Event-related potential (ERP) evidence for sensory-based action memories

International Journal of Psychophysiology 62 (2006) 193 – 202 www.elsevier.com/locate/ijpsycho

Event-related potential (ERP) evidence for sensory-based action memories P. Andrew Leynes a,⁎, Jeremy A. Grey a , Jarret T. Crawford b a

The College of New Jersey, United States b Rutgers University, United States

Received 8 September 2005; received in revised form 11 April 2006; accepted 21 April 2006 Available online 12 June 2006

Abstract Memory for performed and motioned actions was measured on source recognition and source recall tests in order to investigate memory for actions or output monitoring (OM). Event-related potentials (ERPs) were recorded during the source recognition test to provide insight into the basis of OM. Source identification and recall of performed actions was greater than motioned actions indicating that sensory characteristics provide unique information for action memories. The ERP data supported this interpretation because the brain activity elicited by performed actions differed from motioned and new actions. Early parietal ERP differences suggest that sensory information leads to selective recollection of performed actions or that more sensory information was activated by performed actions during remembering. A large late posterior negativity (LPN) was also observed in the absence of frontal ERP differences, which are typically observed during source monitoring. This pattern of ERP differences is evidence that frontal ERPs and the LPN reflect distinct source monitoring processes. Based on the available data, we argue that frontal ERPs reflect general decision processes that evaluate diagnostic information, whereas the LPN reflects processes that are engaged when a detailed inspection of information is required by the context. © 2006 Elsevier B.V. All rights reserved. Keywords: Source memory; ERP; Actions

People must be able to remember past actions (i.e., retrospective memory) and maintain intentions for future actions (i.e., prospective memory) in order to function in many contexts. Failures in these forms of cognition can range from mild (e.g., forgetting to put one’s car in park) to serious (e.g., leaving a sleeping infant in a hot car). Prospective memory has been the subject of a number of studies (see Henry et al., 2004 for a recent review) and a separate body of work has focused on retrospective memory for actions (see Engelkamp, 1998 or Zimmer and Cohen, 2001 for reviews). However, there is obvious overlap between these two areas of research because prospective memory can fail when a search of retrospective memory leads one to believe that the action has been completed (Leynes and Bink, 2002; Leynes et al., 2005b; Schaefer et al., 1998). As a result, one way to integrate the literature is to investigate the factors that affect retrospective ⁎ Corresponding author. Department of Psychology, The College of New Jersey, P.O. Box 7718, Ewing, NJ 08628-0718, United States. Tel.: +1 609 771 2624; fax: +1 609 637 5178. E-mail address: [email protected] (P.A. Leynes). 0167-8760/$ - see front matter © 2006 Elsevier B.V. All rights reserved. doi:10.1016/j.ijpsycho.2006.04.003

memory for actions in an effort to understand how the safety and quality of life may be compromised in many situations. The majority of retrospective action memory studies have used a paradigm in which participants perform simple actions in the laboratory (e.g., roll the ball), which are called subject performed tasks. Recognition or recall of these actions is typically superior to memory for actions that are performed by the experimenter or actions that are learned verbally (the “enactment effect”). Theories explaining this effect have been advanced including a view that the motor information in the memory trace creates this effect (Engelkamp, 2001) or that actions create a rich integration of the information and the person (Kormi-Nouri and Nilsson, 2001). As a result, many studies in this field have attempted to distinguish between competing explanations for the enactment effect, which also furthers an understanding of the basic organization of memory. Although this body of work has yielded many important and interesting findings, Leynes and colleagues (Leynes and Bink, 2002; Leynes et al., 2005b) have focused on the factors that affect retrospective memory for actions. They argue that discriminating

194

P.A. Leynes et al. / International Journal of Psychophysiology 62 (2006) 193–202

between different types of actions (e.g., performed versus imagined actions) can be viewed as a special case of source monitoring (i.e., memory for the origin of information; Johnson et al., 1993). According to this argument, actions that have been performed are one source of information, whereas other types of actions (imagined or other performed actions) are another source of information. Leynes and colleagues referred to these cases of source monitoring as output monitoring (OM) to connect to related work that investigates other situations in which people make judgments concerning their own output (see Marsh and Hicks, 2001). The advantage of this approach is that the theory that describes source monitoring, the Source Monitoring Framework (Johnson et al., 1993), can be used to guide and interpret additional studies of action memory (Hornstein and Mulligan, 2004). The Source Monitoring Framework describes the cognitive processes that govern remembering the origin of information. Source memory is not directly retrieved; rather, source judgments are supported by the information in memory that is activated (qualitative memory characteristics) after which the activated information is evaluated by a complex set of decision processes in order to determine the source. Qualitative memory characteristics include the many features that become bound into memory when the information is encountered (e.g., perceptual details, spatial and temporal information, semantic detail, affective information, motor information, and cognitive operations). After qualitative characteristics are activated, heuristic or systematic decision processes govern the evaluation of memory characteristics to render a source judgment (Johnson et al., 1993). Johnson et al. (1993) suggested that the type of processes used in many source monitoring (and by extension OM situations) may differ somewhat from those used when recalling or recognizing actions because source judgments rely on more differentiated information and the decision processes may vary depending on the context. This hypothesis, as it relates to action memory, has already received some support. For example, Kormi-Nouri (2000) reported that recall of performed and imagined actions did not differ, whereas Leynes and Bink (2002) found that more performed than imagined actions were recognized and recalled when participants determined whether actions were “performed”, “imagined”, or “new.” Furthermore, some novel predictions arise when action memory is considered as source memory. For example, Hornstein and Mulligan (2004) found that enhancing visual information did not affect the enactment effect (item memory), but more visual information led to more source confusions. Because increasing the similarities between the sources creates more source confusions in many other contexts (Johnson et al., 1993), these results suggest source monitoring can successfully guide research on action memory when people are asked to discriminate between different types of actions. Recently, event-related potentials (ERPs) have provided insight into the cognitive processes that are used when discriminating between different types of actions because several studies have examined ERPs during other types of source monitoring (e.g., Rugg et al., 1998; Senkfor and Van Petten, 1998; Wilding, 1999; Wilding and Rugg, 1996). In one study, Leynes

and Bink (2002) asked participants to perform some actions and to imagine and plan to perform another set of actions at study. ERPs were recorded during a source recognition test that asked participants to determine if the test probe was a performed action, an imagined action, or was new. They observed ERP effects that were similar to other source monitoring ERP studies because about 600 ms after the probe both old sources elicited more positive ERPs than new (an old/new effect). Most source monitoring and recognition studies (which have predominantly used verbal materials) report an old/new effect around 600 ms which tends to be largest at parietal electrodes. This effect has been hypothesized to reflect recollection of information stored in memory (e.g., Paller and Kutas, 1992; Paller et al., 1995; Wilding, 2000; Wilding and Rugg, 1996). The old/new difference in the Leynes and Bink study tended to have a more central distribution; consequently, the authors suggested the more central distribution reflected activation of the motor cortex when sensorimotor information associated with imagining or performing actions was reactivated during the test (see Senkfor et al., 2002 for a similar claim). This interpretation is consistent with a positron emission tomography study of memory for actions that reported increased activation in right motor cortex when performed actions were recalled (Nilsson et al., 2000) as well as the claim that motor information contributes to the enactment effect (Engelkamp, 2001). Leynes and Bink (2002) also observed later ERP differences at frontal electrode sites as well as a late posterior negativity (LPN). The frontal ERP differences have been observed in many source monitoring studies and are presumed to reflect post-retrieval processing (Mecklinger, 2000; Ranganath and Paller, 2000; Wilding, 1999; Wilding and Rugg, 1997) or source monitoring decision processes (Leynes, 2002; Leynes and Bink, 2002; Leynes et al., 2003, 2005b,). The LPN has also been observed in many studies and it has been hypothesized to reflect a search for additional attributes in memory (see Johansson and Mecklinger, 2003 for a review). Recently, Leynes, Crawford, and Bink (2005b) investigated the effect of interrupting actions on OM. In their study, participants completed some actions and began to complete another set of actions when the experimenter interrupted the completion of these actions. They argued that initiating (but not completing actions) would make the actions more similar to performed actions because the memory trace would also contain some sensorimotor information, which should lead to more source confusions (Hornstein and Mulligan, 2004; Johnson et al., 1993). The behavioral data supported this prediction because more interrupted actions were misattributed to performed actions. The ERP data suggested that memories for performed actions differed from interrupted actions because amplitude differences in the early parietal ERPs were observed (i.e., performed ERPs were more positive than interrupted ERPs and new ERPs). Although the results of Leynes, Crawford, and Bink (2005b) study suggest that different types of information can be activated and consulted to support OM, our understanding of action memory can be furthered by similar studies that investigate how various kinds of information impact OM. For example, performed

P.A. Leynes et al. / International Journal of Psychophysiology 62 (2006) 193–202

and interrupted actions shared both sensory (i.e., visual, auditory, olfactory, and tactile) and motor memory characteristics in the Leynes, Crawford, and Bink study; therefore, it is unclear how sensory information alone influences OM. This issue is complicated by the various methods used by different laboratories to investigate the enactment effect because the use of real and imagined objects has varied across studies and laboratories (see Zimmer and Cohen, 2001 for a review). Engelkamp and Zimmer (1997) examined this issue by comparing recall of actions that are performed with those that were performed with imaginary objects. Performing actions with real objects lead to better recall overall than actions performed with imaginary objects although the relative size of enactment effect was unaffected. This result suggests that sensory information has a general contribution to memory for actions but does not specifically produce the enactment effect; consequently, this issue has not been a central focus of the action memory literature. However, the role of sensory information remains important from the source monitoring view because many different types of information can serve as a basis for source memory judgments (Hornstein and Mulligan, 2004; Johnson et al., 1993). For example, motor information will be activated when it differentiates actions; however, other qualitative characteristics (e.g., sensory information) will be used when motor information is not diagnostic. This prediction has not yet been tested. The purpose of the present study was to determine whether sensory information affects OM judgments and ERPs. Towards this end, participants performed some actions and motioned others (i.e., performed the action without real objects). In this context, motioned and performed actions would share motor information in the memory trace but differ in the amount of sensory information. The definition of sensory information is broadly defined in the present investigation because performing an action (versus executing the motions) can include a range of sensations, including visual, auditory, olfactory, gustatory, and tactile experiences. ERPs were recorded during a source recognition test and this phase was followed by two source recall tests in order to provide additional insight (cf., Leynes and Bink, 2002; Leynes et al., 2005b). ERPs can provide evidence that sensory information has a unique contribution to OM if performed and motioned actions elicit different ERPs. Conversely, both types of actions might elicit indistinguishable ERPs if the memories for these actions are similar given that some studies have reported no ERP differences between studied sources (e.g., Senkfor and Van Petten, 1998; Wilding and Rugg, 1996). 1. Method 1.1. Participants Twenty-three College of New Jersey students (9 males), aged 17–22, volunteered in exchange for course credit. The data for 4 of these participants was lost due to defective storage media. The data from the other 19 participants (6 males) was included in the data analysis. An incentive of $25 was offered to the subject who had the best memory and fastest reaction time

195

among those tested. All participants were right handed (Oldfield, 1971), and reported that they did not have a history of neurological disease. 1.2. Stimuli A set of 150 actions were used as stimuli in the present experiment. The actions were the same as those used by Leynes, Crawford, and Bink (2005b). Each action was simple, three words long, and contained a unique verb and noun (e.g., “bend the wire”, “crack the egg”, etc.). Each character presented on the monitor extended 0.4° of visual angle vertically and horizontally. For each participant, the computer randomly assigned 50 actions to each of the three item types (performed, motioned, or new). The objects used to perform the actions were in plain view and were located in the central room in the laboratory suite. 1.3. Procedure During the study session, participants encountered 100 actions for a nonspecific test of memory for actions. Participants were instructed — prior to the study phase — that some of the actions would be performed and that some would be motioned. On each study trial, the instruction “Perform” or “Motion” was displayed in the center of the computer monitor, and the action phrase was displayed below the instruction. The software paused after each trial to allow time to complete each action. When the action was to be performed, the participant located the appropriate object and began to perform the action. When the action was to be motioned, the participant located the object but did not touch or manipulate it. Instead, he or she performed the motions that would complete the action. All actions during the study phase were self-paced. Before the study trials began, participants were allowed a few minutes to visually examine the objects to become familiar with the experimental materials. Immediately following the study phase, the experimenter described the ERP recording procedures and attached the electrode cap (see below). Therefore, there was an average delay of 44 min between the study phase and the source recognition test. The testing phase consisted of one source recognition test, which was completed first, and two recall tests (described below). After the electrode cap was attached, the participant entered a separate, sound-attenuated room to record ERPs during the source recognition test (ERPs were not recorded during recall). During the source recognition test, the entire action phrase was presented one-at-a-time in the center of the computer monitor and remained on the screen until a response was registered. All 50 performed, 50 motioned, and 50 new actions were presented in a random sequence. Participants were instructed to decide if the action was performed at study, was motioned, or was new and register their response by pressing one of three marked keys on a computer keyboard using their right-hand index finger, middle finger, or thumb. Following the source recognition test, participants re-entered the room used for the study phase; consequently, all of the objects were in plain view for the two subsequent recall tests. Stimulus materials were returned to their original position

196

P.A. Leynes et al. / International Journal of Psychophysiology 62 (2006) 193–202

before the recall test commenced in order to remove any obvious cues that could be used to determine if actions were previously performed. On the “recall performed” test, participants were asked to verbally recall (in any order) all of the tasks that they performed during the study phase. Participants were asked to verbally recall all of the tasks that were motioned during the study phase on the “recall motioned” test. For both recall tests, the experimenter recorded the actions on a sheet of paper that was obscured from the view of the participant. The order of the recall tests was counterbalanced across participants. The experimental session was concluded, and participants were debriefed after the second recall test. 1.4. ERP recording procedures ERPs were recorded during the source recognition test only. Potentials were sampled at a rate of 150 Hz from twenty-nine Ag/AgCl electrodes mounted in an elastic cap (Neuromedical Supplies, Inc.) referenced to the left mastoid online. Electrode voltages were re-referenced offline to average of the left and right mastoids. Electrodes were placed over the frontal lobes (Fp1, Fp2, F7, F3, Fz, F4, F8, FC3, FCz, FC4), temporal lobes (FT7, FT8, T7, T8, TP7, TP8), parietal lobes (CP3, CPz, CP4, P7, P3, Pz, P4, P8), occipital lobes (O1, O2), and at the central position on the scalp (C3, Cz, C4). Vertical electrooculogram (vEOG) was recorded bipolarly using two Ag/AgCl electrodes affixed above and below the subject's left pupil. Horizontal electrooculogram (hEOG) was recorded bipolarly from identical electrodes and attached to the outer canthi of both eyes. Interelectrode impedance was below 5 kΩ. EEG and EOG signals were recorded with a Contact Precision Instruments amplifier with a 0.03–40 Hz frequency range (− 3 dB attenuation). During the test phase, EEG and EOG were continuously sampled. 1.5. ERP data processing To create ERPs off-line, the continuous data file was divided into epochs that began 300 ms before the presentation of the action phrase and continued for 2800 ms. The data were then digitally filtered using a 30 Hz lowpass filter (− 3 db/oct). Ocular artifacts were corrected using the algorithm developed Table 1 Proportion of responses on the source recognition test (top) and reaction times (bottom) as a function of source type Responses

Source type Performed

Motioned

New

Claim “Performed” “Motioned” “New”

.96 (.05) .03 (.03) .01 (.02)

.05 (.04) .88 (.08) .06 (.06)

b.01 (.01) .01 (.02) .98 (.02)

Reaction time Hits

1818 (429)

2352 (580)

1717 (402)

Note. Standard deviation is in parentheses. Bold numbers represent hits for each source type.

by Semlitsch, Anderer, Schuster, and Presslich (1986). Trials on which ERP amplitudes exceeded ± 150 μV (M = 12%, SD = 14% equally distributed across source type) were excluded from the analyses by computer software. 2. Results Unless specified otherwise, all effects reported in this article are significant at the α = 0.05 level. 2.1. Behavioral data 2.1.1. Source recognition The proportion of performed, motioned, and new responses as a function of source type are displayed in Table 1.1 Although source recognition was at near ceiling levels, these results are entirely consistent with other investigations of memory for actions (Goff and Roediger, 1998; Foley and Ratner, 1998; Thomas and Loftus, 2002) as well as other ERP studies of source monitoring (Leynes and Bink, 2002; Leynes et al., 2005b; Ranganath and Paller, 2000; Senkfor and Van Petten, 1998). The first set of analyses revealed that correct identifications of all three items differed as a function of source, F(2, 36) = 32.32. Post hoc analyses revealed that performed (F(1, 18) = 32.75) and new actions (F(1, 18) = 42.41) were identified at a higher rate than motioned actions, whereas new and performed actions did not differ reliably, F(1, 18) = 3.89, p N .05. Detailed analyses of source monitoring errors were conducted to provide additional insight into source-monitoring processes because two kinds of errors may occur on a source-monitoring test of this type. A source may be misattributed to the other source (i.e., a source confusion) or it may be called “new” (i.e., missed). A series of Bonferroni-corrected F-tests were used to analyze all possible types of errors. None of the results of these analyses were significant, except that motioned actions were missed more often than performed actions (.06 vs. .01: F(1, 18) = 18.08). Response bias, which is reflected in the false alarms, can influence source monitoring in some contexts (Johnson et al., 1993). For example, Leynes and Bink (2002) found a bias to call false alarms “planned” (i.e., an it-had-to-be planned effect). No such bias was detected in the present experiment (F(1, 18) = 5.66, p N .05); however, this result should be cautiously interpreted because the overall false alarm rate was very low in this experiment. 2.1.2. Reaction time Response times for correct source judgments on the source recognition test were analyzed and are displayed in the bottom of Table 1. An analysis of variance revealed that reaction times varied as a function of source, F(2, 36) = 37.12. Post hoc analyses

1 In a few instances, participants made errors when encoding the actions during the study phase. More specifically, some of the motioned actions were actually performed (i.e., 15 out of 1050 total motioned actions studied) and some of the performed actions were inadvertently motioned (i.e., 3 out of 1050 total performed actions studied). These trials were reclassified to the proper category; consequently, proportions were adjusted for each person to reflect the true number of performed and motioned actions.

P.A. Leynes et al. / International Journal of Psychophysiology 62 (2006) 193–202

revealed that identification of performed and new sources were equivalent, F(1, 18) = 3.98, p N .05. Thus, the overall difference in reaction times arose from the fact that motioned action response times were slower than both performed (F(1, 18) = 42.01) and new actions (F(1, 18) = 42.63). 2.1.3. Source recall Recall was measured by verbal recall of both types of actions on separate recall tests. Two methods (i.e., liberal and strict scoring) were used to score verbal recall of both types of actions on the two separate recall tests because participants could verbally reproduce an action but fail to recall the verbatim action phrase. For example, recalling “jingle the chain” when the action was rattle the chain would be scored as a hit using the liberal method but as an error using the strict method. More performed actions (M = .72, SD = .11) were recalled than motioned actions (M = .58, SD = .14; t(18) = 6.14) when the strict scoring method was used and when the liberal scoring method was used (performed M = .86, SD = .10 vs. motioned M = .74, SD = .16; t(18) = 4.30). Thus, the same pattern was observed using both scoring methods. Errors on the source recall tests were separated into retrospective memory and output monitoring errors and were then analyzed for a comparison to earlier OM studies.2 The analysis of recall errors revealed a significant Source Test by Error interaction was also significant, which indicated that the type of error varied as a function of the source test, F(5, 90) = 12.21. Visual inspection of the errors suggested that this interaction arose from more omission errors when recalling motioned actions (M = .238, SD = .164) than when recalling performed actions (M = .127, SD = .077). This observation was confirmed by a post hoc analysis of errors that excluded omissions because the Source Test by Error interaction was not significant in this analysis, F(4, 72) b 1, p N .05. Therefore, output monitoring errors did not differ for these two types of actions. 2.2. ERP data ERPs were averaged from accurate test trials on the source recognition test to form three ERPs (i.e., performed, motioned, and new). The few instances where participants made an error at encoding (completing a motioned action or motioning a performed action) were omitted from the ERP data analysis. A mean number of 45, 41, and 47 trials comprised the performed, motioned, and new item ERPs, respectively. ERPs for incorrect source judgments were not analyzed due to an insufficient number of trials to form reliable ERP averages. To quantify the ERP effects, seven amplitude measures were computed as the average activity over consecutive 200 ms intervals beginning 400 ms after the probe relative to the average activity 300 ms before the probe (i.e., 400–600, 600–800, 800–1000, 1000– 1200, 1200–1400, 1400–1600, 1600–1800 ms after the onset of the test probe). These time intervals were selected to capture the ERP activity leading up to the fastest average response time 2 See Leynes, Crawford, and Bink, 2005b for a detailed description of the types retrospective and output monitoring errors on a recall test.

197

(i.e., new items). These amplitude measures were analyzed separately at 25 electrodes – omitting the most extreme frontal (i.e., Fp1 and Fp2) and parietal electrodes (i.e., O1 and O2). Thus, an analysis of variance model that contained a factor for Source (performed/motioned/new), Anterior/Posterior (AP) electrode placement (5 levels front to back), and Left/Right (LR) electrode placement (5 levels left to right) analyzed the ERP amplitudes. This analysis model was selected to be consistent with a previous ERP study of action memory (Leynes et al., 2005b). Analyses incorporated the Geisser–Greenhouse correction for nonsphericity, and significant effects are reported with corrected degrees of freedom when appropriate. Significant interactions between the factor of Source (i.e., action type) and electrode placement are reported after ERP amplitudes were scaled to eliminate the confounding effects of overall amplitude differences (McCarthy and Wood, 1985). Fig. 1 displays the grand average ERP data recorded during the source recognition test. The results of the omnibus analysis of ERP amplitudes revealed significant effects involving the factor of Source for the 600–800 [Source: F(1.86, 33.48) = 7.91; Source× LR: F(3.68, 66.24)= 2.91; Source × AP × LR: F(5.76, 103.68) = 2.25], 1000–1200 [Source× AP: F(2.16, 38.88) = 10.73; Source× LR: F(4, 72) = 3.28; Source× AP × LR: F(3.52, 63.36)= 4.06], 1200–1400 [Source × AP: F(2, 36)= 6.73; Source× LR: F (3.12, 56.16) = 2.79; Source× AP × LR: F(4.16, 74.88) = 4.95], 1400–1600 [Source: F(1.42, 25.56)= 6.01; Source × LR: F(3.2, 57.60)= 3.27; Source× AP × LR: F(3.84, 69.12)= 2.75], and the 1600–1800 ms time intervals [Source: F(1.42, 25.56) = 7.49; Source× LR: F(3.28, 59.04)= 4.62; Source × AP × LR: F(7.04, 126.72) = 2.34]. The significant effects in the omnibus analyses were explored with pair-wise comparisons of the ERPs elicited by the three sources using the same ANOVA model as the omnibus analysis. Table 2 presents the results of these post hoc analyses as a function of the time interval. As is evident from inspection of Table 2 and Fig. 1, performed ERPs were reliably more positive than both motioned and new during the 600–800 ms interval, whereas motioned and new ERPs did not differ. These differences are most apparent at centrally located electrode sites from the front to the back of the scalp. During later intervals (i.e., 1000–1200, 1200–1400 and 1400–1600 ms intervals), the performed and motioned ERPs did not differ. The significant source by electrode location interactions in the post hoc analyses indicated that both old sources elicited more negative ERPs (an LPN) at central–parietal electrodes (see Fig. 1). The LPN persisted throughout the epoch, and the motioned LPN was larger than the performed LPN during the 1600–1800 ms interval. Fig. 1 and the results described earlier suggest that only two of the three ERP effects that have been reported in previous sourcemonitoring and OM ERP studies were detected in the present experiment. Beginning approximately 600 ms after the onset of the test probe, performed actions elicited more positive ERP activity than both motioned and new items. This effect is similar to the parietal old/new effect described in the introduction. However, both old sources elicited a larger LPN than new items from 1000 to 1800 ms. The motioned LPN was larger than the performed LPN during the 1600–1800 ms interval. Frontal ERP

198

P.A. Leynes et al. / International Journal of Psychophysiology 62 (2006) 193–202

Fig. 1. Grand average ERP activity recorded during the source recognition test at 25 electrode sites. The x-axis is placed at the 0 μV point in all graphs in the figure; therefore, positive microvolts are depicted above the x-axis and negative microvolts are plotted below the x-axis. The graphs are arranged according to the electrode placement on the scalp. Electrodes placed over frontal areas are placed at the top of the figure. Electrodes placed over the left hemisphere appear on the left side of the figure.

differences like those that have been reported in earlier studies were not observed in the present data. 3. Discussion The purpose of this experiment was to investigate whether sensory information contributes to OM judgments. ERPs were

also recorded during a source recognition test in an effort to further our understanding of how source monitoring processes are altered by the kind of information that is bound in the memory trace. To accomplish these goals, participants motioned some actions (i.e., performed the action without a real object) and performed an equal number of other actions. Next, ERP activity was recorded during a source recognition test that

P.A. Leynes et al. / International Journal of Psychophysiology 62 (2006) 193–202

199

Table 2 Significant effects from the pair-wise post hoc comparisons conducted to explore significant effects detected in the omnibus ERP analyses Time interval and factor

Comparison Performed vs. motioned

Performed vs. new

Motioned vs. new

600–800 Source Source × LR Source × AP × LR

F(1, 18) = 11.71⁎⁎ – F(4.8, 86.4) = 3.90⁎⁎

F(1, 18) = 11.08⁎⁎ F(2.4, 43.2) = 7.11⁎⁎

– –

1000–1200 Source × AP Source × LR Source × AP × LR

– – –

F(1.36, 24.48) = 13.20⁎⁎ – F(2.88, 51.84) = 4.80⁎⁎

F(1.28, 23.04) = 13.04⁎⁎ F(2.32, 41.76) = 5.87⁎⁎ F(2.72, 48.96) = 3.68⁎

1200–1400 Source × AP Source × AP × LR

– –

– F(3.84, 69.12) = 5.67⁎⁎⁎

F(1.4, 25.2) = 9.57⁎ F(3.52, 63.36) = 4.67⁎⁎

1400–1600 Source Source × LR Source × AP × LR

– – –

– – F(2.88, 51.84) = 3.21⁎

F(1, 18) = 10.56⁎⁎ F(2.2, 39.6) = 6.08⁎⁎ F(3.68, 66.24) = 2.61⁎

1600–1800 Source Source × LR Source × AP × LR

F(1, 18) = 5.94⁎ – –

– – F(5.6, 100.8) = 3.13⁎⁎

F(1, 18) = 10.89⁎⁎ F(2.2, 39.6) = 8.71⁎⁎ F(5.12, 92.16) = 2.40⁎

Note. AP = Anterior/Posterior electrode placement factor in the analysis, LR = Left/Right electrode placement factor in the analysis. ⁎p b .05, ⁎⁎p b .01, ⁎⁎⁎p b .001.

required performed, motioned, and new actions to be identified. The source recognition test was followed by two source recall tests: one that required recall of actions that were performed at study and one that required recall of actions that were motioned at study. Both the behavioral and ERP data suggest that sensory information provides unique features in action memories. These results are consistent with the Source Monitoring Framework because many different types of memory (sensory, motor, etc.) can serve as a basis for memory judgments. The behavioral results are discussed first in this general context followed by a discussion of the ERP data. The analysis of the behavioral data revealed that source recognition and recall of performed actions was superior to motioned actions. These results are similar to the findings that more performed actions are recognized and recalled when compared to imagined actions (Leynes and Bink, 2002) or interrupted actions (Leynes et al., 2005b). Consequently, these results replicate a memory advantage for performed actions that is similar to the “enactment effect” in an OM context. Furthermore, these results indicate that sensory information adds unique features to action memories because sensory information was the only distinguishing information between motioned and performed actions. Examination of OM errors on both the recognition and recall tests provided important evidence that sensory information is an important factor in making OM judgments particularly when the results are compared to previous findings. More motioned actions were missed on both the source recognition and recall tests; however, misattributions of performed and motioned actions did

not differ on either test.3 This is an important finding because motioned and performed actions were very similar except for the actual sensory information that is gained from manipulating the objects to perform the action. Despite the similarity of these actions, participants did not experience more confusions for either source; instead, more motioned actions were simply forgotten. Previous results have revealed that more confusions are observed when the comparison set of actions contains real or imagined sensory information, such as when actions are interrupted (Leynes et al., 2005b) or imagined (Leynes and Bink, 2002). Similarly, Hornstein and Mulligan (2004) reported more confusions as visual details were increased in discriminations between self and other performed actions. Therefore, the presence of sensory information in all actions considered appears to create a situation that is more confusing when discriminating between several types of self-performed actions or discriminations between self and other performed actions. When these results are considered with the literature on the enactment effect, it appears that motor information is particularly salient when distinguishing actions from non-action information; however, sensory information appears to be more salient when distinguishing between different types of actions that one may or may not have completed. 3 It is important to note that the recall test does not provide a process-pure measure of memory because it followed the source recognition test in the present study. However, this does not pose a problem when the results from the recall test mirror those observed from the source recognition because the evidence converges on the same conclusion. Contamination from the source recognition test would only pose an interpretation problem when the results do not converge.

200

P.A. Leynes et al. / International Journal of Psychophysiology 62 (2006) 193–202

The ERP data provided additional evidence that sensory information has a unique contribution to OM judgments because ERP differences were observed between the different types of actions. Recall from the introduction that previous ERP studies of source and OM have reported two consistent ERP effects: one parietal effect that appears approximately 600 ms after the onset of the probe and one frontal effect that appears approximately 800 ms after the probe. Several other source memory and OM studies have also reported a LPN that tends to onset around 1000 ms after the onset of the probe. Similar ERP effects were observed in the present experiment with some notable differences that provide insight into the cognitive processes that subserve OM judgments. Because these three ERP effects appear to reflect different kinds of processing during remembering, the importance of the early parietal ERP effects are discussed first and then followed with a combined discussion of the frontal ERP effects and the LPN. Performed ERPs were more positive than both motioned and new ERPs during the 600–800 ms interval. Importantly, this pattern of effects is very similar to another OM ERP study (Leynes et al., 2005b) and an ERP study of reality monitoring (Leynes et al., 2005a). This ERP effect has been hypothesized to reflect recollection (e.g., Paller and Kutas, 1992; Paller et al., 1995; Wilding, 2000; Wilding and Rugg, 1996) or activation of sensory information in the memory trace (Leynes et al., 2003) and these results can be interpreted in light of both of these views. The greater parietal ERP amplitudes can be taken as evidence that performed actions were recollected, whereas motioned actions were ascribed to source by an absence of recollection. In other words, the pattern of ERP differences might reflect the selective recollection of performed actions (cf., Herron and Rugg, 2003; Herron and Wilding, 2005). This interpretation is consistent with the behavioral data because reaction times were longer for motioned actions than both performed and new actions. Alternatively, Leynes et al. (2003) argued that early parietal ERPs reflect activation of sensory information that can be more or less activated depending on the testing context (for a similar argument see Dywan et al., 1998, 2002). The early parietal ERP effects in the present experiment are consistent with this hypothesis because more sensory information was bound in the memory trace of performed actions than motioned actions because people manipulated the objects and experienced other sensations from the objects (e.g., object movement, taste, smell, sound).4 Two other ERP effects have been reported in many source monitoring and OM ERP studies that occur later in the epoch: one is observed at frontal electrode sites, whereas the other is a late posterior negativity (LPN). The frontal ERP effect has been hypothesized to reflect post-retrieval processing (Mecklinger, 2000; Ranganath and Paller, 2000; Wilding, 1999; Wilding and 4 A third explanation for this effect is that the early parietal ERPs elicited by motioned actions was attenuated by the larger LPN elicited by these items. Although this is entirely possible, we think it is unlikely because the LPN does not emerge until several hundred milliseconds later. This is evident both in Fig. 1 and in the analysis of the 800–1000 ms time interval that did not reveal any significant differences.

Rugg, 1997) or source monitoring decision processes (Leynes, 2002; Leynes and Bink, 2002; Leynes et al., 2003, 2005b). The LPN has been hypothesized to reflect retrieval of multiple characteristics bound in the memory trace (Johansson and Mecklinger, 2003). In addition, Cycowicz and Friedman (2003) argued that the LPN reflected retrieval of visual information; however, this claim was recently tempered by Friedman, Cycowicz, and Bersick (2005) because the size of the LPN did not vary between accurate and error trials and was unaffected by the color of the test probe. Friedman et al. suggested that the LPN might reflect a “search for and/or retrieval/evaluation of the attributes in modality-specific cortical regions that processed the information during the study phase.” (p. 197). Based on these previous hypotheses, the frontal ERPs and the LPN might reflect different types of decision processes that are used in different testing contexts. The data from the present study support this simple claim in several important ways. The frontal ERP effects observed in the present study were nonexistent or weak. Although these results appear to diverge from the literature, they are consistent with another ERP study of reality monitoring (Leynes et al., 2005a) and two additional ERP studies of source that used an exclusion task (Cycowicz and Friedman, 2003; Friedman et al., 2005). Leynes, Cairns, and Crawford (2005a) suggested that the absence of frontal ERP differences during reality monitoring was due to enhanced frontal positivity in the new ERPs, which they suggested reflected a change in source monitoring decision processes. Unfortunately, this claim cannot be informed by the current data because only one test was used in the present study. Nevertheless, the present study is very similar to the Cycowicz and Friedman (2003) and Friedman et al. (2005) studies because sensory information (i.e., color) formed the basis for the source judgments in those studies. Importantly, these effects are not limited to the electrode reference site used (i.e., nose-tip in Cycowicz and Friedman, 2003 and Friedman et al., 2005; average mastoid reference in the present study) or the type of test judgment (i.e., exclusion task in Cycowicz and Friedman, 2003 and Friedman et al., 2005; three-alternative-forced-choice in the present study). Although frontal ERP effects were not clearly observed in these studies, all of these studies reported a large LPN. In the present investigation, the LPN had a central– parietal maximum, began to emerge about 1000 ms after the onset of the probe, and varied in amplitude between the actions (i.e., motioned N performed N new). Collectively, these studies suggest that the LPN is distinct from frontal ERP effects, despite a similar time course, because different patterns were observed at frontal and parietal electrodes. However, additional work should attempt to dissociate these effects in other ways because it is possible that the LPN activity actually obscured the frontal ERP effects (Cycowicz et al., 2001). Although there have been some general hypotheses regarding the cognitive processes that these two ERP effects represent (described earlier), the present data as well as the studies from Friedman and colleagues allow for a slight refinement of these claims. One way to integrate the previous findings is to appeal to the Source Monitoring Framework because it argues that different processes will be used in different testing contexts. Thus,

P.A. Leynes et al. / International Journal of Psychophysiology 62 (2006) 193–202

it seems reasonable to postulate that the frontal ERPs and the LPN reflect different types of decision processes engaged in different testing contexts. Based on this logic, we suggest that the frontal ERP effects reflect post-retrieval monitoring or decision processes that function in a more general way to evaluate diagnostic information. Alternatively, the LPN might reflect detailed inspection of memory for subtle differences in information. This characterization of these two cognitive processes is also based on results from many previous ERP studies of source that have employed more general manipulations of sources, such as gender of the voice, list modality, and encoding tasks. These studies have not consistently observed the LPN; presumably because the need for more detailed discriminations varied with the test demands. However, the frontal effects are nonexistent or weak and a large LPN is consistently observed when source judgments are based on detailed examination of perceptual information, such as when the color of studied objects was discriminated (Cycowicz and Friedman, 2003; Friedman et al., 2005) or when sensory information distinguished performed and motioned actions (the present study). It is important to point out that our characterization is similar to the previous explanations offered by Johansson and Mecklinger (2003) and Friedman et al. (2005) that were described earlier, but it differs in some important ways. First, our characterization is focused on the processes that evaluate activated memories and does not necessarily implicate additional retrieval of details (cf., Johansson and Mecklinger, 2003). Second, our description does not necessarily assume that the LPN reflects modality-specific processing because sensory information was not limited to one modality in the present study (cf., Friedman et al., 2005). Although these characterizations of the processes that are reflected in frontal ERPs and the LPN are more descriptive at this point, future studies can manipulate the type of information bound into memory in an effort to determine how different decision processes affect ERP data. For example, more general manipulations of sources, such as varying the cognitive operations that distinguish different sources, might affect frontal ERPs (e.g., Wilding, 1999). On the other hand, manipulating detailed aspects of information between sources might alter the LPN. The present study also provides a basis for future studies that can be used to understand source monitoring processes and further our understanding of OM. Sensory information was broadly defined in the present study. Sensory information in an action can be subcategorized into tactile, kinesthetic, visual, auditory, and olfactory information. Visual information may also be separated into static and dynamic visual information, which may contribute differently to memory (Engelkamp, 2001). Therefore, one way to proceed is to carefully manipulate the type of sensory information in a set of actions while observing differences in memory and brain activity. Based on the ideas described earlier, altering the type of sensory information bound in the memory trace should affect the LPN amplitudes observed in the resulting ERP activity. Acknowledgment This research was supported by The College of New Jersey through an internal grant award (SOSA) to the first author and a

201

Summer Research Award to the first and second author. The authors thank three anonymous reviewers for their helpful comments on an earlier draft. The authors also thank Rick Addante, Jenn Burke, Brian Cichon, Crissy Duprey, Brett Bersano, Lia Guimaraes, Lucy Hunt, Eric Klein, Nichole Klindt, Molly McGoldrick, Kelly Rotella, Florencia Szriftgiser, and Ian Zimmerman for their help with collecting the data. References Cycowicz, Y.M., Friedman, D., 2003. Source memory for the color of pictures: event-related brain potentials (ERPs) reveal sensory-specific retrievalrelated activity. Psychophysiology 40, 455–464. Cycowicz, Y.M., Friedman, D., Snodgrass, J.G., 2001. Remembering the color of objects: an ERP investigation of source memory. Cerebral Cortex 11, 322–334. Dywan, J., Segalowitz, S.J., Webster, L., 1998. Source monitoring: ERP evidence for greater reactivity to nontarget information in older adults. Brain and Cognition 36, 390–430. Dywan, J., Segalowitz, S., Arsenault, A., 2002. Electrophysiological response during source memory decisions in older and younger adults. Brain and Cognition 49, 322–340. Engelkamp, J., 1998. Memory for Actions. Psychology Press, Hove, England. Engelkamp, J., 2001. Action memory: a system-oriented approach. In: Zimmer, H.D., Cohen, R.L., Guynn, M.J., Engelkamp, J., Kormi-Nouri, R., Foley, M.A. (Eds.), Memory for Action: a Distinct Form of Episodic Memory? Oxford University Press, New York, pp. 49–96. Engelkamp, J., Zimmer, H.D., 1997. Sensory factors in memory for subjectperformed tasks. Acta Psychologia 96, 43–60. Foley, M.A., Ratner, H.H., 1998. Distinguishing between memories for thoughts and deeds: the role of prospective processing in children's source monitoring. British Journal of Developmental Psychology 16, 465–484. Friedman, D., Cycowicz, Y.M., Bersick, M., 2005. The late negative episodic memory effect: the effect of recapitulating study details at test. Cognitive Brain Research 23, 185–198. Goff, L.M., Roediger, H.L., 1998. Imagination inflation for action events: repeated imaginings lead to illusory recollections. Memory & Cognition 26, 20–33. Henry, J.D., MacLeod, M.S., Phillips, L.H., Crawford, J.R., 2004. A meta-analytic review of prospective memory and aging. Psychology and Aging 19, 27–39. Herron, J.E., Rugg, M.D., 2003. Strategic influences on recollection in the exclusion task: electrophysiological evidence. Psychonomic Bulletin and Review 10, 703–710. Herron, J.E., Wilding, E.L., 2005. An electrophysiological investigation of factors facilitating strategic recollection. Journal of Cognitive Neuroscience 17, 777–787. Hornstein, S.L., Mulligan, N.W., 2004. Memory for actions: enactment and source memory. Psychonomic Bulletin and Review 11, 367–372. Johansson, M., Mecklinger, A., 2003. The late posterior negativity in ERP studies of episodic memory: action monitoring and retrieval of attribute conjunctions. Biological Psychology 64, 91–117. Johnson, M.K., Hashtroudi, S., Lindsay, D.S., 1993. Source monitoring. Psychological Bulletin 114, 3–28. Kormi-Nouri, R., 2000. The role of movement and object in action memory: a comparative study between blind, blindfolded and sighted subjects. Scandinavian Journal of Psychology 41, 71–75. Kormi-Nouri, R., Nilsson, L.G., 2001. The motor component is not crucial! In: Zimmer, H.D., Cohen, R.L., Guynn, M.J., Engelkamp, J., Kormi-Nouri, R., Foley, M.A. (Eds.), Memory for Action: a Distinct Form of Episodic Memory? Oxford University Press, New York, pp. 97–111. Leynes, P.A., 2002. The effect of specific test queries on source monitoring event-related potentials (ERPs). Brain and Cognition 50, 218–233. Leynes, P.A., Bink, M.L., 2002. Did I do that? An ERP study of memory for performed and planned actions. International Journal of Psychophysiology 45, 197–210. Leynes, P.A., Bink, M.L., Marsh, R.L., Allen, J.D., May, J.C., 2003. Test modality affects source monitoring and event-related potentials. American Journal of Psychology 116, 389–413.

202

P.A. Leynes et al. / International Journal of Psychophysiology 62 (2006) 193–202

Leynes, P.A., Cairns, A., Crawford, J.T., 2005a. Event-related potentials (ERPs) indicate that reality monitoring differs from external source monitoring. American Journal of Psychology 118, 497–524. Leynes, P.A., Crawford, J.T., Bink, M.L., 2005b. Interrupted actions affect output monitoring and event-related potentials (ERPs). Memory 13, 759–772. Marsh, R.L., Hicks, J.L., 2001. Output monitoring tests reveal false memories of memories that never existed. Memory 9, 39–51. McCarthy, G., Wood, C.C., 1985. Scalp distributions of event-related potentials: an ambiguity associated with analysis of variance models. Electroencephalography and Clinical Neurophysiology 62, 203–208. Mecklinger, A., 2000. Interfacing mind and brain: a neurocognitive model of recognition memory. Psychophysiology 37, 565–582. Nilsson, L.G., Nyberg, L., Klingberg, T., Åberg, C., Persson, J., Roland, P.E., 2000. Activity in motor areas while remembering action events. NeuroReport 11, 2199–2201. Oldfield, R.C., 1971. The assessment and analysis of handedness: the Edinburgh Inventory. Neuropsychologia 9, 97–113. Paller, K.A., Kutas, M., 1992. Brain potentials during memory retrieval provide neurophysiological support for the distinction between conscious recollection and priming. Journal of Cognitive Neuroscience 4 (4), 375–391. Paller, K.A., Kutas, M., McIsaac, H.K., 1995. Monitoring conscious recollection via the electrical activity of the brain. Psychological Science 6 (2), 107–111. Ranganath, C., Paller, K.A., 2000. Neural correlates of memory retrieval and evaluation. Cognitive Brain Research 9, 209–222. Rugg, M.D., Schloerscheidt, A.M., Mark, R.E., 1998. An electrophysiological comparison of two indices of recollection. Journal of Memory and Language 39, 47–69. Schaefer, E.G., Kozak, M.V., Sagness, K., 1998. The role of enactment in prospective remembering. Memory and Cognition 26, 644–650.

Semlitsch, H.V., Anderer, P., Schuster, P., Presslich, O., 1986. A solution for reliable and valid reduction of ocular artifacts, applied to the P300 ERP. Psychophysiology 23, 696–703. Senkfor, A.J., Van Petten, C., 1998. Who said what? An event-related potential investigation of source and item memory. Journal of Experimental Psychology. Learning, Memory, and Cognition 24, 1005–1025. 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. Thomas, A.K., Loftus, E.F., 2002. Creating bizarre false memories through imagination. Memory and Cognition 30, 423–431. Wilding, E.L., 1999. Separating retrieval strategies from retrieval success: an eventrelated potential study of source memory. Neuropsychologia 37, 441–454. Wilding, E.L., 2000. In what way does the parietal ERP old/new effect index recollection? International Journal of Psychophysiology 35, 81–87. Wilding, E.L., Rugg, M.D., 1996. An event-related potential study of recognition memory with and without retrieval of source. Brain 119, 889–905. Wilding, E.L., Rugg, M.D., 1997. Event-related potentials and the recognition memory exclusion task. Neuropsychologia 35, 119–128. Zimmer, H.D., Cohen, R.L., 2001. Remembering actions: a specific type of memory? In: Zimmer, H.D., Cohen, R.L., Guynn, M.J., Engelkamp, J., Kormi-Nouri, R., Foley, M.A. (Eds.), Memory for Action: a Distinct Form of Episodic Memory? Oxford University Press, New York, pp. 3–24.