The study uses cognitive tests that measure learning and decision-making in win-lose situations. From these tests, they apply pre-determined mathematical models to simulate participants' performance. The advantage of these models over real data is that they allow them to reveal decision patterns that are not obvious from analysing participants' response patterns alone. Using these models they estimate the ‘decisional noise’ which is the degree of stochasticity (changing sequence patterns that do not conform to a strategy of optimising gains and minimising losses) that each participant uses during their execution of the trials.

The main finding of the study is that a reduction in the estimated decisional noise in the tests is a mediator of the positive relationship between age and the use of sophisticated decision strategies such as, for example, the ability to incorporate not only your previous history of gains and losses, but also the possibility of changes in the prediction context (e.g., before booking a hotel, taking into account not only that it is your favourite hotel but also that depending on the time of year it may be too crowded and not a pleasant experience).

The authors interpret that, as the study participants are between 12 and 42 years old, their results suggest that the reduction of decisional noise with age may be a contributing factor to the development of more sophisticated decision-making strategies as we move from adolescence to adulthood.

The study is innovative and rigorous and the computational models are robust and well described. The main limitation, which the authors acknowledge, is that a longitudinal study (in which the same participants complete the tests over several years, allowing us to estimate noise reduction in each individual and understand whether it predicts the development of more sophisticated decisions) with a better segmented sample at different stages of adolescence would be needed to know whether decisional noise reduction really has a ‘maturational function’. This is where readers should be most cautious in interpreting the results.