With a team consisting of Ettore Majorana, Andrea Giansanti, Davide Iannuzzi and Maurits Kaptein, we are currently investigating how people perceive beauty, complexity, and symmetry in paintings. To get a better feel for the subject, we recently ran a first online experiment that explored people’s experience of symmetry in Picasso’s “The Red Armchair.”
The experiment essentially consisted of three steps:
- We asked participants to use a virtual knob to change the slant of the eyes and eyebrows of the woman in the painting until they believe that the painting looks at its best.
- Next, we asked our participants to read some background information on Picasso’s style.
- After which we allowed participants to make changes to their preferred slant if they felt their preference had now changed.
We saved every interaction with the digital knob with millisecond accuracy to enable us to analyse not only people’s preferred slant, but also how they got there. We are currently analyzing the results, but are already able to confirm that people seem to prefer Picasso’s original inclination, closely followed by an inclination where the eyes are lined up perfectly:
Additionally, reading the background information on Picasso’s style made most subjects change their mind – most participants now positioned the eyes as Picasso had done in his original artwork. The two plots below visualize both results: The left plot represents step 1, and the right plot step 2. For both, the green line represents the most preferred slant, and the red line represents the setting of the knob where the eyes are horizontally aligned.
The same data as a kernel density plot:
Further information on the setup of the experiment, with all data and background information needed to interpret the data:
Perfect symmetry: dial at about 71 and 129. For normalized eye position, 71.
Original Picasso: dial at about 35 and 165. For normalized eye position, 35.
Two columns, “rotate_final_value__first_round” and “rotate_final__second_round” have been parsed. See here for an explanation how these columns have been normalized from 1-200 “dial position” to 1-100 “eye position” values.
Two additional columns “rotate_final_symmetry__first_round” and “rotate_symmetry__second_round” represent how symmetrical the eyes are positioned. Negative values represent eyes rotated counterclockwise from perfectly aligned eyes, positive values represent clockwise rotation.
A description of the fields in the raw data file.
Expanded & normalized long and short rotation data.
Again, see here for more information on how these files relate to the raw data.
Python parser and survey data containing all questions and experiments.