An integrative theory relating cognitive aging deficits observed at the beh
avioral level with age-related deficiency in neuromodulation causing less d
istinctive cortical representations is tested in a series of neural network
simulations. Age-related attenuation of catecholaminergic function is simu
lated by lowering the mean gain of the processing unit, which subsequently
reduces responsivity and raises intra-network activation variability. Age d
ifferences in learning rate, asymptotic performance, interference susceptib
ility, complexity cost, intra- and interindividual variability, and ability
dedifferentiation can all be modeled. Together, the simulations illustrate
catecholamine's role in regulating the fidelity of neural information proc
essing and subsequent effects leading to cognitive aging deficits. (C) 2000
Elsevier Science B.V. All rights reserved.