Using the significant clusters in Table 3 as regions of interest, β weights were extracted
for specific AMs vs riser baseline and example generation vs riser baseline. ACC indicates anterior
cingulate cortex; BA, Brodmann area; HC, healthy control subjects; HR, individuals at high familial
risk for developing MDD; L, left; MDD, major depressive disorder; and R, right.
eMethods. Supplemental Methods
eFigure. tSNR Maps for Each Participant Group
eTable 1. Group Motion Statistics
eTable 2. Difference Scores for Fatigue Measures for Each Participant Group
eTable 3. Percentage of Specific and Positive AMs Recalled for Each Group and Task Half
eTable 4. Clinical Characteristics of the MDD Participants
eTable 5. Increases in the BOLD Response for the Different Contrasts Across all Diagnostic
Customize your JAMA Network experience by selecting one or more topics from the list below.
Young KD, Bellgowan PSF, Bodurka J, Drevets WC. Behavioral and Neurophysiological Correlates of Autobiographical Memory Deficits in
Patients With Depression and Individuals at High Risk for Depression. JAMA Psychiatry. 2013;70(7):698–708. doi:10.1001/jamapsychiatry.2013.1189
Copyright 2013 American Medical Association. All Rights Reserved. Applicable
FARS/DFARS Restrictions Apply to Government Use.
Individuals with major depressive disorder (MDD) compared with healthy control subjects (HCs)
consistently recall fewer specific and more categorical autobiographical memories (AMs). This effect
is most pronounced for positive AMs and persists into remission.
To determine whether individuals at high familial risk for developing MDD (HR group) also show an
AM overgenerality bias and to use functional magnetic resonance imaging to assess differences in
functional correlates of AM recall across HR, currently depressed MDD, and HC groups.
While recalling AMs in response to emotionally valenced cue words, study participants underwent
functional magnetic resonance imaging. Control tasks involved generating examples from a given
category and counting the number of risers in a letter string.
Testing was conducted at the Laureate Institute for Brain Research, Tulsa, Oklahoma.
Participants included 16 unmedicated patients with MDD, 16 HR participants, and 16 HCs.
Main Outcomes and Measures
Percentage of specific and categorical AMs recalled and brain regions in which hemodynamic
activity changed during specific and positive AM recall compared with example generation.
Both the MDD and HR groups generated fewer specific, more categorical, and fewer positive AMs
than the HC group (P ≤ .02 for all). During specific AM recall
compared with example generation, neuroimaging results showed between-group differences in the left
cuneus (Talairach space coordinates x, y, z = −7, −71, 18;
F = 7.55), right medial frontal cortex (x, y, z = 7, 59,
12; F = 8.53), right frontal operculum (x, y, z = 23, 23,
12; F = 8.25), and right and left pregenual anterior cingulate cortex
(x, y, z = 9, 37, 10 and x, y, z = −3, 43, 6;
F = 6.84 and F = 7.13, respectively).
Conclusions and Relevance
Autobiographic memory deficits exist in HR individuals, suggesting that these impairments
constitute traitlike abnormalities in MDD. We also found distinct patterns of hemodynamic activity
for each group as they recalled specific AMs. Specifically, the HR and MDD groups showed
differential hemodynamic activity from HCs in medial prefrontal and occipital regions, suggesting
that these groups may use different self-referential focus during successful retrieval of specific
The tendency of individuals with major depressive disorder (MDD) to recall fewer specific and
more categorical autobiographical memories (AMs) compared with healthy control subjects (HCs) is a
consistently replicated finding.1,2 Despite
remission of depressive symptoms,3-5
this abnormality persists and seems unrelated to symptom severity,6,7 suggesting that a pathogenetic construct beyond disturbed mood
underlies overgeneral AM in MDD. These data further suggest the hypothesis that this deficit
reflects a trait marker of depression. Moreover, overgeneral AM may have an antecedent, facilitatory
role in the development of MDD. Researchers have shown that an overgeneral AM response style coupled
with a major environmental stressor increases the likelihood of depressive symptoms at follow-up
assessment.7-9 In addition, studies
focusing on adolescents have found AM overgenerality in adolescents deemed at risk for developing
MDD by virtue of high neuroticism scores or a history of MDD10 and noted that overgeneral AM at age 11 years predicted depressive symptoms at age 12 years in
girls.11 These studies either assessed populations already
experiencing significant pathologic conditions or those who are healthy. To date, no study has
examined whether individuals with a hereditary risk for depression might also demonstrate
overgeneral AM retrieval. Therefore, one objective of the present study was to determine whether
healthy participants at high familial risk for developing MDD (HR group) based on having a
first-degree family relative with MDD also show an AM overgenerality bias.
A second objective of the present study was to examine whether the functional correlates of AM
recall would differ across participants drawn from HR, MDD, and HC samples. Delineating the
functional correlates of AM deficits in HR and MDD samples holds the potential to identify an
endophenotype of MDD.
A substantial amount of research has examined depressive deficits and biases during memory
encoding. For example, MDD samples compared with HC samples show increased amygdala activity during
encoding of negative pictures12 and words,13 as well as increased anterior cingulate cortex (ACC) and
decreased hippocampal activity when encoding positive words.13 Furthermore, increased prefrontal and medial temporal activity is evident during encoding of
positive words in individuals with remitted MDD,14 while
increased hemodynamic activity within the ventrolateral prefrontal cortex, hippocampus, and amygdala
is evident during encoding of both positive and negative stimuli in individuals at risk for
developing MDD.15 These results indicate that individuals
with MDD show altered activity in the network recruited for encoding affective stimuli and that this
abnormality persists into remission and extends to individuals at risk for developing MDD.
Although neurobiological differences during encoding suggest differences during recall, no study
to date has used functional magnetic resonance (fMR) imaging to examine AM recall in HR populations,
and only 3 studies16-18 have used fMR
imaging to examine AM in MDD. Two studies16,17
used a recognition paradigm in which participants were shown AM prompts based on preimaging
interviews. These studies found that activity within the ventrolateral prefrontal cortex decreased
in MDD vs HC samples during negative AM recognition16,17 and that activity within the ventrolateral prefrontal cortex
increased in MDD vs HC samples during positive recognition. Those results should be interpreted with
caution because a recognition paradigm may change the properties of the memories.19,20 To date, only one study18 has used neuroimaging to examine AM recall in MDD and found decreased activity
in the ACC, hippocampus, anterior insula, and parahippocampal gyrus in MDD vs HC samples as they
engaged in AM recall.
We aimed to assess differences in hemodynamic correlates of AM recall across HR, currently
depressed MDD, and HC samples. We predicted that the HR sample would behaviorally show the same
pattern of AM deficits as the MDD sample (fewer specific and more categorical AMs), while the HR
sample and the MDD sample would neurophysiologically show decreased hemodynamic activity in temporal
and prefrontal components of the AM network compared with HCs.
Medically healthy, right-handed individuals aged 18 to 55 years were evaluated for their
eligibility to be included in 1 of the following 3 groups: psychiatrically HCs, psychiatrically
healthy participants having a first-degree relative with MDD, and unmedicated individuals with MDD
in a current major depressive episode according to the DSM-IV-TR.21 Recruitment continued until each group had 16 participants who
underwent imaging and satisfied exclusion criteria for movement during imaging (described in detail
in the next subsection). The total sample sizes entered to reach this recruitment goal were 18 MDD,
17 HC, and 17 HR. Volunteers, recruited from the community via advertisements, underwent medical and
psychiatric screening evaluations at the Laureate Institute for Brain Research, which included the
Structural Clinical Interview for DSM-IV-TR Axis I Disorders.22 Family history was established via the Family Interview for Genetic
Exclusion criteria included psychosis, current pregnancy, serious suicidal ideation, general MR
imaging exclusions, major medical or neurological disorders, and exposure to any medication likely
to influence cerebral function or blood flow within 3 weeks, as well as meeting
DSM-IV criteria for drug or alcohol abuse within the previous year or for alcohol
or drug dependence (except nicotine) within the lifetime. Additional exclusion criteria applied to
the HC and HR groups were current or past Axis I psychiatric conditions and a history of
psychotropic medication use. After receiving a complete explanation of the study procedures, all
participants provided written informed consent as approved by the Western Institutional Review
Board. Participants received financial compensation for their participation. Participant data were
dropped from analysis if absolute movement during fMR imaging exceeded 3 mm in any direction.
Intelligence testing was performed using the 2-subtest version of the Wechsler Abbreviated Scale
of Intelligence.24 Anxiety and depressive symptoms were
rated on imaging day using the State-Trait Anxiety Inventory,25 the 21-item Hamilton Scale for Depression,26 the Montgomery-Åsberg Depression Rating Scale,27 the Snaith-Hamilton Pleasure Scale,28
and the Profile of Mood States.29
Functional MR imaging was performed using a 3-T MR imaging system (GE 750; GE Healthcare) and an
8-channel receiver coil array. Gradient-recalled echoplanar imaging with sensitivity (SENSE) was
used for fMR imaging with the following parameters: 40 × 3-mm sections acquired
axially; repetition time, 2000 milliseconds; echo time, 25 milliseconds; SENSE acceleration, 2; flip
angle, 90°; 96 × 96-pixel matrix; field of view, 24 cm; and voxel size,
3 × 2.5 × 2.5 mm3. A total of 211 echoplanar images
were acquired in each of ten 7-minute runs during the AM task. The first 4 images of each run were
discarded to allow for steady-state tissue magnetization. Also acquired for coregistration with the
echoplanar imaging series were high-resolution T1-weighted anatomical MR images with the following
parameters: 120 × 1.2-mm sections acquired axially; repetition time, 5
milliseconds; echo time, 1.93 milliseconds; flip angle, 8°; 256 × 256-pixel
matrix; field of view, 24 cm; and in-plane resolution, 0.94 mm2.
Before imaging, participants were instructed as to the different types of AMs that they might
retrieve (according to conventional definitions used in the literature)2,30,31 and were informed that their goal was to recall a
specific AM, defined as a memory for an event that occurred at an identified place and lasted up to
1 day. A categorical memory referred to a category of events containing several episodes without
reference to one specific event (eg, all examinations failed without reference to one particular
examination). An extended memory referred to an extended period without reference to a specific
event within the time frame (eg, a weeklong vacation). A semantic memory was defined as a statement
of fact without an associated event (eg, “I’ve never been dancing.”). Participants
were provided these definitions and examples verbally before entering the imaging system and
immediately before the start of each fMR imaging run.
A computerized version of the AM task32 was developed for
use during fMR imaging. Using available software (E-Prime; Psychology Software Tools Inc),
participants were presented with 60 words, including 20 positive (eg, success), 20
negative (eg, danger), and 20 neutral (eg, journal).33 During fMR imaging, participants were presented with a cue word
for 12 seconds and were instructed to recall a past experience. Following the cue, participants
rated the retrieved memory on the specificity (specific, categorical, extended, semantic, repeat, or
no memory) and the valence (negative, somewhat negative, neutral, somewhat positive, positive, or no
memory). Participants had 10 seconds to assign each rating by scrolling to the relevant option and
making their selection using a scroll wheel (Current Designs Inc).
The AM recall condition was compared with a semantic example generation condition to control for
abstract or general knowledge retrieval.34 Participants were
presented with an example generation cue word for 12 seconds and were instructed to think of at
least 7 examples from the presented category. Selected from the same word pool as the memory cue
words, 10 positive categories (eg, flowers), 10 negative categories (eg, villains), and 10 neutral
categories (eg, instruments) were presented. Following an example generation cue word, participants
rated the ease with which they were able to generate examples (very easy, easy, somewhat easy,
somewhat difficult, difficult, or very difficult) and the number of examples they generated (0, 1-2,
3-4, 5-6, 7, or ≥8). Participants had 10 seconds to select each rating.
Following the presentation of each cue and each set of ratings, participants engaged in a riser
detection task as a control for visual input and attention. All example generation and memory cue
words were scrambled into lowercase nonword letter strings, and participants were instructed to
count the number of risers in the string, defined as a letter with a part rising above the tops of
the other letters (eg, “gulmnh” has the risers l and
h). The presentation of each letter string was jittered with a mean presentation
time of 6 seconds. For a randomly selected one-half of these strings, a 2-second period followed
during which participants selected whether the number of risers in the previous string was even or
The order of memory and example cue word presentations was pseudorandomized, with restrictions on
order presentation to prevent sequential presentations of a particular valence. Within each of 10
runs, participants were presented with 6 memory cue words, 3 example generation cue words, and 18
riser letter strings in the following order: cue word, riser (one-half, followed by even or odd
question), ratings, and riser (one-half, followed by even or odd question). Two computers
time-linked to the image acquisition of the MR imaging system controlled stimulus presentation and
behavioral response collection. Participants observed the stimuli using a mirror system attached to
the head coil.
Following the imaging, the experimenter (K.D.Y.) presented participants with all the AM cue words
again in the same order as during imaging. Participants were asked to describe the memory for the
experimenter to corroborate participants’ specificity ratings. The experimenter was blind to
diagnosis at the time of rating. In addition to the definitions provided, a memory was categorized
as “can’t recall” if the participant was unable to recall the memory retrieved
during fMR imaging or the memory provided did not match the participant’s specificity rating
(indicating that he or she had not reported the same memory as during the imaging). Participants
also rated each memory on the following 3 additional properties: arousal (5-point scale ranging from
very low to very high), vividness (5-point scale ranging from not at all vivid to perfectly clear
and vivid), and age when the memory occurred (childhood, adolescence, >18 years but >1 year
before imaging, 6 months to >1 year before imaging, or within 6 months).
Behavioral data were analyzed using available software (SYSTAT 13; Systat Software Inc).
Potential group differences in age, IQ, symptom ratings, performance on the riser detection and
example generation control tasks, and percentage of memories recalled at each specificity level
(specific, categorical, extended, semantic, repeat, no memory, and can’t recall after imaging)
were assessed using a 1-way analysis of variance (ANOVA) (entering diagnosis as the independent
variable and mood, demographic, and specificity variables as the dependent variables), as well as
follow-up independent t tests.
Our a priori hypothesis was that HR participants would exhibit the same AM defects as previously
reported in individuals with MDD using a similar paradigm.18
We expected that both the HR and MDD groups compared with HCs would have fewer positive AMs, fewer
highly arousing AMs, fewer specific and more categorical AMs, and fewer AMs from the most recent
period examined. A priori hypothesis testing focused on the properties of the specific and
categorical memories because too few exemplars of other types of AMs were retrieved to allow
sufficient power to detect group differences. To increase power, the following variables were
collapsed: somewhat positive and positive were considered together to create a positive variable,
somewhat negative and negative were considered together to create a negative variable, low and very
low (arousal and vividness) were considered together to generate low arousal and vividness
variables, and high and very high (arousal and vividness) were considered together to generate high
arousal and vividness variables,
Four repeated-measures ANOVAs were performed to examine group differences in the properties of
the recalled AMs. In each case, the between-group variable was diagnosis, and the repeated measure
was type (specific or categorical). The other repeated measure entered for each ANOVA was arousal
(low, medium, or high), vividness (low, medium, or high), valence (positive, negative, or neutral),
or age when the memory occurred (childhood, adolescence, >18 years but >1 year before imaging,
6 months to >1 year before imaging, or within 6 months) for the dependent variable percentage of
memories recalled. Follow-up independent samples t tests were conducted for
significant results within the ANOVAs. The threshold criterion for significance was set at
P < .05, and post hoc tests were corrected for multiple comparisons
Because participants underwent 90 minutes of fMR imaging, we performed additional behavioral
analyses that ruled out fatigue as a significant factor contributing to the observed results.
Further details are given in the eMethods in Supplement.
Image preprocessing and analysis were performed using available software (AFNI; http://afni.nimh.nih.gov/afni) and consisted of within-group realignment, section
acquisition time correction, coregistration between anatomical and functional images, spatial
normalization to the stereotaxic array by Talairach and Tournoux,35 and smoothing using a 4-mm full-width at half maximum gaussian kernel. Using
AFNI 3dDeconvolve for each participant, the evoked hemodynamic response to each event type was
modeled as a boxcar function convolved with a synthetic hemodynamic response function. Regressors
modeling task and motion parameters were used in the model. The main effects of interest were cue
word presentations that prompted specific AM recall and example generation. In addition to
regressors modeling main effects, each design matrix included regressors modeling rating selection,
even or odd riser question presentations, cue presentation in which nonspecific memories were
recalled, and cue presentation in which the retrieved memory was not recalled. The nonword letter
strings used as stimuli for the riser detection task were modeled as the baseline.
At the group level, AFNI 3dANOVA was used to identify regional differences in the blood oxygen
level–dependent (BOLD) signal between groups for specific memories vs example generation and
example generation vs riser baseline. Because the only behavioral difference found between groups
was for the valence model, additional contrasts of positive memories (specific and categorical) vs
positive example generation and negative memories (specific and categorical) vs negative example
generation also were performed. An additional test was performed to compare the BOLD response for
each contrast for all 3 groups combined. Because the mean number of neutral memories recalled was
fewer than 10 and because the number of categorical AMs recalled by HCs was low, we lacked
statistical sensitivity to justify group comparisons involving these parameters. The significance
criterion for detecting differences was set at corrected P < .05,
determined using AFNI 3dClustSim (cluster size >30 voxels, thresholded at voxel
P < .005). A final analysis indicated that neither the temporal
signal to noise ratio nor motion statistics differed between groups (eMethods in Supplement).
Table 1 gives the demographic and clinical
characteristics for each group (eTable 4 in Supplement summarizes additional clinical characteristics of the
individuals with MDD). A 1-way ANOVA with diagnosis as the independent variable showed no
significant difference across groups in age or IQ
(F2,45 < 1.92, P > .20
for all) but differed in the mean results on the Profile of Mood States, State-Trait Anxiety
Inventory, Hamilton Scale for Depression, Snaith-Hamilton Pleasure Scale, and Montgomery-Åsberg
Depression Rating Scale (F2,45 > 19.5,
P < .001 for all). The MDD group had higher scores on all ratings
compared with the HC group (t30 > 5.04,
P < .001 for all) and the HR group
(t30 > 4.50, P < .001 for
all), while the HC group and the HR group did not differ on any rating
(t30 < 1.06, P > .60 for
The groups did not differ in performance on the riser baseline task, the ratings of ease at
generating the examples, or the number of examples generated for the differently valenced categories
on the example generation task (F2,45 < 2.27,
P > .11 for all). These results are summarized in Table 1.
The percentage of memories recalled for each memory type (Table 1) were assessed using ANOVA with the independent variable diagnosis.
The groups differed in the percentage of memories coded as specific
(F2,45 = 11.1) and categorical
(F2,45 = 16.2) (P < .001 for
both). The HC group had more specific and fewer categorical AMs than both the HR group and the MDD
group (t30 > 3.75,
P < .001 for all), while the HR group and the MDD group did not
differ significantly from each other (t30 < 1.59,
P > .24 for all). The groups did not differ significantly in the
percentage of memories coded as extended, semantic, no memory, or unable to recall after imaging
(F2,45 < 1.30, P > .28
We examined the properties of the specific and categorical memories by performing
repeated-measures ANOVAs with the between-group factor of diagnosis and the repeated measures of
type (specific or categorical) and either valence, arousal, vividness, or age when the memory
occurred. Table 2 gives the percentage of memories
recalled, separated by these properties.
The valence × diagnosis interaction was significant
(F2,44 = 3.23, P < .02).
Follow-up t tests revealed that the HC group had more positive memories than both
the HR group and the MDD group (t30 > 2.91,
P < .02 for all), while the HR group and the MDD group did not
differ from each other (t30 < 1.83,
P > .23 for all). The interaction with diagnosis was nonsignificant
for arousal and for vividness (F2,44 < 0.65,
P > .63 for all), indicating that the groups did not differ on these
ratings. The diagnosis × memory age × type interaction was
significant (F8,42 = 5.67,
P = .002). Follow-up t tests revealed that the MDD
group had fewer specific AMs from the most recent 6 months compared with the HC group and the HR
group (t30 > 2.58,
P < .05 for all), whereas the HR group and the HC group did not
differ from each other on this parameter (t30 = 1.03,
P = .99).
eTable 5 in Supplement
lists regions where BOLD activity increased across the entire sample (all groups combined), and
Table 3 summarizes group differences in regional
BOLD activity. Comparing example generation with riser baseline in the entire sample, BOLD activity
increased in the caudate, precentral gyrus, parahippocampus, orbitofrontal cortex, inferior temporal
gyrus, left posterior cingulate cortex, and right dorsolateral prefrontal cortex. No significant
group difference in regional BOLD activity was identified for this contrast, suggesting that
performance on the control tasks was similar across groups.
Comparing specific AM recall with example generation in the entire sample, BOLD activity
increased in the ACC, amygdala, hippocampus, left precuneus, parahippocampus, temporoparietal
cortex, medial prefrontal cortex, bilateral middle temporal gyrus, bilateral postcentral gyrus and
cerebellum, and left ventrolateral prefrontal cortex and dorsolateral prefrontal cortex. BOLD
activity differed across groups on this contrast (ie, the diagnosis × specific AM
recall × example generation interaction was significant [Figure]), such that the mean BOLD response was higher in the MDD group than
in the HC group and the HR group in the right medial frontal gyrus and bilateral pregenual ACC. The
BOLD response was higher in the HR group than in the MDD group and HC group in the left cuneus
(Brodmann area [BA] 17). Finally, in the right frontal operculum (intrasulcal lateral orbitofrontal
cortex [BA 47s]36), the BOLD activity decreased in the HR
group, did not change significantly in the HC group, and increased in the MDD group.
In the entire sample, positive AM recall vs positive example generation was associated with
increased BOLD activity in the left precuneus, right parahippocampus and orbitofrontal cortex, and
bilateral temporoparietal cortex, caudate, and cerebellum. During recall of negative AMs vs negative
example generation, the combined sample also showed increased BOLD activity in the right
orbitofrontal cortex, occipital cortex and cerebellum, parahippocampus and precuneus, left
dorsolateral prefrontal cortex, and bilateral amygdala, temporoparietal cortex, and caudate. During
positive AM recall, BOLD activity increased in the right middle temporal gyrus of the HR group but
decreased in the MDD group and the HC group. No significant group difference was identified in BOLD
response during negative AM recall vs negative example generation.
We replicated previous behavioral findings of fewer specific and more categorical AMs in MDD1 and for the first time to date demonstrate that these deficits
exist in individuals at risk for developing MDD. Despite the similarity in depression, anxiety, and
anhedonia ratings between the HR group and HC group, HR participants showed the same pattern of AM
deficits seen in the MDD group. We also observed fewer overall positive AMs in both the HR
participants and the MDD group vs the HC group, as well as differential hemodynamic response in
regions recruited during recall of positive AMs in HR participants vs the HC group. In contrast,
these groups showed no significant difference in BOLD activity during recall of negative AMs. These
data support the hypothesis that MDD is associated with impaired recall of positive information
rather than a diathesis toward recalling negative information.37 The extension of this finding to HR participants further supports the role of
alterations in positive processing in the underlying pathology of MDD.
We also replicated previous findings that individuals with MDD had fewer specific AMs from the
most recent 6-month epoch compared with HCs.18 In contrast,
HR participants generated significantly more memories from this epoch than the individuals with MDD
and did not differ significantly from HCs. It remains unclear whether the reduction in recent
memories recalled by individuals with MDD reflects an encoding deficit38 or a paucity of salient life experiences to draw on due to nonspecific factors,
such as inactivity or social isolation.39 The finding that
HR participants do not share the impairment in recent specific AMs manifest in individuals with MDD
raises questions as to whether this abnormality results from recurrent depressive episodes.
The neuroimaging data demonstrated both common and distinct functional correlates of AM recall
between groups. Across groups, BOLD activity changed in core components of the AM network,40 including the amygdala, hippocampus, parahippocampus,
temporoparietal cortex and cerebellum, and ventrolateral, dorsolateral, and medial prefrontal
cortex, as participants recalled specific AMs vs example generation. Differences in BOLD activity
between groups were evident in the ACC, occipital cortex, frontal operculum, and medial prefrontal
In the pregenual ACC, BOLD activity increased to a greater extent in the MDD group than in the HC
group or HR group. This ACC subdivision has been implicated in emotional processing associated with
both positively and negatively valenced stimuli41-44 and seems overactive at rest in
MDD.45 The MDD group also had increased BOLD activity during
specific AM recall compared with the other groups in the frontal polar cortex adjacent to the
pregenual ACC. This area is involved in self-referential processing,46 and increased activity here when watching sad film clips predicts later
rumination and depression relapse.47 The elevated BOLD
response in this area in MDD conceivably may indicate greater self-referential and emotional
processing, as well as depressive rumination. The absence of this abnormal pattern in BOLD activity
in the HR group suggests that these alterations in prefrontal cortex function in the MDD group may
relate more to pathologic emotional processing than specifically to AM recall deficits.
In the right frontal operculum, the HR group had decreased BOLD activity, while the HC group and
MDD group had increased BOLD activity when recalling specific AMs. This area is implicated in
self-focus and rumination and is active during induced sadness.48 Notably, individuals with MDD who have completed cognitive therapy do not
manifest higher BOLD activity in this region during sadness induction.49
Finally, in the occipital cortex, both the MDD group and HR group had increased activity during
specific AM recall, with the HR group showing the greatest increase in BOLD activity. Neuroimaging
studies50-52 of AM recall
demonstrated occipital activation, and this area has been implicated in visuospatial
processing.53 A recent study47 also found that activity in BA 17 correlated with tolerance of distress in both
HC and remitted MDD samples, and increased activity in this area was associated with a lower risk
for depression relapse. This may be particularly relevant to the present study because BOLD activity
was higher in the HR group than in the MDD group.
Despite showing similar behavioral differences during AM recall, the HR group and the MDD group
had distinct neuroimaging profiles. In BA 47s, the BOLD response in the MDD group seemed more
similar to that in the HC group than to that in the HR group. In other regions (BA 10 and the ACC),
BOLD activity in the MDD group differed from that in both the HC group and the HR group. Notably,
the analysis focused on specific AMs, and the properties of these memories (other than age when the
memory occurred) did not differ across groups. Distinct neuroimaging results across groups may be
attributable to different cognitive strategies used by each group to achieve a specific memory.
Conceivably, such strategies may differentially engage self-referential and ruminative processing,
with the MDD group using these processes to a greater extent than the other groups (as suggested by
increased activity in BA 10 and the ACC), while the HR group avoided ruminative processing. Lower
BOLD response in the HR group in BA 47s may reflect a protective mechanism whereby emotions are
interpreted as information associated with a memory rather than as impetus for rumination. In
addition, the ability to recall specific AMs, leading to an increase in BOLD activity of BA 17,
conceivably may aid HR individuals and patients with MDD as they encounter distressing situations.
Previous behavioral studies support this hypothesis by showing that, in remitted patients with MDD,
overgeneral AM recall is more likely manifest when interpersonally relevant content is accessed
before memory recall54 and by demonstrating that increased
exposure to self-relevant AM cues is associated with decreased AM specificity.55
Our results support the assertion that AM overgenerality constitutes a traitlike marker of
MDD.56 However, we hypothesize that this overgenerality may
only become dysfunctional and lead to pathology when coupled with increased rumination or
self-focus. Because the mean age at onset for MDD is the mid-20s57 and because the mean age in our HR group was early 30s, it is likely that our
HR sample predominantly represented a resilient group beyond the age of peak susceptibility.
Therefore, understanding how these participants differ from those with MDD may provide insight into
how to prevent MDD from developing in those at risk. Because several regions where the MDD group
manifested increased BOLD activity compared with the HC or HR group are involved in self-referential
processing, it is conceivable that focusing less on self during AM retrieval provides a protective
mechanism for HR individuals. To our knowledge, assessing whether AMs focus more on self or others
has not been characterized in previous studies of AM in MDD but may prove informative in future
One limitation of our study design is the lack of experimental control over the properties of
recalled AMs and the processes triggered that are unrelated to memory recall or depression
vulnerability. This limitation is inherent to neuroimaging studies of AM recall, which we attempted
to mitigate by carefully selecting relevant control tasks. Although some portion of the response
during memory recall would be attributable to affective processing, this type of processing
presumably occurred for both example generation and memory conditions because both cue words were
matched for valence, arousal, vividness, and frequency of use. The lack of group differences during
the example generation condition supports the conclusion that observed differences during AM recall
relate to different processes of searching for and recalling a memory.
The use of this new control task also may explain why our fMR imaging results did not replicate
some of the neuroimaging findings from a previous study,18
despite the successful replication of the previous behavioral results. The control task used in the
present study (example generation) differed from that used in the previous study (solving
subtraction problems). The strength of the present design is that, by comparing BOLD activity during
recall of specific AMs with activity during example generation, we more specifically characterized
the BOLD response to episodic, sensory, and perceptual representations of the memory.
Overgeneral AM is associated with impaired problem solving both generally58,59 and following negative mood induction,60 impaired executive control,61 and
difficulty imagining the future.62 In addition, AM deficits
interfere with the generation of adaptive responses to social interactions and stressful
circumstances.63 Illuminating the neural mechanisms
underlying AM deficits in MDD and HR samples ultimately may lead to the development of interventions
that enhance the effectiveness of cognitive-based treatments for depression. Existing cognitive
therapies target overgeneralization in patients’ beliefs and perspectives, often accomplished
by having patients keep a diary of significant events and associated feelings.64 Targeting overgeneral AM recall and shifting the focus of AMs from self to
others in cognitive therapy may prove beneficial for treating MDD and preventing those at risk from
developing the disorder.
Corresponding Author: Kymberly D. Young, PhD, Laureate
Institute for Brain Research, 6655 S Yale Ave, Tulsa, OK (email@example.com).
Submitted for Publication: April 24, 2012; final revision received November 1, 2012;
accepted November 3, 2012.
Published Online: May 15, 2013. doi:10.1001/jamapsychiatry.2013.1189.
Author Contributions: Dr Young takes
responsibility for the integrity of the data and accuracy of the data analysis. All authors had full
access to all the data in the study.
Conflict of Interest Disclosures: None
Funding/Support: Funding for this work was provided
by the Laureate Institute for Brain Research through
The William K. Warren Foundation.
Role of the Sponsors: None reported.
Create a personal account or sign in to: