Skip to main content
NCBI home page
NIHPA Author Manuscripts logoLink to NIHPA Author Manuscripts
. Author manuscript; available in PMC: 2009 Jul 1.
Published in final edited form as: Cortex. 2007 Dec 23;44(7):820–833. doi: 10.1016/j.cortex.2007.03.003

Memory for pantomimed actions versus actions with real objects

Ava J Senkfor a,b,*
PMCID: PMC2398730  NIHMSID: NIHMS46541  PMID: 18489962

Abstract

A substantial literature indicates that human actions during object use and pantomimed object use are not identical, and can be differentially affected by brain damage such that apraxic patients can be more impaired in performing actions with objects or at pantomiming such actions. A different literature suggests that memory retrieval can involve reinstating or recapitulating some of the same brain activity that occurred during the original event. The current experiment examines memory for pantomimed actions versus those conducted with real objects to determine if accuracy or brain electrical activity differs during the recollection of episodes involving pantomime versus actual object use. Across two sessions, participants were presented with images of studied objects and judged whether each object was studied by (1) performing an action, (2) watching the experimenter perform an action, (3) imagining an action, or (4) the nonmotoric control task of estimating the object's cost. The study phases preceding this source memory test differed across sessions: in one, participants were presented with real objects and estimated its cost or performed, watched, or imagined typical actions with the objects; in the other they viewed images of each object and estimated its cost or performed, imagined, or watched pantomimes of object use. Although source accuracy was the same across sessions, event-related potentials (ERPs) recorded during memory retrieval differed for memory for actions with real objects versus their pantomime equivalents. Retrieval-phase activity did not differ for cost-encoded objects. The real-object/pantomime difference in brain activity was maximal over left frontal and frontocentral cortex, suggesting differential engagement of motor cortex during memory for real actions versus pantomimed actions.

Keywords: Enactment effect, Pantomime, Memory, Action memory

1. Introduction

The idea that encoding and retrieval are deeply intertwined is well established in psychology, and is at the heart of transfer-appropriate-processing theory: that study conditions foster high retrieval accuracy to the extent that they allow transfer of the knowledge gained during study (Morris et al., 1977). In neuroscience, the idea that memories are stored across distributed neocortical regions and that successful retrieval of an episodic memory involves some degree of recapitulation of the pattern of activity present during learning also has a long history (Hebb, 1949), and is preferred in modern theoretical accounts as well (Damasio, 1989; McClelland et al., 1995). However, the first decade of functional neuroimaging was more successful in revealing differences than similarities in the regions activated during the encoding and retrieval phases of episodic memory experiments (see Haxby et al., 1996, and review by Cabeza and Nyberg, 2000).

I suggest that one reason for this state of affairs is that not all of the cognitive/neural processes engaged during an encoding task are likely to become part of the episodic memory trace. For example, if one is asked to perform a concrete/abstract judgment during the study phase, the meanings of the individual words will be a central focus of study-phase activity and are also likely to be the core elements of the episodic memory trace. Alternatively, other study-phase processes such as holding the task instructions and the definitions of “concrete” and “abstract” in working memory, evaluating each word against the definitions, and monitoring one's task performance are unlikely to form part of the trace. In the same way, some processes engaged during memory retrieval – setting and maintaining a response criterion to differentiate “old” and “new” during a recognition test, strategic search processes during recall tests, etc – may also have no parallel with processes engaged during encoding. Some recent positron emission tomography (PET) studies have, for instance, suggested that some regions of the frontal lobe are active during retrieval of episodic memories from different study modalities (Lepage et al., 2001; Schmidt et al., 2002), but these may reflect executive functions engaged during retrieval rather than content-specific retrieval per se. Given these considerations, a modified recapitulation hypothesis would predict partial overlap between brain activity during encoding and retrieval, overlaid by non-overlapping activity reflecting processes engaged during encoding or retrieval, but not both. A corollary hypothesis is that retrieval of qualitatively different content from memory will yield different patterns of brain activity during a memory test.

Some recent studies lend support to the modified recapitulation hypothesis and to its corollary. Köhler et al. (1998) observed differential brain activity when participants judged objects as old or new versus when they judged their spatial locations as same or different from the study phase, and the test-phase difference partially overlapped the study-phase difference between object and spatial judgments. Two other papers report that visual stimuli elicit activity in auditory cortex during memory retrieval, because they had been paired with sounds during the study phase (Nyberg et al., 2000; Wheeler et al., 2000). Vaidya et al. (2002) similarly report activity in picture-processing regions when encountering test words that are labels for studied pictures. Hayes et al. (2004) report retrieval-phase activity in the parahippocampal gyrus when participants tried to remember spatial information about the study phase, but not temporal order information.

Studies investigating content-specific brain activity during retrieval have tended to focus on broad perceptual differences between study phase conditions: differences between auditory and visual modalities, or between words and pictures. This is, of course, only a subset of what is likely to be stored in memory for a particular episode, which should also include memory for one's own activities – both overt actions and internal cognitive operations. The present study examines differences in brain activity during retrieval of memories about objects, contingent on what participants did during the study phase: performed an action, imagined but did not execute an action, watched the experimenter perform an action, or estimated the cost of the object. The retrieval cues – photos of the objects – were identical across conditions, so that retrieval-phase differences can only indicate content-specific activity due to reinstatement of different study phase activities. One session of the present study replicates a previous experiment in which the study phase included the use of real objects and content-specific brain activity was observed during memory retrieval (Senkfor, 2002; Senkfor et al., 2002). This session is compared to one in which participants view only images of objects, and pantomime actions. Pantomimed object use as defined here means actions conducted as if the object was in the hand, but without any actual contact with the object. As reviewed below, several dissociations between object use and pantomimed object use have been documented in both neurological patients and healthy participants, so that it is reasonable to conclude that there should be differential engagement of some brain regions during these two varieties of action. If such neural differences occur during initial production, they may leave a residue in memory, such that even remembering object use versus pantomimed object use will be accompanied by differential brain activity.

1.1. Varieties of action: pantomime versus interactions with real objects

Deficits in the production of actions are the defining of feature of apraxia after brain damage, but the apraxia literature suggests that actions should be divided into multiple categories (Bartolo et al., 2003): transitive actions with real objects (hereafter called simply “object use”), pantomimes of transitive actions, intransitive symbolic gestures (e.g., waving goodbye), and meaningless gestures. In addition, Bartolo et al. (2003) consider gestures which use body parts to represent an object to be pantomime errors (e.g., showing a cutting action with two fingers as if they were the blades of a scissors), Dissociations among categories are a fairly common observation in apraxic patients. For instance, Buxbaum et al. (2005) report cases in which both recognition and imitation of pantomimed actions are worse than symbolic gestures (see also Dumont et al., 1999), while Cubelli et al. (2000) report both a similar case and the reverse impairment.

For the comparison of interest here, production deficits in apraxic patients are frequently less severe when real objects are used as compared to pantomiming the same actions (Goldenberg et al., 2004; Laimgruber et al., 2004; Poizner et al., 1990). The opposite pattern of impairment has also been reported in three case studies: a deficit in performing actions with real objects with preserved ability to pantomime the same actions (Fututake, 2003; Heath et al., 2003; Motomura and Yamadori, 1994). There has, however, been some dispute as to whether this latter sort of result might be attributable to a visuoperceptual deficit that prevents accurate identification of object parts and thus makes it difficult to interact with real objects (Bartolo et al., 2003).

It has also been suggested that the visual processing which supports object use and pantomimed object use is neurally distinct. Based on dissociations in brain-damaged patients, Goodale and Milner (1992) and Milner and Goodale (1995) suggested that pantomime was more dependent on the ventral stream of visual processing (temporal lobe “what” pathway) and less dependent on the dorsal stream (parietal lobe “where and how” pathway) than object use. However, one fMRI suggested that parietal cortex is critical for pantomimed tool use, as the primary difference between pantomiming and reproducing a meaningless gesture sequence was in parietal rather than temporal regions (Moll et al., 2000).

Not in dispute is the fact that, in both healthy subjects and patients, the actions performed during pantomimed and real-object use are not identical (Goodale et al., 1994; Laimgruber et al., 2004; Milner and Goodale, 1995; Westwood et al., 2000). Picking up a real object includes visually-guided adjustment of both hand shape and the size of the aperture between thumb and fingers (prehension). These adjustments are precise when interacting with real objects, but need not be in pantomimed actions. For instance, when picking up a glass of water, hand aperture is adjusted to the size of the glass when reaching, but then reduced after contact in order to firmly grasp the glass. In pantomimed versions of this action, the final grasping adjustment is frequently omitted. The speed and trajectory of reaching and grasping have also been shown to differ for interactions with real objects versus pantomime (Goodale et al., 1994). Overall, actions with physically-present objects rely on immediate and specific visual, tactile, and proprioceptive sensory input, while pantomimed actions must place greater reliance on some object properties retrieved from memory,1 and largely exclude other object properties (e.g., the muscle force used to pantomime actions with light vs. heavy objects of the same size will not differ, given that no actual weight is moved).

1.2. Previous ERP studies of action memory

The current experiment examines brain electrical activity during a source memory task in which participants view photos of objects, and judge whether they were initially encountered in the context of performing an action, watching the experimenter perform an action, imagining an action, or the control task of cost-estimation. Three varieties of action encoding are included because evidence from a variety of methods (hemodynamic imaging, magnetoencephalography, ERPs, single-unit recording) indicate some overlap in the brain areas engaged during execution of an action, motor imagery, and watching someone else perform an action (Beisteiner et al., 1995; Cunnington et al., 1996; Hari et al., 1998; Grafton et al., 1996; Grezes and Decety, 2001; Rizzolatti et al., 1996; Schnitzler et al., 1997). If episodic retrieval involves some degree of recapitulation of study-phase activity, there is thus reason to expect common brain activity when participants retrieve memories with an action element.

In a previous experiment using these four encoding conditions, content-specific brain activity during memory retrieval was indeed observed (Senkfor et al., 2002). At fronto-central sites overlying premotor cortex, the three conditions that followed action encoding elicited indistinguishable ERPs, but all were distinct from responses to objects that had undergone cost-estimation. This action retrieval effect confirmed the prediction based on the non-mnemonic studies indicating a commonality among action execution, imagination, and observation. At sites overlying visual cortical regions (occipital, temporal, and posterior parietal), electrical activity instead showed very similar responses to Perform- and Watch-encoded objects, which differed from both Imagine- and Cost-encoded objects. This division tracked the split between episodes involving moving hands and moving objects versus episodes with stationary hands and objects, so that we have considered this posterior effect one of motion retrieval. Finally, over prefrontal sites alone, Imagine-encoded objects elicited different ERPs from all other conditions, which did not differ from one another. A partial replication of this design compared only Perform- and Cost-encoded objects, but showed that the differential retrieval-phase activity was contingent on a source memory test in which participants were explicitly asked to remember their encoding phase activities, and did not occur when participants were asked only to judge objects as studied or unstudied (Senkfor et al., in press; cf. Masumoto et al., 2006 for motor cortex activity after perform-encoding as compared to a passive encoding task with no overt judgment).

In the current experiment, ERPs during two source memory tests are compared: one following Perform, Watch, Imagine and Cost-encoding when real objects are present (and handled by the participant or the experimenter in the Perform and Watch conditions), and one in only images of the objects are presented during the study phase and actions are merely pantomimed. If object use and pantomimed object use are neurally distinct, remembering episodes involving these two varieties of action may also involve differential brain activity. The observation of any such differences will suggest that brain activity during memory retrieval is extremely sensitive to differential encoding activity, such as the absence of object-specific tactile and proprioceptive input during pantomime. An a priori prediction is that – if any such differences are observed – they will be evident after action encoding (Perform, Imagine, Watch), but not after cost-encoding. To the extent that differences between the two Cost conditions are observed, these will serve as an assessment of the impact of viewing three-dimensional objects (Real-Object session) versus images of objects (Pantomime session).

2. Methods

2.1. Participants

Twenty healthy young adults (12 women, 8 men), between the ages of 18−27 (mean 21 years), with normal or corrected-to-normal vision, and no history of psychiatric or neurological disorders were included. All participants reported being right-handed, as well as all family members being right-handed (Oldfield, 1971).

2.2. Stimuli

Four hundred and thirty two manipulable real objects (e.g., grass clippers, fork, radio), or toy versions of objects (e.g., slot machine, trumpet, fishing rod) were used, in addition to digital color images of the same objects. An additional 12 objects/images were used for practice. Stimuli were the same as those used in Senkfor et al. (2002).

2.3. Study-phase procedures

The experiment was conducted in two sessions conducted about a week apart. In the Real-object session, actual objects were presented on a table. In the Pantomime session, photographs of the objects were presented on a 20″computer monitor. The assignment of items to session, and the order of sessions were counterbalanced across participants.2 In both study phases, items (objects) were presented one at a time for 7 sec with an interstimulus interval of 11 sec. In the Real-object session, an object was placed on the table to the participant's right or left side; in the Pantomime session, images were presented on the far left or far right of the computer screen. In both sessions, the experimenter stood in front of the participant on the other side of a table. For the pantomime session the experimenter conducted actions on the top of the computer monitor so that full view of the object and action could be achieved. As each item was presented, participants heard one of four randomly intermixed (pre-recorded) encoding instructions: (1) generate and Perform a typical action with the object (or pantomime doing so), (2) Watch the experimenter perform an action with the object (or pantomime doing so), (3) Imagine performing a typical action with the object, or (4) generate and verbalize a likely Cost for the object. In the Real-object session, participants contacted the objects only in the Perform condition. For Perform, Imagine, and Watch trials in both sessions, the spatial location of the object cued which hand should be used by the participant/experimenter, either in reality or in motor imagery; counterbalancing across participants ensured that a given object was presented equally often in right and left spatial locations. After receiving the study phase instructions and prior to a practice section, participants were informed of the upcoming test-phase procedures that would follow the study phase. Each study phase was preceded by a practice set of 12 objects in the Real-object session, or 12 images in the Pantomime session.

Actions pantomimed by the experimenter in the Watch condition were conducted as if the object were actually in hand, and participants were instructed to do the same. During the practice block, participants received feedback on the selection and quality of action production (actual and pantomimed) until sufficient typicality of action and production quality was achieved. Unidentifiable actions or unidentifiable grips on an object were deemed unacceptable pantomimes. An example of an acceptable pantomime for a toothbrush would include a handshape appropriate to a toothbrush handle, moving the hand to the mouth, and conducting a repetitive up/down or side-to-side brushing action. An example of an unacceptable pantomime for a toothbrush was extending the index finger, and using the finger as if it were a toothbrush. As needed, additional instructions on pantomimes were provided, however, participants quickly achieved acceptable levels of pantomime performance.3 After the practice session, participants were reminded of the memory test following the study phase. Study phases were videotaped for later review. Trials with atypical actions, those in which the participant used the wrong encoding task or encoding hand, actions and pantomimes using two hands, unacceptable pantomimes (such as using body parts as tools, unidentifiable actions or unidentifiable grips on an object), and those in which the act/estimate was not produced within the allotted time were dropped from further analyses (∼1% of trials).

2.4. Test-phase procedures

The memory test was identical in the two sessions. Participants were presented with digital color photographs of each studied object in the center of a computer monitor, at 5 sec interval (500 msec duration). No new objects were included at test. Participants pressed one of four keys to indicate the encoding task that had been conducted with the object (Perform, Watch, Imagine, Cost). The mapping of the four response keys (index and middle fingers of the two hands) to the four responses was counterbalanced across participants. Trials in which no response was made were excluded from analysis (∼5%). Each test phase was preceded by a practice set of 12 images (those studied in the practice portion of the study phase). The delay between the end of the study phase and the start of the practice phase was approximately 10−15 min.

2.5. Measures

Accuracy and reaction times (RTs) were recorded via button presses. ERPs were recorded from 26 tin electrodes in a geodesic array, plus two additional prefrontal sites. Horizontal eye movements were monitored using a right-to-left bipolar montage at the external canthi of the two eyes; vertical eye movements and blinks were monitored via an electrode below the right eye. Scalp sites were referenced to the left mastoid during recording, and re-referenced offline to an average of the left and right mastoids. The EEG was amplified with half-amplitude cutoffs at .01 and 100 Hz, and digitized with a sampling rate of 250 Hz. Trials with artifacts due to eye movements, blinks, or amplifier saturation were rejected prior to averaging single trials of EEG into ERPs. ERPs were quantified by mean amplitude measures over a specified time window with analysis of variances (ANOVAs), relative to a 100 msec prestimulus baseline task.

3. Results

3.1. Behavioral performance

Source memory judgments were more accurate for stimuli encoded under the Perform and Watch conditions than the Imagine and Cost conditions, with little difference between the Real-object and Pantomime sessions, as shown in Table 1. Percent accuracies were initially analyzed with an ANOVA taking session (Real-object vs. Pantomime) and encoding task (Perform, Watch, Imagine, Cost) as repeated measures. The main effect of encoding task was significant (F(3,57) = 8.19, p = .001, e = .76),4 but the main effect of session and the interaction between session and encoding task were not (Fs < 1.0). Pairwise comparisons between conditions within each session are shown in Table 2, and confirm the accuracy pattern of Perform = Watch > Imagine = Cost in each session. Pairwise comparisons of accuracy across sessions showed no differences for any of the four encoding tasks (Fs < 1.0).

Table 1.

Accuracy (percent), RTs (msec), and standard error in parentheses, for real object and pantomime sessions

Encoding task Real object
 Pantomime
Accuracy RT Accuracy RT
Perform 91 (1) 1296 (41) 89 (1) 1261 (28.2)
Watch 91 (1) 1297 (40) 91 (1) 1269 (48.3)
Imagine 83 (2) 1603 (58) 85 (2) 1497 (49.8)
Cost 84 (2) 1386 (48) 85 (3) 1239 (33.8)
Open in a new tab

Table 2.

Pairwise comparisons of behavioral performance within session

F(1,19) Watch
 Imagine
 Cost
 
 Accuracy
 RT
 Accuracy
 RT
 Accuracy
 RT
Perform
Real-object <1.0 <1.0 10.8** 39.6*** 12.7** 8.55**
Pantomime <1.0 <1.0 6.21* 26.8*** <1.8 <1.0
Watch
Real-object - - 9.14** 53.4*** 12.2** 7.77**
Pantomime - - 5.11* 31.8*** 4.20* <1.5
Imagine
Real-object - - - - <1.0 19.3***
Pantomime - - - - <1.0 62.5***
Open in a new tab
*

p < .05

**

p < .01

***

p < .001

Table 3 shows response frequencies (raw trial counts) for all possible combinations of encoding task and response types, which elucidate what sorts of errors (source misattributions) participants made. Log-linear models were used to examine the pattern of errors after excluding the correct responses (Brown, 1988). For the Real-object session, the initial model was a saturated one using the factors of encoding task, response at test, and their interaction. The encoding task factor was significant (χ2 = 48.7, df = 3, p < .0001), echoing the accuracy analyses above in showing that some conditions elicited higher accuracy levels than others. A significant effect of the response factor (χ2 = 44.6, df = 3, p < .0001) indicated that errors were not equally distributed across the alternate response options: when in error, participants were most likely to respond “imagine” and least likely to respond “cost”. Finally, a significant interaction between the encoding task and response factors (χ2 = 39.7, df = 5, p < .0001) indicates that some source confusions were more likely than others. In other words, given the accuracy differences among conditions, and the propensity to favor certain responses when in error, some cells of the design still had larger numbers of responses than expected. The two highest discrepancies between observed and expected values were for Cost-encoded items judged as “imagine”, and Perform-encoded items judged as “cost” (note that although “cost” is a rare type of erroneous response overall, a large proportion of these errors occur for Perform-encoded objects). Removing these two cells from the model rendered the interaction between the encoding task and response factors no longer significant (encoding task χ2 = 42.4, df = 3, p < .0001; response χ2 = 35.6, df = 3, p < .0001; interaction χ2 = 6.21, df = 3, p > .10). This last result means that the remainder of the errors in the matrix can be predicted based on the generally high accuracy for Perform- and Watch-encoded items, and the general tendency to say “imagine” and avoid saying “cost” when in error. The overall pattern of accuracy across conditions, response selected when in error, and the most prevalent source confusions are largely similar to the prior study using the same conditions as the Real-object session (Senkfor et al., 2002).

Table 3.

Response frequencies

Encoding task Response at test
Perform Watch Imagine Cost Total-errors
Real-object session
Perform 922 28 32 34 94
Watch 34 893 42 10 86
Imagine 57 78 829 28 163
Cost 43 30 93 855 166
Total-errors 134 136 167 72 509
Pantomime session
Perform 890 31 58 17 106
Watch 21 883 61 10 92
Imagine 49 68 852 35 152
Cost 33 30 87 861 150
Total-errors 103 129 206 62 500
Open in a new tab

Excluded are trials in which the wrong encoding task or no encoding task was conducted during the study phase, and trials in which no answer was offered during the allowed time in the test phase. Frequencies in bold and italics (along the diagonals) are correct responses.

A saturated log-linear model was also applied to the errors in the Pantomime session. This yielded significant effects of both the encoding task (χ2 = 50.0, df = 3, p < .0001) and response factors (χ2 = 107.2, df = 3, p < .0001). Table 2 shows that the origin of these results is the same as in the Real-object session: higher accuracy for Perform- and Watch-encoded items than for Imagine- and Cost-encoded, together with a large number of “imagine” responses and a small number of “cost” responses when in error. However, unlike the Real-object session, the interaction between the encoding task and response factors was not significant in the Pantomime session (χ2 = 4.48, df = 5, p > .40).

The omnibus ANOVA for RTs yielded a main effect of session (F(1,19) = 6.70, p < .05) and encoding task (F(3,57) = 42.0, p < .0001, e = .84). RTs were overall faster in the Pantomime session compared to the Real-object session. RTs for Perform- and Watch-encoded items were overall faster than those for Imagine- and Cost-encoded items. The interaction between session and encoding task tended toward significance (F(3.57) = 3.02, p = .06). Table 2 shows that correct “Imagine” responses were slower than other responses in both sessions. The only aspect of the RTs than distinguished the two sessions was that “Cost” responses were slower in the Real-object than Pantomime session (F(1,19) = 16.6, p < .001).

Overall, the behavioral results show many commonalities between the Real-object and Pantomime sessions: higher accuracy for Perform- and Watch-encoded stimuli as compared to Imagine- and Cost-encoded stimuli, slow RTs for correct “imagine” responses, and a tendency to use the “imagine” option and to avoid the “cost” option on erroneous trials. Only subtle differences emerged between the sessions: faster RTs for “cost” responses, and a more uniform pattern of source confusions in the Pantomime session as compared to the Real-object session.

3.2. Electrophysiological results

The ERP results are presented in two main sections, both based only on trials with correct behavioral responses. First, the data within each session are independently analyzed for signs of content-specific patterns of brain activity during memory retrieval. These analyses are guided by prior results (Senkfor et al., 2002). Specifically, the ERPs are examined for the three effects previously observed: (1) separation of trials encoded with moving objects/hand (Perform and Watch) versus stationary objects (Imagine and Cost) over parietal, temporal, and occipital cortex, (2) separation of action-encoded (Perform, Watch, Imagine) from non-action-encoded (Cost) trials over frontocentral scalp, and (3) a unique effect of remembering motor imagery (Imagine trials vs. all others), previously observed only over prefrontal cortex. Then the two sessions are directly compared to evaluate how memories involving real objects differ from those in which the encoding phase included only images of objects and pantomimed actions. For all comparisons, it is important to keep in mind that the physical stimuli, and the task assigned to the participants are identical across conditions and sessions, so that any observed differences in brain activity can be attributed only to differences in the prior encoding conditions.

3.3. Visual motion and posterior brain activity

3.3.1. Real-object session

Fig. 1 shows ERPs elicited by stimuli from all four encoding conditions in the Real-object session. These are indistinguishable for the first 600−800 msec (depending on scalp site) after stimulus onset. After 600 msec, activity over roughly the entire back half of the head includes more positive ERPs for Perform and Watch trials – those in which visual motion was a critical part of the encoding episode – than Imagine and Cost trials. The relatively late onset of brain activity that is sensitive to retrieval of source information is consistent with other studies using different material (Senkfor and Van Petten, 1998; Van Petten et al., 2000), and the division between trials encoded with or without motion closely resembles that observed in our prior study (Senkfor et al., 2002). ERP amplitudes for Perform- and Watch-encoded trials (collapsed) were compared to those for Imagine- and Cost-encoded trials (collapsed), in the 800−1300 msec latency range. Six parietal, temporal, and occipital scalp sites were selected to represent the effect (depicted in Fig. 3), with ANOVA factors of motion (Perform/Watch vs. Imagine/Cost), parietal/temporal/occipital (three levels), and hemisphere. The main effect of motion during encoding was significant (F(1,19) = 53.2, p < .0001). An interaction between motion and parietal/temporal/occipital reflected the parietal focus of the effect, visible in Fig. 3 (F(2,38) = 13.6, p < .0001, e = .64).

Fig. 1.

Fig. 1

Open in a new tab

Grand average of ERPs elicited by correctly identified studied objects during retrieval from the Real-Object session for Perform, Watch, Imagine, and Cost-encoding tasks. The ERPs are plotted in an approximate two-dimensional representation of the scalp electrode placements, with anterior (prefrontal) at the top and posterior (occipital) at the bottom; left in the figure corresponds to left on the scalp.

Fig. 3.

Fig. 3

Open in a new tab

The top of the figure shows a single electrode site (shown as a double circle on the head below) for ERPs elicited by Perform (black), Watch (green), Imagine (blue), and Cost-encoding tasks (red). The bottom of the figure is a topographic voltage map of the difference between Perform and Watch versus Imagine and Cost for Real objects and Pantomime sessions from 800 to 1300 msec poststimulus onset. The small circles across the scalp reflect the location of electrode sites. The filled black circles correspond to the electrode sites included in the statistical analyses for action retrieval processing.

3.3.2. Pantomime session

Fig. 2 shows ERPs from all four encoding conditions from the Pantomime session. The split between Perform/Watch and Imagine/Cost is very similar to that of the Real-object session, although Perform and Watch are slightly more differentiated from each other. An analysis parallel to that above led to the same results: a main effect of motion, (F(1,19) = 41.6, p < .0001), accompanied by an interaction with the topo-graphic factor of parietal/temporal/occipital (F(2,38) = 16.7, p < .0005, e = .63).

Fig. 2.

Fig. 2

Open in a new tab

Grand average of ERPs elicited by correctly identified studied objects during retrieval from the Pantomime session for Perform, Watch, Imagine, and Cost-encoding tasks.

3.4. Action versus non-action encoding

3.4.1. Real-object session

Over frontal cortex only, ERPs for Cost-encoded objects differ from those to the three conditions with action encoding, which are indistinguishable from one another. Fig. 4 illustrates this effect at a left frontal site; Fig. 1 (second row of electrodes) shows that it was restricted to frontal sites, larger over the left than right hemisphere, and very different from the pattern of activity over more posterior cortex. Fig. 1 also shows that this effect closely resembles that of our previous study with the same four conditions, although slightly more anterior and left-lateralized than previously (in Senkfor et al., 2002, the differentiation between action- and cost-encoding was maximal at frontocentral rather than frontal sites, and more bilaterally distributed). For the 800−1300 msec epoch, the Cost condition elicited less positive ERPs than the three action conditions combines, at left medial and lateral frontal sites (F(1,19) = 4.91, p < .05).

Fig. 4.

Fig. 4

Open in a new tab

Grand average of ERPs of all correct trials, collapsed across encoding tasks, during retrieval from the Real-Object session and Pantomime sessions.

3.4.2. Pantomime session

Figs. 2 and 4 show that the Pantomime session failed to show a differentiation between items encoded in the context of a motor act and the nonmotoric control condition of cost-encoding. An analysis parallel to that above yielded a null difference between the Cost condition and the three action conditions (F < 1.0).

3.5. Motor imagery

In the previous study using real objects, a third aspect of retrieval-phase activity was sensitive to the original encoding condition. At prefrontal sites only, items encoded via motor imagery (the Imagine condition) elicited substantially more positive ERPs than items from the other three encoding conditions. Figs. 1 and 2 show that this effect was not replicated here; comparisons between the Imagine condition and the other three were non-significant for both sessions.

3.6. Retrieval-phase differences contingent on real-object versus pantomime encoding

The analyses above show some broad similarities between memory retrieval after encoding tasks with three-dimensional objects and those with images of the objects. In both sessions, brain activity differences contingent on the specific nature of the encoding task emerged only 600−800 msec after stimulus onset. In both sessions, retrieval of episodes with moving objects and/or hands elicited more positive ERPs than those without movement, especially over parietal cortex. One difference between sessions was also apparent: retrieval of episodes with some motoric element (execution, observation, and imagery) differed from episodes without human actions only in the Real-object session.

Fig. 5 contrasts retrieval phase ERPs from the Real-object and Pantomime sessions, collapsed across the four encoding tasks. Over the back half of the head – sites overlying occipital, temporal, and parietal cortex – the responses are very similar to indistinguishable. At prefrontal, frontal, and fronto-central sites, more positive ERPs were observed in the Real-object session. Over prefrontal cortex, this difference begins as early as 600 msec after stimulus onset and persists until the end of the recording epoch at 1300 msec. At frontal and fronto-central sites, the difference begins some later, and is clearly larger over the left than right hemisphere. Session differences are analyzed with the same latency window as above, 800−1300 msec. An ANOVA with factors of session, medial versus dorsal versus lateral scalp location, and hemisphere (left/right) showed interactions between session and hemisphere (F(1,19) = 4.59, p < .05) and between session, hemisphere, and medial/dorsal/lateral (F(2,38) = 3.57, p < .05, e = .79), reflecting qualitatively different retrieval processing for the two sessions.

Fig. 5.

Fig. 5

Open in a new tab

Grand average of ERP elicited during retrieval for the four encoding tasks from Senkfor et al. (2002) over left premotor and from the current study's Real object and Pantomime sessions over left fronto-central sites.

The topography of the session difference is shown in Fig. 6, separately for each of the four preceding encoding tasks. Six left frontal and fronto-central sites were selected for a region-of-interest analysis (shown in Fig. 6 as filled circles). For the full 800−1300 msec epoch, the difference between items encoded with object-directed actions and pantomimed actions (Perform task in the two sessions) was statistically marginal (F(1,19) = 3.46, p = .08), as was the parallel analysis for imagined actions (F(1,19) = 3.76, p = .07). The session difference was significant for the Watch condition (F(1,19) = 5.03, p < .05), and there was no suggestion of a difference for the Cost condition (F(1,19) = .66). Analysis of a slightly briefer latency range of 1000−1300 msec after stimulus onset showed that Real-object encoding produced larger retrieval-phase ERPs for all of the action encoding conditions (Perform, F(1,19) = 4.77, p < .05; Watch F(1,19) = 4.64, p < .05; Imagine, F(1,19) = 6.54, p < .02). Retrieval activity for Cost-encoded items did not significantly differ between the Real-object and Pantomime sessions (F(1,19) = 1.70).

Fig. 6.

Fig. 6

Open in a new tab

Topographic voltage map of the retrieval differences between Real objects minus Pantomime encoding from 1000 to 1300 msec poststimulus onset for Perform, Watch, Imagine, and Cost-encoding tasks. The red reflects greater amplitude for the Real-object session compared to the Pantomime session. The blue reflects greater amplitude for the Pantomime session compared to the Real-object session. The small circles across the scalp reflect the location of electrode sites. The filled black circles correspond to the electrode sites included in the statistical analyses for action retrieval processing.

4. Discussion

4.1. Content-specific brain activity during memory retrieval

When confronted with digital images of objects, participants were asked to recall the context of their initial exposure to each object: performing an action, watching someone else perform an action, imagining an action, or estimating the object's cost. In one session, the three action tasks were conducted with a real three-dimensional object in hand or in close view; in the other session objects were viewed only as images and actions were pantomimed (or imagined as pantomimes).

In both sessions, electrical brain activity during memory retrieval was influenced by the nature of the initial encoding. Objects encoded in the Perform and Watch tasks elicited more positive ERPs than those encoded in the Imagine and Cost tasks. This difference was of late onset (600−800 msec after stimulus onset), consistent with other ERP studies suggesting that retrieval of contextual detail from an episode occurs after episodic recognition of the item itself (Senkfor and Van Petten, 1998; Senkfor et al., 2002; Van Petten et al., 2000). The “motion retrieval effect” was evident over cortical regions that sub-serve visual processing (occipital, temporal, parietal) and maximal over parietal cortex, consistent with the dominance of parietal regions for motion processing. The posterior motion retrieval effect was very similar to that observed in a previous experiment using only real objects during the study phase (Senkfor et al., 2002).

A second aspect of retrieval-phase brain activity after studying real objects also replicated that prior experiment. Images of all objects studied in the context of action – whether performed, watched, or imagined – elicited very similar ERPs over frontocentral cortex, but all were distinct from the activity elicited by objects studied in the cognitively effortful but nonmotoric task of cost-estimation. Here, as in the previous experiment, source accuracy in the Cost condition was equivalent to one of the action conditions (Imagine), so that the frontocentral effect cannot be attributed to differential memory accuracy.

The spatial locus of the “action retrieval” effect is consistent with an origin in some region of motor cortex, a region engaged when actions are recapitulated in memory in a form abstract enough to encompass those performed, imagined, and observed. The observed leftward asymmetry of the effect is somewhat surprising considering that all of the encoding phase actions (or instructed motor imagery) were equally divided between the left and right hands. However, all of the participants were right-handed, and both ERP and hemodynamic imaging evidence indicates that right-handers engage left motor cortex when using either hand (Kim et al., 1993; Kutas and Donchin, 1974). In right-handed individuals, apraxia follows left-hemisphere damage more commonly than left-hemisphere damage, but causes deficits in behavior with either hand. The leftward asymmetry of the “action retrieval” effect observed here is consistent with the idea that the left-hemisphere is dominant for action by right-handed individuals, and hints that this dominance may persist into memory for actions.

4.2. Pantomimed action versus actions with real objects

The new aspect of the current experiment was the comparison between memory for actions on real objects versus pantomimed actions guided by visual images of the objects. This comparison was motivated by three observations reviewed in Section 1 – Introduction: (1) that pantomimed actions create different tactile and proprioceptive feedback than those performed on real objects; (2) that actions with an object in hand differ in motoric properties from pantomimes of the same actions; and (3) data from apraxic patients show dissociations between object-action and pantomime that are suggestive of differences in brain circuitry. It was hypothesized that, if actions on real objects differ qualitatively or quantitatively during their initial performance, they might also leave different memory traces that would be apparent when participants were asked to mentally re-create a study episode in order to perform a source memory judgment. The Cost-encoding condition serves an important control in this comparison. Because cost-estimation is based on nonmotoric object properties, it was predicted that retrieval of episodes involving cost-estimation would not elicit differential brain activity in the Real-object and Pantomime sessions.

Multiple similarities between the Real-object and Pantomime sessions were observed. Participants were well able to discriminate among the four encoding tasks during the Pantomime session despite the lack of three-dimensional objects in the study phase. Source memory accuracy was generally high, and equivalent in the Real-object and Pantomime sessions for all four conditions. Nonetheless, retrieval-phase brain activity in the Pantomime session lacked what I have called the “action retrieval” effect: over frontal scalp, there was no differentiation between action- and non-action-encoded items. The contrast with the Real-object session indicates that whatever information participants used to discriminate the four encoding tasks was qualitatively different after pantomime encoding.

Direct comparisons between retrieval-phase activity in the two sessions showed no significant difference for Cost-encoded objects, ruling out concerns about the global state of the participants across sessions. For the three varieties of action-encoded objects, the difference between sessions was apparent over prefrontal cortex bilaterally, but largest at left frontal and fronto-central sites (see Figs. 5 and 6). The fronto-central maximum for remembering pantomimed versus object-actions is consistent with an origin within the premotor region of cortex.

The spatial focus of the difference between remembering actions with real objects versus pantomimes is very consistent with the locus of activation in hemodynamic studies comparing pictures of tools to pictures of other sorts of objects (Chao and Martin, 2000; Gerlach et al., 2000; Martin et al., 1996). More recent studies have suggested that this activity is not about tools per se, but instead appears when participants make judgments about manipulable objects (including fruits and vegetables) as compared to objects that are typically viewed without motoric engagement (Gerlach et al., 2002; Kellenbach et al., 2003). The left ventral premotor locus in these hemodynamic studies is also considered to be the human analogue of monkey area F5, in which neurons respond during both performance and observation of actions (Murata et al., 1997; see Grezes and Decety, 2001 for review of human studies suggesting a parallel result). In the current results, the commonality of the real-object/pantomime difference after Perform, Watch, and Imagine encoding strongly suggests an origin in left premotor cortex.

The left frontal/frontocentral difference between remembering actions with real objects versus pantomimed actions suggests stronger engagement of motoric brain regions when remembering actions that were directed by physically-present objects. Observations from healthy individuals using objects versus pantomiming transitive actions indicate that pantomimed actions are less closely tuned to object affordances than are natural actions (Goodale et al., 1994; Milner and Goodale, 1995; Westwood et al., 2000). In the current experiment, photos of the objects were used to cue memory retrieval. One account of the results is thus that the visual appearance of an object was less tightly bound with its motoric affordances after pantomime encoding, so that the visual cue was less likely to reactivate motor brain regions.

A more elaborated version of this account stresses the differential utility of motor reactivation for the source memory task after natural versus pantomimed actions. Because pantomimed actions are less accurately tuned to the properties of a given object, pantomimed actions with a variety of objects are also less differentiated from one other than are object-oriented actions. For an object-oriented action, using a key and a screwdriver are differentiated by the width of the aperture between thumb and index finger (reflecting the different width of a key and a screwdriver handle), a slight difference in the relative location of the other fingers, and a difference in the number of turning cycles (single for a key, multiple for a screwdriver). If these fine details are lost in pantomimed actions, then remembering that one performed a twisting action during the study phase will no longer be especially useful for judging whether one performed that action with the screwdriver depicted on a test-phase trial. If the relatively coarse actions produced during pantomime encoding were less useful in solving the source memory question posed during the retrieval phase here, participants may have strategically deemphasized such information, and been less likely to attempt its retrieval. This account of the pantomime/real-object difference is speculative at present and not directly tested here, but it may be possible to examine whether signs of motor cortex activity during retrieval are dependent on the mnemonic utility of motoric information. If, for instance, actions were performed with different real objects that call for very similar actions (tapping in a number on a telephone keypad vs. a calculator keypad of the same size), the lack of motoric differentiation among objects might also lead to reduced motor cortical activity during a subsequent memory test, as compared to blocks of trials with objects that called for distinctive actions.

The hypothesis proposed here – that left premotor activity is related to the fine details, or veridicality, of an action – may initially appear to be at odds with recent hemodynamic studies comparing judgments about object-oriented actions with those about object function. In a positron emission tomography experiment, Kellenbach et al. (2003) presented photos of objects and asked participants answer a question about an object's function (e.g., “Is it used to attach or hold objects together?”) or how one would use it (e.g., “Does using the object involve a twisting or turning motion?”). Both sorts of judgments about manipulable objects elicited greater left premotor activity than answering a function question about non-manipulable objects. For manipulable objects, however, the difference between the function and action tasks lay in posterior parietal rather than frontal cortex. An fMRI experiment by Boronat et al. (2005) used a similar design, with similar results – that judgments about action properties isolated a region of posterior parietal rather than frontal premotor cortex. The two hemodynamic studies differed from the current study in two major regards. First, the present study investigated episodic memory for actions, rather than online judgments about the action properties of objects. The second difference is that participants actually performed actions (with and without objects) during the study phase. The action task in the hemodynamic studies thus put particular stress on motor imagery, whereas the memory test used here put stress on recalling the specifics of one's prior interaction with an object. I suggest that left premotor cortex is engaged by the latter, but not necessarily the former. It remains to be determined whether it was the mnemonic aspect of the current task, or the presence of real actions that led to a premotor difference between the real-object and pantomime sessions here. A critical gap in our knowledge is how brain activity differs during the object use versus pantomimed actions. Neither ERP nor hemodynamic measures lend themselves to this basic question, as both preclude extensive motion by a participant. Other methods will be required to determine whether actions with real objects more strongly engaged premotor cortex than do pantomimed actions. However, the current results show that brain activity during memory retrieval is exquisitely sensitive to qualitative differences in what occurred during the original episode.

Acknowledgments

Assistance with data collection and video analysis was provided by Anand Patel and Lorena Matei. The research was supported by grants from the National Institutes of Health: AG14792, AG08313, and MH12558.

Footnotes

1

Note that even when pantomimed actions are performed in the presence of an object's image, information about its real-life size and three-dimensional shape will still need to be retrieved from memory.

2

No session-order effects were statistically significant, so that the reported analyses collapse across this factor.

3

Of course, it was not possible to monitor the quality of participants’ motor imagery in the Imagine conditions, or their compliance with the instruction to mentally simulate interacting with the object in the Real-object session but to simulate a pantomime in the Pantomime session. However, the blocking of object use versus pantomime into separate sessions was probably useful in this regard. In the Pantomime session, performing only pantomimes on Perform trials and watching the experimenter conduct pantomimes should have encouraged/reminded the participant to imagine pantomimes as well.

4

Huynh–Feldt correction factor for nonsphericity of variance. Shown are the original degrees of freedom, and the corrected probability level.

REFERENCES

  1. Bartolo A, Cubelli R, Della Sala S, Drei S. Pantomimes are special gestures which rely on working memory. Brain and Cognition. 2003;53:483–494. doi: 10.1016/s0278-2626(03)00209-4. [DOI] [PubMed] [Google Scholar]
  2. Beisteiner R, Hollinger P, Lindinger G, Lang W, Berthoz A. Mental representations of movements: brain potentials associated with imagination of hand movements. Electroencephalography and Clinical Neurophysiology. 1995;96:183–193. doi: 10.1016/0168-5597(94)00226-5. [DOI] [PubMed] [Google Scholar]
  3. Boronat CB, Buxbaum LJ, Coslett HB, Tank G, Saffran EM, Kimber DY, Detre JA. Distinctions between manipulation and functional knowledge of objects: evidence from functional magnetic resonance imaging. Cognitive Brain Research. 2005;23:361–373. doi: 10.1016/j.cogbrainres.2004.11.001. [DOI] [PubMed] [Google Scholar]
  4. Brown MB. Log linear models. In: Dixon WJ, editor. BMDP Statistical Software Manual. University of California Press; Berkeley: 1988. pp. 229–274. [Google Scholar]
  5. Buxbaum LJ, Kyle KM, Menon R. On beyond mirror neurons: internal representations subserving imitation and recognition of skilled object-related actions in humans. Cognitive Brain Research. 2005;25:226–239. doi: 10.1016/j.cogbrainres.2005.05.014. [DOI] [PubMed] [Google Scholar]
  6. Cabeza R, Nyberg L. Imaging cognition II: an empirical review of 275 PET and fMRI studies. Journal of Cognitive Neuroscience. 2000;12:1–47. doi: 10.1162/08989290051137585. [DOI] [PubMed] [Google Scholar]
  7. Chao LL, Martin A. Representation of manipulable man-made objects in the dorsal stream. NeuroImage. 2000;12:478–484. doi: 10.1006/nimg.2000.0635. [DOI] [PubMed] [Google Scholar]
  8. Cubelli R, Marchetti C, Boscolo G, Della Sala S. Cognition in action: Testing a model of limb apraxia. Brain and Cognition. 2000;44:144–165. doi: 10.1006/brcg.2000.1226. [DOI] [PubMed] [Google Scholar]
  9. Cunnington R, Iansek R, Gradshaw JL, Phillips JG. Movement-related potentials associated with movement preparation and motor imagery. Experimental Brain Research. 1996;111:429–436. doi: 10.1007/BF00228732. [DOI] [PubMed] [Google Scholar]
  10. Damasio AR. Time-locked multiregional retroactivation: a systems-level proposal for the neural substrates of recall and recognition. Cognition. 1989;33:25–62. doi: 10.1016/0010-0277(89)90005-x. [DOI] [PubMed] [Google Scholar]
  11. Dumont C, Ska B, Schiavetto A. Selective impairment of transitive gestures: an unusual case of apraxia. Neurocase. 1999;5:447–458. [Google Scholar]
  12. Fututake T. Apraxia of tool use: an autopsy case of biparietal infarction. European Neurology. 2003;49:45–52. doi: 10.1159/000067027. [DOI] [PubMed] [Google Scholar]
  13. Gerlach C, Law I, Gade A, Paulson O. Categorization and category effects in normal object recognition: a PET study. Neuropsychologia. 2000;38:1693–1703. doi: 10.1016/s0028-3932(00)00082-8. [DOI] [PubMed] [Google Scholar]
  14. Gerlach C, Law I, Paulson O. When action turns into words. Activation of motor-based knowledge during categorization of manipulable objects. Journal of Cognitive Neuroscience. 2002;14:1230–1239. doi: 10.1162/089892902760807221. [DOI] [PubMed] [Google Scholar]
  15. Goldenberg G, Hentze S, Hermsdörfer J. The effect of tactile feedback on pantomime of tool use in apraxia. Neurology. 2004;63:1863–1867. doi: 10.1212/01.wnl.0000144283.38174.07. [DOI] [PubMed] [Google Scholar]
  16. Goodale MA, Milner AD. Separate visual pathways for perception and action. Trends in Neurosciences. 1992;15:20–25. doi: 10.1016/0166-2236(92)90344-8. [DOI] [PubMed] [Google Scholar]
  17. Goodale MA, Jakobson LS, Keillor JM. Differences in the visual control of pantomimed and natural grasping movements. Neuropsychologia. 1994;32:1159–1178. doi: 10.1016/0028-3932(94)90100-7. [DOI] [PubMed] [Google Scholar]
  18. Grafton ST, Arbib MA, Fadiga L, Rizzolatti G. Localization of grasp representations in humans by positron emission tomography 2: observation compared with imagination. Experimental Brain Research. 1996;112:103–111. doi: 10.1007/BF00227183. [DOI] [PubMed] [Google Scholar]
  19. Grezes J, Decety J. Functional anatomy of execution, mental simulation, observation, and verb generation of actions: a meta-analysis. Human Brain Mapping. 2001;12:1–19. doi: 10.1002/1097-0193(200101)12:1&#x0003c;1::AID-HBM10&#x0003e;3.0.CO;2-V. [DOI] [PMC free article] [PubMed] [Google Scholar]
  20. Hari R, Forss N, Avikainen S, Kirveskari E, Salenius S, Rizzolatti G. Activation of human primary motor cortex during action observation: a neuromagnetic study. Proceedings of the National Academy of Sciences of the United States of America. 1998;95:15061–15065. doi: 10.1073/pnas.95.25.15061. [DOI] [PMC free article] [PubMed] [Google Scholar]
  21. Haxby JV, Ungerleider LG, Horwitz B, Maisog JM, Rapoport SI, Grady CL. Face encoding and recognition in the human brain. Proceedings of the National Academy of Sciences of the United States of America. 1996;93:922–927. doi: 10.1073/pnas.93.2.922. [DOI] [PMC free article] [PubMed] [Google Scholar]
  22. Hayes SM, Ryan L, Schnyer DM, Nadel L. An fMRI study of episodic memory: retrieval of object, spatial, and temporal information. Behavioral Neuroscience. 2004;118:885–896. doi: 10.1037/0735-7044.118.5.885. [DOI] [PMC free article] [PubMed] [Google Scholar]
  23. Heath M, Almeida QJ, Roy EA, Black SE, Westwood D. Selective dysfunction of tool-use: a failure to integrate somatosensation and action. Neurocase. 2003;9:156–163. doi: 10.1076/neur.9.2.156.15072. [DOI] [PubMed] [Google Scholar]
  24. Hebb DO. The Organization of Behavior. Wiley; New York: 1949. [Google Scholar]
  25. Kellenbach M, Brett M, Patterson K. Actions speak louder than functions: the importance of manipulability and action in tool representation. Journal of Cognitive Neuroscience. 2003;15:30–46. doi: 10.1162/089892903321107800. [DOI] [PubMed] [Google Scholar]
  26. Kim SG, Ashe J, Hendrich K, Ellerman JM, Merkle H, Ugğurbil K, Georgopoulos AP. Functional magnetic resonance imaging of motor cortex: hemispheric asymmetry and handedness. Science. 1993;261:615–617. doi: 10.1126/science.8342027. [DOI] [PubMed] [Google Scholar]
  27. Köhler S, Moscovitch M, Winocur G, Houle S, McIntosh AR. Networks of domain-specific and general regions involved in episodic memory for spatial location and object identity. Neuropsychologia. 1998;36:129–142. doi: 10.1016/s0028-3932(97)00098-5. [DOI] [PubMed] [Google Scholar]
  28. Kutas M, Donchin E. Studies of squeezing: handedness, responding hand, response force, and asymmetry of readiness potential. Science. 1974;186:545–548. doi: 10.1126/science.186.4163.545. [DOI] [PubMed] [Google Scholar]
  29. Laimgruber K, Goldenberg G, Hermsdörfer J. Manual and hemispheric asymmetries in the execution of actual and pantomimed prehension. Neuropsychologia. 2004;43:682–692. doi: 10.1016/j.neuropsychologia.2004.09.004. [DOI] [PubMed] [Google Scholar]
  30. Lepage M, McIntosh AR, Tulving E. Transperceptual encoding and retrieval processes in memory: a PET study. NeuroImage. 2001;14:572–584. doi: 10.1006/nimg.2001.0866. [DOI] [PubMed] [Google Scholar]
  31. Martin A, Wiggs CL, Ungerleider LG, Haxby JV. Neural correlates of category-specific knowledge. Nature. 1996;379:649–652. doi: 10.1038/379649a0. [DOI] [PubMed] [Google Scholar]
  32. Masumoto K, Yamaguchi M, Sutani K, Tsuneto S, Fujita A, Tonoike M. Reactivation of physical motor information in the memory of action events. Brain Research. 2006;1101:102–109. doi: 10.1016/j.brainres.2006.05.033. [DOI] [PubMed] [Google Scholar]
  33. McClelland JM, McNaughton BL, O'Reilly RC. Why there are complementary learning systems in the hippocampus and neocortex: insights from the successes and failures of connectionist models of learning and memory. Psychological Review. 1995;109:419–457. doi: 10.1037/0033-295X.102.3.419. [DOI] [PubMed] [Google Scholar]
  34. Milner AD, Goodale MA. The Visual Brain in Action. Oxford University Press; Oxford: 1995. [Google Scholar]
  35. Moll J, de Oliveira-Souza R, Passman LJ, Cunha FC, Souza-Lima F, Andreiuolo PA. Functional MRI correlates of real and imagined tool-use pantomimes. Neurology. 2000;54:1331–1336. doi: 10.1212/wnl.54.6.1331. [DOI] [PubMed] [Google Scholar]
  36. Morris CD, Bransford JD, Franks JJ. Levels of processing versus transfer-appropriate processing. Journal of Verbal Learning and Verbal Behavior. 1977;16:519–533. [Google Scholar]
  37. Motomura N, Yamadori A. A case of ideational apraxia with impairment of object use and preservation of object pantomime. Cortex. 1994;30:167–170. doi: 10.1016/s0010-9452(13)80332-2. [DOI] [PubMed] [Google Scholar]
  38. Murata A, Fadiga L, Fogassi L, Gallese V, Raos V, Rizzolatti G. Object representation in the ventral premotor cortex (area F5) of the monkey. Journal of Neurophysiology. 1997;78:2226–2230. doi: 10.1152/jn.1997.78.4.2226. [DOI] [PubMed] [Google Scholar]
  39. Nyberg L, Habib R, McIntosh AR, Tulving E. Reactivation of encoding-related brain activity during memory retrieval. Proceedings of the National Academy of Sciences of the United States of America. 2000;97:11120–11124. doi: 10.1073/pnas.97.20.11120. [DOI] [PMC free article] [PubMed] [Google Scholar]
  40. Oldfield RC. The assessment of handedness: the Edinburgh inventory. Neuropsychologia. 1971;9:97–113. doi: 10.1016/0028-3932(71)90067-4. [DOI] [PubMed] [Google Scholar]
  41. Poizner H, Mack L, Verfaellie M, Rothi LJG, Heilman KM. Three-dimensional computergraphic analysis of apraxia: neural representations of learned movements. Brain. 1990;113:85–101. doi: 10.1093/brain/113.1.85. [DOI] [PubMed] [Google Scholar]
  42. Rizzolatti G, Fadiga L, Gallese V, Fogassi L. Premotor cortex and the recognition of motor actions. Cognitive Brain Research. 1996;3:131–141. doi: 10.1016/0926-6410(95)00038-0. [DOI] [PubMed] [Google Scholar]
  43. Schmidt D, Krause BJ, Nottaghy FM, Halsband U, Tellman L, Muller-Gertner HW. Brain systems engaged in encoding and retrieval of word-pair associates independent of their imagery content or presentation modalities. Neuropsychologia. 2002;40:457–470. doi: 10.1016/s0028-3932(01)00102-6. [DOI] [PubMed] [Google Scholar]
  44. Schnitzler A, Salenius S, Salmelin R, Jousmaki V, Hari R. Involvement of primary motor cortex in motor imagery: a neuromagnetic study. NeuroImage. 1997;6:201–208. doi: 10.1006/nimg.1997.0286. [DOI] [PubMed] [Google Scholar]
  45. Senkfor AJ. Episodic action memory: Characterization of the time course and neural circuitry. In: Stamenov M, Gallese V, editors. Mirror Neurons and the Evolution of the Brain. John Benjamins Publishing Co; Amsterdam: 2002. [Google Scholar]
  46. Senkfor AJ, Van Petten C. Who said what: an ERP investigation of source and item memory. Journal of Experimental Psychology: Learning, Memory, and Cognition. 1998;24:1005–1025. doi: 10.1037//0278-7393.24.4.1005. [DOI] [PubMed] [Google Scholar]
  47. Senkfor AJ, Van Petten C, Kutas M. Enactment versus conceptual encoding: equivalent item memory but different source memory. Cortex. doi: 10.1016/j.cortex.2007.12.004. in press. doi:10.1016/j.cortex.2007.12.004. [DOI] [PMC free article] [PubMed] [Google Scholar]
  48. Senkfor AJ, Van Petten C, Kutas M. 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. 2002;14:402–419. doi: 10.1162/089892902317361921. [DOI] [PubMed] [Google Scholar]
  49. Vaidya CJ, Zhao M, Desmond JE, Gabrieli JDE. Evidence for cortical encoding specificity in episodic memory: memory-induced re-activation of picture processing areas. Neuropsychologia. 2002;40:2136–2143. doi: 10.1016/s0028-3932(02)00053-2. [DOI] [PubMed] [Google Scholar]
  50. Van Petten C, Senkfor AJ, Newberg W. Memory for drawings in locations: spatial source memory and event-related potentials. Psychophysiology. 2000;37:551–564. [PubMed] [Google Scholar]
  51. Westwood DA, Chapman CD, Roy EA. Pantomimed actions may be controlled by the ventral visual stream. Experimental Brain Research. 2000;130:545–548. doi: 10.1007/s002219900287. [DOI] [PubMed] [Google Scholar]
  52. Wheeler ME, Petersen SE, Buckner RL. Memory's echo: vivid remembering reactivates sensory-specific cortex. Proceedings of the National Academy of Sciences of the United States of America. 2000;97:11125–11129. doi: 10.1073/pnas.97.20.11125. [DOI] [PMC free article] [PubMed] [Google Scholar]

RESOURCES