Follow-up experiment · input timing
Does input timing change the effect of cutting Roundup’s outputs?
Roundup is a feeding-circuit neuron; MN9 is the measured motor neuron. Compare cutting every Roundup output with cutting only its direct connection to MN9, under two artificial input patterns.
Five artificial templates · one fixed model · one 500 ms measurement window
This exploratory simulation does not establish a biological effect or predict body movement.
Loading results…
Timing comparison
Loading results
The comparison will appear here when the results have loaded.
Result across all five templates
Timing changed the difference between the cuts
Across five artificial templates, shifted timing increased, decreased or left unchanged the difference between the cuts.
E = Ndirect edge cut − Nall outputs cut. Positive E means the all-output cut leaves fewer MN9 spikes than the direct-edge cut. Negative E means it leaves more.
Primary timing interaction: I = Eshifted − Ealigned. Positive I means the extra effect is larger under circularly shifted input. The selected template is highlighted. Each row is one artificial template; these are not five animals.
Compare MN9 spike counts
Choose from precomputed simulations.
Matched scheduled pulses do not guarantee identical applied input pulses or sensory-neuron (GRN) spikes.
Scheduled, applied and GRN input counts
All values sum over the same 21 input neurons during the same 500 ms analysis window. Matched scheduled pulses do not guarantee identical applied pulses or GRN spikes.
Counts use the same 500 ms analysis window; bar lengths share one scale across the grid. “Direct edge cut” preserves every other Roundup output.
Inspect scheduled input pulses
Scheduled candidate input pulses
These are scheduled candidates, not guaranteed applied input pulses or GRN spikes. Matching schedules do not guarantee identical applied input.
Rows are 21 sensory input neuron IDs (GRNs). Marks are scheduled candidate pulses, not guaranteed applied pulses or GRN spikes. Time is relative to the start of the 500 ms analysis window. Scroll the raster horizontally on a small screen.
Inspect input neuron IDs
Methods & sources
What this pilot tests
Data and independent checks
Download the results. These exports support recomputing the displayed outcomes. The full raw NPZ archive is retained locally.
The intact model is compared with two interventions: zero only the direct left Roundup → MN9 edge, or zero all outgoing weights of left Roundup. Incoming connections remain. The direct intervention is source neuron 720575940623211725 to MN9 720575940660219265.
Aligned input shares a template across the 21 input channels. The second pattern uses circularly shifted copies. This is a comparison of artificial timing schedules, not recorded biological input.
The full published model runs at gain 1 and a 0.1 ms step for 1.5 seconds, with a 500 ms periodic input schedule repeated three times. The 500 ms analysis window follows a 500 ms warmup. Only simulated neural output is measured; there is no biological validation or prediction of body movement.
Model: Shiu et al. (2024). Third-party notices. The original connection-strength study and circuit views are separate experiments.