Resource-rational allocation of metacognitive resources in value-based multi-attribute choice

Decision makers overcome computational limitations by allocating limited cognitive resources to the most important information. Here, we investigate whether decision makers also efficiently allocate metacognitive resources when making computationally complex decisions. We randomly assigned participants (n = 815) to complete a value-based, multi-attribute choice task that included four or eight attributes. Participants then self-reported the weight they placed on each attribute. We compared the alignment between their stated weights and weights estimated from their choices – a measure of metacognitive knowledge. Participants in the 8-Attribute condition had less accurate metacognitive knowledge. This effect was driven by less accurate metacognitive knowledge only for low-importance attributes. Metacognitive knowledge for high-importance attributes was equally accurate across conditions. This suggests that, as tasks become more resource intensive, decision makers efficiently allocate their limited metacognitive resources to the most important attributes. These findings open the door to future research into the mechanisms underlying this resource-rational behaviour.