The structural design features of virtual gambling machines are likely to promote specific learning processes in the human reward system that cause gamblers to form erroneous expectations and beliefs. That is the result of a theoretical and overview study by cognition and neuroscience expert Professor Dr. Jan Peters at the University of Cologne, in which he combines results from various empirical and theoretical studies.

Peters explores the links between the design of so-called multi-line electronic gambling machines, and the learning processes that can lead gamblers to form erroneous expectations regarding their control over outcomes and their chances of winning in these games. The work was published in the journal Trends in Cognitive Sciences under the title “A neurocomputational account of multi-line electronic gambling machines.”

It is known that gambling can trigger dopaminergic effects in the brain’s reward system similar to effects elicited by drugs of abuse. In recent years, the segment of so-called “multi-line electronic gambling machines” has grown in the gambling market. These games are played on computers or devices with touchscreens such as smartphones and tablets, but also in arcades or casinos.

The difference between these and classical mechanical slot machines is that gamblers can simultaneously bet on a large number of paylines. This means there are more potential winning combinations as well as other additional features to keep gamblers engaged. These include, for example, losses that are disguised as wins and “near-misses” where the winning symbol closely misses the right spot. This suggests to gamblers that the next win is just around the corner.

The learning model that Peters applies to this situation is so-called latent state inference. According to this cognitive science concept, people try to build an internal model of their environment in order to make predictions. This model is informed by past experience and expectations as well as current events to predict the present state of the environment (the “latent state”). People will therefore attribute the same or similar situations to the same latent state.

For example, if situations differ in terms of reward expectation, this is a very useful learning process. In the case of gambling, however, the outcome of each gambling situation is unpredictable, and the objective expectation of winning is always negative.

Through their design features, including reward uncertainty, near misses, losses-disguised-as-wins and audio-visual features, virtual slot machines activate dopaminergic mechanisms that can reinforce the learning of latent states. This may explain the tendency of gamblers to develop erroneous beliefs and expectations, such as magical thinking about lucky streaks or about “hot” and “cold” states of the machine, which they believe can produce different outcomes.

“Discussions about the potential dangers of gambling often focus on individual risk factors. Unfortunately, they often do not take into account the design features of different gambling products. However, the design of virtual gambling machines could trigger erroneous learning processes in gamblers via the dopamine system. The effect is likely exacerbated in comparison to traditional mechanical slot machines.

“This is also in line with the finding that erroneous beliefs and expectations are likely to be triggered or reinforced by regular gambling,” said Jan Peters. Such beliefs and expectations make people continue to gamble despite high losses, sometimes with dramatic consequences for themselves and their environment.

Further research could directly test these theoretical predictions, for example by experimentally investigating the effects of specific design features. Pharmacological approaches can also contribute to this research by directly assessing the role of dopamine.