Why is this poster making faces at me?

These Chernoff faces are a multivariate technique of visualizing data that relies on the human eye's ability to recognize patterns. Each object (QSO) is represent by a face and each QSO property is represented by a facial feature. In this sample, the facial features correspond to:
Face Width CIV FWHM
Upper Face Eccentricity Radio-loudness
Lower Face Eccentricity Optical Lumiosity
Mouth Optical/X-ray
Curvature Spectral Index
Nose Length Radio Core Fraction
Eye Slant Radio Spectral Index
Eye Height Optical Spectral Index
Eye Separation H-beta FWHM
The first 17 QSOs are the ones that host NALs. How do they differ from the others? Are they happy? Are they sad? Do they scowl at you?

Rajib Ganguly
Last modified: Fri Feb 4 14:54:22 EST 2000