Research map / Vision-language and multimodal / Human and embodied modeling
Connectome-Based Modelling Reveals Orientation Maps in the Drosophila Optic Lobe
Read the paperCheck your idea against 108,967 papers
AI-extracted summary. Written by AI from the paper's full text. Numbers and claims can be wrong or missing context, so check the paper before you cite or build on it.
Simulates the full FAFB fly connectome with a conductance-based LIF model driven by bar stimuli to show spatially organised orientation maps and columns in the Drosophila optic lobe.
Problem
Whether map-like orientation representations (orientation maps and columns) can emerge in the non-cortical Drosophila optic lobe, given only connectome-constrained dynamics.
Approach
Drives a whole-brain FAFB connectome (138,639 neurons, 1.5M synapses) with a conductance-based LIF model using OFF-bar stimuli injected into L1-L3, fits circular Gaussians to firing-rate tuning, and maps preferred orientations spatially to reveal pinwheel-like maps and columns.
Key results
- Mean absolute difference between LIF-predicted and dendritic-structure-predicted preferred orientation is 13.7 degrees.
- Silencing Mi neurons leaves fewer than 5% of well-fit neurons in Dm or Pm; silencing Tm neurons reduces Pm well-fit neurons by 89.6% but Dm by only 9.3%.
- Dm columns (858 extracted) show tighter orientation coherence than shuffled baseline (Mann-Whitney p < 1e-20), with Pm showing weaker columnar correspondence.
Measured effects reported in the paper
| What was changed | Where | Measured | Effect |
|---|---|---|---|
| silence Tm neurons (ablation) | Connectome-constrained LIF model · Dm | well-fit neurons remaining: 100% -> 90.7% | Hurt |
| silence Tm neurons (ablation) | Connectome-constrained LIF model · Pm | well-fit neurons remaining: 100% -> 10.4% | Hurt |
| silence Mi neurons (ablation) | Connectome-constrained LIF model · Dm | well-fit neurons remaining: 100% -> 2.4% | Hurt |
| silence Mi neurons (ablation) | Connectome-constrained LIF model · Pm | well-fit neurons remaining: 100% -> 3.9% | Hurt |
| include Dm3 and Dm15 neurons in the orientation-map analysis (ablation) | Dm orientation-map analysis · Dm | orientation-map spatial organization | No clear effect |
| increase LIF parameter perturbation from 10% to 50% (ablation) | Connectome-constrained LIF model · optic-lobe orientation tuning | orientation selectivity | Hurt |
| perturb LIF parameters by 10% (ablation) | Connectome-constrained LIF model · optic-lobe orientation maps | preferred-orientation map structure | No clear effect |
| decrease L1 input gain A1 by 20% (ablation) | Connectome-constrained LIF model · optic-lobe orientation tuning | preferred-orientation difference | No clear effect |
| decrease L2 input gain A2 by 20% (ablation) | Connectome-constrained LIF model · optic-lobe orientation tuning | preferred-orientation difference | No clear effect |
| decrease L3 input gain A3 by 20% (ablation) | Connectome-constrained LIF model · optic-lobe orientation tuning | preferred-orientation difference | No clear effect |
Surprising findings
- R7 and R8 photoreceptors show orientation selectivity matching Dm9 due to a closed recurrent loop.
- T4 and T5 neurons mostly had near-zero firing and did not meet selection criteria under the static stimulus.
Limitations
- No in vivo validation; results are purely simulation-based.
- Simplified LIF model omits nonlinear dynamics and neurotransmitter effects; w_syn is a free parameter.
- Stimuli injected into L1-L3 rather than photoreceptors R1-R6 because visual column mappings for photoreceptors are unavailable.
Experimental setup
Datasets: FAFB (full adult fly brain) connectome calcium imaging data for L1-L3 gain fitting
Metrics: R^2 of circular Gaussian fit normalised RSS absolute difference between structural and simulated preferr circular standard deviation within columns ablation percentage of well-fit neurons
Baselines: Seung's dendritic-wiring orientation prediction shuffled-orientation null baseline
Builds on: FAFB connectome conductance-based leaky integrate-and-fire model
Closest papers by meaning
- Emergent Orientation Maps —— Mechanisms, Coding Efficiency and Robustness (ICLR 2025 · similarity 0.89)
- Visual Pinwheel Centers Act as Geometric Saliency Detectors (NeurIPS 2024 · similarity 0.86)
- A simple model of co-emergence of grid and place fields (NeurIPS 2026 · similarity 0.85)
- Neurons as Detectors of Coherent Sets in Sensory Dynamics (NeurIPS 2025 · similarity 0.84)
- Brain Mapping with Dense Features: Grounding Cortical Semantic Selectivity in Natural Images With Vision Transformers (ICLR 2025 · similarity 0.83)
- A Computational Framework for Modeling Emergence of Color Vision in the Human Brain (ICLR 2025 · similarity 0.83)
Writing a paper in this area? scime draft is a free LaTeX editor with AI included, real-time co-authors and a check of your references and venue rules before you submit. It also imports OpenAI Prism projects.