Dopamine and memory dedifferentiation in aging

Abstract

The dedifferentiation theory of aging proposes that a reduction in the specificity of neural representations causes declines in complex cognition as people get older, and may reflect a reduction in dopaminergic signaling. The present pharmacological fMRI study investigated episodic memory-related dedifferentiation in young and older adults, and its relation to dopaminergic function, using a randomized placebo-controlled double-blind crossover design with the agonist Bromocriptine (1.25mg) and the antagonist Sulpiride (400mg). We used multi-voxel pattern analysis to measure memory specificity: the degree to which distributed patterns of activity distinguishing two different task contexts during an encoding phase are reinstated during memory retrieval. As predicted, memory specificity was reduced in older adults in prefrontal cortex and in hippocampus, consistent with an impact of neural dedifferentiation on episodic memory representations. There was also a linear age-dependent dopaminergic modulation of memory specificity in hippocampus reflecting a relative boost to memory specificity on Bromocriptine in older adults whose memory was poorer at baseline, and a relative boost on Sulpiride in older better performers, compared to the young. This differed from generalized effects of both agents on task specificity in the encoding phase. The results demonstrate a link between aging, dopaminergic function and dedifferentiation in the hippocampus.

Keywords: Aging; Dedifferentiation; Dopamine; Episodic memory; Hippocampus; Prefrontal cortex.

Graphical abstract

Fig. 1

Paradigm design. Illustrates the mini-block structure of the study and test phases of the task. Note that not all mini-blocks are shown. See Experimental design and task for details.

Fig. 2

Age-related differences in memory specificity (Placebo session). ROIs are overlaid on the T1 MNI template from MRIcron (http://www.mccauslandcenter.sc.edu/mricro/mricron/; sections at x = 30, y = 18, z = 12). A. Plots show accuracy of the ridge regression for predicting the task at retrieval when trained to discriminate the tasks at encoding (chance = 0). Mean accuracy across feature set sizes is shown for each age group. B. Plots show the mean correlation distance metric between encoding and retrieval (within-task correlation–between-task correlation). Error bars represent the within-group standard error of the mean. See Methods for details of measures and Results for details of analyses.

Fig. 3

Dopaminergic modulation of memory specificity in hippocampus assessed using ridge regression. A. Scatter plots show the relation between memory specificity (y-axis) and baseline individual memory performance — (x-axis) in young and older age groups in the 3 drug conditions. Baseline individual memory performance is indexed by Pr on Placebo. Best fit regression lines of memory specificity to Baseline Pr within each age group and drug condition are also shown (note that although raw Pr values are given here, ANCOVA analyses used within-group mean corrected Pr values; see Results). B. The bar graph shows mean memory specificity for each age group and drug condition. Error bars represent the within-group standard errors.