The effects of age on category learning and prototype- and exemplar-based generalization

Abstract

The need to learn new concepts and categories persists through the lifespan, yet little is known about how aging affects the concept learning and generalization. Here, we trained young and older adults to classify typical and boundary category members, and then tested category generalization to new stimuli. During training, older adults had increased difficulty compared to young adults learning category labels for boundary items, but not typical items. At test, categorization performance that included new items at all levels of typicality was comparable across age groups, but formal categorization models indicated that older adults relied to a greater degree on generalized (prototype) category representations than young adults. These findings align with the proposal that older adults are able to form category representations based on central tendency even when they have difficulty learning and remembering individual category members. More broadly, the results contribute to our understanding of multiple categorization strategies and the limited strategy flexibility in older adults. They also highlight how reliance on preserved cognitive functions may sometimes help older adults maintain performance. (PsycInfo Database Record (c) 2022 APA, all rights reserved).

Figure 1.

Category structure. Conceptual depictions of category representations under the assumptions of the A. prototype and B. exemplar models. C. Example stimuli labeled in terms of the number of shared features with the category A prototype (leftmost stimulus). Members of category A share more features with prototype A than prototype B (rightmost stimulus) and vice versa. D. Example training stimuli labeled in terms of their distance from their respective category prototype. Four items that differed from the prototype by two features (distance 2 items) and three items that differed from the prototype by four features (distance 4 items) were included in the training set for each category. The prototypes themselves are depicted in the figure to show all prototypical features of each category but were not included in the training set.

Figure 2.

Task procedure. A. Participants underwent feedback-based training, B. an old/new recognition test, and C. a categorization test where they classified old training items and new generalization items without receiving feedback.

Figure 3.

Training accuracy and reaction times. A. The proportion of correct responses for each block of training separated by age group and training item distance. B. Response times (correct responses only) separated by block, age group, and training time distance. For A-B, responses for items differing from prototypes by two features (distance 2) are presented in black and responses for items differing from prototypes by four features (distance 4) are presented in grey. Solid lines represent responses for the young adult group. Dashed lines represent responses for the older adult group. Error bars represent the standard error of the mean across subjects.

Figure 4.

Categorization accuracy. A. The proportion of correct categorization responses and B. reaction times (correct responses only) for new items differing from prototypes by 0-4 features presented separately for young adults (solid line) and older adults (dashed line). C. The proportion of correct categorization responses and D. reaction times (correct responses only) for distance 2 (dark lines) and distance 4 (lighter lines) items separated for old (training) items versus new items presented separately for young (solid lines) and older adults (dashed lines). Error bars depict the standard error of the mean across subjects.

Figure 5.

Exemplar and prototype model fits to categorization data in terms of the percentage of subjects best fit by each model. Dark grey bar segment = prototype users, medium grey bar segment = exemplar users, light grey bar segment = those with similar fits for the two models, white bar segment = those whose fits did not differ from the null model based on random data (i.e., ‘chance’).

Figure 6.

Means of the maximum attention weight estimated from prototype and exemplar models. Values closer to .1 represent attention distributed evenly across 10 stimulus features, whereas values closer to 1 represent attention focused on only 1 feature. Solid grey bars depict young adult means, stiped grey bars represent older adult means. Error bars depict the standard error of the mean across subjects.

Figure 7.

Recognition accuracy. Recognition accuracy is presented in terms of the corrected hit rate (hits – false alarms) separately for young (solid bars) and older adults (striped bars) for the overall test (including all stimuli), only distance 2 items, and only distance 4 items. Error bars depict the standard error of the mean across subjects.