Mood Dependence 77Or is it? The question was raised in a review paper by
Blaney (1986, p. 237), who noted that
Weingartner et al.’s results—indicating that subjects experienc-
ing strong mood shifts were better able to regenerate associations
first generated in same as opposed to different moods—could be
seen as reflecting either mood congruence or [mood] state de-
pendence. That is, the enhanced ability of subjects to recall what
they had generated when last in a given mood was (a) because
what was congruent with that mood at first exposure was still
congruent with it at subsequent exposure, or (b) because return to
that mood helped remind subjects of the material they were
thinking of when last in that mood, irrespective of content.A study by Eich et al. (1997) sought both to resolve this
ambiguity and to investigate the impact of clinical mood
shifts on the performance of several different tasks. Partici-
pants were 10 patients with rapid-cycling bipolar disorder,
diagnosed according to DSM-IVcriteria (American Psychi-
atric Association, 1994).
Every patient was seen on at least four separate occasions,
the odd-numbered occasions serving as encoding sessions
and the even-numbered occasions representing retrieval ses-
sions. Although the interval separating successive encoding
and retrieval sessions varied from 2 to 7 days between pa-
tients, the interval remained constant within a given patient.
Superimposed on these sessions was a two-by-two design:
mood at encoding—manic or hypomanic (M) versus de-
pressed (D)—crossed with the same two moods at retrieval.
The original plan was to vary these factors within subjects, so
that every patient would participate in all four combinations
of encoding and retrieval moods (viz. M/M, M/D, D/M, and
D/D). This plan proved unworkable, however, as several pa-
tients quit the study prematurely for various reasons (e.g.,
they started a new regimen of drug therapy or they stopped
cycling between moods). Of the 10 patients who took part in
the study, 4 completed all four encoding and retrieval condi-
tions, 3 completed three conditions, and 3 completed two
conditions; the order of completion varied unsystematically
from one patient to the next, the determining factors being
which mood a patient was in when testing began and how
rapidly the patient cycled from one state to the other.
During each encoding session the patients undertook a
series of three tasks, summarized in subsequent paragraphs.
Although the tasks remained constant from one encoding
session to the next, the materials used in these tasks were sys-
tematically varied. The same applied to the tasks and materi-
als involved in the retrieval session, which will be described
shortly.
The first encoding task was autobiographical-event gen-
eration. Paralleling the procedures described earlier, the
patients recollected a maximum of 10 specific events, from
any time in the personal past, that were called to mind by
neutral-noun probes. After recounting the gist of a given
experience (e.g., what happened, who was involved, etc.),
patients categorized the event in terms of its original affective
valence.
The materials for the second encoding task, inkblot rating,
consisted of four Rorschach-like inkblots, printed on large
index cards. Patients viewed each pattern for a few seconds
and then rated its aesthetic appeal.
The final encoding task was letter-association production.
Patients were asked to name aloud 20 words beginning with
one letter of the alphabet (e.g., E) and 20 words beginning
with a different letter (e.g., S).
As was the case at encoding, several different tasks were
administered during each retrieval session. One of these
tasks, autobiographical-event recall, is known to show
strong mood-dependent effects with experimentally induced
moods, and it adhered to the procedures described earlier.
In a second task, inkblot recognition,patients were shown
four sets of six inkblots each. Within each set, one pattern
was an inkblot that the patients had seen during the immedi-
ately preceding encoding session, and the other five were per-
ceptually similar lures. Patients were asked to select the old
(previously viewed) pattern and to rate their confidence in
their recognition decision on a scale ranging from 0 (guess-
ing) to 3 (certain).
On first impression, this task seems ill advised because
several studies (cited earlier) have already sought, without
success, to demonstrate mood-dependent recognition. It is
important to note, however, that most of these studies (a) in-
volved experimentally induced moods (typically happiness vs.
sadness) in normal subjects, and (b) investigated recognition
memory for materials (usually common, unrelated nouns) that
are familiar, simple, meaningful, and unemotional.
Neither these moods nor these materials may be con-
ducive to the occurrence of mood-dependent recognition, the
former because they may be too mild to have much of an im-
pact (see Bower, 1992; Eich, 1995a), the latter because they
allow little latitude for different encodings in different moods
(see Bower & Cohen, 1982; Bower & Mayer, 1989). If so,
then the present study may have stood a better chance than
most at detecting mood-dependent recognition, given that it
(a) involved moods (viz. mania or hypomania vs. depression)
that can reasonably be considered strong, and (b) investigated
recognition memory for novel, complex, and highly abstract
stimuli (viz. Rorschach-like inkblots) that are likely to be
subject to emotional biases at encoding.
The last retrieval task, letter-association retention,was
designed with a view to clarifying the results reported by