Skip to content
neuroflow0.2.31 · docs

← neuroflow

phase-experiment

The experiment phase covers paradigm design, recording setup, and instrument configuration — everything needed before data collection begins.

Approach

  • Identify which sub-task applies (paradigm design, recording setup, instrument config) and handle one at a time
  • Read .neuroflow/ideation/ for the research question and hypothesis before designing anything
  • Confirm modality (EEG, fMRI, eye-tracking, ECG, etc.) early; it constrains all downstream decisions
  • Write a brief design rationale — not just the implementation
  • Keep the /experiment ground rules in view: the ethics gate before any step with participants, participants only see validated software (never Claude Code), no agent on acquisition PCs during recordings, and no software check is a stimulation-safety interlock
  • Audit every generated paradigm with scripts/psychopy_audit.py before handing it over

Relevant skills

  • neuroflow:neuroflow-core — read first; defines the command lifecycle and .neuroflow/ write rules
  • neuroflow:pupil-labs-neon-realtime — load when Pupil Labs Neon eye tracker is part of the recording setup

Workflow hints

  • PsychoPy scripts and config files go to output_path (paradigm/), not inside .neuroflow/
  • Save experiment-plan.md to .neuroflow/experiment/ covering design choices, recording parameters, the marker map, and the allocation seed and schedule hash
  • Log any paradigm design decision that deviates from the pre-registered plan in .neuroflow/reasoning/experiment.jsonl

Scripts and references

  • scripts/psychopy_audit.py — static audit of .py and .psyexp paradigms (frame vs time timing, markers outside win.callOnFlip, unreset ports, hard-coded refresh rates, blocking calls, missing logs) plus the marker map; --csv checks a run's frame-interval log or trial CSV for dropped frames and off-by-a-frame durations. python <skill base dir>/scripts/psychopy_audit.py paradigm/ — exit 0 = clean, 1 = warnings, 2 = unreadable input.
  • scripts/allocation.py — seeded counterbalancing / group allocation schedule with an append-only ledger (generate, next, verify); non-clinical studies only. Exit 0 = done, 1 = refused or findings, 2 = usage error.
  • references/preflight.md — optional preflight() block for the top of a paradigm, plus a manual run-sheet checklist.

Slash command

/neuroflow:experiment — runs this workflow as a slash command.