This experiment tests the hypothesis that the cholinergic nucleus basalis magnocellularis (NBM) is necessary for complex or configural association learning, but not elemental or simple association learning. Male Long-Evans rats with bilateral 192 IgG-saporin lesions of the NBM (n = 12) and sham-operated controls (n = 8) were tested in the transverse patterning problem, which provides a test of both simple and configural association learning. Rats were trained in phases to concurrently solve first one, then two, and finally three different visual discriminations; Problem 1 (A+ vs B- sign) and Problem 2 (B+ vs C-) could be solved using simple associations, whereas solving Problem 3 (C+ vs A-) required the ability to form configural associations. Consistent with our hypothesis, the NBM lesion group solved the simple discriminations in Problems 1 and 2 but showed impaired configural association learning in Problem 3. Additionally, when Problem 2 was introduced, previously high levels of performance on Problem 1 suffered more in the NBM lesion group than in the control group; this finding suggests an impairment in the ability of animals with NBM lesions to divide attention among multiple stimuli or to shift between strategies for solving different problems. Results support our argument that the NBM is critically involved in the acquisition of associative problems requiring a configural solution but not in problems that can be solved using only simple associations. The observed impairments in configural association learning and the apparent loss of cognitive flexibility or capacity are interpreted as reflecting specific attentional impairments resulting from NBM damage.
Copyright 2002 Elsevier Science (USA).