j ourna l homepage: www.e lsev ie r .com/ locate /actpsy
No anticipation without intention: Response–effect compatibilityin effect-based and stimulus-based actions☆
Katharina Zwosta ⁎, Hannes Ruge, Uta WolfenstellerTechnische Universität Dresden, Germany
☆ This workwas supported by the German Research CoWe would like to thank Tatjana Gackstatter, Carmen ScRößler, Amal Kebir, Lydia Heilig and Paula Michels for dat⁎ Corresponding author at: Department of Psychology, T
Goal-directed behavior is characterized by the anticipation of the resulting effect during response selection. Such an-ticipations can be contextualized in the sense that response–effect relationships in one stimulus context are invertedin another stimulus context. The primary study aimwas to test the hypothesis that contextualized effect anticipationmight depend on whether subjects adopt either an effect-based action control style or a stimulus-based controlstyle. Importantly, we hypothesized that the choice of control styles depends on explicit instruction. Effect anticipa-tion during response selection was determined by assessing the behavioral impact of spatial compatibility betweenthe required response and an additional task-irrelevant spatial feature attached to the anticipated effect that wouldbe produced by that response in a given context. In two experiments we found a compatibility effect exclusively inblocks with effect-based instruction but not in stimulus-based blocks. Furthermore, subjects could quickly switchbetween styles without one strategy dominating the others. Together, this shows that contextualized anticipationof distal visual effects is not an automatic process but depends on the intention to produce an effect.
Goal-directed behavior requires knowledge about contingencies be-tween situations, responses and the effects they produce, enabling flexi-ble response selection in different situations according to the anticipatedgoal state. According to ideomotor theory this learning process leads to abidirectional coupling between responses and effects, so that anticipato-ry activation of the effect representation automatically activates thecorresponding response (Hommel, Musseler, Aschersleben, & Prinz,2001; James, 1890). Thus, one critical characteristic of goal-directedbehavior is the anticipation of the effect during response selection.
Indeed, Kunde (2001) demonstrated that responses are faster if theresponse location matches the location of (irrelevant) visual effects itproduces. This means that the location of the effect must have beenactivated before the response was made so that the future event caninteract with ongoing response selection. This effect is not restrictedto spatial compatibility, but compatibility effects were also found forthe duration of action effects (Kunde, 2003), effect intensity (Kunde,Koch, & Hoffmann, 2004) and verbal responses (Koch & Kunde, 2002).
uncil (DFG, SFB 940 project A2).häuffele, Anne Dschietzig, Tinaa acquisition.echnische Universität Dresden,
. Zwosta).
vier B.V.
Importantly, such compatibility effects also occurred if response–effect (R–E) contingencies were contextualized (Kiesel & Hoffmann,2004). In reality, the effect that is produced by a certain action stronglydepends on the context in which it is applied. One response can havedifferent and even contradictory effects if applied in different situationsleading to a hierarchical structure of response–effect relationshipsdepending on the stimulus (Colwill & Rescorla, 1990). As an example,pushing or pulling a door can open or close the door depending on theside of the door where you are standing. So, pursuing the goal to openthe door would lead to contrary actions when you are inside or outsidethe room. This means that R–E contingencies in one situation arereversed in the other situation so that effect anticipation is dependenton the context.
The goal of the present study was to investigate how this contextu-alized effect anticipation is affected by different action control styles. Itwas suggested that there are two fundamentally different ways inwhich people interact with the environment relying either on internalor external stimuli (Herwig, Prinz, & Waszak, 2007): People can eithermanipulate the environment in order to reach certain effects (effect-based action control style) or they can simply react to environmentalstimuli (stimulus-based action control style).
Experimentally, these two action control styles are often induced byusing free-choice vs. forced-choice tasks. However, to date there is onlyone study that compared these two task types specifically with regardto contextualized R–E mappings: Pfister, Kiesel, and Melcher (2010)
Fig. 1. Example of a compatible (left) and incompatible (right) stimulus–response–effectmapping. Effect colors were contingent on task-irrelevant locations. For both effect-based and stimulus-based conditions responding to one stimulus led to spatially compat-ible effects while responding to the other stimulus led to incompatible effects.
629K. Zwosta et al. / Acta Psychologica 144 (2013) 628–634
found that R–E anticipation during action selection (as indexed by thecompatibility effect) occurred only if participants were choosing theirresponses freely but not if responses were spatially predetermined.This particular finding seems to suggest that effect-anticipation duringresponse selection depends on an effect-based action control style.However, it stands to be argued whether the employed forced-choicetask was a ‘fair’ stimulus-based control condition. In fact, subjectsresponded to spatial stimuli in a spatially compatible manner. Hence,the absence of R–E compatibility effects in this condition might simplyreflect that when responses are directly (automatically) activated by aspatial stimulus, anticipated action effects might not have the time toaffect behavior. In fact, by using different response cues implicatingdifferent S–R translation speeds due to dimensional overlap or non-overlap, we could show that the behavioral expression of learnt ac-tion–effect associations is preventedwhen actions are directly activatedby highly over-learnt response cues (Wolfensteller & Ruge, submitted).
To avoid this problem, the present study aimed at manipulating theaction control style differently while keeping a constant forced-choicesetting. To this end, participants were instructed to choose their actionseither according to the effect that should be produced or according toone of two pre-cued stimulus–response rules. Importantly, also underthe pre-cued rule condition, correct responses were followed by effectsaccording to exactly the samehierarchical scheme as in the effect-basedcondition. Hence, we used a pure manipulation of instruction while thehierarchical associative structure that could be used to guide responseselection was identical.
At first glance, this approach resembles previous studies investigat-ing intention–response compatibility. For example, Ansorge (2002)found that spatial R–E compatibility effects only occurred if participantswere instructed to produce a specific spatial effect throughout theexperiment while no R–E compatibility effects occurred if participantswere instructed to respond according to stimulus–response rules (seealso Shin & Proctor, 2012). However, these experiments included anoverlap between the code of the intention and the response as bothwere spatial. For example, the stimulus indicated that it should bemoved to the left and this facilitated left key presses vs. right keypresses. The problem with this type of R–E compatibility effect is thatit could potentially be due to interference between overlapping seman-tic codes (i.e., the stimulus would activate “left” while the requiredresponse would activate “right”) but not between the anticipated effect(left stimulus movement) and the right response.
This latter potential problemwas avoided in the present study as thespatial component of the effect was task-irrelevant. If the instructionsinduce different action control styles, then only in the effect-basedaction control style, when subjects are instructed to produce certain ef-fects (e.g., colors), the action effects (including task-irrelevant features,e.g., the spatial location) should become integrated into responseselection. Importantly, then it should not be necessary for the intentionto dimensionally overlap with the response. Rather, having a specificeffect-related intention (to produce a certain color) alone should besufficient to induce an intentional action control style leading to mea-surable effect anticipation.
2. Experiment 1
We used a novel paradigm where R–E contingencies in one contextwere inverted in the other context so that general R–E associationswere absent (see Fig. 1). Responding to stimulus 1 by pressing the leftkey led to an effect on the left side while pressing the right key led toan effect on the right side. For stimulus 2 this was inverted: Pressing theleft key led to a right effect, pressing right to a left effect. We comparedthree different conditions: In the effect-based condition the instructionwas to produce a certain color which was contingent on a certain (task-irrelevant) location. A second stimulus-based condition only differed interms of the cue that preceded the stimulus. Responses to certain stimuliwere also contingently followed by certain effects but the instructionwas
given in terms of stimulus–response rules not including any aspect of theeffect. Finally, we added a control condition with a stimulus-based in-struction but random effects, thus making effect anticipation impossible.The critical variable was the difference in response times towards stimuliwith response–effect compatible and response–effect incompatible map-pings as this indicates that effects were anticipated before they actuallyappeared.We hypothesized that if being in an effect-based action controlmode alone is sufficient to induce effect anticipation during response se-lection then the spatial component of the future event, even though notexplicitly included in the intention, should interact with the spatial re-sponse. Furthermore, if R–E anticipation is dependent on an effect-based action control style then R–E compatibility effects should be absentin the stimulus-based condition even though there is a perfect contingen-cy between stimuli, responses and effects.
2.1. Methods
2.1.1. SubjectsTwenty subjects participated at the Technische Universität Dresden.
One had to be excluded due to an exceptionally high error rate (36%). Ofthe remaining 19 participants 13 were female and mean age was24years (range: 19–29). All subjectswere right-handed andhadnormalor corrected-to-normal vision. The participants were compensatedwith5€ or received course credit.
2.1.2. Apparatus and stimuliStimuli were presented on a 17"monitor on light background. The ex-
periment was controlled by E-Prime 2.0. Two cues, two stimuli and twoeffect colors were assigned to each condition. There were three pairs ofgenuine black-and-white stimuli that were taken from the Creative Sym-bol Collection of Matton Images. Accordingly, there were also three pairsof effect colors (red–blue, green–purple, orange–pink). In the effect-basedcondition the cue consisted of a colored octagon frame that indicated theeffect color that was to be produced. In the stimulus-based condition andthe control condition the cue frames were black or white indicating therule to be applied. In one of the conditions the frames were circles andin the other one they were squares. Effects were color filled octagonsthat appeared on the right or left side of the cue and stimulus. Responseswere made with the keyboard, pressing the key “D” with the left indexfinger or the key “K”with the right index finger.
For each subject stimuli and effect colors were independentlyrandomly assigned to the three conditions. Then, within effect-basedand stimulus-based conditions, cues were randomly assigned to effectcolors and stimuli to R–E compatibility.
Fig. 2. Examples of trial procedures for the three conditions in Experiment 1. While in effect-based blocks the cue indicated the color to be produced, in stimulus-based and controlcondition blocks the cue indicated the stimulus–response rule to be applied. In both cases responding correctly led to an effect left or right of the stimulus.
Fig. 3. Results of Experiment 1. Response times (RTs, bars) and error rates (ERR, dots) forR–E compatible and incompatible trials depending on instruction. Error bars indicate stan-dard errors.
630 K. Zwosta et al. / Acta Psychologica 144 (2013) 628–634
2.1.3. ProcedureThe trial procedure is pictured in Fig. 2. Each trial started with the
presentation of a fixation cross in the center of the screen for 500ms,after which it was replaced by the cue frame. After 700ms the stimulusappeared in the center of the frame. As soon as a correct response wasmade, additionally the effect was displayed to the right or the left ofthe cue and stimulus. If the response was wrong, the word “Error”was displayed and if no response occurred until 2000 ms then “Tooslow” was displayed. After 1000 ms the next trial started with thefixation cross.
In all conditions a given colored effect always appeared at the samelocation. In the effect-based condition and the stimulus-based conditionthe effect was contingent on the combination of stimulus and response.In the control condition one of the two effects appeared randomly.
The experiment startedwith a practice phase duringwhich the threetasks were first practiced separately in randomized order. In the effect-based condition subjects were instructed to respond with the key thatwould produce a certain effect color. The instruction started with a dis-play showing the first stimulus and the colors that would be producedwhen the right or the left response was made. This was followed byeight trials during which only the first stimulus appeared. This proce-dure was then repeated for the other stimulus, by first instructing theassignments of keys to effect colors followed by eight practice trials.Subsequently, there were 32 trials with both stimuli. For the stimulus-based and the control condition the procedure was the same exceptthat instead of the stimuli the rules were instructed separately. Firstthe cue for rule one was instructed, showing which key to press inresponse to both stimuli followed by the instruction of rule two.When all three taskswere introduced like this, therewere three practiceminiblocks, one per condition, containing 24 trials. During the practicephase error trials were repeated until a correct response was made.
After practice, participantswere instructed that error trials nowwouldnot be repeated anymore and that the actual experiment would begin.There were 21 task blocks (7 per condition) containing 24 trials in ran-domized order. In between the blocks a display saying “next block isabout to start” was presented for 3000 ms. There was a break after 12blocks and the experiment could be resumed by the subject after a self-chosen interval. The whole experiment lasted about 45 to 60min.
2.2. Results
Two repeated measures ANOVAs were conducted on RTs and errorrates with condition (effect-based, stimulus-based and control) and
response-effect compatibility (compatible vs. incompatible) as within-subject factors. For the analysis of RTs only correct trials were included(92.4% of all trials). Response times and error rates for Experiment 1 aredisplayed in Fig. 3.
2.2.1. RTsThere was a main effect of condition on response times, F(2, 36) =
5.94, p= .006, ηp2 = .25, with response times being larger in the effect-based condition (711 ms) than in the stimulus-based (674 ms), t(18)=2.91, p=.009, and the control (664ms), t (18)=2.63, p=.017,conditions.
There was also a main effect of response–effect compatibility, F(1,18)=4.64, p=.045,ηp
2=.21,with responses being faster for compatiblethan incompatible trials. This effect, however, interactedwith the condi-tion factor, F(2, 36)=6.17, p=.005, ηp
2= .26 such that a compatibilityeffect was only present in the effect-based condition (666 ms vs.756 ms for compatible and incompatible trials respectively), t (18) =4.28, pb0.001, but not in the stimulus-based (674ms for both compat-ible and incompatible trials) or the control (667ms vs. 660ms) condi-tions, both tsb1.
2.2.2. Error ratesThe analysis of error rates resulted in a main effect of compatibility,
F(1, 18) = 5.70, p= 0.028, ηp2 = .24, with fewer errors in compatible
(6.8%) than incompatible (8.5%) trials. This compatibility effect was
631K. Zwosta et al. / Acta Psychologica 144 (2013) 628–634
not significantly different in the three conditions, F(2, 36)=1.33, p=.278, although numerically it was larger in the effect-based condition(6.4% vs. 10%) as compared to the stimulus-based condition (7.2% vs.8.3%) and the control condition (6.9% vs. 7.2%).
2.3. Discussion
The results of Experiment 1 clearly demonstrate that effect anticipa-tion only occurred if participants had an explicitly effect-based inten-tion and was not present in a stimulus-based action control style. Thisis in accordance with previous studies that found response–effectcompatibility effects only in an intentional action mode (Ansorge,2002; Pfister et al., 2010; Shin & Proctor, 2012). Importantly, the inten-tion that was instructed was not spatial so that the compatibility effectcannot be explained by interference of a verbalized instruction with re-sponse selection. Furthermore, even though stimulus-based conditionsand effect-based conditions alternated, there was no compatibilityeffect in the stimulus-based condition. This shows that people canquickly change between action control styles and do not stick withone mode once it is adopted, even though both effect-based andstimulus-based trials could have been performed using both strategies.
However, in both conditions the cue reliably predicted the effect.One could argue that in the effect-based condition this relationshipwas far more evident than in the stimulus-based condition becausecues had the effect color while the cues in the stimulus-based conditionwere not semantically related to the effect. Accordingly, the differencesbetween the conditions could have also been due to a different amountof learning of the cue–effect relationship. In order to test this assump-tion in Experiment 2 we employed more comparable cue–effect associ-ations in both conditions.
3. Experiment 2
In Experiment 2wewanted to exclude that the results of thefirst ex-periment were based on more evident cue-color/location associationsin the effect-based condition compared to the stimulus-based condition.For that purpose we used cues with comparable color associations forboth conditions. A cue was for example a drawing of a sun whichcould indicate either to produce the color yellow or to apply the“sun”-rule. In both cases it could be expected that responding wouldproduce a yellow effect.
Another aim of Experiment 2 was to investigate the dynamics ofchanging between action control styles. Effect anticipation in thestimulus-based mode might occur if it is preceded by an effect-basedblock. So in order to check for transfer effects we systematically manip-ulated transition frequencies between the condition blocks. Hence,while in Experiment 1 block order was fully randomized, in Experiment2 each condition block was followed by another condition block withthe same frequency.
3.1. Methods
3.1.1. Subjects26 subjects participated at the TechnischeUniversität Dresden. Six of
themwere excludeddue to error rates larger than25%. Of the remaining20 participants 11 were female and mean age was 22years (range: 18–29). All subjects were right-handed and had normal or corrected-to-normal vision.
3.1.2. Apparatus and stimuliThe apparatus was the same as in Experiment 1. Stimuli were now
three pairs of abstract black-and-white shapes (rectangle–pentagon,circle–semicircle, square–rhomb) that were randomly assigned to theconditions and to R–E compatibility. Cues were black-and-white simpledrawings of objects. In the effect-based and the stimulus-based condi-tions the cues were semantically related to the effect colors. There
were two pairs that were randomly assigned to the two conditions(pair 1: sun–yellow and leaf–green; pair 2: fire–red and wave–blue).In the control condition the cueswere a butterfly and a flower and effectcolors could be either orange or pink. Effects were now colored plus-signs for correct responses and colored minus-signs for incorrectresponses. Again responses were made with the left and right indexfingers using the keys “D” and “K” on the keyboard.
3.1.3. ProcedureThe trial procedure was the same as in Experiment 1 with two ex-
ceptions: The cue disappeared as soon as the stimulus was presented.Also, an error response led to a minus-sign at the location and withthe color which would have been produced if the response was correct(e.g. if the cue indicated to produce a blue effect and thewrong key waspressed a red effect appeared). The trial procedure is depicted in Fig. 4.
During the practice phase before the stimuli or rules were intro-duced participants were presented with both cues presented againstthe background of their corresponding effect color (e.g., sun on a yellowbackground). In the control condition the assignment of a color to acue was random. The rest of the practice phase proceeded just as inExperiment 1.
Themain experiment consisted of 36 task blocks each containing 16trials. Between the blocks a display was presented saying “next block isabout to start” for 3000ms. The order of the task blockswas randomizedwith the limitation that each condition was equally often followed byeach of the three conditions (including repetition). There was a breakafter 18 blocks and the experiment could be resumed by the subjectafter a self-chosen interval. All in all the experiment lasted about60min.
3.2. Results
Response times and error rates for Experiment 2 are shown in Fig. 5.Again RTs and error rates were analyzed separately with two repeatedmeasures ANOVAs with condition and R–E compatibility as within-subject factors.
3.2.1. RTsThere was a significant main effect of condition, F(2, 38) = 14.07,
pb0.001, ηp2=.46. Responses were slower in the effect-based condition
(615ms) than in the stimulus-based condition (544ms), t (19)=4.48,pb .001, and the control condition (569ms), t (19)=3.34, p=.003. Fur-thermore, responses in the stimulus-based conditionwere faster than inthe control condition, t (19) = 2.35, p= .030. The main effect of R–Ecompatibility was not significant, F b 1. Importantly, there was again asignificant interaction between condition and R–E compatibility, F(2,38)=6.62, p= .003, ηp
2= .26. In the effect-based condition responseswere faster in compatible (594ms) than in incompatible (635ms) trials,t (19)= 2.12, p= .048, while there was no compatibility effect in thecontrol condition, t b 1, and in the stimulus-based condition responseswere even faster in incompatible (531 ms) than in compatible(558ms) trials, t (19)=2.96, p=.008.
In order to check for possible effects of the previous condition blockwe additionally calculated an ANOVAwith condition, compatibility andthe previous condition as factors. The compatibility effect was notaffected by the previous condition, and there was also no interactionbetween condition, compatibility and the previous condition, bothFsb1.25.
3.2.2. Error ratesThere was a significant main effect of condition on error rates, F(2,
38) = 4.44, p = .019, ηp2 = .19. Subjects made fewer errors in the
stimulus-based condition (6.5%) than in the effect-based condition(9.6%) and the control condition (9.1%). Themain effect of compatibilitywas not significant, Fb1, just as the interaction between condition andcompatibility, F(2, 38)=1.60, p=.215.
Fig. 4. Trial procedure in Experiment 2. The same set of cues was used for effect-based and stimulus-based condition blocks.
632 K. Zwosta et al. / Acta Psychologica 144 (2013) 628–634
3.3. Discussion
Experiment 2 replicates the findings of Experiment 1 by showingthat the difference between the effect-based and the stimulus-basedconditions remains if cue-color associations aremade similar. Apparent-ly, participants included the future effect location into their responseplanning only if the instruction was effect-based and not if they wereinstructed in terms of stimulus–response rules. Furthermore, theresponse–effect compatibility effect was not influenced by task orderwhich shows that subjects quickly switched between action controlstyles.
In this experiment during stimulus-based blocks participantsshowed an inverted compatibility effect with slightly faster responsesin incompatible compared to compatible trials which led to overallfaster responses in this condition compared to the control condition(and the effect-based condition). We only found this effect in Experi-ment 2 and not in Experiment 1, hence we cannot tell if this effect is re-ally reliable. A possible explanation is that in the stimulus-basedcondition the cue which indicated the future effect color also primedthe associated effect location. In subliminal priming paradigms it wasshown that during a certain period this leads to an automatic inhibitionof a response associated to the prime which behaviorally results in aninverted compatibility effect (see Eimer & Schlaghecken, 2003, for a re-view). However, it was also shown that this happens only if the sublim-inally presented cue is coupled to a certain response, while it is not
Fig. 5.Results for Experiment 2. Response times and error rates for R–E compatible and in-compatible trials depending on the action mode condition. Error bars represent standarderrors.
enough to be semantically associated with a location (Eimer &Schlaghecken, 1998). This explanation does not apply directly to the sit-uation in our experiment as the cue was not associated with a specificresponse but with the future effect location. Nevertheless, as our para-digm is very different from a subliminal priming paradigm it is notclear whether restrictions found in subliminal priming are also applica-ble to our paradigm. Hence, it might still be the case that similarprocesses could have played a role. Importantly, however, this doesnot affect our primary conclusion as negative compatibility effects resultfrompassive primingwhich stands in contrast to an active effect-relatedresponse selection which would lead to a positive compatibility effect.
4. General discussion
The aim of the present study was to investigate how effect anticipa-tion using contextualized stimulus–response–effect contingencies is af-fected by effect-based and stimulus-based action control styles inducedby instruction. Previous research suggested that ideomotormechanismsmight depend on the action control style that is adopted with responseselection being triggered by effect anticipation in an effect-based butnot a stimulus-based action control style. In the present experimentswe manipulated the action control style by instructing participantseither to choose their actions according to a specific effect feature oraccording to stimulus–response rules — without mentioning effect-related aspects.
Two experiments demonstrated that effect anticipation during re-sponse selection occurred only if participants had an effect-based inten-tion. In this case participants included task-irrelevant features of thefuture effect into their response planning. In contrast, response selectionwas not affected by effect anticipation in stimulus-based conditionseven though they included the same stimulus–response–effect contin-gency. Note that the paradigm was designed in such a way that inboth stimulus-based and the effect-based conditions, the effects werecontingent on responses in a certain situation. The absence of effect an-ticipation in the stimulus-based instruction condition persisted wheneffect-based and stimulus-based conditions were made as similar aspossible by using semantically associated cues that reliably predictedthe effect color. Hence, although in both conditions participants knewwhich effect would follow their response, only the instruction to inten-tionally produce this effect feature led to an integration of anticipatedtask-irrelevant features into response selection. Note that based onour results as well as findings from previous studies (e.g. Pfister et al.,2010, who also included neutral trials) response–effect compatibility
633K. Zwosta et al. / Acta Psychologica 144 (2013) 628–634
effects seem to be basedmainly on interference from incompatible trialsand not facilitation by compatible trials making it detrimental to perfor-mance to use task irrelevant redundant effects. Thus, the instruction toselect the response according to one effect feature led to the anticipationof other irrelevant features during response selection even though thisimpaired performance.
Thus, while ideomotor theory claims that repeated experience ofspecific effects following actions leads to (i) a bidirectional couplingbetween both components and (ii) actions actually being initiated byanticipating the future action effect (Hommel et al., 2001; James,1890), we did not find that the latter mechanism was involved if weinstructed participants to adopt a stimulus-based action mode. Insteadthe integration of visual effects into response planning depended onan effect-related intention while experiencing contingent action effectsalone did not trigger ideomotor response selection in this sense.
Our findings extend previous studies comparing free-choice toforced-choice actions. The results show that effect anticipation is notconstricted to free-choice tasks but can be present also in entirelyforced-choice trials if effects are included into the task set. Hence, ourresults demonstrate that the critical point is not the internal selectionof a goal as in free-choice tasks but the conceptualization of thetask as aiming at an effect. In contrast to the operationalization ofstimulus-based conditions in previous studies (e.g. Herwig et al.,2007; Pfister et al., 2010) our operationalization actually made R–Ecompatibility effects more probable by using a task that was lessreflex-like and required a response selection process that did not differin complexity from the effect-based condition. Furthermore, in Experi-ment 2, the trial procedures were exactly identical in stimulus-basedand effect-based conditions. In our opinion this operationalization isthe more direct and more rigorous test (especially in Experiment 2) ofthe hypothesis that intentional and stimulus-based action modes differin regard to ideomotor processes. Hence, the notion that we do not findresponse–effect compatibility effects in the stimulus-based conditionsuggests that the previously found lack of it is not due to otherconfounds but really the absence of an effect-based intention.
This might resolve some of the inconsistencies in the results of previ-ous studies regarding effect anticipation during forced-choice tasks. Forexample, Kiesel andHoffmann (2004) found contextualized R–E compat-ibility effects in entirely forced-choice tasks when they instructed theirparticipants to shoot a ball into the goal, thus using an effect-basedinstruction. Similarly, Pfister and Kunde (2013) found compatibility ef-fects in both forced-choice and free-choice conditions when instructingparticipants according to the effect location. Moreover, compatibilityeffects were observed even if instructions were rule-based in a series ofexperiments using unambiguous R–E contingencies in which one andthe same response always led to the same effect (Kunde, 2001, 2003;Kunde et al., 2004). However, in these studies, participantswere explicitlytold that the effects of their actions could be either congruent or incongru-ent to the response location which might have sensitized participants forR–E compatibility andmight have led to a conceptualization of actions aseffect-based. Thus, based on these results, the usage of R–E contingenciesin stimulus-based actions could either be constricted to tasks with fixedR–E relationships or could be explained by the instruction leading to aneffect-based action mode.
Furthermore, these results go beyond previous findings that showedintention–response compatibility effects (e.g. Ansorge, 2002) by dem-onstrating that having an effect-based intention alone is sufficient fortask-irrelevant effect features to be anticipated and it is not necessaryfor the intention to actually overlap with the response. In Ansorge'sstudy the stimulus indicated to which direction it was supposed bemoved (either right or left) and required a response (either left orright keypress) to move it that was either compatible or incompatiblewith the intended movement direction. Within this paradigm, for theintention to interfere with the response no actual anticipation processwas necessary. Instead, retrieval of the instruction when the stimuluswas displayed could have been the sole cause of the found compatibility
effects. In contrast, in the present study, the instruction was to producea certain effect color while the effect position was task irrelevant. Thusclearly, any spatial compatibility effects must reflect an anticipatoryprocess.
The present study offers another insight concerning the dynamics ofswitching between action control styles. Pfister et al. (2010) found thatif free- and forced-choice trials were equally often presented in random-ized order, R–E compatibility effects also occurred for forced-choice trialswhich they interpreted as a dominance of the effect-based mode whichonce adopted is applied to all trials. Contrary to that, in our experimentswe did not find evidence for a lasting influence of the effect-based style,instead participants quickly switched between action modes. However,there is a critical difference between Pfister et al.'s and our paradigm:While Pfister et al. used the same task for forced-choice and free-choicetrials and manipulated task type on a trial-by-trial basis, in our experi-ments we used three different tasks for the condition blocks with theirown cues, stimuli and effect colors so that learned effect contingenciesin the effect-based condition could not be applied to the stimulus-basedcondition. Further studies are needed to assess what actually influencesthe predominance of effect-based action control which, comparing bothstudies, could be based on task-similarity, ratio of effect-based andstimulus-based trials and the length of condition blocks.
It might be argued that effect-based and stimulus-based conditionsdiffered in terms of the amount of attention directed at the effect. Wetried to counteract this by making the effects very salient and leavingthem on screen for quite a long time. Furthermore in Experiment 2the shape of the effect (plus or minus signs) indicated correct or incor-rect responses. Nevertheless,we cannot exclude that subjects paidmoreattention to the effects in the effect-based condition compared to othertwo conditions. However, as Herwig and Waszak (2009) did not findany effect of attention this seems rather improbable. In their studythey forced people to attend to the effects as they were required todetect deviant effects and respond to them. They did not find that thismanipulation influenced the result that there was no associationbetween responses and effects in the stimulus-based condition.
Another issue concerns the integration of proximal (proprioceptive)effects into response selection. In our task design we investigated onlythe effects of action control styles on the integration of distal visual ef-fects into response planning. We cannot make a statement whether ornot ideomotor response selection using proximal effects like tactilestimulation takes place nevertheless. Hence, it is not to say that in thestimulus-based condition ideomotor mechanisms are not at play at all.Further research is necessary to determine whether the modulation ofaction control styles also influences the integration of proximal effectsinto response selection.
At present it is debated whether the actionmode actually influencesresponse–effect learning or whether only usage of response–effect con-tingencies is influenced by action control styles (Pfister, Kiesel, &Hoffmann, 2011; Wolfensteller & Ruge, 2011). Some authors arguedthat the actual acquisition of response–effect contingencies dependson learning in a free-choice mode (Herwig et al., 2007) while othersdemonstrated that contingencies are also acquired during forced-choice tasks though not necessarily used for selecting the response(Pfister et al., 2011). Thus, in the stimulus-based condition participantsmight indeed have learned the contingencies but we did not findevidence for goal-directed behavior triggered by effect anticipation.While associative learning seems to be a prerequisite of ideomotorcontrol it does apparently not automatically lead to effect-based actioncontrol. Instead, stimulus-based action selection is independent of theaction's consequences even if the consequences are known.
As a conclusion, our results demonstrate that effect anticipation dur-ing response selection relies on the intention to produce an effect anddoes not happen automatically if people continuously experience thattheir actions in a certain situation are contingently followed by a certaineffect. Specifically, we show that it is not the internal selection of a goalthat is necessary for distal effects to be integrated into response
634 K. Zwosta et al. / Acta Psychologica 144 (2013) 628–634
planning but any effect-rated intention leads to an intentional actioncontrol style which automatically makes irrelevant aspects of the effectaffect response selection. On the other hand, if such an intention is notpresent visual effects are not used for response selection even thoughthey can be perfectly anticipated.
References
Ansorge, U. (2002). Spatial intention–response compatibility. Acta Psychologica, 109(3),285–299.
Colwill, R. M., & Rescorla, R. A. (1990). Evidence for the hierarchical structure of instrumen-tal learning. Animal Learning & Behavior, 18(1), 71–82. http://dx.doi.org/10.3758/Bf03205241.
Eimer, M., & Schlaghecken, F. (1998). Effects of masked stimuli on motor activation:Behavioral and electrophysiological evidence. Journal of Experimental Psychology.Human Perception and Performance, 24(6), 1737–1747.
Eimer, M., & Schlaghecken, F. (2003). Response facilitation and inhibition in subliminalpriming. Biological Psychology, 64(1–2), 7–26. http://dx.doi.org/10.1016/s0301-0511(03)00100-5.
Herwig, A., Prinz, W., & Waszak, F. (2007). Two modes of sensorimotor integration inintention-based and stimulus-based actions. Quarterly Journal of ExperimentalPsychology (Hove), 60(11), 1540–1554. http://dx.doi.org/10.1080/17470210601119134.
Herwig, A., & Waszak, F. (2009). Intention and attention in ideomotor learning. QuarterlyJournal of Experimental Psychology (Hove), 62(2), 219–227. http://dx.doi.org/10.1080/17470210802373290.
Hommel, B., Musseler, J., Aschersleben, G., & Prinz,W. (2001). The Theory of Event Coding(TEC): A framework for perception and action planning. Behavioral and Brain Sciences,24(5), 849. http://dx.doi.org/10.1017/S0140525x01000103.
James, W. (1890). The principles of psychology. New York: MacMillan.Kiesel, A., & Hoffmann, J. (2004). Variable action effects: Response control by
Koch, I., & Kunde, W. (2002). Verbal response–effect compatibility. Memory & Cognition,30(8), 1297–1303.
Kunde, W. (2001). Response–effect compatibility in manual choice reaction tasks. Journalof Experimental Psychology. Human Perception and Performance, 27(2), 387–394.
Kunde, W. (2003). Temporal response–effect compatibility. Psychological Research, 67(3),153–159. http://dx.doi.org/10.1007/s00426-002-0114-5.
Kunde, W., Koch, I., & Hoffmann, J. (2004). Anticipated action effects affect the selec-tion, initiation, and execution of actions. Quarterly Journal of ExperimentalPsychology. A, Human Experimental Psychology, 57(1), 87–106. http://dx.doi.org/10.1080/02724980343000143.
Pfister, R., Kiesel, A., & Hoffmann, J. (2011). Learning at any rate: action–effect learning forstimulus-based actions. Psychological Research, 75(1), 61–65. http://dx.doi.org/10.1007/s00426-010-0288-1.
Pfister, R., Kiesel, A., & Melcher, T. (2010). Adaptive control of ideomotor effect anticipa-tions. Acta Psychologica, 135(3), 316–322. http://dx.doi.org/10.1016/j.actpsy.2010.08.006.
Pfister, R., & Kunde, W. (2013). Dissecting the response in response-effect compati-bility. Experimental Brain Research, 224(4), 647–655. http://dx.doi.org/10.1007/s00221-012-3343-x.
Shin, Y. K., & Proctor, R. W. (2012). Testing boundary conditions of the ideomotorhypothesis using a delayed response task. Acta Psychologica, 141(3), 360–372.http://dx.doi.org/10.1016/j.actpsy.2012.09.008.
Wolfensteller, U., & Ruge, H. (2011). On the timescale of stimulus-based action–effect learn-ing. Quarterly Journal of Experimental Psychology, 64(7), 1273–1289. http://dx.doi.org/10.1080/17470218.2010.546417.
Wolfensteller, U., & Ruge, H. (2013). Rapid learning of action effects: response selection andaction-effect stability. Manuscript submitted for publication.
