FlyHack: Stress-Testing Mechanistic Claims in a Connectome-Constrained Visual Model

Key points

• We stress-test these dependencies in flyvis, a Drosophila motion-vision model, using its 50-model ensemble, task-matched causal CNN controls and replicated interventions.1
• At the level of model behaviour, three-level reverse-phi reversal in flyvis is strongly period- and velocity-dependent and also appears in task-matched CNNs; neither family reproduces the fly's stationary-sawtooth response.1
• Adversarial direction reversal requires a perturbation bound of 0.21--0.27 times stimulus contrast in flyvis versus 0.13--0.25 in CNNs, but the gap over a parameter-matched CNN largely closes for larger CNNs.1
• Greedy cell-type sets meet a model-relative T4/T5 tuning criterion in 25 of 50 models, while T4/T5 output silencing reduces the tested optic-flow behaviour to at most two of eight clips in 40.1

Sources (1)

• [1]FlyHack: Stress-Testing Mechanistic Claims in a Connectome-Constrained Visual Model
  arXiv (AI, ML, NLP, CV, robotics, multi-agent) · Oct 9, 05:20 AM

  Connectome-constrained neural networks make biological components directly accessible to intervention, but their mechanistic conclusions can still depend on the comparison model, intervention, circuit representation and trained solution.

  We stress-test these dependencies in flyvis, a Drosophila motion-vision model, using its 50-model ensemble, task-matched causal CNN controls and replicated interventions.

Extractive summary: sentences quoted from the sources.