Advising the ill-advised: Debiasing the conjunction fallacy through gamified training with outcome feedback

Area of Science:

• Cognitive Psychology
• Decision Science
• Behavioral Economics

Background:

• Many decisions involve uncertainty, but logical constraints influence outcome probabilities.
• People often exhibit reasoning errors, notably the conjunction fallacy, judging combined events as more likely than individual ones.
• Previous interventions for conjunction fallacies have limited success and generalizability.

Purpose of the Study:

• To investigate if repeated exposure to decision tasks with minimal feedback can mitigate conjunction fallacies.
• To determine if self-guided learning can improve probabilistic reasoning without explicit instruction.

Main Methods:

• Fifty-six participants were randomly assigned to a training group (conjunction tasks with feedback) or a control group.
• The training group engaged in repeated decision-making tasks.
• Performance was assessed on trained and untrained conjunction tasks, as well as unrelated base-rate tasks.

Main Results:

• The training group showed significant improvement on both trained and untrained conjunction tasks.
• No improvement was observed in the control group or on unrelated base-rate tasks.
• Performance gains were gradual, indicating strategy development over time.

Conclusions:

• Conjunction fallacies can be effectively reduced through self-guided learning with minimal outcome feedback.
• This approach offers a practical method for enhancing probabilistic reasoning in real-world scenarios.
• The findings suggest that learning from experience is key to overcoming cognitive biases.