Hans Eysenck's PEN model describes personality with three independent, biologically-grounded dimensions, each scored roughly -100…+100 relative to the population mean:
P — Psychoticism (tough-minded / aggressive vs empathic / conforming)
E — Extraversion (sociable / stimulation-seeking vs reserved / solitary)
N — Neuroticism (emotionally reactive / anxious vs emotionally stable)
vector magnitude |v| = sqrt(P² + E² + N²)
The classical four temperaments (traced to Galen, reframed by Eysenck as a 2D circumplex of just E and N) fall into quadrants:
E>0, N>0 → Choleric (excitable, assertive)
E>0, N<0 → Sanguine (stable, sociable)
E<0, N>0 → Melancholic (anxious, reserved)
E<0, N<0 → Phlegmatic (calm, reserved)
Sliders move your marker through the P–E–N space in real time. The top canvas is a genuine 3D wireframe cube of that space, projected to 2D — drag it with the mouse (or a finger) to rotate the camera, exactly like orbiting a real 3D scene, only computed and drawn entirely with 2D canvas math (rotation matrices + a perspective divide), no WebGL. The lower canvas is the classic flat E–N circumplex used in psychometrics, with the four temperament quadrants tinted. Random person samples a point from a rough Gaussian around the population mean, the way real trait scores cluster. Add to population drops a dot at your current position in both views — this is exactly how psychometric researchers plot cohorts on trait axes after collecting questionnaire data (e.g. the EPQ-R inventory).
Eysenck proposed a physiological basis for two of the axes: extraverts have chronically lower cortical arousal and seek external stimulation to compensate, while high-N individuals show a more reactive limbic/autonomic system, firing more easily under stress. P was added later and correlates with lower serotonergic and higher dopaminergic activity in some studies. The model remains influential but has been partly superseded by the five-factor (Big Five/OCEAN) model, which adds Openness, Conscientiousness and Agreeableness as further independent axes.