Psychopaths Sociopaths, what is your religion

You should recognize them by the words!

Scientists believe they can use automatic statistical text analysis to identify lies, the structure of terrorist groups or psychopaths

Online publications of all kinds, including and especially personal messages in social networks, are a treasure to be salvaged for many interests, from science to companies and security authorities. Before that, it not only has to be lifted, but above all decrypted in order to reveal its secrets.

It is well known that people not only reveal personal information in what they say, but they also reveal themselves even more through how they say it or what can only be found between the lines. It also depends on where and in which medium you express yourself. The communication scientist Jeff Hancock claims that people tend to lie more when they talk on the phone or when they meet than when they communicate via email, where they cannot assess how their words will be received by others, which apparently makes attempts at deception less attractive .

It is hardly surprising that people on Internet contact exchanges brighten up their pictures and hairdos data about their body such as height, weight or age in order to be more attractive. Even less surprising is that, as Hancock points out in his 2010 study published in Communication Research, people whose physical appearance is judged to be less attractive are more inclined to use deceptions to make them more beautiful. However, according to Hancock, this should only concern the physical appearance; other information such as occupation or income is less misleading.

It could be more interesting that Hancock and colleagues developed a program (Social Language Processing) to use utterances, messages, etc. to automatically capture the structure and dynamics of groups based on what and how is written, for example using language analysis of linguistic features to recognize when to deceive and when to tell the truth, what the hierarchies are and where there is tension in the group. The study Social language processing: A framework for analyzing the communication of terrorists and authoritarian regimes, published in 2010 in the journal Behavioral Sciences of Terrorism and Political Aggression, analyzed letters, communications and other documents from the administration under Saddam Hussein. From this one wants to have gained knowledge of how "status, cohesion and deception" can be recorded in terrorist groups or authoritarian regimes, for example whether the truth was told about any links between the Hussein regime and terrorist groups.

Hancock and his colleagues had previously examined statements by members of the Bush administration in the run-up to the Iraq war (On lying and being lied to: A linguistic analysis of deception). They compared false (Iraq has weapons of mass destruction) with non-false claims (Hussein used poison gas) and came to the conclusion that in false claims significantly fewer words excluding "I" were used, such as "except" or "but", but there were more uses emotional words and verbs of action. But obviously language analysis should only be used against the bad guys, after all the research was funded by the Pentagon, so it's hard to expose your own government.

How do psychopaths speak and write?

The latest study continues this approach to identify psychopathic killers based on their written statements. The intention is not only to be able to make diagnoses with such text analyzes, but also to support law enforcement or even to provide information on treatment or preventive detention.

However, the scientists are not yet ready to simply run their text analysis over the texts on the Internet and social networks in order to identify liars, terrorists or psychopaths. At first they only tried to distinguish "normal" murderers from psychopathic ones. To this end, they had 14 murderers diagnosed as psychopathic and 34 "normal", ie non-psychopathic murderers who had been behind bars for their acts in Canadian prisons, recount their murder. This was then transcribed and the texts analyzed using the Wmatrix program and a dictionary for emotionality in language.

The hypothesis is that psychopathic murderers tell their deed in peculiar words - and that their linguistic expression cannot be consciously controlled, but that their character can be recognized by means of the words. The scientists assumed that psychopaths express an instrumental worldview and hardly express emotions. Using certain words, they would identify themselves as egoists with emotional flatness and distance from their crimes.

That was also the result of the automatic statistical text analysis. Psychopaths use more causal connecting words such as "because", "there" or "so that", as the scientists write in their study, which was prepublished in the journal Legal and Criminological Psychology, compared to the other murderers. This would give the impression that something had to be done to achieve a goal. In addition, they would have used words related to physical needs like food, sex or money twice as often, while the non-psychopaths use more words for social needs such as family, religion or even spirituality. Psychopaths would also provide more details, such as what they ate on the day of their crime. On the other hand, they would rather tell stories in the past or speak less fluently, for example use more "Uh", which the scientists attribute to the fact that they want to appear more positive or have greater difficulty in describing an emotional event such as a murder.

If one were to search for psychopaths in the global network with these statistical probabilities, then many people would probably go online, especially since the scientists apparently believe that they could also find what they are looking for with their method in short Twitter messages and other statements on social networks. It is even suggested that it could be used to find a murderer who contacts his victims via social networks or websites.

Hancock probably wanted to make his research more interesting for the media, because with the method and with only a few test persons, at best, only a comparison between two different types of murderers based on a certain type of text is possible with a certain degree of probability. But it is also irresponsible to feed the media (and perhaps possible financiers) with such highly speculative prospects, it devalues ​​the reputation and credibility of science. In view of a demand from British scientists that journalists should first submit their contributions to research reports to the scientists for a fact check before publication, because the scientific articles have already been checked by peer review and do not require renewed criticism or appreciation, one could almost be tempted to To at least pass the buck back to some scientists. (Florian Rötzer)

Read comments (60 posts) https://heise.de/-3391682Report errorDrucken