Precision and generalization are two fundamental properties of memory function. The former allows individuals to distinguish between similar episodes to reduce interference. The latter allows individuals to link information across episodes and apply past knowledge to new situations. Advanced age is associated with declines in memory function that affect some aspects of memory more than others. Memory specificity, in particular, is known to show substantial age-related decline, whereas the ability to remember the `gist' of past events is typically intact. However, it is unclear if the spared gist memory is related to generalization or simply reflects a lack of specificity. If generalized representations form independently of event-specific representations, older may show intact generalization. Alternatively, if generalization arises from representations of individual events, aging may reduce memory specificity and impair generalization. The proposed research seeks to test the relationship between memory precision and generalization and how that relationship changes with age. Using category learning as a representative domain, three studies will use cognitive modeling of behavioral data, pattern information analyses of functional magnetic resonance imaging (fMRI) data, and model-based fMRI to identify the cognitive and neural bases of memory precision and generalization in young and older adults. Study 1 will test the hypothesis that young but not older adults deploy generalization and precision flexibly depending on the task context. To do so, a category-learning task will be used with training sets that we expect will differ in their demands on representing individual items (i.e., precision) and abstracting to a category prototype (i.e., generalization). Participants will then be tested on memory for individual exemplars presented during training (precision test) and on their ability to classify new exemplars (generalization test). Study 2 will test the hypothesis that older adults have difficulty forming representations of individual items while representations of abstract, prototypical information remain stable with age. FMRI data will be collected while young and older adults complete the training phase of a category-learning task. We will use pattern information analyses (multivariate pattern analysis, representational similarity analysis) and model-based fMRI to track the development of item and category representations over the course of learning in each age group. Study 3 will test the hypothesis that young adults maintain representations of both individual exemplars and category prototypes, and that exemplar representations are used primarily during recognition whereas prototype representations are used primarily during categorization. We propose that older adults will use prototype representations during both categorization and recognition due to difficulties in utilizing exemplar representations. Together, these studies will characterize age-related changes in the relationship between generalization and precision in order to serve as a comparison model for age-related dementias and support development of interventions to aid memory-based cognition.
The proposed research investigates the effects of aging on precision and generalization in memory. Characterizing age-related changes in the relationship between generalization and precision will serve as a comparison model for age-related dementias and support development of interventions that can aid memory accuracy and memory flexibility.
| Bowman, Caitlin R; Zeithamova, Dagmar (2018) Abstract memory representations in the ventromedial prefrontal cortex and hippocampus support concept generalization. J Neurosci : |
| Bowman, Caitlin R; Sine, Shalome L; Dennis, Nancy A (2017) Modulation of target recollection and recollection rejection networks due to retrieval facilitation and interference. Learn Mem 24:607-611 |
