Background: In previous studies, the authors found that patients with spont
aneous confabulation differ from those with nonconfabulating amnesia by 1)
temporal context confusion (TCC) in memory based on an inability to suppres
s intrusions of currently irrelevant memory traces into ongoing thinking; a
nd 2) lesions involving the orbitofrontal cortex, basal forebrain, or amygd
ala and perirhinal cortex. Objectives: To study the long-term clinical cour
se of spontaneous confabulations, determine whether TCC in memory also para
llels the clinical course of spontaneous confabulations, and study the impa
ct of lesion site on clinical course. Methods: Eight patients with spontane
ous confabulation were re-examined 18 months after onset. Tests of memory a
nd executive functioning and measurement of TCC in memory were again applie
d. MRI according to a standard protocol was performed to determine areas of
permanent damage. Results: Seven patients eventually stopped confabulating
. TCC, but not common memory or executive tests, precisely paralleled the c
ourse of spontaneous confabulations. Patients with isolated, less extensive
, orbitofrontal lesions stopped confabulating first and had the best neurop
sychological outcome. Patients with basal forebrain lesions continued to co
nfabulate for several months and remained amnesic. One patient with extensi
ve orbitofrontal damage and perirhinal cortex damage continues to confabula
te after more than 3 years, continuing to confuse memory traces. Conclusion
s: Temporal context confusion in memory is not only the sole feature reliab
ly separating patients with spontaneous confabulation from those with nonco
nfabulating amnesia in the acute stage, it is also the only feature that pr
ecisely parallels the clinical course of spontaneous confabulations. Most p
atients eventually stop confabulating but duration of confabulations depend
s on the lesion site.