Back to Journey
Phase 4 • Current Phase

Making It Real

From piloting to behavioral testing to fMRI. This is where theory meets human decision-making.

01

From Theory to Practice

We have 286 optimized trials that maximize parameter discriminability. We have 16 nested models ready to test. Now comes the real work: testing this framework with real human behavior.

Three stages: Piloting to refine the experiment. Behavioral testing with 100 participants. fMRI to see what the brain is doing.

The Experimental Pipeline

1

Piloting

Currently Running

We're running pilot studies to optimize the experiment. Starting with 286 trials was too long—participants got fatigued. The goal: reduce to 150 trials per condition without losing parameter discriminability.

What we're learning:

  • • Which trials are most informative in practice
  • • How to keep participants engaged throughout the study
  • • What instructions actually make sense to real people
  • • Response time patterns and attention checks that work

Current

286

Target

150

Conditions

2×

2

Behavioral Experiment

Next Phase

Once piloting is complete, we'll run the full behavioral study with 100 participants. This will give us the data to test whether Prospect Theory can explain trust, or if we need additional social preference parameters.

Study Design

  • • N = 100 participants
  • • 300 trials total (150 per condition)
  • • Within-subject design
  • • Real stakes (€8 base + bonus)

Hypotheses

  • • Do α, λ, γ⁺, δ differ?
  • • Which of 16 models fits best?
  • • Individual differences prediction
  • • Betrayal aversion existence

Power analysis: With 100 participants × 300 trials = 30,000 observations, we have >95% power to detect medium effect sizes (Cohen's d = 0.5) in parameter differences.

3

fMRI Study

Future Phase

Once we know which trials and parameters matter behaviorally, we'll run an fMRI study to see where in the brain these computations happen. Shorter design: 70 trials per condition.

Why fMRI?

  • • See if brain regions differ between Trust and Risk
  • • Test if value signals are computed differently
  • • Look for neural signatures of betrayal aversion
  • • Validate computational model neural correlates

Participants

55

Trials / Condition

70

The Experimental Platform

To ensure precise timing and control, we built a custom PsychoPy framework for in-person testing at the Donders Institute.

Dual-Task Paradigm

Trustee Role (Phase 1) & Trustor Role (Phase 2)

Optimized Trials

286 trials selected via Fisher Information

Social vs Non-Social

Human faces vs Lottery objects

experiment_structure.yaml

experiment:

platform: "PsychoPy"

location: "In-Person (Donders)"

phases:

- name: "Reciprocation Task"

role: "Trustee"

- name: "Inverted Trust Game"

role: "Trustor"

conditions: ["Social", "Non-Social"]

stimuli:

faces: "Chicago Face Database (N=300)"

The platform is ready. The design is optimized. Time to test it in the real world.

Explore Phase 5: Data Collection