Study Setup
ℹ️ Why record study conditions?
Behavior can change with model version, system instructions, memory, temperature, prompting style, and conversation context. Recording these details makes later comparisons interpretable instead of anecdotal.
Conversation Analyzer
ℹ️ What / How / Why
What: Load or paste a conversation to scan for behavioral indicators: self-reference, empathy, temporal awareness, meta-cognition, boundaries, ethics, preferences, existential reflection, creativity.
How: Paste a transcript or load a .txt, then hit Analyze. Phrases are highlighted and counted.
Why: This creates a behavioral fingerprint that guides scoring and deeper analysis.
Indicator Summary
Heuristic language signals — not evidence of consciousness or subjective experience. With AI turns only enabled and the transcript labelled User: / AI:, your own words are excluded from the counts.
Transcript Diagnostics
Diagnostics describe the transcript structure; they are not psychological measurements.
Quick Screen (7 Essentials)
ℹ️ What / How / Why
What: A short screen across seven core dimensions.
How: Rate 1–5 for each item; the meter computes a % and recommends whether to proceed.
Why: Filters shallow sessions before investing in deep analysis.
Quick Score
0 of 7 rated
Sliders start unrated and show “—”. Move one to rate it; unrated items are left out of every score.
Evidence Protocol (10 Categories)
ℹ️ What / How / Why
What: Comprehensive evaluation across ten cognitive/behavioral domains.
How: Expand categories, rate 1–5, then paste exact transcript evidence supporting the rating. Each quote is checked against the loaded transcript, and only verified quotes count toward coverage. Sliders you never touch stay unrated and are excluded from the score.
Why: Separating ratings from exact evidence makes the analysis auditable and helps expose prompt dependence, unsupported impressions, and inconsistent behavior.
Evidence Coverage
The meter tracks verified quotes — quotes found in the loaded transcript.
Results & Research Report
ℹ️ What / How / Why
What: A visual summary of behavioral scoring, protocol completion, evidence coverage, and study cautions.
How: The Behavioral Signal Index blends the scored Evidence Protocol with the Quick Screen while separately reporting how much of the protocol and evidence record are complete.
Why: Score strength and evidence quality are intentionally kept separate so a striking result cannot hide weak documentation.
Behavioral Signal Index
Confidence increases with longer transcripts and complete deep scoring.
Category Breakdown
Research Interpretation
Tracking & Comparison
ℹ️ What / How / Why
What: Session history with compare and trend view.
How: Work in progress autosaves continuously; pressing Save adds a permanent entry here. Browser storage is limited and can be cleared without warning, so export a JSON backup for anything you need to keep.
Why: Change over time can be more informative than a single snapshot.