Whether it is about the classification of flora, geometric figures, emotions, colors, etc., anthropologists have noticed for a long time that very different societies and without contact with each other tend to classify phenomena into similar categories.
Newsletter Science&Vie
Receive all science news, three times a week
—-
This has led to the postulation that the principles of categorization are innate and similar for all humans, therefore linked to our brain structure.
A conclusion in total contradiction with what we observe at the individual level: individuals show enormous variations in the appreciation and categorization of phenomena among themselves … However, researchers have removed this contradiction by “tracking” it. way in which individual variability collectively converges towards a consensus that seems identical across groups and cultures.
Categorize unknown shapes
The 1480 volunteers were recruited on-line, and the experiment also took place via networks. The general idea was as follows: to present to these individuals some 1,500 original, varied and random geometric shapes (no squares, triangles, circles) by asking them to classify them into categories.
The experiment was carried out on isolated individuals, on 80 pairs of individuals (able to communicate with each other), on 15 groups of 6 individuals, on 15 groups of 8, on 15 groups of 24 and 15 groups of 50 individuals.
They then observed a very great statistical proximity in the classification proposed by the groups of 50 individuals, while for the isolated individuals and the pairs, the classifications were totally heterogeneous and statistically decorrelated – for the groups of 6, 8 and 24 individuals, the results got closer and closer with the increasing number of people.
“If I assign an individual to a small group, that group is much more likely to come up with a very idiosyncratic and specific category system, Douglas Guilbeault co-signed the article to the press. But if I assign that same individual to a large group, I can predict the system of categories that they will eventually create, regardless of the unique perspective that person brings.“
The collective to regulate networks
Thus, the researchers describe, a large group (here, 50 people) sees a large number of individual suggestions hatch, some of which are reinforced by repetition and interaction between individuals, until one becomes strong enough to statistically dominate.
If therefore the ability to categorize is indeed of a cerebral nature, the convergence towards a standard categorization is the work of collective thought – itself carved out by the evolution of our species. In the present case, there would therefore be a sort of determinism of thought, not individual but collective.
Hence the idea of researchers to use this tool offered by evolution (for our greater good) to regulate certain harmful phenomena in social networks.
So, rather than letting network administrators (Facebook, Tweeter, etc.) find an algorithmic way to recognize hateful or conspiratorial speech, the idea would be to find a collective means of judgment.
Likewise, the idea of reinforcing among physicians the appeal to peers to establish the most accurate diagnosis – thus avoiding biases linked to individual variability.
Source : Nature, January 2021.
–
