3-AFC Metacognitive Sensitivity Task

A Psychtoolbox-powered random-dot kinematogram for measuring how well confidence tracks visual-decision accuracy.

---

Overview

This repository delivers a complete MATLAB experiment that replicates and extends Zehetleitner & Rausch (2013).
Participants view a circular aperture of moving dots, decide whether the dominant motion is left, right, or absent, and then rate their confidence.
Pupil-size changes can be recorded in parallel via any eye-tracker that streams to a serial port (optional).
The code exports trial-wise data ready for Type-2 signal-detection and meta-d′ analysis.

---

What you get

• 9 coherence levels (0 % – 100 %) fully randomised across 441 trials
• 3-AFC response: left-arrow / right-arrow / “N” for pure noise
• 6-point Likert confidence scale (mouse-driven, 1 = very low → 6 = very high)
• Adaptive Gabor mask (45°, 1-s fade) hides stimulus onset
• Automatic breaks every block; run-time ≈ 55 min
• Instant CSV output (PXXX_Data.csv) with fields:
  participantNumber, trial, block, coherence, direction, response, correct, confidence, RT, age, sex

---

Quick start

1. Clone or download the repo
2. Open MATLAB, cd into the folder
3. Run Metacognition_Test.m
4. Enter participant ID, age, sex when prompted
5. Follow on-screen instructions; data file appears in the same folder once the task ends

---

Requirements

• MATLAB R2020b or later
• Psychtoolbox-3 (free)
• Windows / macOS / Linux with ≥ 120 Hz display recommended
• (Optional) eye-tracker streaming pupil diameter to a COM port

---

File map

File	Purpose
Metacognition_Test.m	Main experiment driver
movingDots.m	Stimulus routine (coherent + random dots, Gabor overlay)
getConfidenceRating.m	Interactive Likert scale (mouse hover + click)
generateCircularGabor.m	Creates oriented Gabor patch texture
constructContingencyTables.m	Builds Type-2 SDT table and writes CSV
angle2pix.m	Converts visual angle → screen pixels
displayInstructions.m	Full-screen instruction slides

---

Analysis tips

• Type-2 hits / misses / false-alarms / correct-rejections are already coded by constructContingencyTables
• Feed the CSV into meta-d′ MATLAB toolbox or compute classic Aroc / Broc directly
• Coherence × confidence interaction? A simple mixed ANOVA (confidence as DV, coherence as within factor) usually reveals the expected monotonic relationship when the task works

---

Citation

If you use this code, please cite:
> Choudra, A. (2024). Exploring Metacognitive Sensitivity through a 3-AFC Visual Motion Discrimination Task. MSc Dissertation, University of Reading.

and the original paper:
> Zehetleitner, M., & Rausch, M. (2013). Being confident without seeing: What subjective measures of visual consciousness are about. Attention, Perception, & Psychophysics, 75(7), 1406–1426. https://doi.org/10.3758/s13414-013-0505-2