Imagination inflation in the mirror: Can imagining others' actions induce false memories of self-performance?

Acta Psychologica 158 (2015) 51–60

Contents lists available at ScienceDirect

Acta Psychologica journal homepage: www.elsevier.com/ locate/actpsy

Imagination inflation in the mirror: Can imagining others' actions induce false memories of self-performance? Isabel Lindner a,⁎, Gerald Echterhoff b a b

Department of Psychology, University of Kassel, Holländische Str. 36-38, 34127 Kassel, Germany Department of Psychology, University of Münster, Fliednerstr. 21, 48149 Münster, Germany

a r t i c l e

i n f o

Article history: Received 17 February 2014 Received in revised form 24 February 2015 Accepted 24 March 2015 Available online 1 May 2015 JEL classification: 2343 3040 2330 Keywords: Source memory Self-other confusion False memory Agency

a b s t r a c t Imagining oneself performing a simple action can trigger false memories of self-performance, a phenomenon called imagination inflation. However, people can, and often do, imagine others' behavior and actions. According to a visual-similarity account, imagining another person's actions should induce the same kind of memory error, a false memory of self-performance. We tested this account in three experiments, in which performance was followed by imagination. In the imagination phase, participants were asked to either imagine themselves or to imagine another person performing actions, some of which were not previously performed. Two weeks later, a surprise source-memory test was administered in which participants had to decide whether a depicted action had been performed or not performed. Results revealed that imagining another person can trigger false memories of self-performance. However, visual similarity between performance and imagination predicted the amount of false memories only for other-imagination but not for self-imagination. These findings are consistent with research suggesting that other- and self-imagination rely on different mechanisms: While other-imagination primarily involves visual imagery, self-imagination primarily involves motor imagery. Accordingly, false action memories from other-imagination may result from visual similarity, whereas false action memories from selfimagination may result from motor simulation. © 2015 Elsevier B.V. All rights reserved.

1. Introduction Did you lock the door this morning? Research has shown that imagining yourself locking the door can later make you believe that you have actually locked it when in fact you have not. Such a false memory consisting in the misattribution of merely imagined actions as performed was dubbed the imagination–inflation effect (Goff & Roediger, 1998, see also Garry, Manning, Loftus, & Sherman, 1996). In this type of false action memory, the modality in which an action was originally encoded (performance vs. imagination) is confused in hindsight. However, humans' imaginative capacities allow them not just to imagine their own, but also other people's behaviors (e.g., Decety & Grèzes, 2006). According to a widely held view, imagination inflation stems from the sensory similarity between imagined and performed acts (e.g., Thomas, Bulevich, & Loftus, 2003). Thus, imagining another person performing a simple action, for example locking the door, should lead to the same kind of memory error, a false memory of actually having locked the door oneself. Hence, people may confuse not only the modality but also the agent of actions (self vs. other). In the present studies, we

⁎ Corresponding author. Tel.: +49 561 804 3585; fax: +49 561 804 3584. E-mail addresses: [email protected] (I. Lindner), [email protected] (G. Echterhoff).

http://dx.doi.org/10.1016/j.actpsy.2015.03.008 0001-6918/© 2015 Elsevier B.V. All rights reserved.

thus investigated whether the imagination of another person's acts can produce false memories of self-performance. In the common paradigm for studying false action memories from self-imagination (Goff & Roediger, 1998), participants perform or do not perform simple actions in a first phase (e.g., Unlock the lock), and, in a second phase, imagine or do not imagine themselves performing some of these actions. In this second phase, some of the to-be-remembered actions are usually presented once and others are presented repeatedly. In a later surprise source-memory test, it is typically found that imagination, especially when repeated, leads to a significant proportion of false action memories. That is, participants remember having performed actions when in fact they have not performed these actions in Phase 1, but only imagined themselves performing these actions in Phase 2. This (self-)imagination–inflation effect is robustly found (e.g., Lampinen, Odegard, & Bullington, 2003; Seamon, Philbin, & Harrison, 2006; Thomas & Loftus, 2002). False memories from self-imagination have often been explained by source-monitoring difficulties resulting from the perceptual similarity of imagination and performance (e.g., Lampinen et al., 2003; Thomas et al., 2003). By this account, participants monitor information reactivated during a memory test for cues diagnostic for actual performance (Johnson & Raye, 1981; Johnson, Hashtroudi, & Lindsay, 1993). Because both modalities, performance and imagination, involve the encoding of sensory features, people sometimes have difficulties in

52

I. Lindner, G. Echterhoff / Acta Psychologica 158 (2015) 51–60

deciding whether a vivid sensory memory trace was previously perceived during self-performance or merely imagined. Given the dominance of vision among the human senses (e.g., Posner, Nissen, & Klein, 1976) and its outstanding role in the attribution of agency (e.g., Jeannerod & Pacherie, 2004), it is reasonable to assume that among the sensory impressions generated through imagination visual features are critical in eliciting false action memories (see also Lindner & Henkel, in press). By this view, imagination inflation is primarily due to the similarity of visual representations generated through performance and imagination. Consistent with this notion, research has shown that memory traces from imagination are predominantly visual in nature (e.g., Johnson, Foley, Suengas, & Raye, 1988). Imagination can be conceptualized as a mental simulation of an action or event (e.g., Denis & Kosslyn, 1999; Shepard & Cooper, 1986) aimed at either reliving past or anticipating future events (e.g., Schacter & Addis, 2007; Suddendorf & Corballis, 2007). For instance, an applicant could mentally simulate an upcoming job interview. These simulations are not restricted to the self, but can also include other people (e.g., Decety & Grèzes, 2006). In the example, the applicant may not only imagine her or his own behaviors, but also the interviewer's behaviors, from welcoming the applicant to closing the door behind him or her. These considerations beg the question of whether imagining someone else performing a simple action can later induce a false memory of self-performance. For instance, would the applicant exhibit imagination inflation if s/he had imagined the interviewer rather than her- or himself closing the door? Whereas imagination–inflation studies for these kinds of events have revealed confusions of the modality of action encoding (performance or imagination), it is presently unknown whether the agent of an imagined action can be confused as well. Among the studies using the above described imagination–inflation paradigm, there is only one study that has employed other-imagination: Seamon et al. (2006) asked participants to imagine the experimenter perform actions in Phase 2 (herein called other-imagination). Critically, however, the same participants were also asked to observe the experimenter perform actions in Phase 1 (see also Seamon et al., 2009). This condition was compared to a standard condition in which participants performed actions themselves in Phase 1 and imagined performing the actions themselves in Phase 2. Consistent with this, participants in the first group underwent a source-memory test referring to other-performance (i.e., Did the experimenter perform this action?), while participants in the second group underwent a source-memory test referring to self-performance (i.e., Did you perform this action?). Both imagination conditions induced similar proportions of false action memories, however these memories differed with regard to agency: Other-imagination induced flawed recollections of the experimenter's actual performance whereas self-imagination induced flawed recollections of self-performance. This finding supports the visual-similarity hypothesis inasmuch as similar visual representations are generated when observing as well as imagining someone else versus observing as well as imagining oneself performing an action (see also Seamon et al., 2006). However, because agency was not crossed in the study by Seamon et al. (2006), two questions remain: Can imagining another person performing a simple action not only induce false action memories of other-, but also of self-performance? If so, will both imagining another person and imagining oneself induce comparable amounts of false action memories of self-performance? Based on the visual-similarity account outlined above, one would predict that a self-other confusion is possible. However, this account further predicts that the amount of false action memories will be smaller after other- compared to self-imagination: Visual impressions implicated in mental images of other- and self-performance differ with regard to one key aspect, that is, visual perspective. Visual perspective can crucially change one's perception of an action. For instance, locking a door looks quite different from a first- or a second-person perspective. Thus, visual perspective could moderate the amount of

false action memories from imagination (e.g., Libby, 2003; Marsh, Pezdek, & Lam, 2014). Taken together, imagination of another person's actions should be sufficiently potent to trigger false memories of self-performance. However, the altered visual perspective during other-imagination is likely to serve as a non-self cue, which in turn should reduce the amount of false memories compared to self-imagination. To test this prediction, in Experiment 1 we used the imagination–inflation paradigm explained above and varied the imagined agent in the second, imagination phase: Participants were either instructed to imagine themselves or to imagine another person performing actions. Meanwhile, the agent in Phase 1 was held constant: Actions were performed or not performed by the participants themselves. In Experiment 2, we further examined the role of visual perspective in imagination inflation. Specifically, we manipulated visual input during initial action performance to match perspectives during selfperformance and other-imagination. To this end, we asked participants to observe themselves via a webcam when performing actions in Phase 1. The webcam images created the impression of observing another person facing them. Thus, while perspective was identical between selfperformance and other-imagination, perspective was different between self-performance and self-imagination. If visual perspective is critical for the creation of false action memories, we hypothesized that the pattern expected for Experiment 1 would be reversed: We expected a higher amount of false action memories after other-imagination compared to self-imagination. Cross-experiment comparisons can be used to estimate the influence of visual perspective: Finding a greater effect of otherimagination in Experiment 2 than in Experiment 1 and a smaller effect of self-imagination in Experiment 2 than in Experiment 1 would be in line with such an account. However, cross-experiment comparisons have to be interpreted with caution. To strengthen the empirical basis of our research, we therefore independently manipulated the variables from the first two studies within an extended design in Experiment 3. In this experiment, the agent (self, other) was manipulated both in Phase 1 (performance) and Phase 2 (imagination). Again, if visual perspective is critical for the creation of false memories, imagining oneself should lead to more false performed-responses if the actor is the self (vs. other), but this pattern should be reversed when imagining someone else. 2. Experiment 1 2.1. Method 2.1.1. Participants Thirty-six students of the University of Cologne (30 women) participated for partial fulfillment of curricular requirements. Mean age was 25.33 years (SD = 5.62). 2.1.2. Design We used a 3 (type of encoding in Phase 1: performed vs. read vs. not presented) × 2 (frequency of imagination in Phase 2: 5 × vs. 1× vs. 0×) × 2 (imagined agent in Phase 2: other vs. self) design with the last variable varying between participants. This design was not fully crossed inasmuch as action statements not presented in Phase 1 were never presented in Phase 2, but served as distractors at retrieval. The proportion of performed-responses in a surprise source-memory test served as dependent variable. 2.1.3. Materials and procedure Participants were recruited for a study on mental representations of actions and provided informed consent. The experiment was computerbased. All participants were tested individually in two sessions. The involved objects were hidden from participants' view by a cardboard divider.

I. Lindner, G. Echterhoff / Acta Psychologica 158 (2015) 51–60

The first session comprised two phases. In the first phase, participants either performed or read simple action statements (e.g., Open the book, Stamp the paper). We used the following procedure to equate encoding across the two conditions (perform versus read) as closely as possible, especially to equate object memory and motor activity: First, a picture of the object appeared on the screen for 5 s. Meanwhile, the corresponding object was placed in front of the participant by the experimenter. The participant first indicated on how many days of a typical week she or he would use this kind of object in everyday life. Second, the action statement appeared on the screen for 10 s together with an instruction to perform or to read it. The action statements were derived from Lindner, Echterhoff, Davidson, and Brand (2010). They comprised simple, typical everyday actions, which involved the manipulation of one to three objects and could be performed with one or two hands (e.g., Roll the dice, Put the card in the envelope). Participants were instructed to read the action statement once. In case of a performinstruction, they were asked to perform the particular action once. To match exposure to the object, motor activity, and interstimulusinterval, between the perform- and read-trials (e.g., Lindner et al., 2010), in case of a read-instruction, participants were asked to imitate the experimenter making a hand-movement which was unrelated to the corresponding action. These movements consisted of three to five simple gestures (e.g., touching the table with the palm of the hand), were performed right next to the object(s), and were approximately matched to action performance with regard to length. Last, the object was hidden again by the experimenter after participants had performed the action or hand movements. All in all, 30 action statements were presented in a random order: 15 had to be performed, 15 had to be read. After a 5-min filler task started the imagination phase. Depending on the experimental condition, participants were either instructed to imagine themselves or to imagine another person performing the described actions. To standardize the imagination condition in the latter case, we showed people a picture of another person sitting in front of the to-be-manipulated object(s) (see also Denis, Engelkamp, & Mohr, 1991) for trials of 15 s each (see Fig. 1). These pictures were screenshots taken from videos used in a former observation–inflation study (Lindner et al., 2010; Experiment 3) and only showed the torso, arms and hands of an actor from a second-person perspective in the initial position to perform the action described in the action statement. In order to maximize similarity between the two imagination conditions, participants in the self-imagination group were also shown a picture to standardize imagination (see Fig. 2). These pictures were screenshots from the same observation–inflation study (Lindner et al., 2010; Experiment 3). However, they showed the same scene from a first-person perspective. In order to make sure participants were actually imagining

53

Fig. 2. Example of the pictures used to standardize imagination in the self-condition of Experiments 1 and 2.

themselves, we retouched these pictures so that the hands of the actor which were originally present in these videos were not visible anymore. Specifically, participants were instructed to first read the action statement and then look at the picture and imagine [themselves or the person depicted on the photos] performing the action as vividly and detailed as possible. They were also instructed to repeat this process until the end of the trial, but that the quality of their mental images was more important than their quantity. To ensure that participants did not simply imagine themselves in the other-imagination condition, it was emphasized that it was important for participants to actually imagine the other person rather than themselves and that they were to imagine observing this person while performing the actions in front of them. This was illustrated with an appropriate example. To maintain engagement with the task, after each 15‐sec imagination slot, people were asked to (a) rate the vividness of their mental images (on a 7-point Likert scale), and to (b) indicate how often they had imagined action performance during the entire interval. Following previous studies (Goff & Roediger, 1998; Thomas et al., 2003), five of the action statements which had been performed and five of the action statements which had been read in Phase 1 were presented five times, once, or not at all, respectively, in Phase 2. Therefore, the imagination phase consisted of 60 presentations. Two weeks later, in the second session, we administered the sourcememory test. People were asked to indicate whether action statements were old or new and — if old — whether they had or had not originally performed the corresponding action. In this source-memory test, 60 action statements were presented, 30 of which were old and 30 of which were new. Out of the 30 old action statements, 15 had been performed and 15 had been read in the first phase. Out of the 15 performed as well as read actions, five had been imagined five times, five had been imagined once, and five had not been imagined in Phase 2. Because we had two within-subject factors, all 60 action statements were randomly divided into 12 sets of five action statements, and 45 action statements (nine sets) were counterbalanced across the type of encoding in Phase 1 (performed, read, not-presented) and the frequency of imagination in Phase 2 (5×, 1 ×, 0 ×). 15 action statements (three sets) always served as distractors at retrieval. 2.2. Results and discussion

Fig. 1. Example of the pictures used to standardize imagination in the other-condition of Experiments 1 and 2.

2.2.1. Analyses of performed-responses Alpha was set to .05; ps are reported two-tailed. Descriptive statistics can be found in Table 1. We ran a mixed 2 × 3 × 2-ANOVA with imagined agent in Phase 2 (other vs. self) as between subjects' factor and frequency of imagination in Phase 2 (5× vs. 1× vs. 0×) as well as encoding

54

I. Lindner, G. Echterhoff / Acta Psychologica 158 (2015) 51–60

Table 1 Experiment 1: Mean proportion of performed-responses as a function of encoding in Phase 1 as well as agent and frequency of imagination in Phase 2. Standard deviations are given in parentheses. Imagined agent, Phase 2

Encoding, Phase 1 Performed Read Not presented

Self

Other

Frequency of imagination, Phase 2

Frequency of imagination, Phase 2

5× .86 (.13) .23 (.22) –

5× .96 (.09) .10 (.14) –

1× .90 (.12) .11 (.17) –

0× .61 (.25) .02 (.06) .01 (.01)

1× .87 (.12) .09 (.12) –

0× .72 (.23) .06 (.09) .01 (.02)

Note: Proportions represent the frequency of performed-responses divided by the number of all responses for a corresponding item type. N = 18 in each cell.

in Phase 1 (performed vs. read) as within subjects' factors on performedresponses. We suggested that self- and other‐imagination would differentially affect (false) action memories. That is, we expected to find an interaction between imagined agent and frequency of imagination. Indeed, our analysis did not only yield a main-effect of frequency of imagination, F(2,68) = 32.89, p b .001, η P 2= .492, but also a trend towards an interaction between both these factors, F(2,68) = 2.70, p = .074, η P 2= .074. However, this interaction was moderated by encoding as indicated by a three-way interaction, F(2,68) = 3.34, p = .042, η P 2= .089, and was therefore analyzed separately for actions that had not and those that had actually been performed (i.e., for false vs. correct performedresponses). A 2 × 3-ANOVA with agent and frequency of imagination on false performed-responses only replicated a significant interaction between both factors, F(2,68) = 4.45, p = .015, η P 2= .116. Simple main-effect analyses revealed that neither a single nor repeated other-imagination trials resulted in a significant amount of false memories of selfperformance, ts(17) ≤ 1.29, ps ≥ .215. In contrast, a single selfimagination trial led to a marginal and repeated self-imagination trials led to a clear increase in false performed-responses when separately compared to the baseline (0 ×), ts(17) ≥ 2.05, ps ≤ .057, ds ≥ 0.482. Consistent with this, the increase in false memories from otherimagination was significantly lower than the increase from selfimagination after repeated imagination trials, F(1,34) = 6.63, p = .015, η P 2= .163, but not after a single imagination trial, F = 1.13, p = .296. In contrast, a 2 × 3-ANOVA with agent and frequency of imagination on correct performed-responses failed to show a significant interaction between both factors, F(2,68) = 2.23, p = .116, but yielded a clear main effect of frequency of imagination, F(2,68) = 24.00, p b .001, η P 2= .414. That is, across agents correct performed-responses increased after a single as well as after repeated imagination trials [both separately compared to the baseline (0×), ts(35) ≥ 5.40, ps b .001, ds ≥ 0.90]. Another interesting result from the global analysis including encoding was that while unsurprisingly, actually performing actions in Phase 1 generally led to more performed-responses than just reading the action statements, F(1,34) = 1051.66, p b .001, η P 2= .969, the size of this effect was different for the two imagined agents as indicated by an interaction between encoding and agent, F(1,34) = 5.12, p = .030, η P 2= .131. By trend, other-imagination led to more correct and fewer false performed-responses than self-imagination which is why there was no main effect of agent, F b 1. That is, imagined agent did not generally alter performed-responses, however source discrimination, defined as the difference between correct and false performed-responses across frequencies of imagination, was better after other- (M = .77) compared to self-imagination (M = .67). Encoding also interacted with frequency of imagination, F(2,68) = 5.60, p = .006, η P 2= .141. As was already described with regard to the three-way-interaction, frequency of imagination enhanced correct and false performed-responses differently.

2.2.2. Additional analyses It is arguably easier for people to imagine themselves than to imagine another person performing an action, especially when the other person is unfamiliar and not represented in very detail like on the photos we used. We thus examined whether the differences in false memories between the two groups could be due to differences in task difficulty. To this end, we asked our participants to rate how difficult they found the imagination task on a 7-point Likert scale (1 being not difficult at all, 7 being very difficult). There was no significant difference in reported difficulty between the other‐ (M = 3.78) and the self-imagination (M = 2.94) group; t(34) = 1.39, p = .174. Similarly, we examined a potential confounding influence of vividness. However, we found that the vividness-ratings provided after each imagination slot did not differ between the other- (M = 5.18) and the selfimagination group (M = 5.50); t(34) = 1.21, p = .235. Also, we found no difference in the ratings of how often people had imagined actionperformance (provided after each imagination slot); M = 4.16 in the other-, and M = 3.40 in the self-imagination group; t(34) = 0.91, p = .372. 2.2.3. Discussion In sum, Experiment 1 showed that imagining someone else performing simple actions did not lead to false memories of selfperformance.1 In contrast, there was a higher, and significant, proportion of false memories of self-performance after repeated self-imagination. As in previous studies, the effect was weaker after a single than after repeated self-imagination-trials (e.g., Goff & Roediger, 1998; Lindner et al., 2010; Thomas et al., 2003). Thus, we replicated the (self-)imagination–inflation effect, but found no equivalent after other-imagination. The difference between self- and other-imagination could neither be attributed to differences in task difficulty, nor to the vividness or frequency of the generated mental images. The finding from Experiment 1 is surprising given that the imagination of another person's action performance shares at least some visual representations with self-performance. Why was imagining otherperformance not confused with self-performance? In Experiment 1 false memories of self-performance were not only reduced, but eliminated after other-imagination. We thus suspect that perspective is even more important than we had hypothesized. This view resonates well with extant research: Seamon et al. (2006) found an increase in false memories when participants had imagined another person performing an action; however, he found false memories of other- (and not self-) performance. Moreover, Marsh et al. (2014) just recently showed that confidence-ratings of recent (Life-Events-Inventory) events changed more when participants imagined themselves from a first-person compared to a second-person perspective. Similar to what we found, the latter perspective did not even seem to induce a significant increase in confidence (the mean change score was 0.015 for the second-person perspective and 0.437 for the first-person perspective). Hence, visual perspective not only moderates, but is critical for the occurrence of false action memories. These considerations suggest that matching perspectives between self-performance and other-imagination should increase the chances of finding a self-other confusion. To this end, we replicated Experiment 1 with a twist: In Experiment 2, we asked people to observe themselves via webcam when performing actions in Phase 1, thereby switching visual perspective from the first to the second person (for a similar design, see Hornstein & Mulligan, 2004). Therefore, contrary to Experiment 1 the perspective overtaken during actual performance was identical to the perspective overtaken during imagination in the other-, but not in the self-condition. If visual perspective was indeed 1 As Table 1 shows, a slight increase in false memories is noticeable even in the otherimagination condition which might have turned out to be statistically reliable with a higher-power approach. However, even in this case, this minor potential effect would arguably be meaningless in everyday life.

I. Lindner, G. Echterhoff / Acta Psychologica 158 (2015) 51–60

critical to the occurrence of imagination inflation then we should (a) expect false memories to occur after other-imagination in Experiment 2, and (b) moreover expect imagination inflation to be even greater after other- compared to self-imagination, which should not induce false memories. In other words, we should expect the opposite pattern of results than the one found in Experiment 1.

Table 2 Experiment 2: Mean proportion of performed-responses as a function of encoding in Phase 1 as well as agent and frequency of imagination in Phase 2. Standard deviations are given in parentheses. Imagined agent, Phase 2

3. Experiment 2 3.1. Method 3.1.1. Participants Thirty-six students of the University of Cologne (28 women) participated for partial fulfillment of curricular requirements. Mean age was 22.94 years (SD = 5.93). 3.1.2. Design We used the same design as in Experiment 1. 3.1.3. Materials and procedure We also used the same materials and the same procedure as in Experiment 1 with one exception: When people were performing actions in Phase 1, they observed themselves via webcam on the computer-monitor, that is, as if they watched themselves from a second-person perspective. Therefore, perspective at the time of selfperformance and perspective at the time of imagination were identical in the other-imagination condition and differed in the self-imagination condition. Specifically, a webcam was placed on the computer monitor, which was automatically activated when the action statement, together with the read- or perform-instruction, appeared on the screen. Similar to the photos presented in the other-imagination condition, the webcam focused on the arms and hands of the person. Only one detail of the procedure had to be adjusted due to this manipulation: In Phase 1, the simple gestures which were to perform right next to the object in case of a read-instruction were instructed orally (instead of performed by the experimenter and imitated by the participants). In both, the readand the perform-trials, participants were urged to only watch themselves on the screen and not to look down (i.e., not to overtake a firstperson perspective). The experimenter made sure that participants were complying with this task. In addition, we collected ratings of the perceptual quality (the amount of visual, tactile, and auditory details generated during imagination) of the participants' mental images within this experiment. 3.2. Results and discussion 3.2.1. Analyses of performed-responses Descriptive statistics are provided in Table 2. We ran a mixed 2 × 3 × 2-ANOVA with imagined agent in Phase 2 (other vs. self) as between subjects' factor and frequency of imagination in Phase 2 (5× vs. 1× vs. 0×) as well as encoding in Phase 1 (performed vs. read) as within subjects' factors on performed-responses. We hypothesized that in Experiment 2, we should find a significant interaction between agent and frequency of imagination, indicating more (false) performed-responses after other- compared to selfimagination. However, an inspection of the data showed that the same pattern emerged as in Experiment 1, that is, more false performedresponses after self-compared to other-imagination. To explore this pattern of results, we nevertheless ran the described ANOVA. However, neither an interaction between agent and frequency of imagination, F(2,68) = 1.66, p = .199, nor between agent, frequency and encoding, F b 1, was found. That is, differently from Experiment 1, selfimagination did not result in a (significantly) steeper increase of either false or correct-performed-responses compared to other-imagination. However, a main effect of imagination, F(2,68) = 15.15, p b .001, η P 2= .308, indicated that our manipulation was effective. Yet across

55

Encoding, Phase 1 Performed Read Not presented

Self

Other

Frequency of imagination, Phase 2

Frequency of imagination, Phase 2

5× .86 (.19) .30 (.34) –

5× .82 (.18) .18 (.20) –

1× .84 (.22) .13 (.18) –

0× .63 (.32) .06 (.11) .01 (.01)

1× .87 (.18) .06 (.09) –

0× .72 (.22) .04 (.09) .01 (.02)

Note: Proportions represent the frequency of performed-responses divided by the number of all responses for a corresponding item type. N = 18 in each cell.

agents of imagination, this main effect was again moderated by encoding, F(2,68) = 4.06, p = .022, η P 2= .107: Similar to Experiment 1, a single imagination trial substantially increased correct, t(35) = 3.73, p = .001, d = 0.62, but not false memories for self-performance, t(35) = 1.60, p = .118. However, repeated imagination trials increased both correct and false performed-responses alike, ts(35) ≥ 3.38, ps ≤ .002, ds ≥ 0.56. In other words, independent of imagined agent, we found a substantial amount of false memories after repeated imagination trials in this experiment. Even though the pattern reported above was not moderated by imagined agent, we separately analyzed false performed-responses within each of the two imagination groups to check for imagination– inflation effects. Indeed, the results were identical to the ones found in the analysis collapsed across groups: We found significant increases in false performed-responses after repeated imagination trials, ts(17) ≥ 2.38, ps ≤ .029, ds ≥ 0.56, but not after a single imagination trial, ts ≤ 1.69, ps ≥ .110, within each group. Similar to Experiment 1, the main-effect of encoding, F(1,34) = 595.95, p b .001, η P 2= .946, was by trend moderated by imagined agent, F(1,34) = 3.14, p = .085, η P 2= .085. Again, source discrimination, defined as the difference between correct and false performedresponses across frequencies of imagination, was better after other(M = .71) compared to self-imagination (M = .62), which is why there was again no main-effect of imagined agent, F b 1.

3.2.2. Additional analyses To follow up on the descriptive trend of more false memories after self- (vs. other-) imagination, we again explored participants' ratings concerning their mental images. This time, imagining another person performing the actions was experienced as more difficult (M = 3.83) than imagining oneself (M = 2.83) performing the actions, t(34) = 2.08, p = .046, d = 0.69. However, we found no significant correlation between these ratings and imagination inflation2 within or across groups, rs ≤ ∣‐.134∣, ps ≥ .435. As in Experiment 1, vividness-ratings (M = 5.16 in the other-, and M = 5.33 in the self-imagination group) as well as frequency-ratings (M = 2.86 in the other-, and M = 3.25 in the self-imagination group) provided after each imagination slot did not differ between groups, ts ≤ 0.69, p ≥ .495.3

2 The imagination–inflation measure that we used for our correlational analyses was the difference in false performed-responses after having imagined the actions five times minus zero times. This was done because a substantial amount of false memories was only found after repeated imagination trials and variability was low after a single imagination trial. 3 Please note that for vividness as well as for frequency ratings, data from two people were missing in each group.

56

I. Lindner, G. Echterhoff / Acta Psychologica 158 (2015) 51–60

To further examine the impact of visual details on imagination inflation, we compared both imagination groups with regard to their overall ratings of the amount of visual details generated through imagination. There was a trend towards imagining another person evoking more visual details (M = 6.72) than imagining oneself (M = 6.28), t(34) = 1.87, p = .070, d = 0.62. Because a ceiling-effect emerged with regard to these ratings inasmuch as only three of all 36 participants chose ratings lower than 6 or 7, we did not calculate correlations between these ratings and imagination inflation. We also explored the impact of the generation of acoustical and tactile impressions. Both conditions did not differ with regard to the generation of acoustical impressions, M = 5.67 in the other-, and M = 5.33 in the self-imagination group, t(34) = 0.55, p = .589. Also, we did not find a relationship between the amount to which auditory cues were generated and the amount of false memories within or across groups, rs ≤ .112, ps ≥ .657. However, self-imagination entailed a higher degree of tactile details (M = 6.00) than other-imagination (M = 4.56), t(34) = 2.47, p = .019, d = 0.82. Yet, again, there was no relation between the degree to which tactile details were generated and the amount of false memories within or across groups, rs ≤ ∣‐.296∣, ps ≥ .233. Comparisons within groups revealed that participants in the otherimagination group generated more visual (M = 6.72) than acoustical (M = 5.67) and tactile impressions (M = 4.56), ts(17) ≥ 2.49, ps ≤ .023, ds ≥ 0.59, with no significant difference between acoustical and tactile details, ts(17) = 1.69, p = .109, d = 0.40. Participants in the self-imagination group tended towards generating more visual (M = 6.28) and tactile (M = 6.00) than acoustical impressions (M = 5.33), ts(17) ≥ 2.06, ps ≤ .055, ds ≥ 0.49, with no difference between visual and tactile impressions, t(17) = 1.05, p = .311, although it should be noted that there were clear ceiling effects with regard to both these ratings.

3.2.3. Discussion In sum, when matching perspectives of self-performance and otherimagination in Experiment 2, we found that repeatedly imagining someone else performing simple actions significantly increased false memories of self-performance. That is, we found self-other confusions after repeated imaginations. Contrary to a visual-similarity account, however, self-imagination was, if anything, more likely to evoke false memories than was other-imagination, even though our rating data showed that participants generated a good amount of visual details in both conditions. Taken together, some results from Experiments 1 and 2 were consistent with a visual-perspective account of false memories from imagination. However, other findings were not. Yet, perspective during performance was only manipulated between experiments. The significant effect of other-imagination in Experiment 2 (versus the nulleffect in Experiment 1) points to the critical role of visual perspective in other-imagination. However, comparisons across experiments face methodological limitations. To remedy this shortcoming, we independently manipulated the variables from the first two studies in Experiment 3. In this experiment, we not only varied imagined agent in Phase 2, but also actor in Phase 1. That is in Phase 1, the actions were performed (or only read) by either the participant or the experimenter. And in Phase 2, participants again imagined action performance of either themselves or the experimenter. Experiment 3 therefore employed a fully crossed design and thus allowed to examine the impact of both actor and imagined agent on performed-responses for self and other, respectively. If visual perspective is critical for false performed-responses to occur, then (a) other-imagination should lead to more false performedresponses when the actor was the other (vs. the self), however (b) self-imagination should lead to more false performed-responses when the actor was the self (vs. other).

4. Experiment 3 4.1. Method 4.1.1. Participants Seventy-nine students of the University of Kassel (64 women) participated for monetary compensation. Mean age was 24.28 years (SD = 5.65). 4.1.2. Design We used the same design as in the other experiments, with two changes: We included agent in Phase 1 (self vs. other) as an additional between-subjects' factor and only used two frequencies of imagination in Phase 2 (5 × vs. 0 ×). This reduction from three to two frequencies was not only more parsimonious, but also allowed for one-tailed testing of our hypotheses. 4.1.3. Materials and procedure Materials and procedure were identical with Experiments 1 and 2 with the following exceptions. In Phase 1 of Session 1, either the participants or the experimenters served as actors. To make this situation as agreeable as possible, in case of a read-instruction, we did not ask our participants and experimenters to perform the substitute hand-movements as in Experiments 1 and 2. Instead, both were asked to read the sentences out loud. This procedure has successfully been employed in similar studies (e.g., Lindner, Schain, Kopietz, & Echterhoff, 2012; Schain, Lindner, Beck, & Echterhoff, 2012) and also allowed us to reduce interactions between participants and experimenters. Participants and experimenters were seated across from each other at a table. For the sake of parsimony, we neither presented pictures in Phase 1, nor administered the corresponding rating-task. Likewise, no filler task was administered between Phase 1 and 2. In Phase 2 of Session 1, participants were again either instructed to imagine themselves or to imagine another person performing the described actions. However, this time, we did not use pictures of the objects (and the other person) to standardize imagination. Rather, the name of the action object(s) appeared on the screen, and then the object(s) were brought out by the experimenter who placed them between the participant and him-/herself. Then, the action statement appeared for 12 s and the participant was asked to either imagine him-/herself or the experimenter performing the action. Afterwards, the same ratings were used as in the other experiments with only one change: Frequency of action performance now needed to be rated on a five-point scale ranging from 1× to ≥5×. In Session 2, the memory test either referred to self-performance or other-performance, depending on condition. However, except for the instructions that referred to the different agents, the test was identical for the two groups. Instead of using a two-step memory test like in the first experiments, we directly asked our participants to choose from three options to answer whether a depicted action had been performed, read, or not presented (no mention was made of the subject in order to be able to use the same formulation in both groups). We have used this one-step procedure repeatedly in former studies and obtained comparable results to the two-step memory test used in Experiments 1 and 2 (e.g., Lindner et al., 2012; Schain et al., 2012). Action statements were reformulated from imperative (e.g., Open the book) to descriptive (infinitive plus object, e.g., To open the book). This was done to make the statements fit even if the subject switched between phases (e.g., self-performance, other-imagination). For the sake of parsimony, we omitted actions presented once in the second phase. Therefore, we had a total of 30 action statements in this study. From these actions, 20 were presented in Session 1 and 10 were used as distractors in Session 2. From the 20 statements used in Session 1, 10 were performed and 10 were only read in Phase 1. From each of these 10 action statements, 5 were imagined five times and 5 were not imagined in Phase 2. This time, we did not counterbalance action

I. Lindner, G. Echterhoff / Acta Psychologica 158 (2015) 51–60

statements, but randomly assigned the 30 items to the different itemtypes for each person individually. Three experimenters were involved in this study. They were trained to perform the actions similarly and they were also trained how and where to place the objects in the imagination-phase. All participants and experimenters were unbeknown to each other.

4.2. Results and discussion 4.2.1. Analyses of performed-responses Alpha was set to .05; ps are reported two-tailed for the global analysis and one-tailed in planned contrasts. Descriptive statistics can be found in Table 3. We ran a mixed 2 × 2 × 2 × 2-ANOVA with imagined agent in Phase 2 (other vs. self) as well as actual actor in Phase 1 (other vs. self) as between subjects' factors and frequency of imagination in Phase 2 (5× vs. 0×) as well as encoding in Phase 1 (performed vs. not read) as within subjects' factors on performed-responses. We also ran planned contrasts to test our specific hypotheses. We predicted that while other-imagination would lead to a higher increase in false memories if the actor was the other (vs. the self), this pattern would switch around for self-imagination. Technically speaking, we expected a three-way interaction between frequency of imagination, imagined agent in Phase 2, and actual actor in Phase 1. Indeed, this is what we found, F(1,75) = 4.07, p = .047, η P 2= .051. That is, the interaction between frequency of imagination and imagined agent in Phase 2 depended on actual actor in Phase 1. Because there was no interaction with encoding, in a first step, correct and false performedresponses were collapsed. In line with our prediction, a contrast analysis revealed that imagining someone else led to a higher increase in performed-responses when the actor was another person compared to when the actor was the self, t(75) = 2.20, p = .015, d = 0.46 (one-tailed p). However, different from what we had predicted, actor in Phase 1 did not make a difference when participants had imagined themselves, t(75) = 0.62, p = .262 (one-tailed p). There was also a main-effect of imagination, F(1,75) = 74.79, p b .001, η P 2= .499, with (repeated) imagination trials generally leading to more performed-responses than no imagination. Even though there was a main effect of encoding, F(1,75) = 303.64, p b .001, η P 2= .802, different from Experiments 1 and 2, there was no interaction between encoding and agent in Phase 2, F(1,75) = 1.46, p = .231. Thus,

Table 3 Experiment 3: Mean proportion of performed-responses as a function of agent and encoding in Phase 1 as well as agent and frequency of imagination in Phase 2. Standard deviations are given in parentheses. Imagined agent, Phase 2

Actor, Phase 1 Self

Other

Encoding, Phase 1 Performed Read Not presented Imagination inflation Performed Read Not presented Imagination inflation

Self

Other

Frequency of imagination, Phase 2

Frequency of imagination, Phase 2

5× .75 (.25) .20 (.22) – .17 (.23) .72 (.24) .31 (.28) – .17 (.24)

5× .75 (.28) .16 (.18) – .11 (.20) .90 (.19) .33 (.33) – .23 (.33)

0× .50 (.29) .03 (.07) .01 (.04) .55 (.20) .14 (.16) .02 (.05)

0× .57 (.27) .05 (.15) .01 (.03) .55 (.27) .10 (.23) .02 (.04)

Note: Proportions represent the frequency of performed-responses divided by the number of all responses for a corresponding item type. N = 21 in the self/self, 19 in the self/other, 19 in the other/self, and 20 in the other/other cell with the first notion referring to the actor in Phase 1 and the second to the agent in Phase 2. We added the size of the imagination–inflation effect (defined as the difference in false performed-responses between those items that have and those that have not been imagined) to the table due to the variability in baselines.

57

there was no evidence of better source discrimination after other(compared to self-)imagination in this experiment. However, there was a main effect of actor in Phase 1, F(1,75) = 4.91, p = .030, η P 2= .061, with other-performance leading to overall more performedresponses than self-performance. All other effects were not significant, Fs ≤ 2.89, ps ≥ .093. Taken together, other-imagination enhanced performed-responses more when the actor was the other (vs. the self), but there was no difference between actors when participants had imagined themselves. Because this pattern was not different for actions that were actually performed and those that were not, as indicated by the lack of an interaction with encoding, the analysis did not afford running separate analyses for both item-types. Nevertheless, we separately analyzed false performed-responses only. Even though we found the same trend that we had also found across encoding (i.e., other-imagination leading to more false memories when the actor was another person compared to the self), p was slightly above .05 in the corresponding contrast analysis, t(75) = 1.52, p = .066 (one-tailed p); however, the lack of a difference between actors in the self-imagination condition was clearly replicated, t(75) = .037, p = .485 (one-tailed p). Yet, the interaction between agent and frequency of imagination in Phase 2 and actor in Phase 1 in the corresponding ANOVA was not significant, F = 1.23. Simple-main effect analyses revealed significant inflation effects within all of the four groups, ts ≥ 2.25, ps ≤ .037, ds ≥ 0.52. We also ran the same ANOVA selectively for the two conditions that had already been employed in Experiment 1, that is when the actual actor was the self and the imagined agent was either the self or the other person. Results were different from Experiment 1 in that there was no significant interaction between frequency and imagined agent, F(1,38) = 1.21, p = .278, although there was clearly a main-effect of frequency of imagination, F(1,38) = 32.90, p b .001, η P 2= .464. That is, although imagination increased performed-responses in general and by trend self-imagination enhanced performed-responses more than other-imagination, this difference was not significant. At the same time, contrary to Experiment 1, this increase in false performedresponses was similar to that for correct performed-responses, that is there was no three-way-interaction with encoding, F b 1. Likewise, source discrimination was not better after other-imagination as indicated by the lack of an interaction between imagined agent and encoding, F b 1. 4.2.2. Additional analyses In a first step, we checked whether actual actor made a difference for other- or self-imagination, respectively. Because only for one rating, a significant difference was found, ratings for other- and selfimagination were collapsed across agents. We then compared otherand self-imagination as in Experiments 1 and 2. The other- and self-imagination groups did neither differ with regard to perceived difficulty of the task (M = 2.95 for other- and M = 2.50 for self-imagination), nor with regard to how often (M = 2.79 for other- and M = 2.86 for self-imagination) and how vividly (M = 5.14 for other- and M = 5.26 for self-imagination) they imagined the actions in Phase 2, ts ≤ 1.52, ps ≥ .132.4 There were also no significant differences with regard to the generation of visual (M = 6.23 for other- and M = 5.68 for self-imagination) and auditory (M = 5.26 for other- and M = 5.15 for self-imagination) details, ts ≤ 1.66, ps ≥ .101. A significant difference emerged inasmuch as other-imagination invoked tactile details to a smaller extent (M = 3.63) than self-imagination (M = 5.85), t(73) = 5.78, p b .001, d = 1.32. While there was no difference for other-imagination, self-imagination entailed a higher degree of tactile details when the actor had been another person (M = 6.32)

4 Please note that for these ratings, data from two people were missing and data from two other people were incomplete.

58

I. Lindner, G. Echterhoff / Acta Psychologica 158 (2015) 51–60

compared to the self (M = 5.43), t(38) = 2.17, p = .04, d = 0.70. No significant correlations were found between the increase in performedresponses and the ratings of sensory details across or within groups, rs ≤ .294, ps ≥ .086, and this was also true when the analysis was restricted to false performed-responses only. Comparisons within groups revealed that participants in the otherimagination group generated more visual (M = 6.23) than acoustical (M = 5.26) and tactile impressions (M = 3.63), ts(32) ≥ 2.50, ps ≤ .017, ds ≥ 0.42, and that they also generated more acoustical than tactile details, ts(34) = 3.30, p = .002, d = 0.56. This pattern was very similar to the one found in Experiment 2. Participants in the self-imagination group generated a comparable amount of visual (M = 5.68) compared to tactile (M = 5.85) as well as acoustical impressions (M = 5.15), ts(39) ≤ 1.34, ps ≥ .187. However, tactile details clearly succeeded acoustical details, t(39) = 2.24, p = .031, d = 0.35. This pattern was also similar to the one found in Experiment 2, except for the lack of a clear difference between visual and acoustical impressions. 4.2.3. Discussion Taken together, our results only partially supported the visualsimilarity account: Consistent with this account, when the imagined agent was another person, there was a significantly higher effect of imagination when the actor was another person as well (compared to the self). However, we did not find the same pattern for selfimagination. Here, actor did not make the difference that was predicted by the visual-similarity account. Yet, one could wonder whether the greater proportion of action memories between the actor conditions (self versus other) in the other-imagination group could be explained by differences in memory features per se. Possibly, deciding whether oneself vs. another person has performed an action is based on different features of memories for self- vs. other-performance. If such processes were critical, we should find an analogous difference in the self-imagination group. However, we did not. Similarly, Senkfor, Van Petten, and Kutas (2002) found no evidence for qualitative differences between memories for selfperformed and observed, other-performed actions. In Experiment 1, we did not find a significant amount of false memories when the self was the actor in Phase 1 and the other was the agent in Phase 2. In contrast, we did so in Experiment 3. We believe that this discrepancy is easily explained by the different operationalizations employed in these two experiments. In Experiment 3, we created a “live-imagination condition”, that is, the participants had a much more concrete idea about the to-be-imagined person, the action-object, and the way the person might perform an action. This is different from Experiment 1 (and 2) in which participants only saw a picture of an object and the torso of an unknown person. We propose that the operationalization chosen here might have facilitated imagination. In line with this argument, other-imagination was (descriptively) rated more difficult in Experiment 1 (M = 3.78; notably, in Experiment 2 as well, M = 3.83) than in Experiment 3 (M = 2.95). For selfimagination, the different operationalizations did also seem to alter difficulty of imagination, but not as much as for other-imagination (M = 2.94 in Experiment 1, M = 2.83 in Experiment 2, M = 2.50 in Experiment 3). Consistent with these data, source discrimination was not superior after other- (vs. self-) imagination in Experiment 3, while it had been superior in Experiments 1 and 2. Even though comparisons across experiments should be interpreted with caution, this suggests that the imagination–situation is an important moderator to consider in future research. 5. General discussion A substantial body of research has shown that imagining oneself performing a simple action can later lead to an illusion of actual performance. Our studies demonstrate that imagining another person can also induce false memories of self-performance. In Experiments 1 and 2,

participants were asked to imagine a person depicted on a picture. On this picture, the torso, arms, and hands of this person as well as the action-object were visible. False self-attributions were found when there was a match between the visual perspective at the time of performance and at the time of imagination (Experiment 2). In contrast, no such selfother confusion was found when visual perspective differed between performance and imagination (Experiment 1). In Experiment 3, the imagination–condition was more concrete inasmuch as participants were asked to imagine the experimenter sitting in front of them performing an action with the object lying in front of the experimenter. Here, we also found a significant amount of false memories of selfperformance after other-imagination. According to a prominent account of imagination inflation, the sensory similarity of performance and imagination is critical for the occurrence of the effect (e.g., Lampinen et al., 2003; Thomas et al., 2003). We have argued that visual features, especially perspective, should play a predominant role in the creation of false action memories from imagination (e.g., Lindner & Henkel, in press). Our results are partially consistent with this account: When perspectives were different between performance and other-imagination — as in Experiment 1 — we did not even find a significant amount of false memories of selfperformance while self-imagination reliably triggered false action memories. However, when perspectives between performance and other-imagination were matched — as in Experiment 2 —, the imagination of another person's actions resulted in a significant proportion of false memories of self-performance. Moreover in Experiment 3, otherimagination induced a higher amount of (false) action memories when the actor was another person compared to the self. Furthermore, across Experiments 2 and 3, additional data suggests that otherimagination primarily involved visual impressions. However, other results are inconsistent with the visual-similarity account. When perspectives differed between performance and imagination, the visual-similarity account would predict a lower rate of false memories after self-imagination (vs. other-imagination). Yet the data from Experiment 2 showed that imagination inflation was not lower, but, if anything, higher after self-imagination. Similarly, visual similarity could not explain why self-imagination led to a comparable amount of (false) action memories when the actor was the self compared to the other in Experiment 3. Taken together, while the visual-similarity account was successful in predicting the pattern of results for other-imagination across Experiments 1 and 2 and within Experiment 3, it was not so for self-imagination. Two explanations for the present findings come to mind. On the one hand, it is possible that people use sensory cues when deciding whether an action had or had not been performed, but visual input is less important than other cues, like tactile or auditory cues, especially after selfimagination. Indeed, ratings in the self-imagination group revealed that participants generated visual and tactile impressions to a similar degree.5 However, we did not find any significant correlation between the reported amount of tactile and auditory details in the mental images and the size of the imagination–inflation effect in Experiments 2 and 3. To further address this issue, however, ratings of sensory qualities should be obtained at the item-level rather than a global, comprehensive level as in our experiments. On the other hand, people might base their memory judgments on other cues than visual, acoustical, or tactile ones. In our studies of observation–inflation (e.g., Lindner et al., 2010) we have repeatedly shown that observing another person performing an action can lead to false memories of self-performance. Hence, in these studies people were not asked to imagine another person performing actions, but actually observed another person performing actions in Phase 2. Because our results were not consistent with a sensory-similarity account (Lindner et al., 2010, Exp. 3; Lindner et al., 2012), we have proposed an 5 Although it should be noted that ceiling-effects might have obscured any differences, here.

I. Lindner, G. Echterhoff / Acta Psychologica 158 (2015) 51–60

alternative account building on motor-simulation research. A growing body of literature has shown that observers covertly simulate actions performed by another person as if they are performing these actions themselves (e.g., Grèzes & Decety, 2001; Wilson & Knoblich, 2005). It has also been shown that the motor representations activated during such cortical simulations are reactivated at test (e.g., Senkfor et al., 2002; Wutte, Glasauer, Jahn, & Flanagin, 2012). Based on these findings, we have argued that these mechanisms could result in the described self-other confusion stemming from observation. Interestingly, research on motor simulation has not only shown that observing another person performing an action, but also that imagining oneself performing an action activates motor representations similar to self-performance (e.g., Grèzes & Decety, 2001; Jeannerod, 2001). Moreover, motor simulation appears to be stronger for self-imagination than other-imagination (Lorey et al., 2009; see also Sirigu & Duhamel, 2001; Ruby & Decety, 2001). Taken together, the present pattern of results is consistent with a distinction that has been suggested in the literature: While selfimagination is thought to mainly rely on motor imagery, otherimagination is thought to mainly rely on visual imagery (Denis et al., 1991; Lorey et al., 2009; Sirigu & Duhamel, 2001). By the same token, false action memories from self-imagination might be triggered to a relatively greater extent by motor representations, whereas false action memories from other-imagination might be triggered to a relatively greater extent by visual representations. This account also offers an explanation for our findings. We robustly found false memories after observing (e.g., Lindner et al., 2010, 2012; Schain et al., 2012), but this was not the case after imagining another person performing simple actions in Experiment 1. Also, perspective has not been found to impact false memories from observing (Lindner et al., 2010, Exp. 3), whereas it has been found to impact false memories from imagining someone else (but not oneself) in the present experiments. We suspect that observation (vs. imagination) of another person is more likely to induce motor simulation and think that this view merits further investigation. The substantial appropriation of another person's acts that we demonstrated within our second and third experiment can lead to socially undesirable consequences in real life (as is the case in unconscious plagiarism, Brown & Murphy, 1989). However, our data suggest that agency confusions are overall less likely than are modality confusions, that is, false memories from self-imagination. Moreover, other-imagination enhanced source discrimination compared to self-imagination in Experiments 1 and 2. This was not the case in Experiment 3, however. We have suggested that this was due to the more concrete nature of the imagination–situation in Experiment 3. This is an aspect which should be addressed by future research. A practical conclusion that one could draw from our studies across experiments, however, is that otherimagination seems to be favorable compared to self-imagination when it comes to (true and false) memories of recent self-performance (see also Marsh et al., 2014). Yet, our research also reveals a general tendency for participants to say performed more often when the actor was another person than when the actor was the self. This finding resonates well with literature on the enactment effect showing that people are best at discriminating what they did themselves (Cohen, 1989; Engelkamp & Zimmer, 1994), and with the broader literature on mnemonic benefits of self-referencing (Rogers, Kuiper, & Kirker, 1977; Slamecka & Graf, 1978). In our third experiment, we did not find any other interactions with actor than the predicted three-way interaction, however. That is, self- and other-performance only made a difference after other- but not self-imagination. We have thus proposed that memories for self- and other-performance were unlikely to differ systematically in our study (see also Senkfor et al., 2002). Still, we believe that it is worthy for future investigations to examine under which conditions source memory for self- vs. other-performance do vs. do not differ.

59

One interesting detail that we found across Experiments 1 and 2 was that imagination enhanced correct and false performed-responses differentially. A single imagination-trial was already sufficient to increase correct action memories, but it was not so for false action memories. This might possibly be due to initial performance facilitating subsequent imagination. However, this possibility needs to be addressed by future research. 6. Conclusion Imagination techniques are frequently used to aid our memory. However, imagination is not entirely beneficial for human cognition, but might play tricks on our memories. Beyond replicating false action memories from self-imagination we have demonstrated that it is even possible to mix up imagined actions by another person with one's own actions. Our data suggest that self-imagination and otherimagination rely on different mechanisms. Future research should pinpoint more precisely when and how the imagination of others' actions can induce false self-attributions in action memory. Acknowledgments This work was partially supported by the Zentrale Forschungsförderung (Research Promotion Program) of the University of Kassel. We are grateful to Gerhard Mutz and Moritz Happel for their dedicated technical assistance, Michael Mentzner for patiently serving as the actor for the pictures, as well as Kristina Eichel, Moritz Happel, Sina Leichner, Lisa Oswald, Christina Walther, and Marleen Wiens for their much appreciated help in collecting the data. We also wish to thank Cécile Schain for contributing valuable ideas to the argumentation and Linda Henkel for her remark on correlations between overallratings and false memories. References Brown, A. S., & Murphy, D. R. (1989). Cryptomnesia: Delineating inadvertent plagiarism. Journal of Experimental Psychology: Learning, Memory, and Cognition, 15, 432–442. http://dx.doi.org/10.1037/0278-7393.15.3.432. Cohen, R. L. (1989). Memory for action events: The power of enactment. Educational Psychology Review, 1, 57–80. http://dx.doi.org/10.1007/BF01326550. Decety, J., & Grèzes, J. (2006). The power of simulation: Imagining one's own and other's behavior. Brain Research, 1079, 4–14. http://dx.doi.org/10.1016/j.brainres.2005.12.115. Denis, M., Engelkamp, J., & Mohr, G. (1991). Memory of imagined actions: Imagining oneself or another person. Psychological Research, 53, 246–250. http://dx.doi.org/ 10.1007/BF00941394. Denis, M., & Kosslyn, S. M. (1999). Scanning visual mental images: A window on the mind. Current Psychology of Cognition, 18, 409–465. Engelkamp, J., & Zimmer, H. D. (1994). The human memory. A multi-modal approach. Seattle: Hogrefe & Huber. Garry, M., Manning, C. G., Loftus, E. F., & Sherman, S. J. (1996). Imagination inflation: Imagining a childhood event inflates confidence that it occurred. Psychonomic Bulletin & Review, 3, 208–214. http://dx.doi.org/10.3758/BF03212420. Goff, L. M., & Roediger, H. L. (1998). Imagination inflation for action events: Repeated imaginings lead to illusory recollections. Memory & Cognition, 26, 20–33. http://dx. doi.org/10.3758/BF03211367. Grèzes, J., & Decety, J. (2001). Functional anatomy of execution, mental simulation, observation, and verb generation of actions: A meta-analysis. Human Brain Mapping, 12, 1–19. http://dx.doi.org/10.1002/1097-0193(200101)12:1b1::AID-HBM10N3.0.CO;2-V. Hornstein, S. L., & Mulligan, N. W. (2004). Memory for actions: Enactment and source memory. Psychonomic Bulletin & Review, 11, 367–372. http://dx.doi.org/10.3758/ BF03196584. Jeannerod, M. (2001). Neural simulation of action: A unifying mechanism for motor cognition. NeuroImage, 14, S103. http://dx.doi.org/10.1006/nimg.2001.0832. Jeannerod, M., & Pacherie, E. (2004). Agency, simulation and self-identification. Mind & Language, 19, 113–146. http://dx.doi.org/10.1111/j.1468-0017.2004.00251.x. Johnson, M. K., Foley, M. A., Suengas, A. G., & Raye, C. L. (1988). Phenomenal characteristics of memories for perceived and imagined autobiographical events. Journal of Experimental Psychology: General, 117, 371–376. Johnson, M. K., Hashtroudi, S., & Lindsay, S. D. (1993). Source monitoring. Psychological Bulletin, 114, 3–28. http://dx.doi.org/10.1037//0033-2909.114.1.3. Johnson, M. K., & Raye, C. L. (1981). Reality monitoring. Psychological Review, 88, 67–85. Lampinen, J. M., Odegard, T. N., & Bullington, J. F. (2003). Qualities of memories for performed and imagined actions. Applied Cognitive Psychology, 17, 881–893. http:// dx.doi.org/10.1002/acp.916. Libby, L. K. (2003). Imagery perspective and source monitoring in imagination inflation. Memory & Cognition, 31, 1072–1081. http://dx.doi.org/10.3758/BF03196128.

60

I. Lindner, G. Echterhoff / Acta Psychologica 158 (2015) 51–60

Lindner, I., Echterhoff, G., Davidson, P. S. R., & Brand, M. (2010). Observation inflation: Your actions become mine. Psychological Science, 21, 1291–1299. http://dx.doi.org/ 10.1177/0956797610379860. Lindner, I., & Henkel, L. A. (in press). Confusing what you heard with what you did: False action-memories from auditory cues. Psychonomic Bulletin & Review. Lindner, I., Schain, C., Kopietz, R., & Echterhoff, G. (2012). When do we confuse self and other in action memory? Reduced false memories of self-performance after observing actions by an out-group versus in-group actor. Frontiers in Psychology, 3, 467. http://dx.doi.org/10.3389/fpsyg.2012.00467. Lorey, B., Bischoff, M., Pilgramm, S., Stark, R., Munzert, J., & Zentgraf, K. (2009). The embodied nature of motor imagery: The influence of posture and perspective. Experimental Brain Research, 194, 233–243. http://dx.doi.org/10.1007/s00221-008-1693-1. Marsh, B. U., Pezdek, K., & Lam, S. T. (2014). Imagination perspective affects ratings of the likelihood of occurrence of autobiographical memories. Acta Psychologica, 150, 114–119. http://dx.doi.org/10.1016/j.actpsy.2014.05.006. Posner, M. I., Nissen, M. J., & Klein, R. M. (1976). Visual dominance: An informationprocessing account of its origins and significance. Psychological Review, 83, 157–171. http://dx.doi.org/10.1037/0033-295X.83.2.157. Rogers, T. B., Kuiper, N. A., & Kirker, W. S. (1977). Self-reference and the encoding of personal information. Journal of Personality and Social Psychology, 35, 677–688. http://dx.doi.org/10.1037/0022-3514.35.9.677. Ruby, P., & Decety, J. (2001). Effect of subjective perspective taking during simulation of action: A PET investigation of agency. Nature Neuroscience, 4, 546–550. http://dx. doi.org/10.1038/87510. Schacter, D. L., & Addis, D. R. (2007). The cognitive neuroscience of constructive memory: Remembering the past and imagining the future. Philosophical Transactions of the Royal Society, B: Biological Sciences, 362, 773–786. http://dx.doi.org/10.1098/rstb. 2007.2087. Schain, C., Lindner, I., Beck, F., & Echterhoff, G. (2012). Looking at the actor's face: Identity cues and attentional focus in false memories of action performance from observation. Journal of Experimental Social Psychology, 48, 1201–1204. http://dx.doi.org/10.1016/j. jesp.2012.04. 003.

Seamon, J. G., Blumenson, C. N., Karp, S. R., Perl, J. J., Rindlaub, L. A., & Speisman, B. B. (2009). Did we see someone shake hands with a fire hydrant?: Collaborative recall affects false recollections from a campus walk. The American Journal of Psychology, 122, 235–247. Seamon, J. G., Philbin, M. M., & Harrison, L. G. (2006). Do you remember proposing marriage to the Pepsi machine? False recollections from a campus walk. Psychonomic Bulletin & Review, 13, 752–756. http://dx.doi.org/10.3758/BF03193992. Senkfor, A. J., Van Petten, C., & Kutas, M. (2002). Episodic action memory for real objects: An ERP investigation with perform, watch, and imagine action encoding tasks versus a non-action encoding task. Journal of Cognitive Neuroscience, 14, 402–419. http://dx.doi.org/10.1162/089892902317361921. Shepard, R. N., & Cooper, L. A. (1986). Mental images and their transformations. Cambridge: MIT Press. Sirigu, A., & Duhamel, J. R. (2001). Motor and visual imagery as two complementary but neurally dissociable mental processes. Journal of Cognitive Neuroscience, 13, 910–919. http://dx.doi.org/10.1162/089892901753165827. Slamecka, N. J., & Graf, P. (1978). The generation effect: Delineation of a phenomenon. Journal of Experimental Psychology: Learning, Memory, and Cognition, 4, 592–604. http://dx.doi.org/10.1037/0022-3514.35.9.677. Suddendorf, T., & Corballis, M. C. (2007). The evolution of foresight: What is mental time travel, and is it unique to humans? Behavioral and Brain Sciences, 30, 299–351. http://dx.doi.org/10.1017/S0140525X07001975. Thomas, A. K., Bulevich, J. B., & Loftus, E. F. (2003). Exploring the role of repetition and sensory elaboration in the imagination inflation effect. Memory & Cognition, 31, 630–640. http://dx.doi.org/10.3758/BF03196103. Thomas, A. K., & Loftus, E. F. (2002). Creating bizarre false memories through imagination. Memory & Cognition, 30, 423–431. http://dx.doi.org/10.3758/BF03194942. Wilson, M., & Knoblich, G. (2005). The case for motor involvement in perceiving conspecifics. Psychological Bulletin, 131, 460–473. http://dx.doi.org/10.1037/0033-2909.131.3.460. Wutte, M. G., Glasauer, S., Jahn, K., & Flanagin, V. L. (2012). Moving and being moved: Differences in cerebral activation during recollection of whole-body motion. Behavioural Brain Research, 227, 21–29. http://dx.doi.org/10.1016/j.bbr.2011.09.042.