Beginning with
pioneering studies of Sir Frederic Bartlett, psychologists
have been aware that memory is a constructive process
that is sometimes prone to distortions and illusions.
In contrast, until recent years neuropsychologists and
neuroscientists interested in brain substrates of memory
have paid little attention to illusions and distortions
that illuminate constructive remembering. This presentation
focuses on recent research concerning illusory memories
in amnesic patients and older adults, and concludes by
considering relevant evidence from neuroimaging studies.
To investigate
false recognition in amnesic patients, Schacter, Verfaellie,
& Pradere (J.Mem.&Lang., 1996) used the
Deese paradigm, recently revived by Roediger and McDermott
(JEP:LMC, 1995). After studying such associates
as candy, sour, sugar, bitter and other related
words, people frequently falsely recognize the nonpresented
associate sweet. We found that amnesic patients
showed -- as expected -- reduced levels of veridical recognition
memory compared to a control group, making fewer hits
to studied words and more false alarms to unrelated nonstudied
words than did controls. More importantly, amnesic patients
made fewer false alarms than controls to nonstudied associates
such as sweet.
This latter
finding suggests that false recognition of nonstudied
associates depends on retention of semantic information
that also supports veridical recognition of presented
words. More recent findings indicate that amnesic patients
also show reduced levels of false recognition when tested
with words that are perceptually similar to previously
studied words (Schacter, Verfaellie, & Anes, Neuropsychology,
in press). Taken together, the two sets of studies suggest
that medial temporal/diencephalic structures that are
damaged in amnesic patients play a role in the encoding
and/or retrieval of gist or general similarity information
that drives false recognition.
In contrast
to these findings, we have found that older adults, despite
showing less accurate veridical recognition than younger
adults, are relatively more susceptible to false recognition
of semantic associates (Norman & Schacter, Mem.&Cognit.,
in press). We have also documented an age-related increase
in susceptibility to false recognition of nonpresented
pictures that are perceptually/ conceptually similar to
previously studied pictures (Koutstaal & Schacter,
submitted, 1997).
Recent neuroimaging
data provides further insight into these findings. Using
both PET and fMRI, we have found robust activation of
anterior prefrontal cortex during false recognition, with
some evidence relating these activations to post-retrieval
monitoring activities (Schacter, Reiman, et al., Neuron,
1996; Schacter, Buckner et al.,Proc. Cog. Neurosci.
Soc., 1997). Other PET studies indicate that older
adults sometimes fail to show normal activation of anterior
prefrontal regions (Schacter, Savage et al., NeuroReport,
1996). Thus, age-related increases in susceptibility to
false recognition may depend in part on faulty monitoring
processes that depend on prefrontal regions. In contrast,
PET data indicate that medial temporal regions are related
to successful conscious recollection in both younger and
older adults (Nyberg et al., Nature, 1996; Schacter, Savage
et al., 1996). Combined with the data showing reduced
false recognition on amnesic patients, these results imply
that medial temporal activity is important for both veridical
and illusory recollections.
