Model of the relationship means much more than the people that form them.

The first of its kind study of romantic relationships using artificial intelligence, which provided data on more than 11 thousands of pairs, has allowed to identify the main factors by which partners are positive about such relationships. Article on the study published in the Proceedings of the National Academy of Sciences, writes Naked science.

The analysis based on machine learning algorithms and found that the indicators related to relationships as such, were more powerful predictors of relationship quality in General, the indicators based on the individual characteristics of people. In other words, the type of relationship that you build with a partner can be more important for life satisfaction and happiness than any of your individual circumstances. “The amazing thing was that as soon as you have all the information related to particular relations, individual differences went by the wayside,” says the study’s lead author Samantha Joel from the Western University in Canada.

The science of the relationship exists not one ten years. It generated a huge number of psychological theories about what makes couples happy and that this is prevented. However, scientists agree that to reinforce the significance of the results obtained in the course of a small Autonomous research needed accumulation of these data and their large-scale assessment.

One of the analytical solutions for this may be artificial intelligence that can process vast amounts of data collected from individual institutions and laboratories. This approach was used in the new work, scientists have used an algorithm called random forest (“random forest”) that combines many decision trees.

With the help of this algorithm was analyzed the performance of 11 for 196 pairs of the 43 individual datasets. “The results showed that the characteristics that reflect its own perception of the relationship (e.g., degree of conflict, attachment), predict up to 45 percent of the variance in the assessment of the quality of relations at the beginning of each of the research,” the authors write in their article.

From the point of view of the specific relationship variables that most reliably predicted quality of interaction pairs, the most reliable were: loyalty partner commitments (e.g., “my partner wants our relationship to last forever”); gratitude (“I was very lucky that I have a partner in life”); the perceived satisfaction of the partner (e.g., “our relationship, make my partner very happy”); and the degree of frequency of conflicts and the level of sexual satisfaction.

Factors related to individual characteristics (age, gender, personal characteristics of the respondents) in most cases explained only 21 percent of differences in the quality of the relationship. Among the individual characteristics that most strongly predicted by the quality of the relationship, were: life satisfaction, negative affect (e.g., anxiety or irritation), depression; avoidant attachment (e.g., “I prefer not to be too close to my partner”) and anxious attachment (degree of concern about their relationship with others).

“The experience of negative affect, depression, or insecure attachment, of course, is a risk factor in the relationship — summarized the authors of the work. But if people nevertheless manage to establish a relationship characterized by gratitude, sexual satisfaction and absence of conflict, and they thus consider your partner’s loyal and helpful, these individual characteristics have negligible influence”.

The findings of scientists, in fact, be reduced to a simple truth: the person we choose is not as important as the relationship we build with him. “A model of the relationship that you create with someone, shared norms, jokes, shared experience — [means] a lot more than individuals that shape these relationships,” says co-author Joel INVERS.