Parallel patterns of cognitive aging in marmosets and macaques

Abstract

As humans age, some experience cognitive impairment while others do not. When impairment does occur, it is not expressed uniformly across cognitive domains and varies in severity across individuals. Translationally relevant model systems are critical for understanding the neurobiological drivers of this variability, which is essential to uncovering the mechanisms underlying the brain's susceptibility to the effects of aging. As such, non-human primates are particularly important due to shared behavioral, neuroanatomical, and age-related neuropathological features with humans. For many decades, macaque monkeys have served as the primary non-human primate model for studying the neurobiology of cognitive aging. More recently, the common marmoset has emerged as an advantageous model for this work due to its short lifespan that facilitates longitudinal studies. Despite their growing popularity as a model, whether marmosets exhibit patterns of age-related cognitive impairment comparable to those observed in macaques and humans remains unexplored. To address this major limitation for the development and evaluation of the marmoset as a model of cognitive aging, we directly compared working memory ability as a function of age in macaques and marmosets on the identical working memory task. Our results demonstrate that marmosets and macaques exhibit remarkably similar age-related working memory deficits, highlighting the value of the marmoset as a model for cognitive aging research within the neuroscience community.

Keywords: aging; cognitive impairment; comparative cognition; macaque; marmoset; monkey; non-human primate; working memory.

Figure 1.

Age-dependent impairment in performance on the DRST in macaques and marmosets. A) Depiction of a single DRST trial. B) Marmoset and C) macaque individual learning curves. Each line denotes an individual animal, with color indicating the age during testing. The dashed black line represents chance level performance. Correlations show that increasing macaque (red) and marmoset (blue) age is associated with D) more trials needed to perform above chance in the Novice Phase, E) reduced maximum learning rates in the Learning Phase, and F) smaller working memory capacity. When averaging across ages, no interspecies differences were observed in G) trials to above-chance performance, H) maximum learning rates, or I) working memory capacity. Each circle in D-F represents one individual. Ages in D-F correspond to the age at the time of assessment. Black circles in boxplots in G-I represent outliers.

Figure 2.

Error patterns related to age and trial difficulty level. A) Increased age is correlated with committing a larger number of errors before reaching the performance criterion in the DNMS section of the DRST for both macaques (red) and marmosets (blue). B) There were no significant species-specific differences in errors to reach the criterion. Similar patterns were identified in trials to criterion with C) increased age associated with requiring more trials needed to reach criterion. D) There were no significant interspecies differences in the number of trials to criterion. E) Reduction in perseverative errors across Novice, Learner, and Expert Phases in both species. F) Concurrent increase in primacy errors observed through these Phases for both macaques and marmosets. G) There were significant associations between increasing age and more trials to transition from predominantly perseverative to predominantly primacy errors for macaques (red) and marmosets (blue). H) During the Expert Phase, macaques more frequently misidentified remote (higher n-back) stimuli as novel compared to recent stimuli (lower n-back), suggesting retroactive interference. I) Marmosets exhibit a similar pattern during the Expert Phase, also suggesting vulnerability to retroactive interference; mean ± SEM, *p < 0.05.

Figure 3.

Choice latencies change as a function of DRST Phase and Trial Difficulty Level. Changes in A) correct choice latencies and B) incorrect choice latencies across the Novice, Learner, and Expert Phases for macaques (red) and marmosets (blue). Latencies decreased with increased task experience. In the Expert Phase, significant positive Spearman’s correlations were observed between trial difficulty level and C) correct choice latencies and D) incorrect choice latencies, reflecting increased cognitive load on more challenging portions of trials. mean ± SEM, *p < 0.05.

Figure 4.

Delay-related effects on DRST performance. Marmosets (blue) show significant delay-dependent decreased DRST performance, whereas macaques (red) do not. Also, macaques have significantly higher performance than marmosets at delays longer than 2 seconds. These results are seen on several measures of performance including A) average Final Span Length, B) accuracy on the DNMS (TDL2) portion of the DRST, and C) accuracy on TDL3 trials. D) On TDL3 trials, marmosets’ perseverative errors increased in a delay-dependent manner, whereas macaque perseverative errors remained consistent across varying delays. E) Marmosets’ primacy error rate showed a corresponding delay-dependent decrease, and macaque primacy errors remained consistent across the varying delays. Lightly shaded lines in A-C depict individual animal performance as a function of delay. Bold colored lines in A-C depict species average performance as a function of delay. mean ± SEM, *p < 0.05.