physiognomy
Physiognomy is a pseudoscience based on associating personal characteristics and traits with physical differences and especially with elements of people's faces. The word is derived from the ancient Greek word for nature, physis, and the one for judge, gnomon.
One common example of physiognomy is correlating a high brow with intelligence and a greater affinity for the arts. Other remnants of physiognomy include the expression "stuck up," which comes from the theory that people with upturned noses have a contemptuous attitude the term "thick-headed" to describe stupidity. No valid evidence exists for those associations or for any other claim of physiognomy.
The origins of physiognomy go back to at least 500 BC, when Phythagoras, the Greek philosopher and mathematician, selected his students based on whether or not their physiognomy indicated that they were intellectually gifted. Aristotle wrote that a broad face indicated stupidity, while a round face indicated courage.
In the mid-19th century Casare Lombroso, an Italian scientist, promoted the idea that criminals were genetic throwbacks and that they could be identified by a number of physical characteristics including hawk-like noses, fleshy lips, low and sloped foreheads, flattened or upturned noses and longer than average arms.
Physiognomy is currently enjoying a ressurgence, at least in part because of the development of artifical intelligence and big data analytics technologies. Proponents claim that the ability of advanced technologies to detect minute differences and correlate huge volumes of data could lead to more valid inferences. Critics, on the other hand, argue that human bias is likely to lead to flawed results.
In late 2016, Xiaolin Wu, a professor at McMaster University and Xi Zhang of Shanghai Jiao Tong University published an article, "Automated Inference on Criminality Using Face Images." The paper suggested that machine learning algorithms could determine which individuals in a collection of scanned photos was likely to be a criminal, solely based on their physical characteristics. In the paper's abstract, the authors admitted that the software was unavoidably reliant on human perceptions, which are know to be error-prone. However, they went on to say that “our empirical evidences point to the possibility of training machine learning algorithms, using example face images, to predict personality traits and behavioral propensity.” The paper has since been removed from ArXiv, the repository where it was published, and the authors have posted a response to critics agreeing with, for example, the need for "policing AI research for the good of society."
Phrenology, a variation of physiognomy, is based on associating personal characteristics with measurements of the skull.