You are on page 1of 14

J Happiness Stud (2009) 10:483496

DOI 10.1007/s10902-008-9102-9
RESEARCH PAPER

The Future is Bright? Effects of Mood on Perception


of the Future
Silvia R. Hepburn Thorsten Barnhofer J. Mark G. Williams

Published online: 4 June 2008


Springer Science+Business Media B.V. 2008

Abstract Most people believe that the future will bring them more good things than bad,
and therefore have high hopes for the future (MacLeod et al. Cogn Emot 10:6985, 1996).
However, many patients with mood disorders do not hold this positive belief about the
future. At the extreme, low expectations of positive outcomes in the future can lead to
feelings of hopelessness (OConnor et al. Psychol Health Med 5:155161, 2000). This
paper aims to extend the literature on subjective probability of future events, using a mood
induction paradigm to examine the effects of transient mood change on perceived likelihood of future events in a non-clinical community sample. Participants rated likelihood of
future events from a standardized list and from their own lives. Ratings were made in both
normal and experimentally-induced positive or negative mood. Results show that selfgenerated future events were perceived to be more likely than those from a standardized
list, and that negative mood significantly biased perceived likelihood of other-generated
future events. Participants rating standardized list events saw positive outcomes as less
likely and negative outcomes as more likely in induced negative mood than they did in
normal mood. Mood had no effect on ratings of self-generated events. Possible directions
for future research are discussed.
Keywords Future thinking  Mood induction  Subjective probability 
Hopelessness  Cognitive bias

Our attitude towards the future is influenced by our beliefs about how likely it is that good
things will happen to us relative to bad things (MacLeod and Tarbuck 1994). Pessimism
about the future is one component of Becks depressive thinking triad, along with
S. R. Hepburn  T. Barnhofer  J. M. G. Williams
University of Oxford, Oxford, UK
S. R. Hepburn (&)
Department of Psychology, Institute of Psychiatry, Kings College London, De Crespigny Park,
P.O. Box 78, London SE5 8AF, UK
e-mail: silvia.hepburn@iop.kcl.ac.uk

123

484

S. R. Hepburn et al.

negativity about the self and the world (Beck et al. 1979). Reduced positive expectations
about the future have been shown to relate to feelings of hopelessness in suicidal groups
(Conaghan and Davidson 2002; OConnor et al. 2000). This study investigates the
potential role of mood in biasing future-oriented judgment. Johnson and Tversky (1983)
have shown that manipulating peoples mood affects their perception of the risk of negative future events. This study uses a mood induction paradigm to examine the effect of
mood state on subjective probability ratings of positive and negative self-generated and
other-generated future events made by a non-depressed community sample.
The existing literature on future-oriented judgment consists of two main types of study.
The first examines the ability to generate examples of positive and negative future events
within a given time limit. Fluency for positive future events has been found to be reduced
in suicidal and depressed individuals (e.g. Conaghan and Davidson 2002; Hunter and
OConnor 2003; MacLeod et al. 1997). Future fluency studies are reviewed elsewhere
(Hepburn et al. 2006; MacLeod 1999). The second type of study examines subjective
probability by asking participants to rate how likely they believe particular hypothetical
events are to happen in their personal future. MacLeod et al. (1996) showed that people
without mental health problems believed good things were more likely to happen to them
than bad things, but the same is not true of dysphoric groups (e.g. MacLeod and Cropley
1995). Studies comparing estimates of future event likelihood made by clinical and nonclinical groups have reported mixed findings. For negative future events, studies concur
that individuals with dysphoric mood give higher likelihood estimates than controls
(depressed patients: Butler and Mathews 1983; MacLeod et al. 1997; dysphoric students:
Andersen et al. 1992; Pietromonaco and Markus 1985). For positive future events, some
studies report that individuals with mood disturbance give lower estimates of positive
future events than controls (depressed patients: MacLeod and Cropley 1995; Pysczcynski
and Greenberg 1987; dysphoric students: Andersen et al. 1992) but others found no group
differences (dysphoric students: Pietromonaco and Markus 1985; depressed patients:
Butler and Mathews 1983).
These mixed results are due in part to the fact that participants are rating different lists
of events (MacLeod 1999). Two factors are noteworthy. First, some studies used lists
which consisted of very few items. Second, the studies presumed that events were salient
for participants. This is important because people tend to rate things they feel do not apply
to them as less likely than personally relevant examples (MacLeod and Tarbuck 1994;
Tyler and Lomax Cook 1984). MacLeod and colleagues suggested that event salience may
be more important than valence in informing probability judgments (MacLeod et al. 1997).
One way to ensure that events are personally relevant is to ask each person to generate their
own list of future events.
Another question is left unanswered by the existing literature. The studies mentioned
above found differences in perceived likelihood of future events between depressed and nondepressed individuals. However, such samples also differ dramatically in mood, which was
not controlled by these studies. Hence the possibility remains that differences in the existing
literature are driven simply by differences in prevailing mood (i.e. dysphoric in the depressed
group, euthymic in controls). Mood affects attention, memory and many other cognitive
processes (Williams et al. 1997), and biases decision-making (Schwarz and Clore 2003).
To investigate whether mood has a biasing effect on judgments of future event likelihood, we conducted an analogue study of the effects of experimentally-induced mood on
perceived likelihood in a non-clinical community sample. Manipulation of mood in the
laboratory is well-established in experimental psychology research, using film clips, music,
memories or self-referent statements to alter mood state briefly (Westermann et al. 1998).

123

The Future is Bright?

485

MacLeod and Campbell (1992) looked at differences in likelihood judgments between


groups receiving positive versus negative mood induction, and found differences for both
positive and negative outcomes. However, their study had no baseline comparison. The
current study looks at within-subject changes in likelihood judgments across mood states.
This is analogous to normal transient mood shifts, and therefore has some ecological
validity. Ethical considerations preclude the possibility of exposing clinical patients to
mood induction procedures, so we use an analogue sample. This limits the generalizability
of findings to clinical populations, because laboratory-induced mood is transient and the
relationship between this and depressed mood in clinical patients is uncertain (Clark 1983).
However, the current study takes the preliminary step of establishing whether or not
judgments of future event likelihood in non-depressed individuals are subject to mood
biasing effects.
The current study extends the literature on subjective probability in three ways. First, the
standardized list of events used is longer than in previous studies, and some studies suggest
increased reliability with additional items (Cortina 1993). Second, current participants rate
events from their own personal future as well as events from a standardized list, bringing
increased salience (MacLeod and Tarbuck 1994). Third, instead of comparing a depressed
group to non-clinical controls, a mood induction paradigm is used to elicit within-subject
comparisons of a non-depressed community sample in different mood states. This allows
examination of the extent to which likelihood ratings are susceptible to transient mood change.
Non-clinical volunteers rated self- and other-generated future events following a laboratory-based musical induction of either positive or negative mood (Velten 1968), and these
ratings were compared to ratings of the same events made in normal mood during a follow-up
telephone call. We expected to replicate previous findings (e.g. MacLeod et al. 1996) of
higher likelihood ratings for positive events than for negative ones in this non-clinical sample,
higher likelihood ratings were expected for self-generated events than for the less salient
standardized list events (MacLeod and Tarbuck 1994). Depressed patients show higher
probability ratings for negative events and lower ratings for positive events than controls with
no mood disturbance (MacLeod et al. 1997). Hence we expected our analogue sample to
show higher probability ratings for negative events and lower probability ratings for positive
events in induced negative mood than in normal mood. These hypotheses were informed by
comparisons between control groups and patients with depressed (rather than elevated) mood,
so no hypotheses were made for the positive mood induction. Recent research (Huppert and
Whittington 2003; MacLeod and Moore 2000) supports earlier suggestions (Watson et al.
1988) that positive and negative affect1 are best seen as orthogonal dimensions, rather than
opposite ends of a single dimension. Hence it seems inappropriate to hypothesise that the
effect of positive mood would simply be opposite to that of negative mood.

1 Method
1.1 Participants
Fifty-two non-depressed volunteers responded to advertisements on community notice
boards, websites and university mailing lists. Participants were aged between 18 and 65,
1

The terms affect and mood are used interchangeably by Watson and colleagues, though some
psychologists prefer to distinguish them, with affect referring to the behavioural expression of a mood,
rather than the mood itself.

123

486

S. R. Hepburn et al.

and were in the non-clinical range (M = 5.13, SD = 3.88) on the Beck Depression
Inventory (BDI-II; Beck et al. 1996). They were randomly allocated to receive the positive
mood induction (Positive MI group, n = 26) or the negative mood induction (Negative MI
group, n = 26). Within-subject conditions were used to control for time and repeated
ratings, rather than including a third group undergoing neutral mood induction. This
method is more analogous to real-life mood shifts and may have more ecological validity
than an induced neutral mood state.
1.2 Materials
1.2.1 Depression
The BDI-II (Beck et al. 1996) was used to ascertain that participants were not currently
depressed. This self-report questionnaire has high internal consistency, validity and reliability in clinical and non-clinical samples (Beck et al. 1988).
1.2.2 Mood Ratings
Participants rated mood on visual analogue scales, marking a 100 mm line to indicate their
sadness and happiness at that moment (not at all to extremely). In the follow-up call,
participants were reminded of the line and asked to give a number between 0 and 100
corresponding to where they would put their mark. Mood ratings were made throughout the
study, including immediately before and after mood induction.
1.2.3 Self-generated Events
Personally relevant future events were generated using the Personal Future Task (PFT;
MacLeod et al. 1993). In an adapted version of the original task, participants were asked to
think of events they were looking forward to or not looking forward to over four time
periods in the future (next week, month, year, 510 years). For each trial, participants
generated as many examples as possible in 30 s. The eight trials were divided into blocks
of four (Block A: positive week, negative month, positive year, negative 510 years; Block
B: negative week, positive month, negative year, positive 510 years). Order of trials was
counterbalanced within-group by block, so that half the participants had Block A as their
first set of events (pre-induction) and Block B as their second (post-induction), while the
others had the reverse. Fluency for this task (number of events generated per trial) is
reported elsewhere, along with the effects of mood on fluency (Hepburn et al. 2006).
Participants gave ratings of the valence of their self-generated events, in order to check
they were positive or negative as required. Overall, negative events had a larger spread of
valence ratings than positive ones, perhaps reflecting a tendency to categorize more neutral
items as negative rather than positive. The self-generated items were similar to the events
on the standardized list, with some overlap in content. In order to make other-generated
events applicable to most participants, their tone was more general than the self-generated
events (e.g. self: going to see my friend John in London at the weekend versus other:
meeting up with friends). As in previous studies, events expected in the distant future
were less specific than imminent events, but a preliminary analysis found no effects of time
period.

123

The Future is Bright?

487

1.2.4 Other-generated Events


A standardized list of 48 hypothetical events (6 positive and 6 negative events for 4 time
periods) was developed as follows. A written version of the PFT was administered to a
pilot sample of 15 non-depressed volunteers (graduate students aged 2435, 8 females, 13
white). Responses were adapted to be less idiosyncratic, and then combined with those
used in MacLeods work (MacLeod et al. 1991, 1996, 1997; MacLeod and Cropley 1995).
A shortlist of 64 items was then rated for likelihood and valence on 7-point scales, by a
convenience sample of eight non-depressed volunteers (university employees aged 1955,
all white females). The final list of 24 positive and 24 negative items had clear positive or
negative valence (events were cut if mean positivity and negativity ratings differed by less
than 1 point), and were perceived as possible but not inevitable (events were cut if mean
likelihood lay below 2.5 or above 5.5). The final list was divided into two sub-lists. Postinduction, participants rated the likelihood of events from one of the two sub-lists. At
follow-up all participants rated both sub-lists of events in recovered mood.
1.2.5 Likelihood Ratings
Participants rated how likely each event was to happen, using a scale of 1 (not at all) to 7
(extremely). The experimenter read items aloud and wrote down participant responses.
Mean ratings were calculated for each category of events (positive self-generated, negative
self-generated, positive other-generated, negative other-generated) for each phase of the
experiment.
1.3 Procedure
1.3.1 Pre-induction Phase (Laboratory, Before Mood Induction)
Volunteers were sent an information sheet before their assessment, and the experimenter
went through this with them in person before consent was taken. The sheet explained that
they were equally likely to be allocated to receive a positive versus a negative mood
induction. Participants answered socio-demographic questions and completed the BDI-II.
In the first block of the PFT, they generated their first set of self-generated future events,
which they rated for likelihood. Then the mood induction was administered for 8 min.
1.3.2 Mood Induction
The mood induction procedures involved participants reading 30 uplifting or depressogenic
self-statements (Velten 1968) while listening to appropriate music (Positive: Gigue from
Corellis Violin Sonata, opus 5.9; Negative: Russia Under the Mongolian Yoke by Prokofiev, at half speed). Participants reflected on the statements in their own time for eight
minutes while music played. Music was played at low volume throughout the postinduction phase. A booster was administered for two minutes (after the second block of
PFT), during which the volume was increased and participants were asked to focus again
on the statements. The purpose of the mood induction was made explicit; participants were
asked to go with the feelings it evoked. Although this raises possible demand effects,
Research Ethics Committees in the United Kingdom prefer instructions to be explicit, and

123

488

S. R. Hepburn et al.

meta-analysis suggests that demand effects cannot fully account for observed mood shifts
(Westermann et al. 1998).
1.3.3 Post-induction Phase (Laboratory, After Mood Induction)
Music was played at reduced volume throughout the post-induction phase. Participants
generated their second set of self-generated future events on the second block of the PFT,
and rated these for likelihood. A mood booster was then administered for two minutes.
Participants rated the likelihood of other-generated events, from the standardized list.
Finally, arrangements were made for a follow-up telephone call no more than 24 h after the
end of the session (M = 15.3 h, SD = 8.16). The experimenter checked how participants
were feeling before they left the laboratory and discussed the experience of the mood
induction with them, but a full debrief was left until the end of the follow-up phone call. A
positive mood induction was made available for all members of the negative MI group, but
none took up the offer.
1.3.4 Follow-up (Telephone Call, Within 24 h of Laboratory Session)
In the follow-up call, participants rated their mood. They gave likelihood ratings for selfgenerated events (first, then second, set), and other-generated events (list not yet rated, then
list rated during post-induction phase). They were thanked and debriefed.
1.3.5 Rationale for the Procedure
The follow-up phase was an integral part of the study design. The current procedure
was designed to accommodate two research questions on mood bias. As well as the
current question of mood effects on perceived likelihood of future events, we were also
interested in its effects on fluency for generating future events (i.e. number of events
generated in a given time period). Fluency for future events was measured using the
PFT pre- and post-mood induction (as described in Hepburn et al. 2006). The current
study used ratings of these self-generated events. It was important to compare ratings of
the same events made in different mood states: a difference in ratings of one set of
events rated pre-mood induction and ratings of a different set of events post-induction
might be attributable to different events being rated. It was important to present the
standardized list of events after the PFT was complete, as priming from the standardized
list might have contaminated subsequent events generated by participants. These complications necessitated the use of a follow-up phase in normal mood, so that
standardized items could be rated in normal mood subsequent to the PFT. For consistency and comparison, the same procedure was used for self-generated items, with
ratings in induced mood made post-induction, and re-ratings made at follow-up. This
also eliminated the risk of participants remembering their initial responses from preinduction when re-rating events a few minutes later post-induction. The telephone call
was arranged soon enough after the session that self-events events were still relevant,
but long enough afterwards for mood to return to normal. A statistical comparison
found that mood at follow-up was not significantly different from mood at baseline (all
p [ .90).

123

The Future is Bright?

489

2 Results
2.1 Sample Characteristics
Group means for demographic variables are given in Table 1, along with tests of significance for group differences. The groups were comparable on age, education and depression
scores, and there were no group differences in distribution of gender, students or relationship status.
2.2 Mood Manipulation Check
To check the effectiveness of the mood inductions, happiness and sadness ratings over time
were examined in two repeated-measures analyses of variance. These followed the form
Group (positive MI, negative MI) 9 Time (pre-induction, post-induction, follow-up).
Means are given in Fig. 1.
Significant Group 9 Time interactions were found for both happiness,
F(1,50) = 26.41, p \ .001, g2 = .35, and sadness, F(1,50) = 13.77, p \ .001, g2 = .50.
The interactions were followed up with pair-wise post-hoc comparisons (Bonferronicorrected) which indicated three main things. First, both mood inductions were effective in
changing mood, though the negative one resulted in more mood change than the positive.
The positive MI significantly increased ratings of happiness compared to baseline,
Mi-j = 9.46, SE = 3.18, p \ .02, but did not affect sadness, Mi-j = 2.35, SE = 2.59,
p [ .90. The negative MI significantly decreased happiness, Mi-j = 22.27, SE = 3.18,
p \ .001, as well as increasing sadness, Mi-j = 20.19, SE = 2.59, p \ .001, compared to
baseline. Second, the interactions indicate group differences in mood over time. Postmood induction, the positive MI group was significantly happier than the negative MI
group, Mi-j = 33.19, SE = 5.49, p \ .001, and the negative MI group were significantly
sadder than the positive MI group, Mi-j = 30.04, SE = 4.64, p \ .001. The groups did not
differ in happiness either pre-induction or at follow-up (all p [ .10). Pre-induction, there
was a trend for higher baseline sadness in the negative MI group (perhaps a short-term
result of being informed of their group allocation), but this was not significant,
Mi-j = 7.50, SE = 4.07, p = .07. The mean group difference in sadness was identical
between pre-induction and follow-up, Mi-j = 7.50, SE = 5.31, p = .16, so pre-induction
and follow-up mood can be seen as equivalent. Third, the analysis showed that happiness

Table 1 Group means and comparisons on demographic variables


Positive MI
group n = 26

Negative MI
group n = 26

Test of difference

Age (yrs)

25.46 (6.04)

25.92 (8.41)

F(1,51) = .05, p = .82, r = .03

Age completed
education (yrs)

23.85 (3.39)

23.00 (3.46)

F(1,51) = .80, p = .38, r = .13

Depression

4.65 (3.46)

5.62 (4.26)

F(1,51) = .80, p = .38, r = .13

Gender

10 Male

9 Male

v2(1, n = 52) = .08, p = .77, Cramers V = .04

Student or not

18 Students

17 Students

v2(1, n = 52) = .09, p = .77, Cramers V = .04

Relationship
status

13 In relationship 15 In relationship v2(1, n = 52) = .08, p = .78, Cramers V = .07

123

490

S. R. Hepburn et al.

100

Happiness (Pos MI)

Happiness (Neg MI)

Sadness (Pos MI)

Sadness (Neg MI)

90

Mood Rating

80
70
60
50
40
30
20
10
0

Pre- mood induction

Follow-up

Post- mood induction

Fig. 1 Visual analogue scale ratings of happy and sad mood over time

and sadness at the time of the follow-up phone call did not significantly differ from
happiness and sadness in the pre-induction phase (all p [ .90).
2.3 Likelihood Ratings of Future Events

Likelihood rating (1=not at all, 7 = extremely)

The main analysis compared likelihood ratings made by the two groups over time. Ratings
of individual items were averaged to give mean ratings for self- and other-generated events
of positive and negative valence. Means are shown in Fig. 2. Analysis compared ratings
made post-induction with re-ratings made at follow-up. It was predicted that self-generated

6.5
6.0
5.5
5.0
4.5
4.0
3.5
3.0

Post-induction

Follow-up

Time
Positive MI self-gen pos event

Positive MI self-gen neg event

Negative MI self-gen pos event

Negative MI self-gen neg event

Positive MI other-gen pos event

Positive MI other-gen neg event

Negative MI other-gen pos event

Negative MI other-gen neg event

Fig. 2 Mean perceived likelihood of self-generated and other-generated positive and negative future events
rated post-mood induction and at follow-up

123

The Future is Bright?

491

events would be rated more likely than other-generated events. Compared to normal mood,
we predicted that negative mood would cause participants to rate positive events as less
likely and negative events as more likely.
Preliminary analysis of likelihood ratings used a 4-way repeated-measures ANOVA:
Group (positive MI, negative MI) 9 Time (post-induction, follow-up) 9 Event (selfgenerated, other-generated) 9 Valence (positive, negative). This yielded significant main
effects of event, F(1,50) = 149.26, p \ .001, g2 = .75, and valence, F(1,50) = 48.40,
p \ .001, g2 = .49, and an Event 9 Valence interaction, F(1,50) = 5.31, p \ .05,
g2 = .10. This interaction indicated that both groups deemed positive self-generated events
most likely, followed by negative self-generated events, positive other-generated events,
and negative other-generated events. To investigate the other significant findings from the
4-way analysis further (Group 9 Valence interaction, F(1,50), p \ .05, g2 = .08,
Time 9 Valence interaction, F(1,50) = 23.25, p \ .001, g2 = .32, Group 9
Time 9 Valence interaction, F(1,50) = 7.35, p \ .01, g2 = .13), we examined ratings of
self-generated and other-generated events separately. Two analyses of variance were
performed, in the form Group (positive MI, negative MI) 9 Time (post-induction, followup) 9 Valence (positive, negative).
2.3.1 Other-generated Events
For other-generated events (on Fig. 2, the lower four lines), analysis found a main effect of
valence, F(1,49) = 49.59, p \ .001, g2 = .50, and interactions between group and
valence, F(1,49) = 4.95, p \ .05, g2 = .09, and time and valence, F(1,49) = 13.98,
p \ .001, g2 = .22. These were qualified by a Group 9 Time 9 Valence interaction,
F(1,49) = 7.75, p \ .01, g2 = .14, which was followed up with post-hoc pair-wise comparisons (Bonferroni). These showed that induced positive mood had no significant effect
for either positive or negative events (p [ .10). However, induced negative mood
decreased perceived likelihood of positive events, Mi-j = .27, SE = .08, p \ .01, and
increased perceived likelihood of negative events, Mi-j = .33, SE = .12, p \ .01. The
analysis was re-run with sub-list as a factor to control for the different sub-lists used, but
results remained virtually unchanged.
2.3.2 Self-generated Events
For self-generated events (on Fig. 2, the higher four lines), results were less clear-cut.
There was a significant Time 9 Valence interaction, F(1,49) = 5.82, p \ .05, g2 = .11,
and main effect of valence, F(1,50) = 5.42, p \ .05, g2 = .10, but no effect of time
(p = .43). Pair-wise post-hoc comparisons (Bonferroni) of the Time 9 Valence interaction
showed that positive events were viewed as more likely than negative events, both at postinduction, Mi-j = .14, SE = .05, p \ .01, and at follow-up, Mi-j = .61, SE = .18,
p \ .01. This finding was predicted, as the sample was non-depressed (MacLeod et al.
1996). Positive events were rated less likely post-induction than at follow-up, Mi-j = .14,
SE = .05, p \ .01, but mood induction did not significantly affect ratings of negative
events (p = .36). This may reflect the fact that these non-clinical participants generated
relatively few negative events in the fluency task (indeed, some participants generated
none, reducing the power of this particular test). Perhaps such events were relatively
inaccessible for them during the generation task (Hepburn et al. 2006; Newby-Clark and
Ross 2003). To control for re-rating the events, a change score was calculated for repeat

123

492

S. R. Hepburn et al.

ratings of self-generated events from pre-induction to follow-up (i.e. irrespective of mood).


Re-running the analysis with this change score as a covariate did not affect results.

3 Discussion
This study aimed to extend the literature on subjectivity probability of future events.
Previous research shows that dysphoric individuals typically perceive future bad outcomes
as more likely than controls (e.g. Butler and Mathews 1983; MacLeod et al. 1997). Some,
but not all, studies also report that mood-disturbed groups perceive future good events as
less likely (e.g. MacLeod and Cropley 1995, but c.f. Pietromonaco and Markus 1985).
Inconsistencies may result from studies: (a) using lists of events of differing length and
sometimes low reliability, (b) using events which may not have been salient for participants, and (c) not controlling for mood effects. The current study addressed these issues by
asking non-dysphoric individuals to rate idiosyncratic, personally-relevant future events
they had generated themselves, as well as an extended standardized list of events, under
different mood conditions. Within-subject comparisons showed significant effects of
experimentally-induced negative mood on estimates of future event likelihood. We first
discuss findings from the comparison between self- and other-generated events, before
considering the effects of mood, the limitations of the study, and directions for future
research.
The analysis showed that participants perceived self-generated future events to be more
likely than events from the standardized list, just as MacLeod et al. (1997) predicted. This
may be because other-generated events are harder to imagine, being less personally relevant. In this study, mood had differential effects on ratings of other-generated and selfgenerated events. Compared to their ratings under normal mood, negative mood made
participants perceive negative other-generated events as more likely, but no such effect
was found for negative self-generated events. This discrepancy should be interpreted with
caution, given the studys limitations, outlined below. However, there are some interesting
potential explanations for this finding.
One would be that self-generated events are buffered from mood effects by the contextual details surrounding them, compared to events from the standardized list. Buehler
and McFarland (2001) found that participants tend to focus attention narrowly on events
when provided with hypothetical situations, neglecting to consider contextual information.
With their participants, a narrow focus resulted in stereotypical, script-based responses
which were relatively automatic. When depressed people make judgments about the future,
they engage in low-effort, semi-automatic thinking (Andersen et al. 1992; Andersen and
Limpert 2001), so their probability judgments may begin as a direct read-out of their mood.
In judging the likelihood of other-generated events in negative mood, our participants may
also have been responding semi-automatically rather than considering the event in a reallife context. In contrast, if their self-generated events were more grounded in contextual
detail, making likelihood judgments more certain, they would be less susceptible to mood
bias. Future studies which test this theory would be welcomed, and if replicated, the
observed difference would be extremely interesting. However, it is possible that the difference is due to a ceiling effect: as perceived likelihood of self-generated events was
higher than other-generated events, there may have been less scope for changes in mood to
bring about changes in perceived likelihood of self-generated events. Higher likelihood of
self-generated events may have been due to the instruction to think of things they were
looking forward to or not looking forward to, or may reflect higher accessibility of more

123

The Future is Bright?

493

probable events during the generation process. Some participants generated very few
negative future events (none in some cases), so this category was relatively small (Hepburn
et al. 2006). In a related study, Newby-Clark and Ross (2003) found that non-clinical
participants typically recalled a mixture of positive and negative past events, but anticipated a homogeneously ideal future (i.e. exclusively non-negative events). In this respect,
future studies with depressed samples (who may be able to generate a larger number of
potentially negative future events) may be informative. Further work is needed to fully
establish the differences between ratings of self- and other-generated events, but the
inclusion of self-generated events in the current study nevertheless represents a step forward towards more ecologically valid findings.
The data shows significant effects of negative mood on likelihood judgments of othergenerated events. Both mood inductions were effective in changing mood, but the positive
induction was relatively weak, and it is not clear whether the absence of an effect of
positive mood on likelihood ratings reflects the weakness of the procedure, or whether a
strong increase in positive mood would still have no effect on likelihood ratings. Many
previous studies have found positive mood inductions to be less effective than negative
mood inductions (Westermann et al. 1998). Negative mood induction influenced ratings
both of mood and of perceived likelihood of events from the standardized list. Participants
in induced negative mood rated negative outcomes as more likely, and positive outcomes
as less likely, than they rated the same events in normal mood at follow-up. Ratings were
generally lower the first time than the second, irrespective of mood, but when this increase
was controlled, the effects remained significant.

4 Limitations
The study has limitations which must be taken into account when findings are interpreted.
The sample size was small, resulting in modest effect sizes, so the findings should be
viewed as preliminary, and requiring replication in a larger sample. In addition, the results
cannot be extrapolated to apply to clinical groups, as study participants were members of
the community who reported no difficulties with mood. The study did not measure
extraversion or introversion, which have been found to predict likelihood judgments
(Zelenski and Larsen 2002). There is also some controversy surrounding experimental
mood induction. Apparent mood effects may result from demand (i.e. participants trying
less hard to generate mood-incongruent events). However, we consider this unlikely. The
procedure is believed to successfully mimic the effects of organic negative mood on many
variables, and has been shown to influence variables which cannot be voluntarily controlled (Clark 1983; Westermann et al. 1998). As this study employs new dependent
variables, we cannot be sure the results are not due to demand, but if they were, we would
expect to see more consistent increases in perceived likelihood of negative future events
under induced sad mood. There was a trend for the negative MI group to have higher
sadness scores at baseline than the positive MI group, probably a response to the bad luck
of being allocated to this group. This leaves open the possibility that the reported differences following mood induction were due to pre-induction mood. However, this seems
unlikely given that the groups did not differ in pre-induction depression, happiness scores,
or future fluency (the number of positive or negative events they were able to generate for
themselves pre-induction, Hepburn et al. 2006). Finally, because the follow-up data was
collected over the phone rather than in the laboratory, it is possible that the within-subject
differences in likelihood estimates were due to differences in setting rather than differences

123

494

S. R. Hepburn et al.

in mood. However, the fact that there were no significant differences between ratings of
mood at baseline and at follow-up suggests that despite the change of setting, the follow-up
data reflects a realistic measure of the participants normal non-induced mood state.

5 Future Directions
Even taking into account the studys limitations, our finding that experimentally induced
mood biases future thinking in individuals without mental health issues is an important
insight which points to some questions for future work, particularly for clinical researchers.
The current data follow our previous report (Hepburn et al. 2006) in showing that nondepressed participants become more pessimistic about the future after a minor and shortterm shift towards negative mood. People with a depressive illness experience intense and
prolonged low mood. Is intensity and duration of negative mood all that drives the difference in future-oriented judgment observed between depressed and non-depressed
individuals (e.g. Butler and Mathews 1983)? To answer this question, future studies would
need to examine the disparities that remain between clinical and non-clinical samples once
mood is adequately controlled. Finally, research on memory retrieval in analogue samples
has found that non-depressed participants spontaneously engage in mood repair after
recalling a sad memory (Joormann and Siemer 2004; Rusting and DeHart 2000). Similar
studies could help to establish whether similar repair processes occur for negative thinking
about the future in non-depressed individuals.
Acknowledgements This research was supported by a Wellcome Trust Prize Studentship to Silvia
Hepburn. Thanks to Catherine Crane, Danielle Duggan, Melanie Fennell and Wendy Swift.

References
Andersen, S. M., & Limpert, C. (2001). Future-event schemas: Automaticity and rumination in major
depression. Cognitive Therapy and Research, 25(3), 311333. doi:10.1023/A:1026447600924.
Andersen, S. M., Spielman, L. A., & Bargh, J. A. (1992). Future-event schemas and certainty about the
future: Automaticity in depressives future-event predictions. Journal of Personality and Social Psychology, 63, 711723. doi:10.1037/0022-3514.63.5.711.
Beck, A. T., Steer, R. A., & Brown, G. K. (1996). BDI-II manual. San Antonio, TX: The Psychological
Corporation, Harcourt Brace & Company.
Beck, A. T., Steer, R. A., & Garbin, M. G. (1988). Psychometric properties of the Beck depression inventory:
25 years of evaluation. Clinical Psychology Review, 8, 77100. doi:10.1016/0272-7358(88)90050-5.
Beck, A. T., Weissman, A. W., Lester, D., & Trexler, L. (1979). Cognitive therapy for depression. New York:
Guilford.
Buehler, R., & McFarland, C. (2001). Intensity bias in affective forecasting: The role of temporal focus.
Personality and Social Psychology Bulletin, 27, 14801493. doi:10.1177/01461672012711009.
Butler, G., & Mathews, A. (1983). Cognitive processes in anxiety. Advances in Behaviour Research and
Therapy, 5, 5162. doi:10.1016/0146-6402(83)90015-2.
Clark, D. M. (1983). On the induction of depressed mood in the laboratory: Evaluation and comparison of
the Velten and musical procedures. Advances in Behaviour Research and Therapy, 5, 2749. doi:
10.1016/0146-6402(83)90014-0.
Conaghan, S., & Davidson, K. M. (2002). Hopelessness and the anticipation of positive and negative future
experiences in older parasuicidal adults. British Journal of Clinical Psychology, 41(3), 233242. doi:
10.1348/014466502760379208.
Cortina, J. M. (1993). What is coefficient alphas? An examination of theory and applications. Journal of
Applied Psychology, 78, 98104. doi:10.1037/0021-9010.78.1.98.

123

The Future is Bright?

495

Hepburn, S. R., Barnhofer, T., & Williams, J. M. G. (2006). Effects of mood on how future events are
generated and perceived. Personality and Individual Differences, 41, 801811. doi:10.1016/j.paid.
2006.03.022.
Huppert, F. A., & Whittington, J. E. (2003). Evidence for the independence of positive and negative wellbeing: Implications for quality of life assessment. British Journal of Health Psychology, 8(1), 107122.
doi:10.1348/135910703762879246.
Hunter, E. C., & OConnor, R. C. (2003). Hopelessness and future thinking in parasuicide: The role of
perfectionism. British Journal of Clinical Psychology, 42, 355365. doi:10.1348/014466503322528900.
Johnson, E. J., & Tversky, A. (1983). Affect generalization and the perception of risk. Journal of Personality
and Social Psychology, 45, 2031. doi:10.1037/0022-3514.45.1.20.
Joormann, J., & Siemer, M. (2004). Memory accessibility, mood regulation and dysphoria: Difficulties in
repairing sad mood with happy memories? Journal of Abnormal Psychology, 113, 179188. doi:
10.1037/0021-843X.113.2.179.
MacLeod, A. K. (1999). Prospective cognitions. In T. Dalgleish & M. J. Power (Eds.), Handbook of
cognition and emotion. New York, NY: John Wiley & Sons Ltd.
MacLeod, A. K., Byrne, A., & Valentine, J. D. (1996). Affect, emotional disorder, and future-directed
thinking. Cognition and Emotion, 10, 6985. doi:10.1080/026999396380394.
MacLeod, C., & Campbell, L. (1992). Memory accessibility and probability judgments: An experimental
evaluation of the availability heuristic. Journal of Personality and Social Psychology, 100, 478486.
MacLeod, A. K., & Cropley, M. L. (1995). Depressive future-thinking: The role of valence and specificity.
Cognitive Therapy and Research, 19, 3550. doi:10.1007/BF02229675.
MacLeod, A. K., & Moore, R. (2000). Positive thinking revisited: Positive cognitions, well-being and mental
health. Clinical Psychology and Psychotherapy, 7(1), 110. doi:10.1002/(SICI)1099-0879(200002)7:
1\1::AID-CPP228[3.0.CO;2-S.
MacLeod, A. K., Pankhania, B., Lee, M., & Mitchell, D. (1997). Parasuicide, depression and the anticipation of positive and negative experiences. Psychological Medicine, 27, 973977. doi:10.1017/
S003329179600459X.
MacLeod, A. K., Rose, G., & Williams, J. M. G. (1993). Components of hopelessness about the future in
parasuicide. Cognitive Therapy and Research, 17, 441455. doi:10.1007/BF01173056.
MacLeod, A. K., & Tarbuck, A. F. (1994). Explaining why negative events will happen to oneself: Parasuicides are pessimistic because they cant see any reason not to be. British Journal of Clinical
Psychology, 33, 317326.
MacLeod, A. K., Tata, P., Kentish, J., Carroll, F., & Hunter, E. (1997). Anxiety, depression, and explanation-based pessimism for future positive and negative events. Clinical Psychology and
Psychotherapy, 4, 1524. doi:10.1002/(SICI)1099-0879(199703)4:1\15::AID-CPP112[3.0.CO;2-#.
MacLeod, A. K., Williams, J. M. G., & Bekerian, D. A. (1991). Worry is reasonable: The role of explanations in pessimism about future personal events. Journal of Abnormal Psychology, 100, 478486.
doi:10.1037/0021-843X.100.4.478.
Newby-Clark, I. R., & Ross, M. (2003). Conceiving the past and future. Personality and Social Psychology
Bulletin, 29(7), 807818. doi:10.1177/0146167203029007001.
OConnor, R. C., Connery, H., & Cheyne, W. M. (2000). Hopelessness: The role of depression, future
directed thinking and cognitive vulnerability. Psychology, Health and Medicine, 5, 155161. doi:
10.1080/713690188.
Pietromonaco, P., & Markus, H. (1985). The nature of negative thoughts in depression. Journal of Personality and Social Psychology, 48, 799807. doi:10.1037/0022-3514.48.3.799.
Pysczcynski, T., & Greenberg, J. (1987). Self-regulatory perseveration and the depressive self-focusing
style: A self-awareness theory of reactive depression. Psychological Bulletin, 102, 122138. doi:
10.1037/0033-2909.102.1.122.
Rusting, C. L., & DeHart, T. (2000). Retrieving positive memories to regulate negative mood: Consequences for mood-congruent memory. Journal of Personality and Social Psychology, 78(4), 737752.
doi:10.1037/0022-3514.78.4.737.
Schwarz, N., & Clore, G. (2003). Mood as information: 20 Years later. Psychological Inquiry, 14, 296303.
doi:10.1207/S15327965PLI1403&4_20.
Tyler, T. R., & Lomax Cook, F. (1984). The mass media and judgment of risk: Distinguishing impact on
personal and societal levels of judgement. Journal of Personality and Social Psychology, 47, 693708.
doi:10.1037/0022-3514.47.4.693.
Velten, E. (1968). A laboratory task for induction of mood states. Behaviour Research and Therapy, 6, 473
482. doi:10.1016/0005-7967(68)90028-4.

123

496

S. R. Hepburn et al.

Watson, D., Clark, L. A., & Tellegen, A. (1988). Development and validation of brief measures of positive
and negative affect: The PANAS Scales. Journal of Personality and Social Psychology, 54, 1063
1070. doi:10.1037/0022-3514.54.6.1063.
Westermann, R., Spies, K., Stahl, G., & Hesse, F. W. (1998). Relative effectiveness and validity of mood
induction procedures: A meta-analysis. European Journal of Social Psychology, 26, 557580.
doi:10.1002/(SICI)1099-0992(199607)26:4\557::AID-EJSP769[3.0.CO;2-4.
Williams, J. M. G., Watts, F. N., MacLeod, C., & Mathews, A. (1997). Cognitive psychology and emotional
disorders. Chichester: Wiley.
Zelenski, J. M., & Larsen, R. J. (2002). Predicting the future: How affect-related personality traits influence
likelihood judgments of future events. Personality and Social Psychology Bulletin, 28(7), 10001010.
doi:10.1177/01467202028007012.

123

You might also like