381
Implementation Intentions and Shielding
Goal Striving From Unwanted Thoughts
and Feelings
Anja Achtziger
Universität Konstanz, Germany
Peter M. Gollwitzer
Universität Konstanz, Germany, and New York University, USA
Paschal Sheeran
University of Sheffield, United Kingdom
realization of goal intentions by specifying when, where,
and how goal-directed responses should be initiated.
Implementation intentions have the format of if-then
plans: “If situation X arises, then I will do Y!”
(Gollwitzer, 1993, 1999). Thus, in the case of the goal
intention “I want to eat healthily!” a supporting if-then
plan could be “If I order something to eat in my favorite
restaurant, then I will take a vegetarian meal!”
Implementation intentions create a mental link
between a selected cue or situation (e.g., the occurrence
of a certain stimulus such as the number 3 on a com-
puter screen) and a goal-directed response (e.g., pressing
a response key as fast as possible; Brandstätter,
Lengfelder, & Gollwitzer, 2001). Forming an imple-
mentation intention commits one to perform this goal-
directed response as soon as the specified situation is
encountered. For instance, Bayer and Gollwitzer (2007,
Study 1) observed that female high school students per-
formed better in a mathematics test if they supported
Authors’ Note: We thank the Center for Research on Intentions and
Intentionality (Universität Konstanz) for funding this research and
Norbert Michalski for the acquisition of the data for Study 2.
Moreover, we thank two anonymous reviewers for their valuable
comments on an earlier version of this article. Please address corre-
spondence to Anja Achtziger, Universität Konstanz, Universitätsstr. 10,
D-78457 Konstanz, Germany; phone: +49-7531-883287; fax: +49-
7531-883286; e-mail address: anja.achtziger@uni-konstanz.de.
PSPB, Vol. 34 No. 3, March 2008 381-393
DOI: 10.1177/0146167207311201
© 2008 by the Society for Personality and Social Psychology, Inc.
Forming an implementation intention or “if-then plan”
promotes the attainment of different types of goals
(Gollwitzer, 1999; Gollwitzer & Sheeran, 2006). So far,
research on implementation intentions has focused on the
initiation of goal striving, whereas the issue of shielding
of ongoing goal striving has been largely neglected. In two
field experiments concerned with dieting (Study 1) and ath-
letic goals (Study 2), goal shielding was supported by
implementation intentions geared at controlling potentially
interfering inner states (i.e., cravings for junk food in
Study 1, and disruptive thoughts, feelings, and physiolog-
ical states in Study 2). In both experiments, forming if-
then plans enhanced the rate of goal attainment. Thus,
implementation intention formation can be used to pro-
mote the realization of desired outcomes not only by
facilitating getting started with goal striving but also by
preventing goal striving from straying off course.
Keywords: implementation intentions; goal achievement;
goal shielding; dieting; sport
G
ollwitzer (1993, 1999) differentiates two kinds of
intentions: goal intentions and implementation
intentions. Goal intentions are defined as end states an
individual wants to attain; they turn desires into bind-
ing goals. Therefore, goal intentions can be thought of
as goals in the common sense as they have the format of
“I want to attain X!” (e.g., “I want to eat healthily!”).
Implementation intentions are formed to support the
© 2008 Society for Personality and Social Psychology, Inc.. All rights reserved. Not for commercial use or unauthorized distribution.
at Bobst Library, New York University on February 23, 2008 http://psp.sagepub.comDownloaded from
First publ. in: Personality and Social Psychology Bulletin 34 (2008), 3, pp. 381-393
Konstanzer Online-Publikations-System (KOPS)
URL: http://www.ub.uni-konstanz.de/kops/volltexte/2008/5708/
URN: http://nbn-resolving.de/urn:nbn:de:bsz:352-opus-57086
their goal intention “I will correctly solve as many prob-
lems as possible!” with the implementation intention
“And if I start a new problem, then I will tell myself: I
can solve this problem!” As a consequence of forming
this implementation intention, a strong mental link is
created between the external situation of starting a new
problem on the test and the goal-directed response of
motivating oneself with respect to performing well on
the mathematics test.
The mental link created by an implementation inten-
tion is expected to facilitate goal attainment on the basis
of the following psychological processes: By forming an
implementation intention, the mental representation of
the specified cue becomes highly activated and therefore
more easily accessible (Gollwitzer, 1999). This height-
ened cognitive accessibility makes it easier to detect the
critical cue in relevant situational contexts and to readily
attend to it even if one is busy with other things.
Moreover, because of its heightened cognitive accessibil-
ity, the recall of the specified situation is facilitated. The
hypothesis of a heightened cognitive accessibility of the
situation specified in the if-component of the implemen-
tation intention is supported by many studies (e.g., Aarts,
Dijksterhuis & Midden, 1999; Webb & Sheeran, 2004).
For instance, Achtziger, Bayer, and Gollwitzer (2007)
observed that participants holding implementation inten-
tions attended to and recalled the situation specified in an
implementation better than did participants who only
hold mere goal intentions. Next to the heightened cogni-
tive accessibility of the cue specified in the implementa-
tion intention, automatic initiation of the response specified
in the then-component was also found to be responsible for
the effectiveness of implementation intentions. In the pres-
ence of the critical cue, the respective response is executed
immediately (Gollwitzer & Brandstätter, 1997), without fur-
ther conscious intent (Bayer, Achtziger, Malzacher,
Moskowitz & Gollwitzer, 2007), and efficiently (Brandstätter
et al., 2001).
In the past 10 years, research in different fields of psy-
chology showed that forming an implementation inten-
tion by specifying an anticipated cue as a condition for
initiating goal-directed responses leads to more frequent
goal attainment compared to setting mere goal inten-
tions. This is especially true when goal intentions were
hard to attain or somewhat unpleasant to act on (reviews
by Achtziger & Gollwitzer, 2008; Gollwitzer & Sheeran,
2006). Most of these studies investigated the self-regulatory
issue of getting started with goal striving by executing
relevant goal-directed responses. Accordingly, the imple-
mentation intentions under study specified a good
opportunity to act in the if-component and an appropri-
ate instrumental goal-directed response in the then-
component of the if-then plan. However, as soon as a per-
son has succeeded with initiating goal striving, it still
needs to be brought to a successful ending to guarantee
goal attainment. For this purpose, people need to protect
ongoing goal striving from getting derailed. This becomes
a particularly pressing issue when the individual faces the
following impeding conditions (Gollwitzer & Sheeran,
2006; Gollwitzer, Bayer, & McCulloch, 2005): (a) con-
flicting attention and behavioral responses that make
people stray off course (e.g., distractions and tempta-
tions), (b) obstacles and barriers that are not anticipated
by the individual (e.g., deindividuating performance con-
texts may lead to social loafing), (c) detrimental inner
states that threaten goal attainment (e.g., fears, moods,
ego depletion), and (d) situationally activated antagonis-
tic goals that produce goal conflicts.
The present research investigates whether people can
use implementation intentions to protect goal striving in
one such critical situation (i.e., detrimental inner states).
Specifically, we tested whether detrimental inner states
such as disruptive thoughts (i.e., cravings) and feelings
(i.e., performance anxiety) can be controlled by imple-
mentation intentions so that goal striving can proceed
undisturbed. Such implementation intentions would
have to specify a potentially disruptive inner state in the
if-component and link it to an instrumental coping
response in the then-component. By facilitating the exe-
cution of the coping response specified in the then-
component, such implementation intentions are expected
to effectively shield the ongoing goal striving from get-
ting derailed by the detrimental inner state specified in
the if-component.
INNER STATES AND GOAL SHIELDING
Certain inner states are known to strongly imperil goal
attainment. For instance, Herman and Polivy (2004)
showed that desire-related thoughts (cravings) for
unhealthy foodstuffs imperiled the goal of self-regulating
eating behavior. Williams, Vickers, and Rodrigues (2002)
observed that performance anxiety led to decrements in
efficiency and effectiveness during table tennis training.
Accordingly, we investigated whether these inner states
(i.e., cravings and fear) can be specified in the if-component
of implementation intentions and be linked to an effective
shielding response in the then-part. If people could make
use of these inner states as initiators of goal-shielding
responses, they would have a self-regulating strategy
at their disposal that could be used across all kinds of
goal strivings alike as well as across various disruptive
situational contexts. One does not need to anticipate
what kind of external disruptive event will ultimately
threaten one’s goal striving at hand to form a protective
implementation intention. Rather, the disruptive inner states
(e.g., performance anxiety) generated by these negative
382 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN
© 2008 Society for Personality and Social Psychology, Inc.. All rights reserved. Not for commercial use or unauthorized distribution.
at Bobst Library, New York University on February 23, 2008 http://psp.sagepub.comDownloaded from
events can be used as a summary label and placed in the
if-component of a protective implementation intention.
To investigate this possibility, we ran two studies,
using disruptive inner states as if-components of protec-
tive implementation intentions: Study 1 investigated
whether specifying the negative inner state of craving for
high-fat food in the if-component of an implementation
intention and linking it to an ignore response can protect
striving toward the goal of eating healthily. Study 2 inves-
tigated whether implementation intentions to ignore neg-
ative inner states of irritation and performance anxiety
can be used to shield the goal of performing well in
tennis competitions.
STUDY 1: CONTROLLING CRAVINGS BY
MEANS OF IMPLEMENTATION INTENTIONS
Research indicates that people’s efforts to control their
food intake are plagued by frequent failure (e.g., Garner
& Wooley, 1991; Heatherton, Mahamedi, Striepe, Field,
& Keel, 1997). A key factor in such self-regulatory fail-
ures are desire thoughts (cravings) for unhealthy food-
stuffs that overwhelm people’s good intentions (e.g.,
Herman & Polivy, 2004; Kavanagh, Andrade, & May,
2005; Loewenstein, 1996). The elaborated intrusion
theory of desire (EITD; Kavanagh et al., 2005) proposes
that the intrusive effects of desire result from the interplay
of automatic and deliberative processes. External cues trig-
ger spontaneous thoughts about the target that are per-
ceived as intrusive. These intrusive thoughts are the result
of learned cue–target associations and place little demand
on cognitive resources. However, when the intrusive
thoughts elicit a strong affective reaction or provoke the
experience of deprivation, then the thought will become
elaborated, which requires controlled processes. Progressive
elaboration of the target-related thoughts is the mecha-
nism that explains the persistence and influence of desire.
Elaboration alters the prioritization of attentional and
working memory processes, making it likely that addi-
tional internal/external cues are accessed and that further
intrusive thoughts are generated. The processing priority
that is accorded to elaborating thoughts about the target
explains why progress toward superordinate goals (e.g.,
the dieting goal) often is undermined during craving
episodes.
The distinction between intrusive and elaborated
thoughts has important implications for efforts to control
desire. If people do not attend to intrusive thoughts or do
not retain those thoughts in working memory, then elab-
oration of desire-relevant information should be pre-
vented, and thus goal striving should not get disrupted.
Study 1 tested this idea. Participants were asked about
their goal intentions to reduce consumption of a high-fat
snack food. In addition, a subset of participants was asked
to form an implementation intention to ignore thoughts
about that foodstuff. The prediction tested is that partici-
pants who form an if-then plan to shield their dieting goal
from unwanted food cravings will show greater reduction
in snack food consumption compared with control partic-
ipants who do not form an if-then plan.
Method
Participants and Design
Undergraduates at a U.K. university (N = 92; 74
women and 18 men; M age = 19.50, SD = 3.79) took
part in a study on snack food consumption. Participants
were randomly assigned to one of two conditions (con-
trol vs. implementation intention).
Procedure
Baseline questionnaire. Participants completed a base-
line questionnaire that asked them to specify the follow-
ing: “Which high-fat food (e.g., chocolate, pizza, French
fries) have you eaten most of in the last week?” and “How
many times did you consume this food in the last week?”
Several other variables were assessed based on Ajzen’s
(1991) theory of planned behavior. All answer scales
ranged from 1 to 7. Goal intention was measured by “I
intend to halve my consumption of this food in the next
week” (definitely no–definitely yes) and “I will try to
halve my consumption of this food in the next week” (def-
initely no–definitely yes) (α=.92). Perceived control was
measured by the following: “For me to halve my con-
sumption of this food in the next week would be . . .”
(easy–difficult), “I am confident I can halve my consumption
of this food in the next week” (strongly disagree–strongly
agree), and “How certain are you that you can halve
your consumption of this food in the next week?” (very
uncertain–very certain) (α=.82). Subjective norm was
measured by a single item: “Most people who are impor-
tant to me think that I should halve my consumption of
this food in the next week” (definitely no–definitely yes).
Attitudes were measured by responses to the stem, “For
me to halve my consumption of this food in the next week
would be . . .” on five bipolar scales (pleasant–unpleasant,
healthy–unhealthy, important–unimportant, enjoyable–
unenjoyable, satisfying–unsatisfying) (α=.77). Finally,
social desirability was measured using the 10-item version
of the Marlowe-Crowne scale (Crowne & Marlowe,
1960) developed by Strahan and Gerbasi (1972).
Reliability was modest among the present sample (α=.47).
Implementation intention manipulation. The imple-
mentation intention manipulation came right after the
Achtziger et al. / IMPLEMENTATION INTENTIONS AND GOAL SHIELDING 383
© 2008 Society for Personality and Social Psychology, Inc.. All rights reserved. Not for commercial use or unauthorized distribution.
at Bobst Library, New York University on February 23, 2008 http://psp.sagepub.comDownloaded from
assessment of social desirability. Participants were
asked, “Please tell yourself: ‘And if I think about my
chosen food, then I will ignore that thought!’ Please say
this line to yourself three times and commit yourself to
acting on it. When you have said the line to yourself
three times, tick this box.” In the control condition,
participants received no instructions to form implemen-
tation intentions after they had completed the social
desirability items.
Follow-up questionnaire. One week later, all partici-
pants completed a follow-up questionnaire asking for
the number of times they had consumed their specified
foodstuff during the previous week.
Results
Equivalence of Conditions
We tested whether implementation intention partici-
pants differed from control participants on background
variables. A multivariate analysis of variance (MANOVA)
on the variables age, goal intention, perceived control,
attitude, norm, past consumption, and social desirabil-
ity showed no overall effect of condition, F(7, 84) =
0.63, ns (η
2
= .05), and none of the univariate tests were
significant. Chi-square tests showed equivalent propor-
tions of men and women in the two conditions and that
both conditions specified equivalent foodstuffs (ps >
.13). Social-desirability scores were not associated with
reported snack food consumption at follow-up among
implementation intention or control participants (rs = –.10
and .03, respectively, ps > .47).
Impact of Implementation Intention
Formation on Reducing Consumption
To test whether implementation intention formation
reduced consumption of specified high-fat snacks, we con-
ducted a 2 (Implementation Intention: Formed vs. Not
Formed) × 2 (Time: Baseline vs. Follow-up) repeated mea-
sures ANOVA. The main effect of Implementation
Intention was not significant, F(1, 90) = 0.71, ns (η
2
=
.001). There was a highly significant main effect of Time,
F(1, 90) = 27.99, p < .001 (η
2
= .24), such that all partici-
pants reduced their consumption of snack food from base-
line to follow-up (Ms = 4.17 and 2.96, SDs = 2.47 and
2.49, respectively). This main effect was qualified by the
expected interaction between Implementation Intention
and Time, F(1, 90) = 3.86, p = .05 (η
2
= .04). Figure 1 pre-
sents the mean number of snacks consumed by the groups
at both time points. Planned comparisons showed that the
implementation intention and control groups did not differ
on either baseline or follow-up consumption, Fs < 1.16.
Moreover, there was a significant reduction in consumption
among control participants, F(1, 40) = 3.99, p = .05 (η
2
=
.09), and participants who formed implementation inten-
tions, F(1, 50) = 36.92, p < .001 (η
2
= .43). Thus, the inter-
action is explained by the significantly greater reduction in
snack food consumption among implementation intention
participants (M = –1.60, SD = 1.88) as compared to con-
trol participants (M = –0.73, SD = 2.35).
Impact of Implementation Intention
Formation on Goal Attainment
We also examined to what extent implementation
intention versus control participants succeeded in reach-
ing the goal of halving their consumption of chosen
snack foods. One-sample t tests showed that control
participants consumed significantly more snacks at
follow-up (M = 3.27) compared to one-half of their
baseline consumption (M = 2.00), t(40) = 3.21, p <
.001. In contrast, there was no significant difference
between half of baseline consumption (M = 2.15) and
consumption at follow-up (M = 2.71) among partici-
pants who formed implementation intentions, t(40) =
1.62, p > .11. Thus, participants who formed if-then
plans achieved their goal, unlike participants with
equivalent goal intentions who did not plan.
Moderated Regression Analyses
Participants in the present study generally had weak
goal intentions to halve their consumption of high-fat
snacks (M = 3.60, SD = 1.82); in fact, 54.3% of the sam-
ple scored below the midpoint (4) on the intention scale.
384 PERSONALITY AND SOCIAL PSYCHOLOGY BULLETIN
12345
Control Implementation Intention
Conditions
Consumption of Specified Snack Food
Baseline Follow-Up
Figure 1 Consumption of specified snack food at baseline and
follow-up by condition.
© 2008 Society for Personality and Social Psychology, Inc.. All rights reserved. Not for commercial use or unauthorized distribution.
at Bobst Library, New York University on February 23, 2008 http://psp.sagepub.comDownloaded from
Because strong implementation intention effects are only
expected to emerge when participants hold strong goal
intentions (e.g., Sheeran, Webb, & Gollwitzer, 2005),
we conducted a moderated regression analysis to test
whether implementation intention formation especially
benefited participants who strongly intended to reduce
their consumption of high-fat snacks. Snack food con-
sumption at follow-up was regressed on past consump-
tion and social desirability at Step 1, the strength of goal
intention and other theory of planned behavior variables
at Step 2, and dummy-coded implementation intention
(not formed = 0, formed = 1) at Step 3. Goal intention
scores were standardized before computing the Goal
Intention Strength × Implementation Intention interac-
tion term (Aiken & West, 1991); this variable entered
the regression equation at Step 4.
Table 1 shows the results of these analyses. At Step
1, baseline consumption emerged as a strong predictor
of consumption at follow-up and explained 41% of the
variance. The inclusion of goal intention strength at
Step 2 did not improve the fit of the model, and base-
line consumption remained the only significant predic-
tor. At Step 3, implementation intentions produced a
marginally significant increase in the variance explained
in behavior (β=–.15, p = .083). Most important, how-
ever, the Goal Intention Strength × Implementation
Intention interaction was associated with a significant
increment in the variance explained in behavior (∆R
2
=
.04, p < .01). In the final equation, baseline consump-
tion and the interaction term were the only significant
predictors of behavior.
We decomposed the interaction term in the manner
specified by Aiken and West (1991). In particular,
simple slopes for implementation intentions were com-
puted at three levels of goal intention: low (M –1 SD),
moderate (M), and high (M + 1 SD). Findings showed
that when participants had low or moderate goal inten-
tions, then implementation intention formation was not
associated with consumption at follow-up (Bs = .39 and
–.69, respectively, ns). However, when participants’
intention to reduce their snack food consumption was
high, then implementation intention formation was
associated with reduced consumption of participants’
chosen snack food (B = –1.76, p < .05). Thus, forming
an implementation intention was particularly beneficial
in reducing snack food consumption among participants
who strongly intended to decrease their intake of snacks.
Discussion
Implementation intentions that specify thinking
about a tempting food in the if-component and ignoring
that thought in the then-component effectively shielded
goal striving to halve respective food intake. This find-
ing goes beyond past research on implementation inten-
tions (see meta-analysis by Gollwitzer & Sheeran, 2006)
as it specifies a negative inner state (i.e., a critical
thought), whereas most prior research specified external
cues (i.e., a critical external context). Most important,
the present study focused on shielding ongoing goal
striving from getting derailed by disruptive thoughts
(i.e., craving for high-fat snacks), whereas past research
on implementation intentions focused on getting on with
goal attainment by initiating goal-directed behaviors.
Study 1 drew upon a recent theory of desire (EITD;
Kavanagh et al., 2005) to specify what if- and then-
components of an implementation intention would be
effective in reducing desire thoughts—and thereby pro-
mote the achievement of the goal of reducing consump-
tion of high-fat snacks. According to the EITD,
Achtziger et al. / IMPLEMENTATION INTENTIONS AND GOAL SHIELDING 385
TABLE 1: Regression of Snack Food Consumption at Follow-Up on Predictor Variables (Study 1)
Betas
Step Variable Entered Step 1 Step 2 Step 3 Step 4
1. Baseline consumption .64*** .57*** .59*** .59***
Social desirability –.10 –.05 –.04 –.04
2. Goal intention –.11 –.12 .11
Attitude –.04 –.04 –.05
Subjective norm .10 .08 .08
Perceived control –.11 –.09 –.08
3. If–then plan –.15
†
–.15
†
4. Goal intention × If-Then Plan .31**
R
2
.41 .44 .46 .50
Model F 30.26*** 10.76*** 9.90*** 10.23***
∆R
2
— .03 .02 .04
∆F — 1.01 3.08† 7.25**
NOTE:
†
p < .09. **p < .01. ***p < .001.
© 2008 Society for Personality and Social Psychology, Inc.. All rights reserved. Not for commercial use or unauthorized distribution.
at Bobst Library, New York University on February 23, 2008 http://psp.sagepub.comDownloaded from
