The induction of a specific mental set for problem solving is accompanied by increased neural representational similarities.

A mental set is a cognitive bias that results from repeatedly applying a stereotypical problem-solving strategy, which can hinder new and creative ideas. However, the neural mechanisms underlying mental set induction remain unclear. We hypothesized that repeated use of a strategy increases neural representation similarity, thus producing a mental set. To test this, we conducted representational similarity analysis across continuously presented mindset-induction trials using the same loose chunk decomposition problem-solving strategy. This induction was expected to impair the solving of subsequent probe trials requiring a more insightful tight chunk decomposition strategy. Results showed that neural representational similarity increased across loose chunk decomposition induction stages, particularly in regions related to cognitive control, mental operations, and the default mode network. These increases predicted activation in the anterior cingulate cortex and medial frontal lobe during tight chunk decomposition trials. Furthermore, the medial frontal lobe moderated the impact of increased similarity during mental set induction, affecting representational changes from loose chunk decomposition to tight chunk decomposition in perceptual, operative, and executive processing regions. In summary, our findings suggest that increased neural representational similarity in goal-directed mental manipulation networks and the default mode network supports mental set induction, enhancing conflict resolution and representational change during creative problem-solving with different strategies.