ArticlePDF Available

Abstract

Our actions cause manifold environmental changes. Monitoring these action effects serves at least two vital functions: While the validation of currently relevant effects assesses goal-achievement, screening for currently irrelevant effects accumulates knowledge about potential action-effect relationships. However, monitoring the perceptual consequences of our actions presumably impairs performance in concurrent tasks. Here, we investigated how effect relevance modulates monitoring costs by manipulating instructions in three dual-task experiments. We found performance decreases not only after validation of goal-relevant action effects but to a smaller extent also after screening of goal-irrelevant action effects. These results suggest that effect monitoring is a rather fundamental limitation of dual tasking.
Acta Psychologica 224 (2022) 103522
0001-6918/© 2022 Published by Elsevier B.V. This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).
Monitoring goal-irrelevant effects interferes with concurrent tasks
Moritz Schaaf
*
, Wilfried Kunde, Robert Wirth
Department of Psychology, Julius Maximilians University of Würzburg, R¨
ontgenring 11, 97070 Würzburg, Germany
ARTICLE INFO
Keywords:
Cognitive exibility
Action effects
Effect monitoring
Cognitive control
Psychological refractory period
ABSTRACT
Our actions cause manifold environmental changes. Monitoring these action effects serves at least two vital
functions: While the validation of currently relevant effects assesses goal-achievement, screening for currently
irrelevant effects accumulates knowledge about potential action-effect relationships. However, monitoring the
perceptual consequences of our actions presumably impairs performance in concurrent tasks. Here, we investi-
gated how effect relevance modulates monitoring costs by manipulating instructions in three dual-task experi-
ments. We found performance decreases not only after validation of goal-relevant action effects but to a smaller
extent also after screening of goal-irrelevant action effects. These results suggest that effect monitoring is a rather
fundamental limitation of dual tasking.
1. Introduction
Humans change their perceptions by their actions. While some per-
ceptions are intended consequences of the action (i.e., they are relevant,
they are goals), other perceptual changes might occur as a foreseeable
byproduct (i.e., they are irrelevant, they are not goal-related). For
example, when writing a text, the letters on the screen are relevant, they
are the typist's goal. In contrast, the sound of the keys or the feeling of
the forearm moving underneath the sleeves are irrelevant, the typist
would consider the goal accomplished even without those perceptions.
Agents monitor the perceptual changes caused by their actions. This
monitoring of action effects serves at least two vital functions. First, by
processing the currently relevant effects, agents can assess whether the
intended and actual outcomes match (i.e., whether they achieved their
goals, Miller et al., 1960). This has been called the validation function
(Wirth, Janczyk, & Kunde, 2018). Second, by processing currently
irrelevant effects, agents can acquire new action-effect links (Elsner &
Hommel, 2001). If a system is monitoring presently goal-unrelated ef-
fects, it can detect reliable co-occurrences of own behavior and envi-
ronmental consequences that might become relevant later. This has been
called the screening function. While the validation of action-effect links
presumes goals that the produced effects are compared with, screening
for action-effect relations could potentially take place constantly.
Recent evidence shows that monitoring of action effects interferes
with concurrent tasks. Consider a dual-tasking study by Wirth, Janczyk,
and Kunde (2018). In Task 1, participants were instructed to produce an
object (i.e., an action effect) on a screen by pressing a key. During the
display of this action effect, the imperative stimulus for Task 2 came up.
As the only temporal overlap between both tasks was the display of Task
1 effects, variations in the performance of Task 2 arguably had to
originate from the monitoring of these effects. The duration of the
proposed monitoring process was manipulated by instructing partici-
pants to produce effects that were spatially incompatible (rather than
compatible) to a keypress. Lengthening the effect monitoring process in
Task 1 delayed (Δ =39 ms, Exp. 1) responding in Task 2 (for a similar
inuence by delaying effect onset, see Kunde et al., 2018; Δ =30 ms,
Exp. 1). Several aspects of this delay have been scrutinized, like that it
stems from a postponed start of the second task (Wirth, Janczyk, &
Kunde, 2018), that it is inuenced by expectations (Wirth, Steinhauser,
et al., 2018), or that apparently the same neural system is engaged that is
also engaged in error monitoring (Steinhauser et al., 2018).
Yet, in almost all previous studies participants were explicitly
instructed to produce these effects on purpose, and hence, the monitored
action effects were goal relevant. This renders it unclear whether the
detrimental impact of effect monitoring on another concurrent task is
due to the validation or the screening function. Put differently, does
effect monitoring delay other tasks only when the action effect is
intended, and goal achievement is checked, or also when irrelevant side
effects are explored? It is hard to judge from existing literature which
function is most important. On one hand, there is evidence that task-
relevant feedback and task-irrelevant action contingent effects evoke
similar event-related potentials (Band et al., 2009). On the other hand,
visual action effects do not grab attention when they are completely
unpredictable (Kumar et al., 2015) and thus, their intentional
* Corresponding author.
E-mail address: moritz.schaaf@uni-wuerzburg.de (M. Schaaf).
Contents lists available at ScienceDirect
Acta Psychologica
journal homepage: www.elsevier.com/locate/actpsy
https://doi.org/10.1016/j.actpsy.2022.103522
Received 19 April 2021; Received in revised form 11 January 2022; Accepted 25 January 2022
Acta Psychologica 224 (2022) 103522
2
production cannot be a goal of the actor (Hommel & Wiers, 2017).
To disentangle these functions, we conducted three dual-tasking
experiments. In the rst task, manual actions produced visual effects
that varied either regarding the compatibility to the action (Exp. 1) or
expectancy (Exp. 2 and 3). The relevance of these effects was manipu-
lated by altering the instructions. In one condition, participants were
asked to produce these effects as a goal. In another condition, they were
told that the effects are irrelevant byproducts of the task. We assumed
that the monitoring of action effects draws on scarce attentional re-
sources, thereby interfering with the processing of a concurrent task
(Pashler, 1994; Welford, 1952, 1967). Consequently, the second task
served to measure the ongoing monitoring of these effects. The central
question was whether the effect features of the rst task would differ-
entially impact the performance in the second task, depending on the
instructions.
2. Experiment 1
To test the potential contribution of effect relevance on previously
reported monitoring costs, we held the supercial aspects of stimuli and
responses constant but framed the task in two different ways. Partici-
pants responded to the color of a xation cross in a puzzle piece by a left
or right keypress (Task 1; see Fig. 1). This response added another puzzle
piece on the screen: either on the side of the keypress (compatible effect)
or on the respectively other side (incompatible effect). Shortly after the
presentation of this perfectly foreseeable effect, the imperative stimulus
for another choice reaction task came up (Task 2). We expected
incompatible action effects in Task 1 to lengthen effect monitoring and
thus to delay processing in Task 2, as observed before.
Crucially, instructions varied. In one condition, participants were
told to add a puzzle piece to the [left/right] side by pressing the [left/
right] key(Effect-Instruction). In another condition, they were told to
press the [left/right] key, which will produce a task-irrelevant [left/
right] puzzle piece(Response-Instruction). Hence, while the tasks were
supercially identical, the goal of Task 1 (and thereby the relevance of
the action effect) was manipulated via instructions. If only the validation
function (i.e., relevant effects) impaired subsequent performance,
spatially incompatible effects should impact Task 2 only with the Effect-
Instruction, leading to a statistical interaction between Task 1 instruc-
tion and Task 1 compatibility in Task 2 performance. If, however, the
screening function (i.e., all effects) caused monitoring costs, spatially
incompatible effects should impact Task 2 irrespective of the
instructions.
2.1. Participants
Forty-eight participants were recruited (M
age
=26.5 years, SD =
6.5), provided written informed consent and received monetary
compensation. This sample size allows for counterbalancing of block
order and at an alpha of .05, it provides a power of >.95 to detect
monitoring costs in Task 2, assuming effect sizes as found in previous
research (Wirth, Janczyk, & Kunde, 2018; Exp. 5, d
z
=0.55). One
participant was removed from the nal sample due to unusually slow
responses and was replaced.
2.2. Apparatus and stimuli
For Task 1, the stimuli were pictures of puzzle pieces with connectors
at both sides and a centrally presented colored xation cross. The color
of the xation cross (S1; red, green, blue, or yellow) required a middle
nger response (R1) on the Sor K keys of a QWERTZ keyboard. As
action effects (E1), these keypresses produced puzzle pieces at either the
left or the right side of S1 with the horizontal location in alignment with
the position of the Sand Kkeys on the keyboard. Hence, these action
effects were either spatially compatible (e.g., a right keypress producing
a puzzle piece on the right side) or spatially incompatible (e.g., a right
keypress producing a puzzle piece on the left side). Before each block,
participants were instructed about the spatial compatibility and rele-
vance of the puzzle piece in the next block (add a puzzle piece to the
[left/right] side by pressing the [left/right] keyor press the [left/
right] key which will produce a task-irrelevant [left/right] puzzle
piece).
For Task 2, participants had to categorize a letter (S2; Hvs. S). S2
was presented centrally below the puzzle piece in white font and
required an index nger response on the X or Mkey (R2).
2.3. Procedure
The onset of a puzzle piece with a grey xation cross marked the
beginning of a trial. After 500 ms, the xation cross changed color.
Depending on the task context, one of two colors appeared. The
Response-Instruction task context mapped two colors to specic re-
sponses (yellow =press the left button, blue =press the right button),
the Effect-Instruction task context mapped two colors to specic effects
(red =produce a left puzzle piece, green =produce a right puzzle piece).
Immediately after R1 was given, E1 appeared and stayed on screen
until the trial ended. If no response key was pressed 2000 ms after S1
target onset, the trial counted as an omission, no E1 appeared, and the
outline of the central puzzle piece turned red with onset of Task 2 as
error feedback.
S2 was presented 50 ms after E1 onset and called for R2. The two
tasks were always presented in this order with no temporal overlap
(except for the display of E1) between the tasks. Again, if no key was
pressed for 2000 ms, the trial counted as an omission. If both tasks were
answered correctly, the next trial started immediately after R2. Other-
wise, written feedback was presented at the end of the trial for 500 ms in
red color.
Participants completed 24 blocks consisting of 40 trials, with each
combination of the two possible S1 per block (red vs. green with the
Effect-Instruction; blue vs. yellow with the Response-Instruction) and
the two possible S2 (Hvs. S) presented ten times. Task context
(Effect-Instruction vs. Response-Instruction) and R1-E1 relationship
(compatible vs. incompatible) were manipulated within participants,
between blocks (6 consecutive blocks per combination). To counter-
balance the order of the four possible condition blocks between partic-
ipants, task context was counterbalanced between the rst and second
half of the experiment, and R1-E1 relationship was then counter-
balanced within each task context. Additionally, the S2-R2 mapping was
counterbalanced between participants.
2.4. Results
2.4.1. Data treatment
Raw data and analysis scripts for all experiments are publicly
available at osf.io/q48ay. For reaction time (RT) analyses, we excluded
Fig. 1. Trial procedure. For Task 1, participants responded to the color of the
xation cross by pressing the S or K key. This response added either a spatially
compatible or an incompatible puzzle piece. After this visual action effect, a
white letter appeared to which participants had to respond with the X or M key
(Task 2). This trial shows the Effect-Instruction (red xation cross) and the
incompatible mapping (right button press left effect). (For interpretation of
the references to color in this gure legend, the reader is referred to the web
version of this article.)
M. Schaaf et al.
Acta Psychologica 224 (2022) 103522
3
trials with errors (Task 1: 6.9%, Task 2: 10.6%). The remaining trials
were screened for outliers, and we removed trials in which RTs for Task
1 or Task 2 deviated more than 2.5 SDs from the corresponding cell
mean, computed separately for each participant and condition (5.6%).
The nal sample for RT analyses consisted of 79.9% of the original trials.
Data were analyzed via 2 ×2 ANOVAs with R1-E1 relationship
(compatible vs. incompatible) and task context (Effect-Instruction vs.
Response-Instruction) as within-subjects factors (see Fig. 2).
2.4.2. Task 1
2.4.2.1. RTs. Task 1 RTs were lower in Response-Instruction blocks
(502 ms) than in Effect-Instruction blocks (522 ms), F(1, 47) =7.54, p =
.009,
η
p2
=.14. No other inuences were observed, all Fs <1.
2.4.2.2. Error rates. Task 1 error rates were higher in Effect-Instruction
blocks (7.8%) than in Response-Instruction blocks (5.9%), F(1, 47) =
7.31, p =.010,
η
p2
=.13. Further, there were more errors in blocks with
an incompatible R1-E1 mapping (7.7%) than in blocks with a compat-
ible mapping (6.1%), F(1, 47) =7.07, p =.011,
η
p2
=.13. The inuences
of task instructions and compatibility of R-E mapping did not interact, F
<1.
2.4.3. Task 2
2.4.3.1. RTs. Task 2 RTs were not inuenced by the task context, F <1,
but were faster after compatible action effects (491 ms) than after
incompatible action effects (503 ms), F(1, 47) =10.60, p =.002,
η
p2
=
.18. The experimental manipulations did not interact, F(1, 47) =0.56, p
=.458,
η
p2
=.01.
2.4.3.2. Error rates. No inuences on Task 2 error rates were observed,
all Fs <2.32, all ps >.133.
2.5. Discussion
Exp. 1 addressed possible inuences of effect relevance on previously
observed effect monitoring costs. Task 1 required participants to
respond to the color of a xation cross and Task 2 had them categorize a
letter. In Task 1, participants produced a foreseeably spatially compat-
ible or incompatible effect with their response. While the demand of
monitoring E1 was varied by this compatibility of the R1-E1 relation-
ship, the reason for monitoring E1 was manipulated by altering the task
context via the instructions. Task 2 served to measure the monitoring
costs.
We found that the task context inuenced the ease of producing a
response to Task 1. While this could reect that the pursuit of two goals
is harder than the pursuit of only one goal (Hommel & Wiers, 2017;
Janczyk & Kunde, 2020), it could likewise reect the choice of xation
cross colors. Further, and in line with previous studies, producing a
foreseeable incompatible effect was more difcult (Kunde et al., 2012;
Pster et al., 2014; Wirth et al., 2015). Also replicating previous results,
responses in the unrelated Task 2 took longer after incompatible effects
than after compatible effects, reecting longer effect monitoring in the
rst task. Crucially, while this demand of monitoring inuenced Task 2
RTs, the relevance of the monitored effects had no detectable inuence
on Task 2 performance.
In Exp. 1 we increased monitoring costs by manipulating effect
compatibility. Effects that violate long-term links between actions and
effects (left actions rarely produce changes on the right side) take longer
to monitor. Exp. 2 aimed at replicating the main results of Exp. 1 by
another means to manipulate effect monitoring duration, by presenting
effects that did or did not violate action-effect links established in the
experiment itself.
3. Experiment 2
The instructions were varied as in Exp. 1, but responses in Task 1
were now mapped to effects that had no pre-experimental or natural
(spatial) relationship. Instead, one response made a stimulus grow,
whereas the alternative response made the stimulus shrink. This map-
ping was kept constant for participants during the experiment. Based on
previous research we expected that participants acquire these short-term
action-effect links even if the action effect is irrelevant to the task (Elsner
& Hommel, 2001). Occasionally, the R1-E1 mapping was violated: In
10% of the trials, responses produced effects that were usually produced
by the respectively other response. We expected that such violations
would lengthen effect monitoring and thus delay responding in the
second task.
3.1. Method
Forty-eight new participants (M
age
=27.5 years, SD =7.9) were
recruited. Apparatus, stimuli, and procedure were as in Exp. 1, but the
action effect was changed: R1 no longer added a spatially (in)compatible
puzzle piece; rather, the central puzzle piece was scaled up or down as
an action effect (E1). This R1-E1 relationship was no longer manipulated
between blocks, but on a trial-to-trial basis: In 90% of the trials, the
expected action effect was displayed, while in the remaining 10%, the
unexpected action effect was displayed. Task 2 remained unchanged.
Participants again completed 24 blocks consisting of 40 trials, with
each of the four possible combinations of S1 and S2 presented nine times
for expected and one time for unexpected action effects. Block order, R1-
E1 mapping and S2-R2 mapping were counterbalanced between
participants.
3.2. Results
The data were treated as in Exp. 1. After excluding trials with errors
(Task 1: 4.5%, Task 2: 9.6%) and outliers (5.4%), the nal sample for RT
analyses consisted of 82.6% of the original trials. The data were again
analyzed via 2 ×2 ANOVAs with R1-E1 relationship (expected vs. un-
expected) and task context (Effect-Instruction vs. Response-Instruction)
as within-subjects factors (see Fig. 3).
3.2.1. Task 1
3.2.1.1. RTs. Task 1 RTs were lower in Response-Instruction blocks
(484 ms) than in Effect-Instruction blocks (505 ms), F(1, 47) =11.44, p
=.001,
η
p2
=.20. No other inuences were observed, all Fs <1.75, all ps
>.192.
3.2.1.2. Error rates. No inuences on Task 1 error rates were observed,
all Fs <2.29, all ps >.137.
3.2.2. Task 2
3.2.2.1. RTs. Task 2 RTs were not inuenced by the task context, F(1,
47) =1.28, p =.264,
η
p2
=.03, but were faster after expected action
effects (508 ms) than after unexpected action effects (520 ms), F(1, 47)
=15.56, p <.001,
η
p2
=.25. The experimental manipulations did not
interact, F(1, 47) =1.73, p =.195,
η
p2
=.04.
3.2.2.2. Error rates. No inuences on Task 2 error rates were observed,
all Fs <1.
3.3. Discussion
The results of Exp. 2 replicate the key nding of Exp. 1: Again, Task 2
reaction times increased with higher effect monitoring demands in a
M. Schaaf et al.
Acta Psychologica 224 (2022) 103522
4
previous task, which this time originated from unexpected action effects.
Contrary to Exp. 1, participants now had to build this expectancy based
on newly emerging R1-E1 relationships. Despite this adjustment, we
again found no modulation of the monitoring costs by the task context,
as indicated by the nonsignicant interaction in Task 2 reaction times.
1
While in both Exp. 1 and Exp. 2 the action effects in the Response-
Instruction condition were evidently irrelevant for performing the
task, the Task 2 RTs show that they are evidently not irrelevant for task
performance. Hence, increased effect monitoring, whatever its cause,
affects other tasks even if the effects are not relevant.
4. Experiment 3
Although in neither experiment manipulations of task instructions
and effect monitoring duration (by R1-E1 compatibility in Exp. 1 and
R1-E1 expectancy in Exp. 2) interacted signicantly, there were
descriptively larger monitoring costs for the conditions that emphasize
the production of E1. This suggests that even though the relevance of
effects is not a necessary precondition of monitoring, it might still
modulate its magnitude. Consequently, Exp. 3 investigated action effects
that are (or are not) directly relevant for performing the task. To make
monitoring of the produced action effects part of the task requirement,
participants now had to count the number of unexpected action effects
in the Effect-Instruction condition. Further, to rule out any possible
carry-over effects, task context was manipulated between participants,
and the sample size was doubled.
4.1. Method
Ninety-six new participants (M
age
=30.4 years, SD =10.2) were
Fig. 2. Results of Experiment 1. Response Times (RTs;
top) and Error Rates (bottom) for Task 1 (left) and Task 2
(right). Yellow triangles represent trials with compatible
effects, whereas blue points represent trials with incom-
patible effects. Error bars denote the standard error of
paired differences, computed separately for each com-
parison of compatibility (Pster & Janczyk, 2013). (For
interpretation of the references to color in this gure
legend, the reader is referred to the web version of this
article.)
Fig. 3. Results of Experiment 2. Response Times (RTs;
top) and Error Rates (bottom) for Task 1 (left) and Task 2
(right). Yellow triangles represent trials with expected
effects, whereas blue points represent trials with unex-
pected effects. Error bars denote the standard error of
paired differences, computed separately for each com-
parison of expectedness (Pster & Janczyk, 2013). (For
interpretation of the references to color in this gure
legend, the reader is referred to the web version of this
article.)
1
When pooling the data from both experiments (N =96) and conducting a 2
×2 ×2 mixed ANOVA with R1-E1 relationship (low monitoring demand vs.
high monitoring demand) and task context (Effect-Instruction vs. Response-
Instruction) as within-subjects factors and experiment (compatibility manipu-
lation vs. expectancy manipulation) as between-subjects factor, R1-E1 rela-
tionship and task context also did not interact, F(1, 94) =1.86, p =.176,
η
p2
=
.02.
M. Schaaf et al.
Acta Psychologica 224 (2022) 103522
5
recruited online. 6 participants were removed from the nal sample due
to high error rates (>30%) and were replaced.
Stimuli and procedure were comparable
2
to Exp. 2, but the Effect-
Instruction condition now required participants to report the number
of unexpected events at the end of each block. Further, task context was
manipulated between participants. This allowed mapping each R1 to
only one S1 (red vs. green). Participants again completed 24 blocks
consisting of 40 trials. To achieve a variable number of odd trials per
block, trials were now randomized over the whole experiment instead of
within each experimental block. Task context, S1-R1 mapping, R1-E1
mapping and S2-R2 mapping were counterbalanced between
participants.
4.2. Results
The data were treated as in Exp. 1 and 2. After excluding trials with
errors (Task 1: 4.3%, Task 2: 8.1%) and outliers (5.8%), the nal sample
for RT analyses consisted of 84.0% of the original trials. The data were
analyzed via 2 ×2 ANOVAs with R1-E1 relationship (expected vs. un-
expected) as within-subjects and task context (Effect-Instruction vs.
Response-Instruction) as between-subjects factor (see Fig. 4).
4.2.1. Task 1
4.2.1.1. RTs. No inuences on Task 1 RTs were observed, all Fs <1.
4.2.1.2. Error rates. No inuences on Task 1 error rates were observed,
all Fs <1.
4.2.2. Task 2
4.2.2.1. RTs. Task 2 RTs were lower with the Response-Instruction
(552 ms) than with the Effect-Instruction (724 ms), F(1, 94) =53.43,
p <.001,
η
p2
=.36. Further, Task 2 responses were faster after expected
action effects (585 ms) than after unexpected action effects (692 ms), F
(1, 94) =108.36, p <.001,
η
p2
=.54. The experimental manipulations
interacted, F(1, 94) =82.43, p <.001,
η
p2
=.47, with a large inuence of
the R1-E1 relationship on Task 2 RTs with the Effect-Instruction, t(47) =
9.88, p <.001, d =1.43, Δ =201 ms, and a smaller, but still highly
signicant inuence with the Response-Instructions, t(47) =3.97, p <
.001, d =0.57, Δ =14 ms.
4.2.2.2. Error rates. Task 2 PEs were lower with the Response-
Instruction (7.4%) than with the Effect-Instruction (9.4%), F(1, 94) =
4.70, p =.033,
η
p2
=.05. Further, Task 2 responses were more accurate
after expected action effects (8.0%) than after unexpected action effects
(8.8%), F(1, 94) =4.13, p =.045,
η
p2
=.54. The experimental manip-
ulations interacted, F(1, 94) =8.79, p =.004,
η
p2
=.09, with an inu-
ence of the R1-E1 relationship on Task 2 PEs with the Effect-Instruction,
t(47) =2.89, p =.006, d =0.42, Δ =1.9%, and no inuence with the
Response-Instructions, t <1.
4.3. Discussion
Exp. 3 extends the results of the rst two experiments. In line with
Exp. 1 and 2, explicitly irrelevant action effects in Task 1 inuenced
performance in Task 2, implying that screening for action-effect con-
tingencies is a cause for dual-tasking costs. These screening costs were
similar in magnitude to those observed in the rst two experiments,
suggesting that the previously observed slowing after irrelevant effects
was not caused by carry-over effects from the within-manipulation of
task context. In contrast to Exp. 1 and 2, however, we observed signif-
icantly larger monitoring costs for relevant effects in Exp. 3, where we
employed a stronger relevance manipulation. Hence, although all action
effects were processed, the intensity of this processing depends largely
on the goals currently pursued. Considering the descriptive decrease in
Task 2 performance we found in the rst two experiments, this can be
seen as a tentative hint that the monitoring costs observed in previous
research represent a mixture of both screening and validation of action
effects, and that their share to the observable slowing in Task 2 might be
dependent on the exact wording of the instructions.
5. General discussion
The present experiments investigated the origin of previously re-
ported performance decrements emerging after action effects with
increased monitoring demand. As the action effects were goal-relevant
in most previous research, it was unclear whether performance decre-
ments originate from validating goal-relevant effects or from screening
for further goal-irrelevant effects.
To disentangle these two possibilities, we manipulated the task in-
structions: While participants were instructed to produce a certain effect
in one half of the experiment, this action effect was nominally task-
irrelevant in the other half. In the rst two experiments, altering the
relevance of action effects did not mitigate the monitoring costs in Task
2: Performance decrements (Δ =12 ms) following effects with high
monitoring demand were not reduced signicantly by different in-
structions. This data pattern suggests that previously reported effect
monitoring costs are not constrained to goal-relevant action effects, but
also emerge when irrelevant effects are screened. In the third experi-
ment, explicitly requiring participants to monitor the action effects
increased the observed monitoring costs. This suggests that while
screening takes place constantly, certain goals entail the need to validate
the produced perceptual changes, thereby impairing performance in
concurrent tasks further (Δ =14 ms for screening vs. Δ =201 ms for
screening plus validation of action effects).
A previous observation already hinted at the detrimental impact of
effect screening. Spatially incompatible action effects (as compared to
compatible effects) delay processing in a second task even when this
effect is uncontrollable (Wirth, Janczyk, & Kunde, 2018, Exp. 5). It is
tempting to assume that unpredictable effects can barely be a goal of the
actor, indicating that any ensuing inuences of these effects were due to
screening rather than validation. Yet, whether participants nonetheless
pursue these action effects as relevant goals, despite only being suc-
cessful at chance level, remained unsettled with that study. The present
instruction manipulation directly compares relevant and irrelevant ef-
fects and thus, we can now conclude on solid empirical footing that
effect screening is a cause of dual task decrements.
What exactly causes this observable impact on subsequent tasks is
yet to be investigated. While the present study rules out that monitoring
is evoked only by relevant effects, feature binding has likewise been
rejected as primary mechanism (discussed in Wirth & Kunde, 2020). On
this spot, we would like to advance two further possible candidates. One
viable explanation is that a generic monitoring system triggers an un-
specic stop-signal after detecting an expectancy-violation (Wessel,
2018), even if the violated expectancy is irrelevant for the current goal.
Hence, monitoring costs could be restricted to action effects that are
unexpected. This notion receives support from reports that oddball
stimuli (Steinhauser & Kiesel, 2011; Wessel et al., 2012) and spatially
incompatible action effects (Steinhauser et al., 2018) engage the same
neural system that is engaged in error monitoring.
Likewise, the modality overlap between the action effect of Task 1
and the imperative stimulus of Task 2 could be crucial. As both E1 and
S2 rely on spatially separated visual information, potentially possible
parallel perceptual processing might be peripherally constrained by
longer xations on unexpected effects (Brockmole & Boot, 2009). This
would delay the perceptual stage of Task 2, explaining why monitoring
2
To facilitate online data collection, the instructions were in English instead
of German.
M. Schaaf et al.
Acta Psychologica 224 (2022) 103522
6
of visual action effects seems to postpone visual second tasks (Wirth,
Janczyk, & Kunde, 2018; Wirth, Steinhauser, et al., 2018). Hence,
monitoring costs could occur even if the monitored action effect is
completely expected (Kunde et al., 2018).
6. Conclusion
We set out from the notion that agents must check what they cause,
that they must monitor their action effects. Previous research reported
that performance in a concurrent task deteriorates when participants
monitor goal-relevant action effects. Here, we found smaller (but reli-
able) performance decrements when participants monitor irrelevant
action effects. Thus, our ndings indicate that effect monitoring is a
rather fundamental limitation of dual tasking. Having an eye on what
you cause seems to be a process that is not easy to switch off. Having an
eye on this effect monitoring process is certainly worthwhile when it
comes to further explaining why doing two things at once is hard.
Funding source declaration
This work was supported by the German Research Foundation (Grant
KU 1964/11-2) within the priority program SPP 1772. The funders had
no bearing on study design, data collection and analysis, decision to
publish, or preparation of the manuscript.
Declaration of competing interest
None.
References
Band, G. P., van Steenbergen, H., Ridderinkhof, R. K., Falkenstein, M., & Hommel, B.
(2009). Action-effect negativity: Irrelevant action effects are monitored like relevant
feedback. Biological Psychology, 82(3), 211218. https://doi.org/10.1016/j.
biopsycho.2009.06.011
Brockmole, J. R., & Boot, W. R. (2009). Should I stay or should I go? Attentional
disengagement from visually unique and unexpected items at xation. Journal of
Experimental Psychology: Human Perception and Performance, 35(3), 808815. https://
doi.org/10.1037/a0013707
Elsner, B., & Hommel, B. (2001). Effect anticipation and action control. Journal of
Experimental Psychology: Human Perception and Performance, 27(1), 229240. https://
doi.org/10.1037/0096-1523.27.1.229
Hommel, B., & Wiers, R. W. (2017). Towards a unitary approach to human action
control. Trends in Cognitive Sciences, 21(12), 940949. https://doi.org/10.1016/j.
tics.2017.09.009
Janczyk, M., & Kunde, W. (2020). Dual tasking from a goal perspective. Psychological
Review, 127(6), 10791096. https://doi.org/10.1037/rev0000222
Kumar, N., Manjaly, J. A., & Sunny, M. M. (2015). The relationship between action-effect
monitoring and attention capture. Journal of Experimental Psychology: General, 144
(1), 1823. https://doi.org/10.1037/xge0000032
Kunde, W., Pster, R., & Janczyk, M. (2012). The locus of tool-transformation costs.
Journal of Experimental Psychology: Human Perception and Performance, 38(3),
703714. https://doi.org/10.1037/a0026315
Kunde, W., Wirth, R., & Janczyk, M. (2018). The role of feedback delay in dual-task
performance. Psychological Research, 82(1), 157166. https://doi.org/10.1007/
s00426-017-0874-6
Miller, G. A., Galanter, E., & Pribram, K. H. (1960). Plans and the structure of behavior.
Henry Holt and Co. https://doi.org/10.1037/10039-000
Pashler, H. (1994). Dual-task interference in simple tasks: Data and theory. Psychological
Bulletin, 116(2), 220244. https://doi.org/10.1037/0033-2909.116.2.220
Pster, R., & Janczyk, M. (2013). Condence intervals for two sample means:
Calculation, interpretation, and a few simple rules. Advances in Cognitive Psychology,
9(2), 7480. https://doi.org/10.5709/acp-0133-x
Pster, R., Janczyk, M., Wirth, R., Dignath, D., & Kunde, W. (2014). Thinking with
portals: Revisiting kinematic cues to intention. Cognition, 133(2), 464473. https://
doi.org/10.1016/j.cognition.2014.07.012
Steinhauser, M., & Kiesel, A. (2011). Performance monitoring and the causal attribution
of errors. Cognitive, Affective, & Behavioral Neuroscience, 11(3), 309320. https://doi.
org/10.3758/s13415-011-0033-2
Steinhauser, R., Wirth, R., Kunde, W., Janczyk, M., & Steinhauser, M. (2018). Common
mechanisms in error monitoring and action effect monitoring. Cognitive, Affective, &
Behavioral Neuroscience, 18(6), 11591171. https://doi.org/10.3758/s13415-018-
0628-y
Welford, A. T. (1952). The psychological refractory period and the timing of high-speed
performance - A review and a theory. British Journal of Psychology, 43(1), 219.
https://doi.org/10.1111/j.2044-8295.1952.tb00322.x
Welford, A. T. (1967). Single-channel operation in the brain. Acta Psychologica, 27, 522.
https://doi.org/10.1016/0001-6918(67)90040-6
Wessel, J. R. (2018). An adaptive orienting theory of error processing. Psychophysiology,
55(3), Article e13041. https://doi.org/10.1111/psyp.13041
Wessel, J. R., Danielmeier, C., Morton, B. J., & Ullsperger, M. (2012). Surprise and error:
Common neuronal architecture for the processing of errors and novelty. Journal of
Neuroscience, 32(22), 75287537. https://doi.org/10.1523/jneurosci.6352-11.2012
Wirth, R., Janczyk, M., & Kunde, W. (2018). Effect monitoring in dual-task performance.
Journal of Experimental Psychology: Learning, Memory, and Cognition, 44(4), 553571.
https://doi.org/10.1037/xlm0000474
Wirth, R., & Kunde, W. (2020). Feature binding contributions to effect monitoring.
Attention, Perception & Psychophysics, 82(6), 31443157. https://doi.org/10.3758/
s13414-020-02036-9
Wirth, R., Pster, R., Janczyk, M., & Kunde, W. (2015). Through the portal: Effect
anticipation in the central bottleneck. Acta Psychologica, 160, 141151. https://doi.
org/10.1016/j.actpsy.2015.07.007
Wirth, R., Steinhauser, R., Janczyk, M., Steinhauser, M., & Kunde, W. (2018). Long-term
and short-term action-effect links and their impact on effect monitoring. Journal of
Experimental Psychology: Human Perception and Performance, 44(8), 11861198.
https://doi.org/10.1037/xhp0000524
Fig. 4. Results of Experiment 3. Response Times (RTs;
top) and Error Rates (bottom) for Task 1 (left) and Task 2
(right). Yellow triangles represent trials with expected
effects, whereas blue points represent trials with unex-
pected effects. Error bars denote the standard error of
paired differences, computed separately for each com-
parison of expectedness (Pster & Janczyk, 2013). (For
interpretation of the references to color in this gure
legend, the reader is referred to the web version of this
article.)
M. Schaaf et al.
... 1 However, baseline and operant conditions differ in a crucial detail beyond the presence or absence of such action-effect sequences: baseline conditions feature one event to draw attentional resources, whereas operant conditions feature two events that can both draw attentional resources. Indeed, typical action-effect sequences draw sustained attention, from attentional orienting over performance monitoring and, specifically, effect monitoring, even if effects are irrelevant (Kok et al., 2006;Schaaf et al., 2022;Wirth et al., 2018). In contrast, singular events, i.e., actions without subsequent effects (aside from proprioceptive feedback) or stimulus appearance without preceding actions, represent shorter time frames and/or lack some of the attention-drawing features of a combined action-effect or two-event sequence. ...
... Moreover, for visual attention, evidence suggests attentional breadth to be extraordinarily malleable and adaptable (e.g., Gable & Harmon-Jones, 2010;Golubickis & Macrae, 2021;Goodhew & Plummer, 2019). Controlled actions especially seem to draw attentional processes easily, in spatial as well as temporal dimensions (e.g., Schaaf et al., 2022;Schneider & Shiffrin, 1977;Wirth et al., 2018), and indeed attention seems an important factor in establishing metacognitive phenomena accompanying actions as well, such as the sense of agency (Cheyne et al., 2009;Hon, 2017). In further support of our interpretation, a recent review article also focuses on attentional processes as a defining aspect of temporal binding (Hon, 2022). ...
... For one, our secondary event was also relevant to the participants, and thus attention was deliberately focused on this task in contrast to the operant classic task in which the secondary event was irrelevant. However, previous studies indicate that action-effect sequences grab attention so reliably that we cannot help but monitor even irrelevant aspects of this sequence such as irrelevant action effects (e.g., Schaaf et al., 2022). That is, while the attentional processes at play may differ between the two tasks in this experiment, the fact that both events in both tasks may grab attentional resources is well established. ...
Article
Full-text available
In the last years, it has become general consensus that actions change our time perception. Performing an action to elicit a specific event seems to lead to a systematic underestimation of the interval between action and effect, a phenomenon termed temporal (or previously intentional) binding. Temporal binding has been closely associated with sense of agency, our perceived control over our actions and our environment, and because of its robust behavioral effects has indeed been widely utilized as an implicit correlate of sense of agency. The most robust and clear temporal binding effects are typically found via Libet clock paradigms. In the present study, we investigate a crucial methodological confound in these paradigms that provides an alternative explanation for temporal binding effects: a redirection of attentional resources in two-event sequences (as in classical operant conditions) versus singular events (as in classical baseline conditions). Our results indicate that binding effects in Libet clock paradigms may be based to a large degree on such attentional processes, irrespective of intention or action-effect sequences. Thus, these findings challenge many of the previously drawn conclusions and interpretations with regard to actions and time perception.
... The monitoring of effects of actions and subsequent adjustments in behaviour was thematized in two articles of this special issue. Schaaf et al. (2022) conducted three experiments to test if the instructed relevance of any action effects influences the monitoring of the effects of an action in a dual-task setting. Although irrelevant action effects showed lower monitoring costs, the results indicate that irrelevance cannot serve as an exclusion criterion from action effect monitoring. ...
Article
Full-text available
In many if not all situations humans are engaged in more than one activity at the same time, that is, they multitask. In laboratory situations, even the combination of two simple motor tasks generally yields performance decrements in one or both tasks, compared with corresponding single task conditions. In contemporary models of dual tasking, these dual task costs are attributed to a capacity-limited stage of mentally specifying required responses. Ideomotor theory suggests that the generation of responses is a process of specifying goals, that is, desired future perceptual states (= effect anticipation). Based on this, we argue that effect anticipation is the process responsible for dual task costs. We substantiate this suggestion with results from several lines of research, showing that (a) effect anticipation coincides with a capacity-limited process in dual task experiments, (b) no dual task costs arise if no effects are to be anticipated in one of the tasks, (c) dual task costs vary as a function of a how well effects from two tasks fit together, and (d) monitoring the occurrence of effects also adds additional costs. These results are discussed in a common framework and in relation to other observations and fields. (PsycInfo Database Record (c) 2020 APA, all rights reserved).
Article
Full-text available
Monitoring the perceptual effects of body movements is supposed to be a capacity-limited process that can interfere with processing of a concurrent task. Here we studied the contribution of feature binding to such effect monitoring interference. In three experiments, we varied the possibility of feature overlap between responses and effects in a primary task and responses in a secondary task. We show that responses in a secondary task are delayed when they partially, rather than completely, alternate or repeat features of responses/effects of a primary task. Yet, these partial feature repetition/alternation costs are small, and they occur on top of other factors that lengthen the critical effect monitoring process, such as the spatial compatibility of responses and effects in the primary task. The results thus show that feature binding contributes to, but cannot fully account for, delays in a secondary task caused by monitoring effects of a primary task.
Article
Full-text available
In the present study, we considered error-related brain activity in event-related potentials, to investigate the relationship between error monitoring—that is, the detection and evaluation of erroneous responses—and action effect monitoring—that is, monitoring of the sensory consequences of behavior. To this end, participants worked on a task-switching paradigm that consisted of a free-choice task, in which a puzzle piece had to be attached to an existing one (the prime task), and a subsequent color flanker task (the probe task). We examined whether unexpected action effects in the prime task would affect the subsequent error monitoring in the probe task. We found the neural correlates of error monitoring during the probe task, the error-related negativity as well as the error positivity, to be increased after unexpected action effects in the prime task. In contrast, the neural correlates of visual attention were decreased after unexpected action effects, in line with recent findings on an attenuation of sensory processing after errors. Our results demonstrate a direct link between monitoring processes in the two tasks. We propose that both error monitoring and action effect monitoring rely on a common generic monitoring system related to novelty detection or affective processing. Preactivating this system by means of unexpected action effects increases the sensitivity for detecting an error in the subsequent task.
Article
Full-text available
People aim to produce effects in the environment, and according to ideomotor theory, actions are selected and executed via anticipations of their effects. Further, to ensure that an action has been successful and an effect has been realized, we must be able to monitor the consequences of our actions. However, action-effect links might vary between situations, some might apply for a majority of situations, while others might only apply to special occasions. With a combination of behavioral and electrophysiological markers, we show that monitoring of self-produced action effects interferes with other tasks, and that the length of effect monitoring is determined by both, long-term action-effect links that hold for most situations, and short-term action-effect links that emerge from a current setting. Effect monitoring is fast and frugal when these action-effect links allow for valid anticipation of action effects, but otherwise effect monitoring takes longer and delays a subsequent task. Specific influences of long-term and short-term links on the P1/N1 and P3a further allow to dissect the temporal dynamics of when these links interact for the purpose of effect monitoring.
Article
Full-text available
Actions aim to produce effects in the environment. To accomplish this properly, we not only have to recruit the appropriate motor patterns, but also we must be able to monitor whether an intended effect has ultimately been realized. Here, we investigated the impact of such effect monitoring on performance in multitasking situations: Multitasking basically means to produce and monitor multiple actions and effects in fast succession. We show that effect monitoring cannot run in parallel, without causing processing decrements, with a second task. Also, monitoring of effects that are spatially incompatible to a response seems to take longer than the monitoring of spatially compatible action effects (Experiments 1 through 4). We further argue that effect monitoring is essential toward learning of response-effect associations, as it captures not only anticipated action effects, but also unpredictable occurrences in the environment (Experiments 5a and 5b).
Article
Full-text available
Doing two things at once is hard, and it is probably hard for various reasons. Here we aim to demonstrate that one so far barely considered reason is the monitoring of sensory action feedback, which detracts from processing of other concurrent tasks. To demonstrate this, we engaged participants in a psychological refractory period paradigm. The responses in the two tasks produced visual action effects. These effects occurred either immediately or they were delayed for the first of the two responses. We assumed that delaying these effects would engage a process of monitoring visual feedback longer, and delay a concurrent task more, as compared to immediate effects. This prediction was confirmed in two experiments. We discuss the reasons for feedback monitoring and its possible contribution to dual tasking.
Article
The ability to detect and correct action errors is paramount to safe and efficient goal-directed behaviors. Existing work on the neural underpinnings of error processing and post-error behavioral adaptations has led to the development of several mechanistic theories of error processing. These theories can be roughly grouped into adaptive and maladaptive theories. While adaptive theories propose that errors trigger a cascade of processes that will result in improved behavior after error commission, maladaptive theories hold that error commission momentarily impairs behavior. Neither group of theories can account for all available data, as different empirical studies find both impaired and improved post-error behavior. This article attempts a synthesis between the predictions made by prominent adaptive and maladaptive theories. Specifically, it is proposed that errors invoke a nonspecific cascade of processing that will rapidly interrupt and inhibit ongoing behavior and cognition, as well as orient attention toward the source of the error. It is proposed that this cascade follows all unexpected action outcomes, not just errors. In the case of errors, this cascade is followed by error-specific, controlled processing, which is specifically aimed at (re)tuning the existing task set. This theory combines existing predictions from maladaptive orienting and bottleneck theories with specific neural mechanisms from the wider field of cognitive control, including from error-specific theories of adaptive post-error processing. The article aims to describe the proposed framework and its implications for post-error slowing and post-error accuracy, propose mechanistic neural circuitry for post-error processing, and derive specific hypotheses for future empirical investigations.
Article
From its academic beginnings the theory of human action control has distinguished between endogenously driven, intentional action and exogenously driven, habitual, or automatic action. We challenge this dual-route model and argue that attempts to provide clear-cut and straightforward criteria to distinguish between intentional and automatic action have systematically failed. Specifically, we show that there is no evidence for intention-independent action, and that attempts to use the criterion of reward sensitivity and rationality to differentiate between intentional and automatic action are conceptually unsound. As a more parsimonious, and more feasible, alternative we suggest a unitary approach to action control, according to which actions are (i) represented by codes of their perceptual effects, (ii) selected by matching intention-sensitive selection criteria, and (ii) moderated by metacontrol states. Human action control and decision-making are traditionally understood as emerging from the competition between will (representing intentionality and rationality) and habit (representing automatized and stimulus-driven tendencies). This dual-route concept is still dominant in behavioral, clinical, and neuroscientific research. Criticism has been accommodated by allowing some degree of continuity between the two routes, but without specifying clear-cut criteria to locate behavior on this continuum, however. We suggest replacing the dual-route model by a unitary model of action control which assumes that goal-directed actions are represented in terms of expected action effects that are selected according to a match between intended and expected action effects.