• conscious perception
• psychophysics
• pupillometry
• time-series analysis

This thesis investigates the use of pupil behavior to automatically classify stimuli as perceived or not-perceived during psychophysical procedures. Psychophysics traditionally relies on subjective reports, which can be time-consuming and prone to bias. This work proposes a framework using pupil dilation response (PDR) to objectively measure perception.

A custom experimental setup was developed, including a pupillometer, a stimulation device, and software for data acquisition and analysis. The pupillometer uses off-the-shelf components and 3D printing. The stimulation device employs a magnetic linear stepper motor controlled through a custom h-bridge circuit and an Arduino microcontroller.

Experiments were conducted on 10 participants. The results show that the developed system can successfully acquire pupil data and statistically differentiate between perceived and non-perceived stimuli. An ensemble of two k-Nearest Neighbors (KNN) classifiers based on Dynamic Time Warping (DTW) achieved a positive predictive value (PPV) and negative predictive value (NPV) of around 75\% for all participants.

The thesis highlights the potential of pupil behavior analysis for psychophysics. Obtained results suggest that the adopted dynamic approach is suited for PDR analysis and classification. However, limitations exist. The classification performance depends on participant focus and sustained attention. Further research is needed to account for attention and improve classification accuracy.