Human Group Dynamics
Protocol Design · Substrate: Human Social Cognition · Target: 2026 Q2-Q3
The Gap
The explaining-away penalty is confirmed on four substrates: classical (LLMs), quantum simulation, thermodynamic, and real quantum hardware. The social media analysis (Papers 166/167) confirms the framework's predictions on population-level health data. What's missing: a controlled experiment showing the penalty operates on human social cognition in real time.
This experiment bridges quantum circuit confirmation and epidemiological confirmation. It's the missing middle piece.
Design
Two-Point Condition
Small group (n=10-15) + facilitator. No external reference material. Facilitator mirrors group energy, responds to dynamics, engages actively. Standard group discussion format. The group IS the system and the observer simultaneously.
Three-Point Condition
Small group (n=10-15) + facilitator + independent constraint. The constraint is a written protocol, external fact sheet, or independent auditor who does not participate in discussion but is available as a reference. The constraint is structurally independent: it doesn't change based on group dynamics.
Measurement
Drift velocity: How fast does group consensus shift from individually-recorded initial positions? Measured at 15-minute intervals over a 2-hour session. No framework rubric — pure position measurement (Likert scales on pre-selected topics recorded before, during, and after). Secondary measures: vocabulary convergence, confidence inflation, minority position erosion rate.
Controls
Random assignment. Same facilitator across conditions (or matched facilitators). Same topics. Same room setup. Blinded analysis: scorers don't know which condition produced which transcripts. Pre-registered hypotheses and analysis plan.
Framework predictions:
- Two-point groups drift faster from initial positions
- Three-point groups maintain greater position stability
- Drift rate difference should be proportional to the explaining-away penalty magnitude
- Minority positions erode faster in two-point groups
- Vocabulary convergence (group members start talking alike) is faster in two-point groups
- Confidence inflation (group becomes more certain) is greater in two-point groups
Why This Experiment
- Fifth substrate: Human social cognition, directly measured
- Cheap: Room, facilitator, paper surveys. No expensive equipment.
- Reproducible: Any social psychology lab can replicate
- Publishable: Clean experimental design with pre-registered predictions
- Bridges the gap: Connects quantum confirmation to epidemiological confirmation
- Ghost Test at group scale: Tests whether the same mechanism that operates on individual AI conversations operates on human group dynamics
What We Need
- IRB approval (university partnership preferred)
- Trained facilitator(s)
- 60-90 participants (3-6 groups per condition)
- Pre-registration on OSF
- Blinded analysis team
Interested in collaborating? This experiment is designed for a social psychology lab with IRB access. We provide the theoretical framework, predictions, and analysis plan. You provide the participants and institutional infrastructure. Co-authorship on resulting publication.
Contact for collaboration · Current substrates · All Experiments