Curiosity, error correction, and confidence in younger and older adults

The ability to learn from errors is a cornerstone of adaptive cognition, as errors mark salient points of breakdown in our knowledge and thus provide potent opportunities for learning. Although older adults show age-related declines in episodic memory, inhibitory control, and error awareness, they nonetheless demonstrate preserved-and in some cases equivalent-ability to correct errors in semantic knowledge when they are provided with feedback. This apparent paradox raises a fundamental question about how older adults successfully update knowledge despite declines in cognitive processes that are traditionally thought to support error correction. The present study examined whether age-related differences in curiosity and confidence shape patterns of semantic error correction.

Participants answered general information questions, indicated how confident they were in their answer and how curious they were to find out the correct answer, and then were told the correct answer. Later, they were retested on all of the questions. Following the task, participants completed a self-report questionnaire that assessed intellectual humility.

Older adults had greater first-test accuracy on the general information questions than younger adults; their confidence was higher than younger adults; their curiosity was greater than younger adults; and, despite having answered more question correctly to begin with, and thus having more difficult questions remaining, they did equally well on error correction. While younger adults preferentially corrected their high confidence errors (i.e., they hypercorrected), older adults did not, and instead corrected both high and low confidence errors about equally. Older and younger adults were equally 'open to modifying their views' - an aspect of the intellectual humility related to trait curiosity that, here, predicted both curiosity ratings and error correction.

This research contributes to a growing body of research showing older adults' spared memorial processing, including preserved error correction, for information about which they have greater knowledge. We posit that a difference in the density of the semantic landscape-favoring the older over the younger adults-may be responsible for the effects observed here.