A computational model of cognitive control deficits in healthy aging

This proposal tests specific predictions regarding the relationship between cognitive and neurobiological impairments observed in healthy aging. These predictions were derived from a neural-network computational model which postulates specific functional roles for the dopamine (DA) neurotransmitter system and dorsolateral prefrontal cortex (DL-PFC) in cognitive processes related to the control of thoughts and behavior. A growing literature has suggested that healthy aging is associated with disturbances in both DA and PFC function, and with impairments in tasks requiring cognitive control. However, despite accumulating evidence about these age-related cognitive and neurobiological changes, there is still little understanding of whether or how they are associated. Based on work with our computational model, we propose three hypotheses regarding the mechanisms by which the functional interaction of the DA system in DL-PFC influences specific cognitive processes, and how these may breakdown in healthy aging. First, we suggest that DL-PFC subserves the representation and maintenance of task-relevant context. Second, we suggest that the DA system serves to regulate the flow of information into DL-PFC. Third, we suggest that as a result of DA disturbances in DL-PFC among older adults, context information represented in DL-PFC cannot be actively sustained and will instead decay over time. These hypotheses lead to specific, detailed, and often counter-intuitive predictions regarding age-related changes in both brain activity and behavior during performance of cognitive control tasks that rely on the representation and maintenance of context. We propose to investigate these hypotheses in both a behavioral and functional neuroimaging study (using functional magnetic resonance imaging, or fMRI, methods) comparing healthy older and younger adults. Both studies will use a single cognitive task designed and validated to be sensitive to the processing of context. To our knowledge, the studies we propose represent the first use of computational models of cognition to make explicit and detailed predictions regarding regional brain activation and behavioral performance in healthy aging. As such, this project promises to generate valuable new empirical data regarding the neurobiological underpinnings of age-related cognitive decline, to significantly advance our theoretical understanding of this process, and to establish the groundwork for developing powerful new behavioral and neuroimaging measures that may eventually have direct clinical applicability.

NIA 1 R03 AG18138-01

[CCP Home Page] [CCP Research] [Aging]