
- Daily Summary for 2026-11-19
- 2504.03590-Towards a Unified Framework for Determining Conformational Ensembles of Disordered Proteins
- 2603.15217-A multiscale discrete-to-continuum framework for structured population models
- 2605.22145-Persistent Homology as a Morphological Signature of Fibrin Networks