Researchers have uncovered how people make decisions in situations they have never encountered before. In a study published in the journal Nature, they created 121 new strategic games and examined how participants assessed the rules, made their initial moves, and predicted opponents’ actions without any prior experience.
Most previous work in this field focused on how seasoned players refine their skills in familiar games or how computers process a vast array of potential options. This new study zeroed in on a different question: how a person reasons when navigating an unfamiliar system of rules.
The developed games were digital strategies set on a grid-based board, similar in principle to tic-tac-toe, but featuring varying board sizes, rules, and win conditions. In some versions, forming a line meant victory, while in others it meant defeat. Over 1,000 people took part in the experiment, divided into several groups.
The first group examined an empty game board and a description of the rules, then assessed how fair and interesting the game seemed without making a single move. Participants in the second group played the unfamiliar games directly, while the third group watched the newcomers and tried to predict their next moves.
To explain the principles of human thinking, the scientists created a computer model called Intuitive Gamer. It simulated the human decision-making process: the model tested a few possible moves and evaluated only the immediate outcome—such as how to get closer to winning or hinder the opponent.
When the researchers compared the model’s decisions to those of the experiment participants, they found similarities. According to the study, when faced with an unknown situation, people do not try to calculate all possible outcomes. Instead, they use a limited number of quick mental simulations of options to select a sufficiently good course of action.
The study authors note that this behavior can be described as a systematic and adaptive way of making decisions: a person does not act randomly, but also does not engage in an exhaustive search of all possibilities, which would require too much time and resources.
The researchers believe that the Intuitive Gamer approach could be useful in creating artificial intelligence models. Models that employ similar strategies of simplified prediction can make sound decisions in novel situations while using significantly fewer computational resources than systems designed for a full analysis of all possible options.
About the Author
