Biopsych

Biopsychology Preregistration Base Skill

Description

Base skill for analyzing biopsychology research projects and generating preregistration documents. Routes to modality-specific skills based on detected research type.

Objective

Intelligently detect biopsychology research modality and delegate to appropriate modality-specific preregistration skill.

Workflow

Step 1: Initial Scan

Scan project directory for characteristic files:

• EEG/ERP indicators: .set, .fdt, .vhdr, .eeg, .bdf, EEGLAB scripts, MNE, FieldTrip
• fMRI indicators: .nii, .nii.gz, sub-*/ BIDS structure, SPM, FSL, fMRIPrep
• Online behavioral: .html, .js, .json, jsPsych, Qualtrics, Pavlovia, Prolific
• Psychophysiology: .acq, .edf, .ibi, .rr, AcqKnowledge, Kubios, Ledalab, NeuroKit2, ECG/EDA/HRV/EMG keywords

Step 2: Modality Detection

Determine primary research modality from file patterns:

• Count file type matches per modality
• Check for BIDS structure (indicates neuroimaging)
• Search scripts for software keywords
• Check README/documentation keywords

If multiple modalities detected, ask user which is primary.

Step 3: Route to Modality Skill

Delegate to appropriate skill:

• EEG/ERP → skills/eeg/SKILL.md
• fMRI → skills/fmri/SKILL.md
• Online → skills/online/SKILL.md
• Psychophysiology → skills/psychophys/SKILL.md

Detection Logic

Scan directory recursively for:
  - File extensions matching modality patterns
  - Script content (grep for software names, parameters)
  - BIDS-compliant structure (sub-XXX folders)
  - Documentation keywords
  
Scoring:
  eeg_score = count(.set, .vhdr, .eeg) + keywords("electrode", "ICA", "ERP")
  fmri_score = count(.nii, .nii.gz) + keywords("BOLD", "TR", "fMRI")
  .js) + keywords("jsPsych", "Qualtrics")
  psychophys_score = count(.acq) + keywords("ECG", "EDA", "HRV")
  
Modality = highest_score
Confidence = winner_score / total_score

Input Detection

From directory structure:

• Project organization patterns
• Data folder naming
• Script locations

From files:

• Study title (README.md, protocol.txt)
• Sample size (power analysis scripts, N reported in docs)
• Design info (experimental paradigm descriptions)

Output Format

Route to modality skill with detected information:

Detected Modality: [modality]
Confidence: [high/medium/low]
Key files found: [list]
Extracted info: [title, N estimate, design]

Behavior Rules

1. Always scan before deciding - don't assume from project name
2. Show detection confidence - explain reasoning to user
3. Allow manual override - let user specify modality if detection wrong
4. Preserve files - never modify existing project data
5. Provide fallback - ask modality if detection ambiguous

Example Interactions

User: Create a preregistration

Agent:
1. Scans project directory
2. Finds: 3 .set files, EEGLAB scripts, "electrode" mentions
3. Detects: EEG/ERP (confidence: high)
4. Routes to: skills/eeg/SKILL.md
5. Skill continues with EEG-specific extraction and template

User: Create a preregistration

Agent:
1. Scans project directory
2. Finds: 2 .nii.gz files, SPM scripts, BIDS structure (sub-01, sub-02)
3. Detects: fMRI (confidence: high)
4. Routes to: skills/fmri/SKILL.md
5. Skill continues with fMRI-specific extraction and template

User: Create a preregistration

Agent:
1. Scans project directory
2. Finds: Equal files for EEG and online (.js, .html, .set)
3. Ambiguous detection (medium confidence)
4. Asks: "I found both EEG and online experiment files. Which is primary?"
5. User: "Online"
6. Routes to: skills/online/SKILL.md

Configuration

Load from configs/biopsych-config.json:

{
  "modalities": ["eeg", "fmri", "online", "psychophys"],
  "confidence_threshold": 0.6,
  "scan_depth": 3,
  "file_patterns": {
    "eeg": [".set", ".vhdr", ".eeg", ".bdf"],
    "fmri": [".nii", ".nii.gz"],
    "online": [".html", ".js", ".json"],
    "psychophys": [".acq", ".edf", ".ibi", ".rr"]
  }
}

Related Skills

• skills/eeg/SKILL.md - EEG/ERP preregistration
• skills/fmri/SKILL.md - fMRI preregistration
• skills/online/SKILL.md - Online behavioral preregistration
• skills/psychophys/SKILL.md - Psychophysiology preregistration

Templates

All templates in templates/ directory:

• templates/biopsych/common/METADATA_SCHEMA.md - Required for all
• templates/biopsych/common/ETHICS_PRIVACY_TEMPLATE.md - Required for all
• Modality-specific templates (routed by skill)