Four groups of well-functioning senior citizens (6 males and 18 females per group) over the age of 65 were presented with a series of discrimination-learning problems. All were provided four pretraining problems, appropriate to their individual condition, and encouraging self-pacing. Two levels of problem complexity (four-dimensional vs. eight-dimensional problems) were crossed with two levels of a memory-aid variable (limited memory displays provided vs. none provided). Solution on four-dimensional problems was superior to that on eight-dimensional problems; memory aids significantly aided performance on eight-dimensional problems. Also, performance on four-dimensional problems deteriorated over the series of experimental problems while solution rates for eight-dimensional problems remained stable. The improvements with memory aids suggest that performance on highly complex problems may reflect deficiencies in component processes, notably memory. The deterioration of performance on problems of only moderate complexity may well reflect boredom. Both factors have practical implications which may enable the elderly to utilize more of their cognitive capabilities. (Author)