Counting on Memory: How Expertise Shapes Our Numerical Judgments of Associations

Accurate numerical estimation underlies many aspects of cognition, from basic quantity judgments to complex decision-making. One domain where numerical reasoning is especially critical is memory, where individuals often must estimate the likelihood that one event or idea is associated with another. In this study, participants completed a free association task across multiple sessions to generate their own individualized word-pair norms. Later, they provided numerical probability judgments (0–100%) of how often they had produced each pair. These judgments were compared to collective free association norms, a matched group evaluating others’ pairs, and a traditional control group. Results showed that participants who judged their own pairs were significantly more accurate in estimating associative probabilities than control or matched groups, reflecting the benefits of expertise derived from repeated interaction with stimuli. However, systematic overestimation bias persisted, especially for weak associations, indicating that metacognitive sensitivity to probability differences remains limited. These findings highlight how expertise improves, but does not perfect, the ability to translate memory associations into numerical judgments, offering new insights into the intersection of numerical cognition, metacognition, and memory.