Individual Differences in the Intention Superiority Effect

advertisement
Intention-superiority effect
Short title: Intention-superiority effect
When does the intention-superiority effect occur? Activation patterns before and after task
completion, and moderating variables
Suzanna L. Penningroth
Department of Psychology, Dept. 3415, University of Wyoming
1000 E. University Ave., Laramie, WY, 82071, U.S.A.
Corresponding author:
Suzanna L. Penningroth, Ph.D.
Email: spenning@uwyo.edu
Phone: 307-755-5669 (in the U.S.)
Fax: 307-766-2926
Department of Psychology, Dept. 3415, University of Wyoming
1000 E. University Ave., Laramie, WY, 82071, U.S.A.
1
Intention-superiority effect
Acknowledgements
I would like to thank Stephanie Worthen, Katie Rudoff, Gary King, Kevin Malatesta,
Nathan Jones, Rachel Holt, Ashley Gilken, Heather Fleck, Joanna Morris, Aminda O’Hare, and
Kathy McGuire for assistance in data collection, data entry, and data compilation; Walter D.
Scott for editorial comments; and Rich Marsh for sharing his lexical decision task materials.
2
Intention-superiority effect
3
Abstract
This experiment investigated the intention-superiority effect, potential moderators of the
effect, including personality disposition (i.e., state vs. action orientation) and cue source (self
versus experimenter), and the fate of the effect after performance. The intention-superiority
effect was defined as greater accessibility for intention-related words than words to be
remembered. Accessibility was assessed with lexical decision latencies for words from short
action scripts, and activation patterns were measured before and after scripts were performed. A
general preperformance intention-superiority effect was replicated, but this effect depended on a
personality disposition by cue interaction. For the state-oriented group, the effect occurred when
participants expected performance to be self-cued, but not experimenter-cued. For the actionoriented group, the effect occurred when participants expected performance to be experimentercued, but not self-cued. After script performance, a sustained intention-superiority effect was
found. Possible reasons for the differences in results found in this study and past studies are
discussed.
Intention-superiority effect
4
When does the intention-superiority effect occur? Activation patterns before and after task
completion, and moderating variables
Completing a prospective memory task (an intention), such as filling the car’s fuel tank
on the way home, requires remembering both that there is something to be done and what that
something is. In other words, performance relies on memory for intent (e.g., do something on
the drive home) and memory for content (e.g., fill up the fuel tank) (e.g., Einstein & McDaniel,
1990; Kvavilashvili, 1987). Increasingly, researchers have focused their attention on
understanding the content of intentions, especially how activation persists or fluctuates (e.g.,
Freeman & Ellis, 2003b; Marsh, Hicks, & Bink, 1998). The intention-superiority effect refers to
the finding that unfulfilled intentions show greater accessibility than other memory contents
(e.g., Goschke & Kuhl, 1993). Greater accessibility has been demonstrated as (1) better recall
(e.g., Koriat, Ben-Zur, & Nussbaum, 1990, Maylor, Chater, & Brown, 2001), (2) faster
recognition latencies (e.g., Goschke & Kuhl, 1993), (3) faster lexical decision latencies (e.g.,
Marsh et al., 1998), and (4) greater interference in a Stroop task (Cohen, Dixon, & Lindsay,
2005).
The intention-superiority effect specifically refers to the heightened accessibility of an
intention’s content during the second phase of a prospective memory task, the retention interval,
which occurs after the intention has been formed but before the performance interval (Ellis,
1996; Kliegel, Martin, McDaniel, & Einstein, 2002). Investigators have studied this effect both
because it might be linked to better prospective memory performance (e.g., Freeman & Ellis,
2003a) and because heightened accessibility of pending intentions outside of the performance
interval might interfere with other cognitive tasks (e.g., Kuhl & Helle, 1986).
Although many studies have replicated the general intention-superiority effect for
uncompleted intentions (e.g., Marsh et al., 1998; Maylor et al., 2001), little is known about the
conditions that produce this effect. Based on the action control theory developed by Kuhl and
colleagues (e.g., Goschke & Kuhl, 1993; 1996; Kuhl & Beckmann, 1994b), two variables likely
to act as moderators in intention-superiority effects are state versus action orientation and
performance cue source (e.g., self cued or externally cued). Therefore, one major purpose of the
current study was to test these moderating variables in the preperformance intention-superiority
effect. The second major purpose was to investigate intention activation after task completion.
Few studies have investigated this issue, and the results have been mixed. Some studies have
shown an inhibition effect for completed intentions (e.g., Marsh et al., 1998) and other studies
have shown persisting activation (e.g., Nowinski, Holbrook, & Dismukes, 2005). Also, there
has been almost no research on the possible moderating effects of state versus action orientation
and performance cue source on postcompletion activation of intentions. However, these
moderating effects would be consistent with action control theory (e.g., Goschke & Kuhl, 1993;
Kuhl & Goschke, 1994).
The Intention-Superiority Effect
Many studies have shown that intentions show greater accessibility in long term memory
(LTM) compared to other memory contents, and this intention-superiority effect has been
demonstrated with different methods. Most studies have used experimenter-assigned intentions.
For instance, in an early demonstration of the intention-superiority effect, Koriat et al. (1990)
compared recall for noun-verb pairs under two encoding conditions: intended for verbal report
or future enactment. Recall was superior for items that were intended for enactment. The
intention-superiority effect has also been measured as faster RTs for words associated with
intentions than for neutral words. For example, Goschke and Kuhl (1993) used a postponedintention task in which participants memorized two action scripts (e.g., setting a table and clear
a desk) and then received instructions telling them which script they would later perform. In a
Intention-superiority effect
5
word recognition task, latencies were faster for words from scripts to be performed than for
words from scripts that were not to be performed. In a series of studies, Marsh, Hicks, and
colleagues adapted the procedure used by Goschke and Kuhl (1993) by using a lexical decision
task instead of a recognition task and replicated the general intention-superiority effect (Marsh et
al., 1998; Marsh, Hicks, & Bryan, 1999; Marsh, Hicks, & Watson, 2002). More recently, Cohen
et al. (2005) demonstrated the intention-superiority effect using a Stroop task. That is, color
naming was slower for words related to a pending intention than for words related to an intention
participants were told to forget. Marsh, Cook, Meeks, Clark-Foos, and Hicks (2007) also found
an intention-superiority effect for material participants were trying to ignore. During visual
presentation of an ongoing task and prospective memory task, participants were also exposed to
intention-related material and neutral material in an auditory channel they were told to ignore. In
a surprise recognition test administered later, participants showed superior recognition for
intention-related material. Finally, the intention-superiority effect has also been demonstrated
using real intentions reported by participants (Freeman & Ellis, 2003a; Maylor et al., 2001). For
instance, Maylor et al. (2001) demonstrated the effect as higher recall rates for real prospective
memories than retrospective memories.
Although the intention-superiority effect has been demonstrated in numerous studies,
researchers disagree on the underlying mechanism for the effect. For instance, Goschke and
Kuhl (1993) concluded that intentions have a slower rate of decay than non-intention memory
contents. They suggested that intentions be considered a LTM analogue of the goal nodes that
act as source nodes in Anderson’s (1983) ACT* theory. In ACT*, only goal nodes are capable
of maintaining activation in WM without rehearsal or external stimulation. Alternatively, the
intention-superiority effect might be described as an enhanced revival rate for intention-related
material, perhaps because of a commitment marker (Kuhl & Kazén-Saad, 1988) or because the
first cue encountered serves as a reminder, followed by reflexive retrieval of related contents
(Cohen et al., 2005). However, both types of explanation refer to greater accessibility of
intention-related contents over other memory contents (Goschke & Kuhl, 1993; Marsh et al.,
1998). In contrast, Freeman, Ellis, and collaborators (e.g., Freeman & Ellis, 2003b) have
suggested that it is not the intentional status that affords intentions greater accessibility, but the
motor response that is planned. They argue that the effect would be more accurately labeled an
action superiority effect. Several studies have supported this proposed mechanism. For
instance, Freeman and Ellis (2003b) found that the intention-superiority effect disappeared when
motor programming for intended actions was blocked by another motor task. Using fMRI data,
Eschen et al. (2007) found that brain areas associated with motor planning for hand movements
showed greater activation when verbs were encoded for later enactment versus verbal report.
However, as Freeman and Ellis (2003b) note, motor planning might be more important for the
types of simple actions they studied (e.g., clap) than for more complex intentions, which might
be maintained in a predominantly verbal code.
Potential Moderating Variables in the Intention-Superiority Effect
There are theoretical reasons to predict moderating variables in the intention-superiority
effect. First, the theory of action control developed by Kuhl and colleagues (e.g., Goschke &
Kuhl, 1993; Kuhl & Beckmann, 1994a; 1994b) leads to predictions about a dispositional
personality variable that affects self regulation (Klinger & Murphy, 1994) and intention
activation. Individuals who are state oriented tend to hesitate in taking action. In contrast,
individuals who are action oriented tend to take action when needed. For example, on the scale
used to measure action versus state orientation, the Action Control Scale (ACS, Kuhl, 1994b), in
response to the item “When there are two thing that I really want to do, but I can’t do both of
them,” the state-oriented choice would be “It’s not easy for me to put the thing that I couldn’t do
Intention-superiority effect
6
out of my mind.” The action-oriented response would be “I quickly begin one thing and forget
about the other thing I couldn’t do.” As these different responses show, state-oriented
individuals are more likely to ruminate about goals that are currently irrelevant. This
characteristic of state orientation is thought to affect intention activation (Goschke & Kuhl, 1996;
Kuhl & Kazén-Saad, 1988). Generally, state orientation is thought to be characterized by the
persistence of declarative representations for unfulfilled intentions, and this is true even when the
intention cannot be performed in the current context (Kuhl & Goschke, 1994). Therefore, stateoriented individuals are predicted to show a larger intention-superiority effect than actionoriented individuals.
Several studies have provided evidence of a greater intention-superiority effect in statethan action- oriented individuals. In the 1993 experiments by Goschke and Kuhl described
earlier (Exps. 1, 2, and 4), a greater intention-superiority effect was found for state- than actionoriented participants. Penningroth (2005b) replicated the moderating effect of state versus action
orientation, but the intention-superiority effect was measured as higher recall rates for real-life
prospective memories than retrospective memories.
However, based on action control theory (e.g., Goschke & Kuhl, 1993, 1996), state
versus action orientation would also be predicted to interact with cue source in affecting the
intention-superiority effect. Specifically, state-oriented individuals are characterised as showing
the intention-superiority effect more generally, even in conditions where it would not be adaptive
to do so, such as when a reliable cue is available to aid remembering (e.g., a computer-provided
prompt). In contrast, action-oriented individuals are characterized as showing more adaptive use
of the intention-superiority effect, such as when the performance cue is less reliable (e.g., a selfinitiated time-based cue).
In their 1993 study, Goschke and Kuhl found evidence for this predicted pattern of results
although the pattern was shown by a combination of results from four experiments. The
postponed-intention paradigm (Goschke & Kuhl, 1993; Kuhl & Goschke, 1994) was used in all
four experiments, with the intention-superiority effect defined as faster recognition latencies for
words from action scripts to be performed than from scripts to be remembered. In three of the
experiments, participants expected to be cued (by the computer) when it came time to perform
the script. Under these external cueing conditions, state-oriented groups showed a larger
intention-superiority effect than action-oriented groups. However, in the other experiment,
participants had to remember to stop the recognition task after 15 minutes and perform the
intended script. Thus, in this case, the task was a time-based prospective memory task (Einstein
& McDaniel, 1990) with a self-provided performance cue. With this type of cue, both actionand state-oriented groups showed an intention-superiority effect. Freeman and Ellis (2003b,
Exp. 1) also examined cue source as a possible moderator for the intention-superiority effect
(state vs. action orientation was not assessed). They compared self-cueing and experimentercueing conditions, but results showed an intention-superiority effect for both cue-source
conditions. In summary, very few studies have examined the influence of cue source or
interactions with cue source in the intention-superiority effect.
Yet, the distinction between self-initiated and other-initiated cueing conditions (e.g.,
experimenter or computer cueing) may be an important one. It might be argued that only selfcueing conditions characterise true prospective memory tasks (i.e., tasks incorporating the intent
to remember in addition to the content to be remembered; incorporating “prospective memory
proper;” Graf & Uttl, 2001). From this view, self-initiated conditions are the most relevant when
studying the intention-superiority effect. However, the intention-superiority effect has also been
demonstrated when participants expected to be told when to perform (Freeman & Ellis, 2003b;
Intention-superiority effect
7
Goschke & Kuhl, 1993). Therefore, there is still a need for research that investigates the
intention-superiority effect within both of these cue-source conditions.
Postperformance Activation Levels
The fate of intention activation after task completion has also been investigated. Results
from several studies have shown a postperformance inhibition effect (e.g., Badets, Blandin,
Bouquet, & Shea, 2006; Freeman & Ellis, 2003a; Marsh et al., 1998). For example, Marsh and
colleagues (e.g., Marsh et al., 1998, 1999) have reported a reversal in activation levels after
intention completion (slower RTs for words from prospective scripts than from neutral scripts).
As Marsh et al. (1998) note, this pattern seems adaptive because inhibiting completed intentions
would allow a person to then devote attentional resources to other tasks. Similarly, Freeman and
Ellis (2003a) used a speeded retrieval task to assess the accessibility of everyday intentions
before and after completion (Maylor, Darby, & Della Sala, 2000) and found that completed
intentions were less accessible than uncompleted intentions.
In contrast, results from other studies have shown that after task performance, the
intention-superiority effect persists or activation returns to baseline levels (e.g., Cohen et al.,
2005; Nowinski et al., 2005). For instance, Nowinski et al. (2005) assessed intention activation
across two phases of an experiment that participants believed to be separate experiments.
Reaction times for a pleasantness rating task in the second “experiment” were slower for the
word that had been the prospective memory target in the first “experiment.” The authors argued
that lingering activation for completed intentions caused partial retrieval of the prospective
memory task that was no longer relevant. As previously described, Cohen et al. (2005) measured
intention accessibility as interference in a Stroop task. For completed intentions, they found a
decrease in activation levels but no evidence of inhibition. Finally, studies reporting the subject
performed task effect are also relevant when trying to characterise the pattern of intention
activation after performance. This effect refers to greater accessibility (e.g., higher recall and
recognition rates) for action phrases that were performed during encoding rather than heard or
read during encoding (e.g., Cohen, 1981; Freeman & Ellis, 2003b; Jahn & Engelkamp, 2003).
Thus, in the subject performed task effect, the demonstrated heightened accessibility of
performed actions is relative to action words that were not performed.
Therefore, past research has shown three different outcomes for the activation levels of
just-completed intentions: inhibition, sustained activation, or a return to baseline. These
different outcomes also suggest different mechanisms for deactivation. That is, an inhibition
effect (relative inaccessibility) would require either deliberate inhibition of just-completed
intentions or selective activation of other memory contents because a decay mechanism alone
would not explain this effect. In contrast, observed outcomes of sustained activation or a return
to baseline activation can be explained with either a decay mechanism or deliberate processing
mechanisms.
Potential Moderating Variables in Postperformance Intention Activation
The theoretical description of state orientation leads to the prediction that these
individuals will show sustained activation of intentions after task completion (e.g., Kuhl &
Goschke, 1994). That is, state-oriented individuals are characterized as showing excessive
perseveration of intentions. They have difficulty deactivating intentions that are no longer
relevant, and this is thought to be true for both successful and unsuccessful outcomes. In
contrast, action-oriented individuals are characterized as showing more flexible control over
intention activation, which includes the ability to “turn off” completed intentions. Beckmann
(1994) reported findings consistent with this prediction. In one unpublished study, state-oriented
participants showed a trend for less deactivation of completed intentions than action-oriented
participants.
Intention-superiority effect
8
Cue source (self vs. experimenter) might also affect postperformance intention activation.
That is, self-cued performance might lead to quicker de-activation because self-cued
performance relies on carrying out both the intent and content of the prospective memory task
whereas experimenter-cued performance only requires retrieval of the content. However,
Freeman and Ellis (2003b) tested whether the subject performed task effect differed for
conditions of self-cueing and experimenter-cueing and found no difference. That is, for both cue
source conditions, accessibility was higher if action phrases were performed at encoding than if
they were verbally encoded. Given that this effect has only been tested in one study with a
resulting null effect, more research is needed on how cuing expectations affect postcompletion
intention accessibility.
The Present Research
In brief, the main purpose was to extend past findings on the intention-superiority effect
by examining activation patterns before and after task completion, but also to test potential
moderating influences of cue source (self vs. experimenter) and dispositional orientation (state
vs. action orientation). The postponed intention procedure (Goschke & Kuhl, 1993) was adapted
so that accessibility was measured with a lexical decision task (Marsh et al., 1998, Exp. 3).
Specifically, in each of two blocks, participants memorised two short action scripts. Then they
were told which script they would have to perform (the prospective script) and which they would
just have to remember (the neutral script). An intention-superiority effect would be
demonstrated as a faster RT for prospective than neutral script words during a lexical decision
task. In one block, the lexical decision task occurred before script performance, and in the other
block, the lexical decision task occurred after script performance.
There were several research questions. First, is the preperformance intention-superiority
effect moderated by the personality disposition of state versus action orientation or by cue
source? The intention-superiority effect was predicted to be larger for state- than action-oriented
individuals. State versus action orientation was also predicted to interact with cue source (self
vs. experimenter). If the pattern of results Goschke and Kuhl (1993) found across experiments is
replicated here within a single experiment, results would show an intention-superiority effect
only with self-cued conditions for the action-oriented group, but for both experimenter-cued and
self-cued conditions for the state-oriented group. Second, what happens to intention activation
after performance? Past findings have been mixed. However, based on action control theory
(e.g., Kuhl & Beckmann, 1994b), a higher maintained intention-superiority effect was predicted
for state-oriented individuals specifically.
Method
Participants
The final sample comprised 191 undergraduate students at the University of Wyoming
(61 males) who participated in exchange for partial fulfillment of a course research requirement1.
Participants ranged in age from 17 – 30 years with a mean age of 20.0 years.
Materials
Most of the materials and procedures were adapted from those used by Marsh et al.
(1998, Exp. 3), who had adapted much of their procedure from Goschke and Kuhl (1993),
substituting a lexical decision task for a recognition task to measure accessibility.
Scripts. Four scripts were developed, two to be memorised in each of the two blocks.
Each script contained a title (e.g., Setting a table) followed by five action phrases (e.g., Spread
the tablecloth, Set the plates, Polish the glasses, Fold the napkins, Light the candles).
Designation of the two scripts within a block as either the prospective or neutral script was done
randomly, as were the pairings of scripts, to create a counterbalanced set of materials.
Intention-superiority effect
9
Lexical decision task. Construction of the lexical decision task materials followed the
same plan in both blocks. Four sets of items were used as stimuli. Two sets of items were
words from the action phrases in the memorised scripts. From each script, the five verbs (e.g.,
fold) and five nouns (e.g., napkins) were used, resulting in 10 prospective script words and 10
neutral script words. The third set of items, the nonscript words (e.g., human), comprised 20
English words that had not appeared in the scripts and that were matched to the 20 script words
on number of syllables and frequency (Francis, Kučera, & Mackie, 1982) and were not
synonyms or homophones of script words. Finally, the fourth set of items included 40
pronounceable nonwords (e.g., plame) that were matched to the 40 words on number of syllables
and were not pseudohomophones. In each block, the lexical decision task began with 12 buffer
items (half words and half nonwords).
The lexical decision tasks (and much of the experiment) were administered through
Superlab programs. The programs randomised the presentation order for the 80 non-buffer items
in the lexical decision task. One major difference between the current study and past studies was
the addition of an anagram task before script performance. This filler task made two
performance cue conditions possible. That is, in addition to experimenter cueing, the filler task
made possible a self-cueing condition. The answers to the anagrams contained no words from
the memorized scripts and no words that appeared to be related to the words from these scripts.
Dispositional orientation: State versus action orientation. Participants were categorised
on orientation (i.e., as state or action oriented) according to their scores on the prospective and
decision-related versus hesitation (AOD) subscale of the ACS (Kuhl, 1994b; see Goschke &
Kuhl, 1996). The prospective subscale is a 12-item scale with two choices for every item. For
example, one item is “When I am facing a big project that has to be done: (A) I often spend too
long thinking about where I should begin; (B) I don’t have any problems getting started.” For
this item, the action-oriented response is choice B. The score is computed by adding up the
action-oriented responses. This subscale has good internal reliability (Cronbach’s α = .78, Kuhl,
1994b) and construct validity (Klinger & Murphy, 1994; Kuhl, 1994b). Goschke and Kuhl
(1993) classified participants above the sample median as action oriented and those below the
sample median as state oriented. In the current study, participants who scored at or above the
sample median (which was 7) were categorised as action-oriented2 (n = 106) and participants
who scored below the sample median were categorised as state-oriented (n = 85).
Design
The dependent variable was script-word RT in the lexical decision task. A mixed
factorial design was used. Script (prospective vs. neutral) and completion status (before vs. after
script performance) were within-groups variables. Orientation (state vs. action orientation) and
cue source (self vs. experimenter) were between–groups variables. Participants were randomly
assigned to self-cued (n = 93) and experimenter-cued (n = 98) conditions. Thus, four betweengroups conditions were created: state-oriented/self-cued (n = 45), state-oriented/experimentercued (n = 40), action-oriented/self-cued (n = 48), and action-oriented/experimenter-cued (n =
58).
Procedure
In brief, participants completed the postponed-intention paradigm used by Marsh et al.
(1998, Exp. 3), but with an added anagram task before script performance and a between-groups
manipulation of cue-type. Participants were run individually. They first entered a main room
with doors to three smaller rooms. They were told they would memorise actions and then
perform them. To perform the actions, they were to go to the two rooms on the end and use
Room 1 for the first set (in the first half) and Room 2 for the second set (in the second half).
Intention-superiority effect
10
Specifically, when it was time to perform the actions, they were to walk to the room, open the
door, and perform the designated actions using the materials in the rooms.
The experiment began in the closest room. The computer administered many of the
instructions, but the experimenter remained in the room (behind participants) to provide some
instructions. Participants first heard an overview of the experiment. That is, in each half of the
experiment, they would learn two pairs of scripted actions for a memory test later. They would
have to perform one of the scripts from each pair from memory but would not learn which was to
be performed until after learning both scripts. Three other tasks were then explained: counting
backward by threes, the lexical decision task, and the anagram task. Participants practised the
counting backward and lexical decision tasks. For the lexical decision task, participants placed
their index fingers on the outermost buttons of a four-button response box. They were instructed
to respond “Yes” (right button) if the letter string was an English word and “No” (left button) if
it was not. For each trial, a warning tone appeared for 500 ms, followed by a fixation point (the
“+” sign) that remained for 500 ms. A letter string then appeared and remained until a response
occurred. The screen provided feedback for errors during practice but not during the main task.
After a 200 ms delay, the next trial began. Participants were instructed to respond as accurately
and quickly as possible. For the anagram task, participants were shown a paper with anagrams
and told that the task was to find an English word that consisted of all the letters in the string.
The order of completion status condition was counterbalanced across participants. That
is, approximately half of the participants were randomly assigned to do the lexical decision task
before task performance in Block 1 and after task performance in Block 2 (Block order 1, which
is depicted in Figure 1). The order was reversed for the remaining participants (Block order 2),
with the lexical decision task occurring after task performance in Block 1 and before task
performance in Block 2.
Procedure for Block order 1. The block began with script learning. Participants were
instructed that the computer would present two action scripts to be memorised, each containing
five actions. One of the scripts would be performed later and the actions were to be performed in
the same order as they were presented. Then participants were told what to expect for script
presentations. That is, the title of the first script would remain on the screen for 10 s. Then, each
action would be added sequentially, with each updated screen remaining on for 10 s. Then, the
entire script would remain on the screen for an additional 30 s. This process would be repeated
for the second script. Last, the entire sequence would be repeated. The screen would then
prompt participants to recall the two scripts by writing them down on paper, and the
experimenter would check them. Participants were also told to do two things to help with
memorising the scripts: Read each line out loud the first time it appeared, and visualise each
action3. If recall was not perfect for both scripts, there was one more presentation of each script,
followed by another recall test.
Next, participants received instructions for the next five tasks. The first task would be
counting backward by threes from the number shown on the screen until the computer beeped
after 30 s. The second task would be reading the screen to find out which script would be
performed later. That screen would stay on for 10 s. The third task would be the word-decision
task (lexical decision task), which would start immediately. Participants were briefly reminded
of the lexical decision task instructions. The fourth and fifth tasks would be the anagram task
and performing the designated script. The screen would say Do anagrams for 3 minutes, then
perform script. Further explanation of this task varied by cue condition and was provided by the
experimenter. The experimenter-cued group was told I will stop you after three minutes and tell
you to perform the designated script. The self-cued group was instead told You’ll do anagrams
for exactly three minutes. See this digital clock? I’ll start it. Then, you’ll use it to watch for
Intention-superiority effect
11
when three minutes are up. Then you’ll need to stop yourself and go perform the script4. The
clock was positioned so that participants had to turn their heads to check the clock.
Next, the experimenter started the computer program, which presented the counting
backward task, the script performance instructions, and the lexical decision task. For the script
performance instructions, the screen presented the instruction Read this screen out loud., then
You will perform, followed by the prospective script title (e.g., Setting a table). Below that, the
screen presented the instruction You will not perform, followed by the neutral script title (e.g.,
Brewing coffee). The order of these two performance instructions on the screen was
counterbalanced across participants. This counterbalancing and the requirement that participants
read the screen out loud were included to equate processing of the prospective and neutral script
contents as much as possible. After the lexical decision task, the experimenter instructed
participants to do the anagram task for three minutes, briefly summarising the earlier instructions
(which varied by cue condition) for stopping the anagram task to perform the designated script.
When participants rose to perform the script, the experimenter followed and recorded the order
of the actions performed.
Upon returning to the computer room, the second half of the experiment (Block 2) began
with the presentation of two new scripts to be memorised. The script learning procedure was the
same as in Block 1. The summary instructions for the five tasks were the same as in Block 1,
except task order was changed for the last three tasks. That is, in Block 1, the order was: lexical
decision task, anagrams, and then script performance. In Block 2, the order was: anagrams,
script performance, and then lexical decision task. Thus, the lexical decision task latencies were
recorded after script performance. At the end of the experiment, participants completed a survey
that included assessment of state versus action orientation. No final memory test for the two
scripts was administered. However, it was clear that participants expected a test because in a
postexperimental survey given to approximately one third of the participants, 100% answered
“yes” when asked if they had expected a memory test on the scripts they had not performed.
Procedure for Block order 2. For participants who received Block order 2, the procedure
was the same as with Block order 1 except the lexical decision task occurred after task
performance in Block 1 and before task performance in Block 2.
In summary, the procedure was similar to that of Marsh et al. (1998, Exp. 3), but with the
following major changes. First, an anagram task was added before script performance in order to
add the manipulation of cue source (self vs. experimenter) expected. Marsh et al. noted that they
had not specified the type of cue participants should expect, and it was unknown whether
participants expected an external or self-initiated cue. The new anagram task also added a threemin delay before script performance. Second, presentation of the script performance instructions
varied in two ways from Marsh et al. In Marsh et al., the instructions remained on the screen for
three s, but in the current experiment, they remained for 10 s because in pilot testing several
participants required this presentation length to read all the instructions. Finally, in the current
study, the order of prospective and neutral script performance instructions on the screen was
counterbalanced, and participants were required to read the entire screen out loud.
Results
An alpha level of .05 was used, and all significance levels reported are two-tailed. To
further reduce Type I error rates, for post hoc comparisons, F values were calculated using the
MSE term from the original analyses (Field, 2005). The main dependent variables were
calculated from lexical decision latencies for prospective and neutral script words. Data
trimming involved deleting RTs for erroneous responses and RTs greater than 3 SDs from an
individual’s mean (for each of the four dependent variables: preperformance prospective words,
preperformance neutral words, postperformance prospective words, and postperformance neutral
Intention-superiority effect
12
words). Then, for each of the four dependent variables, outliers were curbed to within 3 SDs. In
addition, all major analyses were repeated twice: first, using log-transformed RTs (Tabachnick
& Fidell, 1996), and second, after equating group size for the variables of orientation and cue
source by randomly deleting participants. The pattern of results did not change with either
modification.
Conditions Associated with the Preperformance Intention-Superiority Effect
Preliminary analyses. Before examining the conditions under which the intentionsuperiority effect occurs, effects of block and gender were tested with preperformance intentionsuperiority effect (mean neutral RT – mean prospective RT) as the dependent variable.
Therefore, a 2 X 2 X 2 X 2 (Gender X Block X Orientation X Cue source) ANOVA was
conducted. There was a main effect of block, with a larger intention-superiority effect in Block 2
(M = 19.19 ms, SD = 53.80) than in Block 1 (M = 2.96 ms, SD = 47.14), F(1, 174) = 6.32, p =
.013,p2 = .035.5 There was no main effect of gender, F(1, 174) = 0.91, p = .342,p2 = .005.
Further, none of the interactions involving block or gender were significant. That is, there were
no two-way interactions (Block X Gender: F(1, 174) = 2.30, p = .131,p2 = .013; Block X Cue
source: F(1, 174) = 1.81, p = .181,p2 = .010; Block X Orientation: F(1, 174) = 0.16, p =
.692,p2 = .001; Gender X Cue source: F(1, 174) = 0.20, p = .658,p2 = .001; Gender X
Orientation: F(1, 174) = 3.66, p = .057,p2 = .021). There were also no three- or four-way
interactions (Block X Gender X Cue source: F(1, 174) = 0.44, p = .506,p2 = .003; Block X
Gender X Orientation: F(1, 174) = 0.57, p = .451,p2 = .003; Gender X Cue source X
Orientation: F(1, 174) = 0.12, p = .731,p2 = .001; Block X Gender X Cue source X
Orientation: F(1, 174) = 0.25, p = .618,p2 = .001).
Main analyses. A 2 X 2 X 2 (Orientation X Cue source X Script) ANOVA was
conducted with RT as the dependent variable. The effects of interest were the main effect of
script, the interaction between orientation and script, the interaction between cue source and
script, and the three-way interaction between orientation, cue source, and script. As predicted,
there was a main effect of script (i.e., a general intention-superiority effect) with faster RTs for
prospective script words (M = 533.18 ms, SD = 92.58) than neutral script words (M = 544.38,
ms, SD = 93.62), F(1, 187) = 8.43, p = .004,p2 = .043. There were no other main effects6.
The predicted two-way interaction between orientation and script was not significant,
F(1, 187) = 0.35, p = .553,p2 = .002. However, given the theoretically-based prediction that the
state-oriented group was more likely to show an intention-superiority effect than the actionoriented group (e.g., Beckmann, 1994; Kuhl & Goschke, 1994), simple-effects analyses were
conducted (Rosnow & Rosenthal, 1989). Results showed a significant intention-superiority
effect for the state-oriented group, with faster RTs for prospective (M = 526.91 ms, SD = 91.80)
than neutral words (M = 540.25 ms, SD = 95.81), F(1, 84) = 5.89, p = .017, p2 = .070. In
contrast, the intention-superiority effect for the action-oriented group was only marginally
significant, F(1, 105) = 3.71, p = .057, p2 = .032, with no significant difference between RTs for
prospective words (M = 538.22 ms, SD = 93.33) and neutral words (M = 547.70, SD = 92.15).
Therefore, both orientation groups showed a nominal intention-superiority effect, but this effect
was only significant in the state-oriented group. In addition, the interaction between cue source
and script was not significant, F(1, 187) = 0.18, p = .671,p2 = .001, indicating that the self-cued
and experimenter-cued groups did not differ on the intention-superiority effect. No other twoway interactions were predicted, and none were found.
The Orientation X Cue source X Script interaction was significant, F(1, 187) = 5.14, p =
.025,p2 = .027. Results showed an intention-superiority effect for the state-oriented group in
the self-cued condition (neutral M = 550.98 ms, SD = 114.41; prospective M = 531.22 ms, SD =
Intention-superiority effect
13
98.10), F(1, 44) =7.90, p = .012,p2 = .152, but not the experimenter-cued condition (neutral M
= 528.17 ms, SD = 68.72; prospective M = 522.05 ms, SD = 85.13), F(1, 39) = 0.59, p = .447,
p2 = .015 (see Figure 2). The action-oriented group showed the opposite pattern, with an
intention-superiority effect in the experimenter-cued condition (neutral M = 556.34 ms, SD =
102.18; prospective M = 537.84 ms, SD = 95.76), F(1, 57) = 7.75, p = .007, p2 = .119, but not
the self-cued condition (neutral M = 537.25 ms, SD = 78.12; prospective M = 538.67 ms, SD =
91.32), F(1, 47) = 0.04, p = .842, p2 = .001 (see Figure 2) 7.
Postperformance Intention Activation
Preliminary analyses. First, effects of block and gender were tested with
postperformance intention-superiority effect as the dependent variable. Therefore, a 2 X 2 X 2 X
2 (Gender X Block X Orientation X Cue source) ANOVA was conducted. Neither the main
effect of block, F(1, 174) = 1.05, p = .308, p2 = .006, nor the main effect of gender, F(1, 174) =
0.08, p = .783, p2 = .001, was significant. An unanticipated Block X Gender interaction was
found, F(1, 174) = 5.97, p = .016, p2 = .033. Post hoc analyses were conducted with RT as the
dependent variable to find out which combinations of the block and gender variables showed a
postperformance intention-superiority effect. The effect occurred for females in Block 2 (neutral
M = 523.89 ms, SD = 78.58; prospective M = 508.93 ms, SD = 67.56), F(1, 56) = 6.98, p = .011,
p2 = .111, and for males in Block 1 (neutral M = 569.21 ms, SD = 97.51; prospective M =
543.79 ms, SD = 79.94), F(1, 23) = 6.05, p = .022, p2 = .208. However, the effect did not occur
for females in Block 1 (neutral M = 546.63 ms, SD = 84.03; prospective M = 542.01 ms, SD =
81.19), F(1, 72) = 0.92, p = .341, p2 = .013, or for males in Block 2 (neutral M = 535.54 ms, SD
= 78.68; prospective M = 540.73 ms, SD = 83.90), F(1, 36) = 0.46, p = .502, p2 = .013. Because
there has been no previous research on gender differences in the postperformance intentionsuperiority effect, this novel finding is difficult to interpret and will not be discussed further. For
the dependent variable postperformance intention-superiority effect, none of the other
interactions involving block or gender were significant. That is, there were no two-way
interactions (Block X Cue source: F(1, 174) = 1.72, p = .191,p2 = .010; Block X Orientation:
F(1, 174) = 1.32, p = .252,p2 = .008; Gender X Cue source: F(1, 174) = 0.11, p = .740,p2 =
.001; Gender X Orientation: F(1, 174) = 0.98, p = .324,p2 = .006). There were also no threeor four-way interactions (Block X Gender X Cue source: F(1, 174) = 0.11, p = .742,p2 = .001;
Block X Gender X Orientation: F(1, 174) = 1.39, p = .239,p2 = .008; Gender X Cue source X
Orientation: F(1, 174) = 0.82, p = .366,p2 = .005; Block X Gender X Cue source X
Orientation: F(1, 174) = 0.93, p = .338,p2 = .005).
Main analyses. Unless stated otherwise, the dependent variable was RT. A 2 X 2 X 2
(Orientation X Cue source X Script) ANOVA was conducted. The effects of interest were the
main effect of script, the interaction between orientation and script, the interaction between cue
source and script, and the three-way interaction between orientation, cue source, and script.
There was a main effect of script with faster RTs for prospective (M = 532.12 ms, SD = 78.62)
than neutral (M = 540.53 ms, SD = 83.82) script words, F(1, 187) = 6.85, p = .010,p2 = .035.
Therefore, there was no evidence for a postcompletion inhibition effect for intentions but there
was evidence for a maintained intention-superiority effect. There were no other main effects.
The interaction between orientation and script was not significant, F(1, 187) = 0.22, p =
.638,p2 = .001). However, based on specific theoretically based predictions for intention
accessibility patterns for the two orientation groups (e.g., Beckmann, 1994; Kuhl & Goschke,
1994), simple-effects analyses were conducted (Rosnow & Rosenthal, 1989). These results
showed a significant maintained intention-superiority effect for the state-oriented group, with
Intention-superiority effect
14
faster RTs for prospective (M = 525.80 ms, SD = 77.83) than neutral words (M = 535.59 ms, SD
= 85.02), F(1, 84) = 4.06, p = .047, p2 = .048. In contrast, the intention-superiority effect for
the action-oriented group was only marginally significant, F(1, 105) = 2.83, p = .095, p2 = .026,
with no significant difference between RTs for prospective words (M = 537.18 ms, SD = 79.25)
and neutral words (M = 544.50, SD = 83.03). In summary, both orientation groups showed a
postperformance intention-superiority effect of roughly the same size although the effect was
slightly larger, and significant, in the state-oriented group.
The second two-way interaction of interest, the interaction between cue source and script,
was not significant, F(1, 187) = 1.69, p = .195,p2 = .009. Thus, the self-cued and experimentercued groups did not differ on the postperformance intention-superiority effect. No other twoway interactions were predicted, and none were found. Finally, the three-way interaction
between orientation, cue source, and script was also not significant, F(1, 187) = 0.94, p =
.759,p2 = .001.
Comparison of preperformance and postperformance intention-superiority effects. To
test whether the intention-superiority effect was the same size before and after script
performance, a 2 X 2 (Pre/post X Script) ANOVA was conducted with RT as the dependent
variable. The interaction was not significant, F(1, 190) = 0.31, p = .578, p2 = .02, establishing
that the size of the intention-superiority effect did not decrease after performance.
Discussion
The current experiment investigated the intention-superiority effect, which was measured
as greater accessibility for intention-related words than for words to be remembered. Previous
studies have shown a general intention-superiority effect before task performance. However,
there has been less research examining potential moderating variables or intention activation
after performance.
Preperformance Intention Activation
An overall intention-superiority effect was found for preperformance RTs. This result
adds to the growing body of literature that has demonstrated greater accessibility for intentionrelated contents than for other memory contents (e.g., Freeman & Ellis, 2003a; Marsh et al.,
1998).
However, based on action control theory (e.g., Goschke & Kuhl, 1993; Kuhl &
Beckmann, 1994b) and previous research (Goschke & Kuhl, 1993; Penningroth 2005b), a larger
intention-superiority effect was predicted for the state- than the action-oriented group. There
was only weak support for this prediction, with simple-effects analyses revealing slightly
different outcomes for these two personality orientations. Specifically, an intention-superiority
effect was observed in the state-oriented group, but this effect was only marginally significant in
the action-oriented group.
Why was this group difference smaller than in past studies (Goschke & Kuhl, 1993;
Penningroth, 2005b)? One explanation might involve methodological differences across studies,
including differences in how the intention-superiority effect was measured. In the current study,
the effect was measured with a lexical decision task, which presumably assessed activation in a
more implicit way than the recognition and recall tasks used by Goschke and Kuhl (1993) and
Penningroth (2005b). In a set of unpublished experiments reported in Goschke and Kuhl (1996),
state- and action-oriented groups showed an equally large intention-superiority effect when an
implicit measure (word-fragment completion) was used. Therefore, the relatively small group
difference found in the current study might be attributed to the use of a relatively implicit
measure of accessibility. There is scant research on implicit measures of the intention-
Intention-superiority effect
15
superiority effect, but research on this type of effect would have important theoretical and
practical implications.
Another possible explanation concerns the conditions that promote a state-oriented mode
of control in state-oriented individuals (Kuhl, 1981; Kuhl & Kazen-Saad, 1988). For example,
both situational and dispositional state-orientation might be needed to produce a large intentionsuperiority effect. Past research has shown that various types of state-oriented responses require
both dispositional state orientation and situational factors that promote state orientation, such as
thinking about past failures or about current emotional states (Kuhl, 1981; Kuhl & Kazen-Saad,
1988). Therefore, it might be useful to replicate the current experiment with an added
manipulation of situational state orientation to test this possibility.
In addition, a general orientation group effect might not be expected if the effect
depended on cue source. In Goschke and Kuhl (1993), state-oriented groups showed a larger
intention-superiority effect than action-oriented groups in the three experiments in which
participants expected an external cue. However, both action- and state-oriented groups showed
an intention-superiority effect when participants had to cue themselves after 15 minutes. In the
current study, the interaction between orientation and cue source occurred within a single
experiment. However, the state-oriented group showed an intention-superiority effect in the selfcued condition but not the experimenter-cued condition. The action-oriented group showed an
intention-superiority effect in the experimenter-cued condition but not the self-cued condition.
Thus, although the current results matched the result from Goschke and Kuhl (1993) for the
state-oriented group in the self-cued condition, they did not match for the other three conditions.
Several possible explanations for the interaction between orientation and cue source were
evaluated. Because these are post hoc explanations, they require further research to evaluate
their merits. First, considering that the experimenter-cued condition might represent an unnatural
situation for participants, one approach would be to focus on comparing the state- and actionoriented groups in the self-cued condition, as the more ecologically valid condition. If this is
done, the predicted orientation group difference is found: a larger intention-superiority effect for
the state- than action-oriented group. Also, this result replicates the orientation group difference
Penningroth (2005a) found using real-life intentions, which were most likely self cued, not
directed by another person. Additionally, focusing on the results for the self-cued condition
allows measurement of the intention-superiority effect when the intent or prospective component
of the prospective memory task is included (Einstein & McDaniel, 1996). That is, although both
cue conditions (self cued and experimenter cued) required maintaining the content or
retrospective component of the scripts to be performed, it might be argued that only the self-cued
condition required maintaining the prospective component (i.e., remembering that there was
something to do at a specific time). Graf and Uttl (2001) have advocated for increased research
on this aspect of prospective memory, which they term “prospective memory proper,” as a
potentially unique aspect of cognitive functioning and a feature likely to show moderating effects
of experimental manipulations and subject variables. Therefore, adopting this rationale and
focusing on the self-cued condition leads to the conclusion that only state-oriented individuals
show an intention-superiority effect, at least when the definition of prospective memory task is
restricted to situations that include prospective memory proper8.
Second, it is possible that the preperformance intention-superiority effect is larger when
there are high working memory (WM) demands. This explanation is supported by research that
showed that a higher WM load caused a larger intention-superiority effect (Penningroth, Graf, &
Gray, 2010). In that study, the intention-superiority effect was assessed in the same way as in
the current study. However, during the lexical decision task, central executive WM load was
manipulated by adding either a random number generation task (high load) or an articulatory
Intention-superiority effect
16
suppression task (lower load) (Marsh & Hicks, 1998; Teasdale et al., 1995). Therefore, an
experimentally induced high WM caused a larger intention-superiority effect. Additional
analyses from the current study also fit with this second explanation. That is, conditions that
showed a preperformance intention-superiority effect (i.e., self-cueing for the state-oriented
group and experimenter-cueing for the action-oriented group) also showed nominally higher RTs
for nonscript items (i.e., RTs for nonwords and nonscript words) as well as nominally higher
overall RTs.
This second explanation only accounts for the interaction pattern observed if the cue
conditions differentially affected WM load in state- and action-oriented groups. In fact, for stateoriented individuals, there are theoretical reasons to predict that self-cueing might cause a higher
WM load than experimenter-cueing (Kuhl, 1994a). For example, in the current study, selfinitiated performance can be thought of as requiring two different intentions, to perform the
intended script and to keep track of the time elapsed in order to perform on time. Compared to
action-oriented individuals, state-oriented individuals have more difficulty coordinating two
goals (Kuhl, 1994a) and therefore might experience a higher WM load with self-cueing
expectations (Kuhl & Helle, 1994). It is less clear why action-oriented individuals would
experience a high WM load in the experimenter-cueing condition. Perhaps this unnatural cueing
expectation (to be cued by another person) caused a higher WM load in action-oriented
individuals. Alternatively, it might have been the easy nature of this task condition that actually
led to higher RTs and an increased intention-superiority effect in the action-oriented group.
According to Kuhl and Beckmann (1994a), action-oriented individuals have shown “stateoriented” response patterns (e.g., in ERP data and negative priming in a Stroop task) when
conditions were relaxing and not challenging. The experimenter-cued condition might have
lacked sufficient challenge for action-oriented individuals and therefore led to a situational state
orientation that included persistence of declarative representations for unfulfilled intentions. In
summary, although these explanations are post hoc and speculative, they do suggest interesting
avenues for future research.
Postperformance Intention Activation
Results showed a general intention-superiority effect after script performance. That is,
RTs for prospective scripts were faster than for neutral scripts, indicating sustained intention
activation. These results provide the first evidence of sustained activation with a postponed
intention paradigm. In addition, the postperformance intention-superiority effect was just as
large as the preperformance effect. Previous studies have reported (1) an inhibition effect, (2)
persisting activation, or (3) a return to baseline activation. The current results are consistent with
studies that found persisting activation (e.g., Freeman & Ellis, 2003b; Jahn & Engelkamp, 2003;
Nowinski, et al., 2005) or a return to baseline levels (e.g., Cohen et al., 2005). However, other
studies have reported a postperformance intention inhibition effect (e.g., Badets et al., 2006;
Freeman & Ellis, 2003a). For example, Marsh et al. (1998) found a reversal in relative activation
levels after intention completion (i.e., slower lexical decision task RTs for prospective than
neutral words).
In trying to understand these discrepant results, it might be informative to directly
compare the current study to Marsh et al. (1998) because there were many similarities in the
procedures and materials used. In both studies, postperformance activation was measured with
lexical decision latencies. However, in the current study, participants had an added three-minute
delay (for the anagram task) between the instructions designating the script to perform and script
performance. In contrast, in Marsh et al. (1998), participants received performance instructions
and then immediately performed the intended script. Therefore, one explanation for differing
results might hinge on the delay: When the intention is held longer, people do not show an
Intention-superiority effect
17
inhibition effect just after performance, but instead show sustained activation. This
interpretation is also supported by the results from Nowinski et al. (2005). In that study,
participants held intentions for a long period (192 trials of semantic matching), and results
showed persisting activation after performance. This explanation could be evaluated by
systematically varying the retention interval for intentions and then observing the effects on
postperformance activation.
Another possible explanation for different postperformance outcomes involves the
opportunity to deactivate the intention. Based on results from several studies, Beckmann (1994)
concluded that deactivation does not necessarily occur automatically. Instead, deactivation
might benefit from, and perhaps require, deliberate postactional processing. For example,
completed intentions showed greater activation (less deactivation) when postactional processing
was prevented with a 15-s counting backwards task than when postactional processing was
allowed (Beckmann, 1994). In fact, a longer period of postcompletion free time was associated
with greater deactivation of the intention. Beckmann did not report whether deactivated
intentions showed an inhibition effect or simply returned to baseline activation levels. However,
a deliberate processing mechanism that deactivates intentions could potentially explain all three
outcomes (inhibition, sustained activation, or return to baseline), with the specific outcome being
partly dependent on the amount of time and cognitive resources available after completing the
task. More cognitive “free time” might lead to greater deactivation, and perhaps even an
inhibition effect. Unfortunately, this explanation can not be evaluated as a reason for the
different outcomes observed in the current study and in Marsh et al. (1998) because the exact
nature of the postperformance interval was not controlled in either study. Also, the deactivation
patterns reported by Beckmann followed a performance failure, so research is needed to test
whether the same patterns follow performance success. Specifically, a worthwhile task for
future research would be to manipulate the attentional resources and time available for intention
deactivation after task completion in order to map out the time-course and final level of
activation.
In examining postperformance intention activation, two possible moderators were tested:
cue source and state versus action orientation. In short, cue source (self vs. experimenter) did
not affect the intention-superiority effect after task performance. This finding replicated the null
effect reported in the one past relevant study (Freeman & Ellis, 2003b). Based on action control
theory (e.g., Beckmann, 1994; Goschke & Kuhl, 1993; Kuhl & Beckmann, 1994b), a larger
intention-superiority effect was predicted for state-oriented than action-oriented individuals.
Results showed weak support for this prediction. That is, simple-effects analyses revealed
slightly different outcomes for the two orientation groups. In particular, the state-oriented group
showed a sustained intention-superiority effect after task performance, but for the action-oriented
group, this effect was only marginally significant. These effects are congruent with descriptions
of these two personality types (e.g., Kuhl & Goschke, 1994, Beckmann, 1994): Action-oriented
individuals can adapt activation levels to the current context, but state-oriented individuals show
“context-blind maintenance” (Beckmann, 1994, p. 160). In the current study, the group
difference in deactivation was small. However, when combined with similar results reported by
Beckmann (1994), this set of findings provides provisional support for these theoretical
depictions of state- versus action-orientation.
Conclusions
In conclusion, the current findings extend research on the intention-superiority effect in
several ways. A general preperformance intention-superiority effect was replicated. However,
moderating variables were also examined. Although the intention-superiority effect was
predicted to be larger for individuals with dispositional state orientation than dispositional action
Intention-superiority effect
18
orientation, there was only weak support for this prediction. The intention-superiority effect also
depended on the interaction of cue source and state versus action orientation. In the stateoriented group, participants showed an intention-superiority effect when they expected
performance to be self-cued, but not when they expected performance to be experimenter-cued.
In the action-oriented group, participants showed an intention-superiority effect when
participants expected performance to be experimenter-cued but not when they expected
performance to be self-cued. In addition, a general postperformance intention-superiority effect
was demonstrated, providing the first evidence of sustained activation with a postponed intention
paradigm. Although both orientation groups showed persisting activation, the effect was slightly
larger in the state-oriented group.
Given the accumulating evidence that pending intentions show heightened accessibility,
two related issues would benefit from further research. First, heightened accessibility of an
intention does not necessarily translate into increased task performance (e.g., Freeman & Ellis,
2003a; Graf & Mandler, 1984; Penningroth, 2005a). More research is needed to understand
other variables that impact the relationship between activation level and performance, such as
attentional demands, characteristics of the performance cue, individual difference variables, and
even motivational factors. Second, other researchers have suggested a careful analysis of the
conditions under which the intention-superiority effect would be adaptive or maladaptive (e.g.,
Goschke & Kuhl, 1993). For example, heightened activation might improve prospective
remembering under some conditions (Freeman & Ellis, 2003a; Kvavilashvili, 1987). However,
heightened activation of a delayed intention might compromise performance on an intervening
task (e.g., Kuhl, 1994a; Kuhl & Goschke, 1994; Kuhl & Helle, 1994).
Intention-superiority effect
19
References
Anderson, J. R. (1983). The Architecture of Cognition. Cambridge: Harvard University Press.
Badets, A., Blandin, Y., Bouquet, C. A., & Shea, C. H. (2006). The intention superiority effect
in motor skill learning. Journal of Experimental Psychology: Learning, Memory, and
Cognition, 32, 491 – 505.
Beckmann, J. (1994). Volitional correlates of action versus state orientation. In J. Kuhl and J.
Beckmann (Eds.), Volition and Personality: Action versus State Orientation (pp.155 166). Seattle, WA: Hogrefe & Huber.
Cohen, A.-L., Dixon, R. A., & Lindsay, D. S. (2005). The intention interference effect and
aging: similar magnitude of effects for young and old adults. Applied Cognitive
Psychology, 19, 1177 – 1197.
Cohen, R. L. (1981). On the generality of some memory laws. Scandinavian Journal of
Psychology, 22, 267 – 281.
Einstein, G. O., & McDaniel, M. A. (1990). Normal aging and prospective memory. Journal of
Experimental Psychology: Learning, Memory, and Cognition, 16, 717-726.
Einstein, G. O., & McDaniel, M. A., (1996). Retrieval processes in prospective memory:
Theoretical approaches and some new empirical findings. In M. Brandimonte, G. O.
Einstein, and M. A. McDaniel (Eds.), Prospective memory: Theory and applications
(pp.115 - 141). Mahwah, NJ: Lawrence Erlbaum Associates, Inc.
Ellis, J. (1996). Prospective memory or the realization of delayed intentions: A conceptual
framework for research. In M. Brandimonte, G. O. Einstein, and M. A. McDaniel (Eds.),
Prospective memory: Theory and applications (pp. 1 - 22). Mahwah, NJ: Lawrence
Erlbaum Associates, Inc.
Eschen, A., Freeman, J., Dietrich, T., Martin, M., Ellis, J., Martin, E., et al. (2007). Motor brain
regions are involved in the encoding of delayed intentions: A fMRI study. International
Journal of Psychophysiology, 64, 259 - 268.
Field, A. (2005). Discovering Statistics Using SPSS (2nd ed.). London: Sage.
Francis, W. N., Kučera, H., & Mackie, A. W. (1982). Frequency Analysis of English Usage:
Lexicon and Grammar. Boston: Houghton Mifflin.
Freeman, J. E., & Ellis, J. A. (2003a). The intention-superiority effect for naturally occurring
activities: The role of intention accessibility in everyday prospective remembering in
young and older adults. International Journal of Psychology, 38, 215 – 228.
Freeman, J. E., & Ellis, J. A. (2003b). The representation of delayed intentions: A prospective
subject-performed task? Journal of Experimental Psychology: Learning, Memory, and
Cognition, 29, 976 – 992.
Goschke, T., & Kuhl, J. (1993). Representation of intentions: Persisting activation in memory.
Journal of Experimental Psychology: Learning, Memory, and Cognition, 19, 1211 –
1226.
Goschke, T., & Kuhl, J. (1996). Remembering what to do: Explicit and implicit memory for
intentions. In M. Brandimonte, G. O. Einstein, and M. A. McDaniel (Eds.), Prospective
Memory: Theory and Applications (pp. 53 - 91). Mahwah, NJ: Lawrence Erlbaum
Associates, Inc.
Graf, P., & Mandler, G. (1984). Activation makes words more accessible, but not necessarily
more retrievable. Journal of Verbal Learning and Verbal Behavior, 23, 553 – 568.
Graf, P., & Uttl, B. (2001). Prospective memory: A new focus for research. Consciousness and
Cognition, 10, 437 – 450.
Jahn, P., & Engelkamp, J. (2003). Design-effects in prospective and retrospective memory for
actions. Experimental Psychology, 50, 4 – 15.
Intention-superiority effect
20
Kliegel, M., Martin, M., McDaniel, M. A., & Einstein, G. O. (2002). Complex prospective
memory and executive control of working memory: A process model. Psychologische
Beiträge, 44, 303 – 318.
Klinger, E., & Murphy, M. D. (1994). Action orientation and personality: Some evidence on
the construct validity of the Action Control Scale. In J. Kuhl and J. Beckmann (Eds.),
Volition and personality: Action versus state orientation (pp. 79 - 92). Seattle, WA:
Hogrefe & Huber.
Koriat, A., Ben-Zur, H., & Nussbaum, A. (1990). Encoding information for future action:
Memory for to-be-performed tasks versus memory for to-be-recalled tasks. Memory &
Cognition, 18, 568 – 578.
Kuhl, J. (1981). Motivational and functional helplessness: The moderating effect of state
versus action orientation. Journal of Personality and Social Psychology, 40, 155 – 170.
Kuhl, J. (1994a). A theory of action and state orientation. In J. Kuhl and J. Beckmann (Eds.),
Volition and Personality: Action versus State Orientation (pp.9 - 46). Seattle, WA:
Hogrefe & Huber.
Kuhl, J. (1994b). Action versus state orientation: Psychometric properties of the Action
Control Scale (ACS – 90). In J. Kuhl and J. Beckmann (Eds.), Volition and Personality:
Action versus State Orientation (pp. 47 - 59). Seattle, WA: Hogrefe & Huber.
Kuhl, J., & Beckmann, J. (1994a). Introduction: Action versus state orientation in the context
of personality and volition. In J. Kuhl and J. Beckmann (Eds.), Volition and Personality:
Action versus State Orientation (pp. 1 - 5). Seattle, WA: Hogrefe & Huber.
Kuhl, J., & Beckmann, J. (1994b). Volition and Personality: Action versus State Orientation.
Seattle, WA: Hogrefe & Huber.
Kuhl, J., & Goschke, T. (1994). State orientation and the activation and retrieval of intentions in
memory. In J. Kuhl and J. Beckmann (Eds.), Volition and Personality: Action versus
State Orientation (pp. 127 – 153). Seattle, WA: Hogrefe & Huber.
Kuhl, J., & Helle, P. (1986). Motivational and volitional determinants of depression: The
degenerated-intention hypothesis. Journal of Abnormal Psychology, 95, 247 – 251.
Kuhl, J., & Helle, P. (1994). Motivational and volitional determinants of depression: The
degenerated-intention hypothesis. In J. Kuhl and J. Beckmann (Eds.), Volition and
Personality: Action versus State Orientation (pp. 283 - 296). Seattle, WA: Hogrefe &
Huber.
Kuhl, J., & Kazén-Saad, M. (1988). A motivational approach to volition: Activation and deactivation of memory representations related to uncompleted intentions. In V. Hamilton,
G. H. Bower, & N. H. Frijda (Eds.), Cognitive Perspectives on Emotion and Motivation
(pp. 63 – 85). Dordrecht: Kluwer Academic Publishers.
Kvavilashvili, L. (1987). Remembering intention as a distinct form of memory. British Journal
of Psychology, 78, 507 – 518.
Marsh, R. L., Cook, G. I., Meeks, J. T., Clark-Foos, A., & Hicks, J. L. (2007). Memory for
intention-related material presented in a to-be-ignored channel. Memory & Cognition,
35, 1197 – 1204.
Marsh, R. L., & Hicks, J. L. (1998). Event-based prospective memory and executive control of
working memory. Journal of Experimental Psychology: Learning, memory, and
Cognition, 24, 336 – 349.
Marsh, R. L., Hicks, J. L., & Bink, M. L. (1998). Activation of completed, uncompleted, and
partially completed intentions. Journal of Experimental Psychology: Learning, Memory,
and Cognition, 24, 350 - 361.
Intention-superiority effect
21
Marsh, R. L., Hicks, J. L., & Bryan, E. S. (1999). The activation of unrelated and canceled
intentions. Memory & Cognition, 27, 320 –327.
Marsh, R. L., Hicks, J. L., & Watson, V. (2002). The dynamics of intention retrieval and
coordination of action in event-based prospective memory. Journal of Experimental
Psychology: Learning, Memory, and Cognition, 28, 652 - 659.
Maylor, E. A., Chater, N., & Brown, G. D. A. (2001). Scale invariance in the retrieval of
retrospective and prospective memories. Psychonomic Bulletin & Review, 8, 162 – 167.
Maylor, E. A., Darby, R. J., & Della Sala, S. (2000). Retrieval of performed versus to-beperformed tasks: A naturalistic study of the intention-superiority effect in normal aging
and dementia. Applied Cognitive Psychology, 14, S83 – S98.
Nowinski, J. L., Holbrook, J, & Dismukes, R. K. (2005, May). Effects of lingering intentions on
subsequent tasks. Poster presented at the American Psychological Society Conference,
Los Angeles, CA.
Penningroth, S. L. (2005a). Effects of attentional demand, cue typicality, and priming on an
event-based prospective memory task. Applied Cognitive Psychology, 19, 885 – 897.
Penningroth, S. L. (2005b). Free recall of everyday retrospective and prospective memories:
The intention-superiority effect is moderated by action vs. state orientation and by
gender. Memory, 13, 711 - 724.
Penningroth, S. L., Graf, P., & Gray, J. (2010). The effect of a working memory load on the
intention-superiority effect: Examining features of automaticity. Manuscript submitted
for publication.
Rosnow, R. L., & Rosenthal, R. (1989). Statistical procedures and the justification of
knowledge in psychological science. The American Psychologist, 44, 1276 – 1284.
Tabachnick, B. G., & Fidell, L. S. (1996). Using Multivariate Statistics (3rd ed.). New York:
HarperCollins College Publishers.
Teasdale, J. D., Dritschel, B. H., Taylor, M. J., Proctor, L., Lloyd, C. A., Nimmo-Smith, I., et al.
(1995). Stimulus-independent thought depends on central executive resources. Memory
& Cognition, 23, 551 – 559.
Intention-superiority effect
Author Note
Portions of these data were reported at the second International Conference on
Prospective Memory, Zurich, Switzerland, July, 2005.
22
Intention-superiority effect
23
Footnotes
1. The original sample included 218 participants, but the data for 27 participants were not
included (8 participants either could not learn the scripts or understand instructions, 11
participants forgot which script to perform, and 2 participants were omitted because of
experimenter or computer error).
2. This cutoff value was reinforced by analysing a large unpublished set of scores (N = 1063) on
the prospective (AOD) subscale of the ACS that represented data from five studies run in our
lab. The median for that large sample was also seven. Also, scores of 7 (at the median) were
categorised as action oriented instead of state oriented because this created a more balanced
design and score ranges more similar to Goschke and Kuhl (1993).
3. These instructions were included for three reasons: (a) in pilot testing, some participants had
difficulty memorising the scripts, (b) in pilot testing, some participants reported using these
strategies, so standardising instructions controlled for this possible confound, and (c) Goschke
and Kuhl (1993, Exp. 4) found that visualising both scripts (prospective and neutral) did not
decrease the intention-superiority effect.
4. On average, self-cued participants stopped themselves after 3 min,13 s (SD = 36 s).
5. The main analyses were also done separately within each block to see whether the intentionsuperiority effect patterns (e.g., the Orientation X Cue interaction effect on the intentionsuperiority effect) differed across blocks. However, the patterns did not change from Block 1 to
2.
6. Using a Stroop task, Cohen et al. (2005) found that the intention-superiority effect was not
evident for the first stimulus participants encountered but was evident for subsequent stimuli.
Therefore, in the current study, the lexical decision latency difference for the first prospectiveneutral word pair was compared to the mean latency for later pairs. However, there was no
difference in the intention-superiority effect (in preperformance or postperformance blocks).
7. Following the recommendation of an anonymous reviewer, I also tested the three-way
interaction after redefining state- and action- oriented groups as the lower and upper thirds of the
ACS score distribution (rather than with a median split). The pattern of results was replicated
(i.e., the same pattern as in figure 2), but two effects were no longer statistically significant
(presumably because of reduced power). That is, the omnibus interaction effect was no longer
significant, and the intention-superiority effect for the action-oriented group in the experimentercued condition became only marginally significant. For the other three comparisons, the results
replicated those from the larger sample.
8. I would like to thank an anonymous reviewer for this suggestion.
Intention-superiority effect
Figure Captions
Figure 1. Order of tasks for block order 1.
Figure 2. Preperformance mean RTs, in ms, as a function of script (prospective memory (PM)
vs. neutral), orientation (state-oriented vs. action-oriented), and cue source (self vs.
experimenter). Error bars represent one standard error above and below the mean.
24
BLOCK 1
Learn
scripts
Told
which
to perform
LDT
BLOCK 2
Anagrams
Perform
script
Learn
scripts
Told
which
to perform
Anagrams
Perform
script
LDT
Download