Abstract

Event sequence knowledge is necessary to learn, plan, and perform activities of daily life. Clinical observations suggest that the prefrontal cortex (PFC) is crucial for goal-directed behavior such as carrying out plans, controlling a course of actions, or organizing everyday life routines. Functional neuroimaging studies provide further evidence that the PFC is involved in processing event sequence knowledge, with the medial PFC (Brodmann area 10) primarily engaged in mediating predictable event sequences. However, the exact role of the medial PFC in processing event sequence knowledge depending on the frequency of corresponding daily life activities remains obscure. We used event-related functional magnetic resonance imaging while healthy volunteers judged whether event sequences from high- (HF), moderate- (MF), and low-frequency (LF) daily life activities were correctly ordered. The results demonstrated that different medial PFC subregions were activated depending on frequency. The anterior medial Area 10 was differentially activated for LF and the posterior medial Area 10 for HF activities. We conclude that subregions of the medial PFC are differentially engaged in processing event sequence knowledge depending on how often the activity was reportedly performed in daily life.

Introduction

Event sequence knowledge contains information about activities of daily life; for example, the activity “get ready for work” consists of a sequence of events such as waking up, getting out of bed, using the bathroom, taking a shower, getting dressed, eating breakfast, etc. Retrieving stored event sequences is essential for planning and performing daily life activities. Clinical observations suggest that the prefrontal cortex (PFC) is crucial for goal-directed behavior such as carrying out plans, controlling a course of actions, or organizing everyday life routines (Shallice 1982; Stuss and Benson 1984; Eslinger and Damasio 1985; Janowsky et al. 1989; Shallice and Burgess 1991). For example, when patients with PFC lesions were asked to evoke a plan of a daily activity, for example, “organize your morning routine,” the patients generated as many prototypical events as normals but had difficulties in temporal ordering the events of the same activity (Sirigu et al. 1995).

The human PFC (Brodmann's area 10 in particular) is proportionally twice as large in humans as in any of the great apes (Rilling and Insel 1999; Semendeferi et al. 2002). In addition, there is evidence that PFC-regulated specialized human behavior must be due to a more sophisticated internal and differentially organized neural architecture (Elston 2000; Elston and Rosa 2000; Chiavaras et al. 2001). A key principle of neurons in the human PFC is their ability to sustain firing and to code the temporal and sequential properties of ongoing events in the environment or in mind over longer periods of time (Fuster et al. 2000; Levy and Goldman-Rakic 2000). Furthermore, pyramidal cells in the PFC are significantly more spinous, as compared with pyramidal cells in other cortical areas, making them capable of handling a larger amount of excitatory inputs (Elston 2000; Elston and Rosa 2000). These evolutionary advances from primates to humans in the structure and connectivity of the PFC and the physiological properties of its neurons enable the human PFC to code, store, and retrieve long and complex sequences of behavior (Nichelli et al. 1995).

Medial and lateral PFC belong to 2 distinct architectonic trends within the human PFC (Pandya and Yeterian 1996). The medial trend is phylogenetically and ontogenetically older than the lateral trend, which is especially well developed in humans (Stuss and Benson 1986). There is also strong evidence for a functional dissociation between medial and lateral PFC (Koechlin et al. 2000; Burgess et al. 2003; Gilbert, Spengler, Simons, Frith, et al. 2006; Gilbert, Spengler, Simons, Steele, et al. 2006). In sequence learning, for example, lateral PFC regions are engaged in performing sequences contingent upon unpredictable events, whereas the medial PFC region is engaged in situations when predictable sequences of stimuli are encountered (Koechlin et al. 2000). Functional neuroimaging studies provide evidence that the PFC is involved in mediating event sequence knowledge (Partiot et al. 1996; Crozier et al. 1999; Ruby et al. 2002; Knutson et al. 2004). In particular, the medial PFC (Brodmann area [BA] 10, MPFC) is primarily engaged in mediating event sequence knowledge that has a predictable relationship with action sequences (Wood et al. 2005). Despite this large body of evidence, the exact role of the MPFC in processing event sequence knowledge depending on the frequency of corresponding daily life activities remains obscure.

We have proposed a “representational” account of PFC function that seeks to establish the format in which context-dependent information is stored in the PFC (Grafman 2002; Wood and Grafman 2003). According to this approach, storage and processing are integrally related and dependent on the same neuronal infrastructure. The PFC processes goal-oriented event sequence knowledge by encoding and retrieving the sequence of event components. In parallel, the PFC interacts with knowledge stored in posterior cortical regions (e.g., objects, faces, and words) to bind these components into the context of an episode by placing objects, faces, and words into a certain time, place, and action (via the hippocampus and related structures) (O'Reilly and Rudy 2000). We argue that the representation of event sequence knowledge is critically dependent on how frequently a person enacts these sequences. For example, an activity such as “getting ready for work” includes a sequence of events that are executed more often than the events executed in an activity such as “having your picture taken.” There is neuropsychological evidence that leads to the hypothesis that low-frequency (LF) and high-frequency (HF) event sequence knowledge are mediated by different neural substrates. Patients with anterior PFC damage show frequency effects in event knowledge tasks, with HF event sequence knowledge being better preserved than LF event sequence knowledge (Sirigu et al. 1995).

If frequency is a crucial parameter, then variation in frequency should be closely linked to brain regions recruited during tasks that require event sequence knowledge. In this study, we used event-related functional magnetic resonance imaging (fMRI) to investigate the pattern of brain responses when healthy volunteers were engaged in event order judgments about the sequential organization of LF (e.g., having your picture taken), moderate frequency (MF) (e.g., doing the laundry), and HF (e.g., getting ready for work) activities according to normative data. Based on evidence that the MPFC (BA 10) is engaged in processing predictable event sequence knowledge (Wood et al. 2005), we hypothesized that event sequences will evoke activation in different subregions of the MPFC depending on how often persons were engaged in such daily life activities.

Material and Methods

Participants

Eighteen healthy volunteers (9 women, mean age 29.4 ± 6.0 years, range 23–40, mean education level 17.2 ± 2.1 years, range 15–22) participated for financial compensation in the fMRI experiment. All were native English speakers and strongly right-handed as determined from the Edinburgh Handedness Inventory (97.1 ± 6.1, range 76–100). They underwent a neurological examination by a board-certified neurologist during the previous 12 months and had normal or corrected-to-normal vision, no history of medical, psychiatric, or neurological diagnoses, and were not taking medication. Informed consent was obtained according to procedures approved by the National Institute of Neurological Disorders and Stroke (NINDS) Institutional Review Board.

Stimuli

Stimuli in the fMRI experiment were derived from 2 normative studies. Daily life activities were taken from a normative study conducted by Rosen et al. (2003). In part one of that study, individuals (n = 120) recorded their daily activities for 7 consecutive days. The activities ranged along a frequency continuum from activities reported only once by a single individual during the week (e.g., going to an audition) to activities reported more than once by most or all of the individuals during the week (e.g., getting ready for work). The authors selected 5 activities from the beginning, middle, and ending of the frequency continuum. Those activities were grouped into LF, MF, and HF categories. In part 2 of the study, a new set of individuals (n = 60) was asked to generate events belonging to the 15 selected activities. The authors reported the 18 most frequently generated events for each activity.

For our fMRI experiment, we used the same 15 daily life activities (5 activities per category): HF activities (e.g., getting ready for work), MF activities (e.g., doing laundry), and LF activities (e.g., having your picture taken). Each activity was composed of a header and 16 sequentially structured events (supplementary appendix). Note that each of these activities represents a predictable event sequence that stores both the goals and boundaries of events. However, each of these event sequences varies in terms of frequency depending on how often the event sequence was reportedly performed in daily life. Eight event pairs, adjacent in sequence, were created for each activity, which were either ordered in chronological sequence (e.g., wake up—get out of bed) or inverse sequence (e.g., get in car—leave house). Each frequency group consisted of 40 event pairs balanced for order of sequence. Headers contained 2–5 words (mean ± standard error of the mean [SEM] 3.47 ± 0.46), and event pairs contained 2–7 words (mean ± SEM 3.02 ± 0.36).

In another normative study we conducted, 40 participants (20 women, mean age 31.5 ± 8.6 years, range 22–48, mean education level 17.5 ± 3.4 years, range 14–23) rated the emotion and scene-likeness for all the events from each activity to control for emotion processing (positive and negative valence) and autobiographic retrieval (episodic and semantic), respectively. Participants rated the emotion and scene-likeness for the events of each activity while seeing both the header and the event of the activity on a 7-point Likert-scale: 1) Emotion: We would like you to rate each event according to whether it has an intense emotional association for you (positive and negative valence scale: 1 = not at all and 7 = extremely intense); and 2) Scene-likeness: We would like you to rate how intensely are you reminded of a specific episode or scene that you have experienced during your life while reading about the event (1 = not at all and 7 = extremely intense). Emotion (positive valence) and scene-likeness ratings were equal among frequency categories (emotion [positive valence]: F2,78 = 2.06, P = 0.134; scene-likeness: F2,78 = 2.10, P = 0.145), whereas emotion ratings (negative valence) decreased significantly from the LF to HF category (emotion [negative valence]: F2,78 = 30.95, P < 0.001).

In addition, participants rated the frequency and familiarity for the events of each activity on a 7-point Likert-scale: 1) Frequency: We would like you to rate how often you are usually involved in each of the following events (1 = never and 7 = daily); and 2) Familiarity: We would like you to rate how familiar you are with each of the following events, that is, how often you have performed or observed these events (1 = extremely unfamiliar and 7 = extremely familiar). Because frequency and familiarity ratings were highly correlated for each frequency category (LF: r = 0.93, P < 0.001; MF: r = 0.94, P < 0.001; HF: r = 0.89, P < 0.001), stimuli were not further analyzed separately for the frequency and familiarity dimensions.

Presentation Conditions and Procedure

An experimental condition (sequence order judgment task) and a control condition (font discrimination task) were employed. The latter condition served as a baseline that controlled for display, motor responses, and attentional mechanisms, thought to be common to the sequence order judgment task across frequency conditions. At the start of each trial an instruction indicating the type of task (order task or font task) was presented for 1600 ms on the screen. After a blank screen of 400 ms, a header for an activity (e.g., get ready for work) appeared. After a delay of 1000 ms the header disappeared. For the sequence order task, a pair of events (e.g., wake up—get out of bed) belonging to this activity was displayed vertically in the middle of the screen. For the font task, a pair of events (e.g., enter a store—brush teeth) belonging to different activities was displayed on the screen. Within a fixed time of 5000 ms, participants had to make a decision on a hand-response pad with their right index (left button) or middle finger (right button). Stimulus presentation was event-related and trials were separated by a randomly assigned jittered interstimulus interval of 4 s (range: 2–6 s) (Fig. 1).

Figure 1.

Experimental design. Each 8-s trial began with an instruction indicating the type of task (sequence order task [SOT] or font discrimination task [FDT]) for 1600 ms, followed by a 400-ms blank interval. Afterward the task began with the presentation of the activity header. After 1000 ms, the header disappeared and a pair of events was displayed. For the SOT (experimental condition), participants had to decide which one of the events occurs first in the chronological sequence of this activity by pressing an assigned key. For the FDT (control condition), participants had to decide which of the events had the same font (Swiss or Helvetica) as the header. The same headers and event pairs were used for both tasks but for the FDT, event pairs from one activity were assigned to headers of other activities. Within a fixed time of 5000 ms, participants had to make a decision on a hand-response pad with their right index or middle finger. After their response, participants saw a blank screen for the rest of the 5000 ms. Trials were separated by randomly assigned jittered intervals (mean 4 s, range 2–6 s). RT and ER were recorded for each trial.

Figure 1.

Experimental design. Each 8-s trial began with an instruction indicating the type of task (sequence order task [SOT] or font discrimination task [FDT]) for 1600 ms, followed by a 400-ms blank interval. Afterward the task began with the presentation of the activity header. After 1000 ms, the header disappeared and a pair of events was displayed. For the SOT (experimental condition), participants had to decide which one of the events occurs first in the chronological sequence of this activity by pressing an assigned key. For the FDT (control condition), participants had to decide which of the events had the same font (Swiss or Helvetica) as the header. The same headers and event pairs were used for both tasks but for the FDT, event pairs from one activity were assigned to headers of other activities. Within a fixed time of 5000 ms, participants had to make a decision on a hand-response pad with their right index or middle finger. After their response, participants saw a blank screen for the rest of the 5000 ms. Trials were separated by randomly assigned jittered intervals (mean 4 s, range 2–6 s). RT and ER were recorded for each trial.

Instructions for the experiment were as follows. For the sequence order task, participants had to decide which of the events occurs first in the chronological sequence of the activity. Half of the participants were instructed to press the left button if the event on the top occurred first (e.g., wake up—get out of bed) or to press the right button if the event on the bottom occurred first (e.g., get in car—leave house). The other half of the participants was instructed to do the converse. For the font discrimination task, participants had to decide which event had the same font (Swiss or Helvetica Font) as the header. If the event on the top had the same font, of the participants had to press the left button or if the event on the bottom had the same font as the header the right button had to be pressed. For the remaining participants, the converse was requested. Participants were asked to perform the tasks to the best of their ability and to respond as quickly and accurately as possible. After their response, participants saw a blank screen for the rest of the 5000 ms.

The fMRI experiment consisted of 3, 12-min runs counterbalanced across participants. Each run included both trials for the sequence order task (n = 40) and trials for the font discrimination task (n = 20) randomized across each run. The experiment consisted of 180 trials (font discrimination task: 60 trials and sequence order task: 120 trials). Response time (RT) and error rate (ER) were recorded for each trial. All stimuli were carefully matched for word length across frequency conditions. The same headers and event pairs were used for both the experimental and control conditions, but for the control condition, event pairs from one activity were assigned to headers of other activities.

Data Acquisition

Before being scanned, participants were first trained on a separate set of stimuli to familiarize them with the experiment. Stimulus presentation was synchronized with the scanner pulses using the Experimental Run Time System (Berisoft Cooperation, Germany, http://www.erts.de) software package running on an IBM computer. With a magnetically shielded LCD video projector, stimuli (18-point font type) were back-projected onto a translucent screen placed at the feet of the participant. During scanning, participants had to lie flat on their back in the MRI scanner and viewed the screen by a mirror system attached to the head coil. Head motion was restricted using foam pads placed around the participants' head.

Imaging was performed on a 3-T General Electric whole-body scanner equipped with a standard circularly polarized head coil. Anatomical (T1-weighted 3D Magnetization-Prepared Rapid Acquisition Gradient Echo sequence: time repetition [TR], 9.7 ms; time echo [TE], 4.0 ms; flip angle, 12°; field of view, 240 mm; matrix size, 256 × 256; thickness, 1.2 mm; in-plane resolution, 0.8594 × 0.8594 mm2) and functional images (T2*-weighted 2D gradient echo-planar image [EPI] sequence: TR, 2 s; TE, 30 ms; flip angle, 90°; thickness, 6 mm; number of slices, 22; field of view, 240 mm; in-plane resolution, 3.75 × 3.75 mm2) were acquired. Per run 365 volume images were taken parallel to the anterior commissure–posterior commissure line. The first 5 volumes were discarded to allow for T1 equilibration effects.

Data Analysis

Behavioral data analysis was carried out using SPSS (SPSS Inc., Chicago, IL, http://www.spss.com). Alpha was set to P < 0.05 for all behavioral analyses. Image analyses were performed using Brain Voyager 2000 and Brain Voyager QX (Brain Innovation, Maastricht, The Netherlands, http://www.BrainVoyager.com). The following data preprocessing steps were applied: slice scan time correction (using sinc interpolation), linear trend removal, temporal high-pass filtering to remove low-frequency nonlinear drifts of 3 or fewer cycles per time course, spatial smoothing (8-mm full width at half maximum), and 3D motion correction to detect and correct for small head movements by spatial alignment of all participants to the first volume by rigid body transformation. Estimated translation and rotation parameters were inspected and never exceeded 2 mm or 2°. Functional slices were coregistered to the anatomical volume using position parameters from the scanner and manual adjustment to obtain optimal fit and transformed into Talairach space (3 × 3 × 3 mm3) (Talairach and Tournoux 1988).

A general linear model (GLM) corrected for first-order serial correlation was applied. Multisubject analyses were performed by multiple linear regressions of the blood oxygen level–dependent (BOLD)-RT course in each voxel using 4 predictors: LF, MF, HF, and control (C). Predictor time courses were adjusted for the hemodynamic response delay by convolution with a hemodynamic response function (delta 2.5, tau 1.25) (Boynton et al. 1996). Random-effect analyses were performed on the group data (n = 18), enabling generalization of the statistical inferences to the population level. Due to a priori hypothesis regarding the role of the MPFC (BA 10) in mediating daily life activities (Wood et al. 2005) an uncorrected P value adjusted to a per-voxel false-positive probability of P < 0.0005 with a cluster size threshold of 300 mm3 was applied (Forman et al. 1995). Due to a lack of a priori hypothesis for other areas of the brain a corrected P < 0.05 was used for whole-brain analysis.

Time courses of each frequency category were extracted from each ROI to compute their respective percentage signal changes with the condition-per-file-based series as baseline. Statistical images were overlaid onto Brain Voyager's single subject canonical T1 image in Talairach space. BAs were determined by using the Talairach Daemon Client software (Research Imaging Center, San Antonio, TX, http://ric.uthscsa.edu/).

Results

Behavioral Data

The RTs (mean ± SEM) and ERs (mean ± SEM) from the control (C) condition differed significantly from those in the experimental conditions (LF, MF, and HF) (RT: t(17) = 13.57, P < 0.001; ER: t(17) = 9.63, P < 0.001) (Table 1). The RTs and ERs of the experimental conditions were submitted to a one-way analysis of variance (ANOVA) with frequency as a within-subject factor. The ANOVA revealed no significant main effect for the factor frequency (RT: F2,34 = 0.83, P = 0.445, ER: F2,34 = 2.14, P = 0.133); that is, RTs and ERs were equal across frequency conditions.

Table 1

RTs (mean ± SEM in ms) and ERs (mean ± SEM in %) for control and experimental conditions

 RT (ms) ER (%) 
 Control condition 
Control 1060 ± 362 2.6 ± 3.9 
 Experimental condition 
LF 2369 ± 458 9.3 ± 7.5 
MF 2380 ± 453 10.3 ± 7.5 
HF 2334 ± 461 12.2 ± 4.4 
 RT (ms) ER (%) 
 Control condition 
Control 1060 ± 362 2.6 ± 3.9 
 Experimental condition 
LF 2369 ± 458 9.3 ± 7.5 
MF 2380 ± 453 10.3 ± 7.5 
HF 2334 ± 461 12.2 ± 4.4 

Imaging Data

To identify brain regions commonly activated by processing event sequence knowledge independently of the factor frequency, we first contrasted the control (C) from the experimental conditions (LF, MF, and HF). The task contrast (LF + MF + HF > C) yielded activation in the medial frontal gyrus (MFG) (BA 10; x = 3, y = 54, z = −5), supporting our hypothesis that the MPFC is involved in processing of event sequence knowledge (Fig. 2A). Additional activations were found in the left inferior frontal gyrus (BA 9) and the left supplementary motor area (BA 6) (Table 2). The activation of the supplementary motor area is consistent with findings of other neuroimaging studies dealing with sequencing information (Crozier et al. 1999; Ruby et al. 2002), whereas the activation of the inferior frontal gyrus is consistent with findings dealing with semantic judgment (Demb et al. 1995; Poldrack et al. 1999) and response selection (Rowe et al. 2000; Badre and Wagner 2004) relevant for the experimental task.

Figure 2.

Brain activation for task contrast. (A) When contrasting control from experimental condition activation was observed in the MPFC (BA 10; x = 3, y = 54, z = −5). Brain activation for frequency contrasts. (B) When each frequency condition was contrasted against the other 2 frequency conditions activation was observed in the anterior MPFC (BA 10; x = 3, y = 68, z = 5) for the LF contrast, in the middle MPFC (BA 10; x = 8, y = 53, z = −5) for the MF contrast, and in the posterior MPFC (BA 10; x = −5, y = 43, z = −9) for the HF contrast. Time courses of activation profiles for each frequency condition. x axis, poststimulus onset times measured from header onsets. y axis, adjusted BOLD signal changes expressed in relative percentage of the mean.

Figure 2.

Brain activation for task contrast. (A) When contrasting control from experimental condition activation was observed in the MPFC (BA 10; x = 3, y = 54, z = −5). Brain activation for frequency contrasts. (B) When each frequency condition was contrasted against the other 2 frequency conditions activation was observed in the anterior MPFC (BA 10; x = 3, y = 68, z = 5) for the LF contrast, in the middle MPFC (BA 10; x = 8, y = 53, z = −5) for the MF contrast, and in the posterior MPFC (BA 10; x = −5, y = 43, z = −9) for the HF contrast. Time courses of activation profiles for each frequency condition. x axis, poststimulus onset times measured from header onsets. y axis, adjusted BOLD signal changes expressed in relative percentage of the mean.

Table 2

Brain areas activated during the task and frequency contrasts

Regions of activation Cluster size (mm3Laterality Talairach coordinates t-score 
   x y z  
Task contrast       
LF + MF + HF > control       
MFG (10) 1686 54 −5 3.43* 
Supplementary motor area (6) 620 −5 53 8.06** 
Inferior frontal gyrus (9) 1332 −45 15 20 7.40** 
Frequency contrasts       
LF contrast: LF > MF + HF       
MFG (10) (ant MPFC) 347 68 3.79* 
MF contrast: MF > LF + HF       
MFG (10) (mid MPFC) 486 53 −5 3.58* 
HF contrast: HF > LF + MF       
MFG (10) (post MPFC) 315 −5 43 −9 3.59* 
Regions of activation Cluster size (mm3Laterality Talairach coordinates t-score 
   x y z  
Task contrast       
LF + MF + HF > control       
MFG (10) 1686 54 −5 3.43* 
Supplementary motor area (6) 620 −5 53 8.06** 
Inferior frontal gyrus (9) 1332 −45 15 20 7.40** 
Frequency contrasts       
LF contrast: LF > MF + HF       
MFG (10) (ant MPFC) 347 68 3.79* 
MF contrast: MF > LF + HF       
MFG (10) (mid MPFC) 486 53 −5 3.58* 
HF contrast: HF > LF + MF       
MFG (10) (post MPFC) 315 −5 43 −9 3.59* 

Note: Brodmann's areas are depicted in parentheses. Cluster size (volume in mm3), laterality (right and left hemisphere), and t-score are also given. The stereotaxic coordinates of the peak of the activation are given to Talairach space. Abbreviations: ant, anterior; mid, middle; post, posterior

*

P < 0.0005, cluster size of minima 300 mm3.

**

Pcor < 0.05, whole-brain analysis.

In order to explore brain regions that differed in their activation according to frequency, we performed 3 frequency contrasts in which each frequency category was contrasted against the other 2 frequency categories (LF contrast: LF > MF + HF; MF contrast: MF > LF + HF, and HF contrast: HF > LF + MF). We refer to these effects as content dependent because they reflect the content of the event sequence knowledge about daily activities without the interference of task contingencies (which were balanced across conditions). These contrasts revealed 3 distinct regions that varied along an anterior-to-posterior axis of the MPFC dependent on frequency: the right anterior MPFC (MFG; BA 10; x = 3, y = 68, z = 5) for the LF contrast (LF > MF + HF), the right middle MPFC (MFG; BA 10; x = 8, y = 53, z = −5) for the MF contrast (MF > LF + HF), and the left posterior MPFC (MFG; BA 10; x = −5, y = 43, z = −9) for the HF contrast (HF > LF + MF) (Table 2, Fig. 2B). The whole-brain analysis gave no additional regions of activation.

To further confirm the specificity of the anterior-to-posterior axis within the MPFC subregions we employed a region of interest (ROI) analysis. For each participant, parameter estimates (mean beta weights) for each frequency condition were derived from each MPFC location after identifying the peak of activation and surrounding voxels encompassing 50 mm3 (Fig. 3A). A 2-way ANOVA with parameter estimates was performed to explore the interaction between the factor frequency (low and high) and ROI (anterior MPFC and posterior MPFC). The main effects of frequency and ROI were not significant (frequency: F1,17 = 0.62, P = 0.439; ROI: F1,17 = 0.43, P = 0.523). Importantly, the frequency by ROI interaction was significant (F1,17 = 18.33, P < 0.001). Planned follow-up paired t-tests indicated a significantly higher mean activation level for the LF compared with HF condition in the anterior MPFC (t(17) = 2.35, P < 0.05) and a significantly higher mean activation level for the HF compared with LF condition in the posterior MPFC (t(17) = −3.79, P < 0.001); that is, the anterior MPFC is more engaged in LF activities and the posterior MPFC more in HF activities (Fig. 3B). On the subject-by-subject level, parameter estimates in 14 of 18 participants were higher for LF than MF and HF condition for the anterior MPFC, whereas parameter estimates in 15 of 18 participants were higher for HF than LF and MF condition for the posterior MPFC (Fisher's Exact test: χ2(1) = 13.11, P < 0.001) (Fig. 3C).

Figure 3.

ROI analysis. (A) Parameter estimates (mean beta weights) for each frequency condition (low, moderate, and high) derived from each functional ROI (anterior MPFC, middle MPFC, and posterior MPFC). (B) On the group level, a 2-way ANOVA with parameter estimates showed a significant interaction between the factor frequency (low and high) and ROI (anterior MPFC and posterior MPFC). (C) On the subject-by-subject level, participants' parameter estimates were plotted separately for the factor frequency (low and high) and ROI (anterior MPFC and posterior MPFC). For the anterior MPFC parameter estimates in 14 of 18 participants were higher for LF than MF and HF condition, whereas for the posterior MPFC parameter estimates in 15 of 18 participants were higher for HF than LF and MF condition. Both the group and subject-by-subject analysis provide support for a frequency gradient along the anterior-to-posterior axis of the MPFC, in which the anterior MPFC is more engaged in LF activities and the posterior MPFC in HF activities.

Figure 3.

ROI analysis. (A) Parameter estimates (mean beta weights) for each frequency condition (low, moderate, and high) derived from each functional ROI (anterior MPFC, middle MPFC, and posterior MPFC). (B) On the group level, a 2-way ANOVA with parameter estimates showed a significant interaction between the factor frequency (low and high) and ROI (anterior MPFC and posterior MPFC). (C) On the subject-by-subject level, participants' parameter estimates were plotted separately for the factor frequency (low and high) and ROI (anterior MPFC and posterior MPFC). For the anterior MPFC parameter estimates in 14 of 18 participants were higher for LF than MF and HF condition, whereas for the posterior MPFC parameter estimates in 15 of 18 participants were higher for HF than LF and MF condition. Both the group and subject-by-subject analysis provide support for a frequency gradient along the anterior-to-posterior axis of the MPFC, in which the anterior MPFC is more engaged in LF activities and the posterior MPFC in HF activities.

Finally, to control our findings for possible confounds with emotion processing and autobiographic retrieval, we adopted a parametric approach by defining additional predictors for the GLM using ratings from the dimensions emotion (positive and negative) and scene-likeness as modulation parameters. We tested for linear relationships between changes in emotion and scene-likeness ratings and changes in the BOLD signals. Analyses for both emotion and scene-likeness ratings did not modulate activations in any of the previous identified MPFC regions. Instead, ratings with positive emotional valence were associated with activations in the amygdala (x = 21, y = −3, z = −21) and superior frontal gyrus (BA 10; x = 11, y = 60, z = 18), ratings with negative emotional valence were associated with activations in the anterior insula (BA 13; x = 30, y = 20, z = 15), lateral middle frontal gyrus (BA 10; x = 41, y = 52, z = 12), and retrosplenial cortex (BA 31; x = 2, y = −57, z = 26), and scene-likeness ratings were associated with activations in the retrosplenial cortex (BA 31; x = −5, y = −63, z = 21) (supplementary Fig. S1).

Discussion

We used event-related fMRI to investigate the pattern of brain responses when healthy volunteers were engaged in event order judgments about the sequential organization of LF, MF, and HF activities according to normative data. The analyses revealed a frequency gradient along the anterior-to-posterior axis of the MPFC, in which the anterior medial Area 10 is engaged in LF activities, and the posterior medial Area 10 is engaged in HF activities.

But were the identified MPFC subregions selectively related to the variation in frequency? First, it could be argued that the MPFC regions were responding to task difficulty associated with higher cognitive control or conflict (Carter et al. 1998). However, because RTs and ERs did not differ among frequency conditions, this is unlikely. Second, it could be argued that the results reflect activation of a multipurpose working memory system rather than being specific to event sequence knowledge. Against this interpretation is the fact that working memory load was kept constant across conditions. Moreover, according to accumulated evidence from neuroimaging studies, brain activation in working memory tasks is primarily associated with dorsolateral regions of the PFC (Smith and Jonides 1999; Cabeza and Nyberg 2000). Third, it could be argued that differences in emotions could lead to different MPFC activations because the MPFC is also engaged in emotion processing (Lacroix et al. 2000; Kalisch et al. 2005). The regression analysis with the parameter emotion did not modulate activation in any of the previously identified frequency-dependent MPFC regions. Instead, emotion ratings showed associations with brain regions such as amygdala and anterior insula, in agreement with the crucial role of these regions in emotion processing (Zalla et al. 2000; Calder et al. 2001; Cardinal et al. 2002). Finally, it could be argued that judgments about LF event sequences primarily require retrieval of episodic autobiographical memories, whereas judgments about HF event sequences require primarily semantic autobiographical memory (Levine et al. 2004). Therefore, activation in the identified MPFC regions would reflect the monitoring of information retrieved from episodic or semantic autobiographical memory. We think this is unlikely because scene-likeness ratings were equal among frequency conditions. In addition, the regression analysis with scene-likeness ratings did not modulate activation in the frequency-dependent MPFC regions. Instead activation was found in the retrosplenial cortex in agreement with the role of this region in retrieval of autobiographical memories (Fink et al. 1996; Piefke et al. 2003).

Altogether, the results support the idea that the MPFC modulates the processing of event sequence knowledge with an anterior–posterior gradient according to the frequency of use in real life. Note that during the experimental task participants were not actually involved in carrying out these activities. However, to solve the task participants had to access the stored representations about the sequential organization of LF, MF, and HF activities. Therefore, it appears reasonable to assume that performing the task taps into the same sorts of representations that are activated during planning, monitoring, and executing those daily life activities.

A previous functional imaging study corroborates our findings. In another parametric event-related fMRI study, multidimensional scaling was used to establish the psychological structure of event sequence knowledge (Wood et al. 2005). The same normative data sample was employed as in the present study with several activities from the MF and HF condition. Using a different group of participants, experimental paradigm, and analysis procedure a selective association between activation in the middle MPFC (BA 10; x = 4, y = 54, z = −9) and a psychological dimension termed “experience” was demonstrated. This dimension was best described by the variables of rule knowledge, commonality, and frequency. In our experiment, we found an almost identical location in the middle MPFC (BA 10; x = 8, y = 53, z = −5) for the MF contrast. Our results replicate the earlier finding but they provide additional evidence that the MPFC modulates the processing of event sequence knowledge according to its frequency of execution in real life.

How can the frequency gradient along the anterior-to-posterior axis of the MPFC be explained? According to our representational account of PFC function, the PFC processes goal-oriented event sequence knowledge by encoding and retrieving the sequence of event components (Grafman 2002; Wood and Grafman 2003). The MPFC represents event sequences that have a predictable relationship with action sequences. We argue that the anterior MPFC codes more complex cognitive information about an event sequence. As an event sequence becomes more frequently used an economy of representation develops, in which the posterior MPFC, activated in parallel with the anterior MPFC, codes sparser cognitive information about the same event sequence. This coding format leads to different profiles of MPFC activation depending upon the frequency of the event sequence the person executes. It would allow simpler representational codes to rapidly instruct lower level systems (e.g., motor) to implement action sequences. In other words, the same activity can be performed quickly (e.g., by using a heuristic based on sparser coding) or slowly (e.g., by using deliberate reflection based on detailed coding) depending upon situational demands on the stored event sequence knowledge.

Interestingly, each of the frequency-dependent MPFC regions falls onto one of the 3 architectonic subdivisions of human BA 10 proposed by Ongur et al. (2003) (anterior MPFC: polar area 10p, middle MPFC: rostral area 10r, and posterior MPFC: medial area 10 m). Area 10 is probably the single largest cytoarchitectonic area of the PFC (Ramnani and Owen 2004). The subregions have a similar cellular pattern but vary in the degree of granularity and the development of cortical layer III (and layer IV), with the most prominent and well-developed layer III located in the polar area, which is not observed in nonhuman primates. Each subregion has a module-like organization with specific input–output relations. Because layers III and IV contain intracortical association fibers (Creutzfeldt 1995), it is reasonable to assume that integration of increasingly large neuronal assemblies is supported as a result of increasing numbers of association fibers. This architectural complexity increase along the medial axis toward the frontopolar cortex may be an indication of the underlying frequency-dependent knowledge represented in each of the medial subregions.

In conclusion, our findings demonstrate that subregions of the MPFC are differentially engaged in processing event sequence knowledge depending on how often the corresponding daily life activities are performed. The anterior medial Area 10 was differentially activated for LF and the posterior medial Area 10 for HF daily life activities. Because BA 10 is one of the few brain regions in humans that shows a proportional increase in size compared with other primates (Semendeferi et al. 2001, 2002), one of the driving features of this size increase may be its special capability to store a large number of experiences composed of many events, enabling humans to have access to event sequences that span the past, present, and future. Being able to represent this kind of knowledge would confer humans with a great advantage in carrying out plans, controlling a course of actions, or organizing everyday life routines.

Supplementary Data

Supplementary material can be found at: http://www.cercor.oxfordjournals.org/.

This research was supported by the Intramural Research Program of the National Institutes of Health/National Institute of Neurological Disorders and Stroke/Cognitive Neuroscience Section. The authors are grateful to Eric Wassermann for his help with performing the neurological exams.

Conflict of Interest: None declared.

References

Badre
D
Wagner
AD
Selection, integration, and conflict monitoring; assessing the nature and generality of prefrontal cognitive control mechanisms
Neuron
 , 
2004
, vol. 
41
 (pg. 
473
-
487
)
Boynton
GM
Engel
SA
Glover
GH
Heeger
DJ
Linear systems analysis of functional magnetic resonance imaging in human V1
J Neurosci
 , 
1996
, vol. 
16
 (pg. 
4207
-
4221
)
Burgess
PW
Scott
SK
Frith
CD
The role of the rostral frontal cortex (area 10) in prospective memory: a lateral versus medial dissociation
Neuropsychologia
 , 
2003
, vol. 
41
 (pg. 
906
-
918
)
Cabeza
R
Nyberg
L
Imaging cognition II: an empirical review of 275 PET and fMRI studies
J Cogn Neurosci
 , 
2000
, vol. 
12
 (pg. 
1
-
47
)
Calder
AJ
Lawrence
AD
Young
AW
Neuropsychology of fear and loathing
Nat Rev Neurosci
 , 
2001
, vol. 
2
 (pg. 
352
-
363
)
Cardinal
RN
Parkinson
JA
Hall
J
Everitt
BJ
Emotion and motivation: the role of the amygdala, ventral striatum, and prefrontal cortex
Neurosci Biobehav Rev
 , 
2002
, vol. 
26
 (pg. 
321
-
352
)
Carter
CS
Braver
TS
Barch
DM
Botvinick
MM
Noll
D
Cohen
JD
Anterior cingulate cortex, error detection, and the online monitoring of performance
Science
 , 
1998
, vol. 
280
 (pg. 
747
-
749
)
Chiavaras
MM
LeGoualher
G
Evans
A
Petrides
M
Three-dimensional probabilistic atlas of the human orbitofrontal sulci in standardized stereotaxic space
Neuroimage
 , 
2001
, vol. 
13
 (pg. 
479
-
496
)
Creutzfeldt
O
Cortex cerebri: performance, structural and functional organization of the cortex
1995
Oxford
Oxford University Press
Crozier
S
Sirigu
A
Lehericy
S
van de Moortele
PF
Pillon
B
Grafman
J
Agid
Y
Dubois
B
LeBihan
D
Distinct prefrontal activations in processing sequence at the sentence and script level: an fMRI study
Neuropsychologia
 , 
1999
, vol. 
37
 (pg. 
1469
-
1476
)
Demb
JB
Desmond
JE
Wagner
AD
Vaidya
CJ
Glover
GH
Gabrieli
JD
Semantic encoding and retrieval in the left inferior prefrontal cortex: a functional MRI study of task difficulty and process specificity
J Neurosci
 , 
1995
, vol. 
15
 (pg. 
5870
-
5878
)
Elston
GN
Pyramidal cells of the frontal lobe: all the more spinous to think with
J Neurosci
 , 
2000
, vol. 
20
 pg. 
RC95
 
Elston
GN
Rosa
MG
Pyramidal cells, patches, and cortical columns: a comparative study of infragranular neurons in TEO, TE, and the superior temporal polysensory area of the macaque monkey
J Neurosci
 , 
2000
, vol. 
20
 pg. 
RC117
 
Eslinger
PJ
Damasio
AR
Severe disturbance of higher cognition after bilateral frontal lobe ablation: patient EVR
Neurology
 , 
1985
, vol. 
35
 (pg. 
1731
-
1741
)
Fink
GR
Markowitsch
HJ
Reinkemeier
M
Bruckbauer
T
Kessler
J
Heiss
WD
Cerebral representation of one's own past: neural networks involved in autobiographical memory
J Neurosci
 , 
1996
, vol. 
16
 (pg. 
4275
-
4282
)
Forman
SD
Cohen
JD
Fitzgerald
M
Eddy
WF
Mintun
MA
Noll
DC
Improved assessment of significant activation in functional magnetic resonance imaging (fMRI): use of a cluster-size threshold
Magn Reson Med
 , 
1995
, vol. 
33
 (pg. 
636
-
647
)
Fuster
JM
Bodner
M
Kroger
JK
Cross-modal and cross-temporal association in neurons of frontal cortex
Nature
 , 
2000
, vol. 
405
 (pg. 
347
-
351
)
Gilbert
SJ
Spengler
S
Simons
JS
Frith
CD
Burgess
PW
Differential functions of lateral and medial rostral prefrontal cortex (area 10) revealed by brain-behavior associations
Cereb Cortex
 , 
2006
, vol. 
16
 (pg. 
1783
-
9
)
Gilbert
SJ
Spengler
S
Simons
JS
Steele
JD
Lawrie
SM
Frith
CD
Burgess
PW
Functional specialization within rostral prefrontal cortex (area 10): a meta-analysis
J Cogn Neurosci
 , 
2006
, vol. 
18
 (pg. 
932
-
948
)
Grafman
J
Grafman
J
The human prefrontal cortex has evolved to represent components of structured event complexes
Handbook of neuropsychology
 , 
2002
Amsterdam
Elsevier
(pg. 
157
-
174
)
Janowsky
JS
Shimamura
AP
Squire
LR
Source memory impairment in patients with frontal lobe lesions
Neuropsychologia
 , 
1989
, vol. 
27
 (pg. 
1043
-
1056
)
Kalisch
R
Wiech
K
Critchley
HD
Dolan
RJ
Levels of appraisal: a medial prefrontal role in high-level appraisal of emotional material
Neuroimage
 , 
2006
, vol. 
30
 (pg. 
1458
-
66
)
Knutson
KM
Wood
JN
Grafman
J
Brain activation in processing temporal sequence: an fMRI study
Neuroimage
 , 
2004
, vol. 
23
 (pg. 
1299
-
1307
)
Koechlin
E
Corrado
G
Pietrini
P
Grafman
J
Dissociating the role of the medial and lateral anterior prefrontal cortex in human planning
Proc Natl Acad Sci USA
 , 
2000
, vol. 
97
 (pg. 
7651
-
7656
)
Lacroix
L
Spinelli
S
Heidbreder
CA
Feldon
J
Differential role of the medial and lateral prefrontal cortices in fear and anxiety
Behav Neurosci
 , 
2000
, vol. 
114
 (pg. 
1119
-
1130
)
Levine
B
Turner
GR
Tisserand
D
Hevenor
SJ
Graham
SJ
McIntosh
AR
The functional neuroanatomy of episodic and semantic autobiographical remembering: a prospective functional MRI study
J Cogn Neurosci
 , 
2004
, vol. 
16
 (pg. 
1633
-
1646
)
Levy
R
Goldman-Rakic
PS
Segregation of working memory functions within the dorsolateral prefrontal cortex
Exp Brain Res
 , 
2000
, vol. 
133
 (pg. 
23
-
32
)
Nichelli
P
Clark
K
Hollnagel
C
Grafman
J
Duration processing after frontal lobe lesions
Ann N Y Acad Sci
 , 
1995
, vol. 
769
 (pg. 
183
-
190
)
Ongur
D
Ferry
AT
Price
JL
Architectonic subdivision of the human orbital and medial prefrontal cortex
J Comp Neurol
 , 
2003
, vol. 
460
 (pg. 
425
-
449
)
O'Reilly
RC
Rudy
JW
Computational principles of learning in the neocortex and hippocampus
Hippocampus
 , 
2000
, vol. 
10
 (pg. 
389
-
397
)
Pandya
DN
Yeterian
EH
Damasio
AR
Damasio
H
Christen
Y
Neurobiology of decision making
1996
Berlin
Springer
(pg. 
13
-
46
)
Partiot
A
Grafman
J
Sadato
N
Flitman
S
Wild
K
Brain activation during script event processing
Neuroreport
 , 
1996
, vol. 
7
 (pg. 
761
-
766
)
Piefke
M
Weiss
PH
Zilles
K
Markowitsch
HJ
Fink
GR
Differential remoteness and emotional tone modulate the neural correlates of autobiographical memory
Brain
 , 
2003
, vol. 
126
 (pg. 
650
-
668
)
Poldrack
RA
Wagner
AD
Prull
MW
Desmond
JE
Glover
GH
Gabrieli
JD
Functional specialization for semantic and phonological processing in the left inferior prefrontal cortex
Neuroimage
 , 
1999
, vol. 
10
 (pg. 
15
-
35
)
Ramnani
N
Owen
AM
Anterior prefrontal cortex: insights into function from anatomy and neuroimaging
Nat Rev Neurosci
 , 
2004
, vol. 
5
 (pg. 
184
-
194
)
Rilling
JK
Insel
TR
The primate neocortex in comparative perspective using magnetic resonance imaging
J Hum Evol
 , 
1999
, vol. 
37
 (pg. 
191
-
223
)
Rosen
VM
Caplan
L
Sheesley
L
Rodriguez
R
Grafman
J
An examination of daily activities and their scripts across the adult lifespan
Behav Res Methods Instrum Comput
 , 
2003
, vol. 
35
 (pg. 
32
-
48
)
Rowe
JB
Toni
I
Josephs
O
Frackowiak
RS
Passingham
RE
The prefrontal cortex: response selection or maintenance within working memory?
Science
 , 
2000
, vol. 
288
 (pg. 
1656
-
1660
)
Ruby
P
Sirigu
A
Decety
J
Distinct areas in parietal cortex involved in long-term and short-term action planning: a PET investigation
Cortex
 , 
2002
, vol. 
38
 (pg. 
321
-
339
)
Semendeferi
K
Armstrong
E
Schleicher
A
Zilles
K
van Hoesen
GW
Prefrontal cortex in humans and apes: a comparative study of area 10
Am J Phys Anthropol
 , 
2001
, vol. 
114
 (pg. 
224
-
241
)
Semendeferi
K
Lu
A
Schenker
N
Damasio
H
Humans and great apes share a large frontal cortex
Nat Neurosci
 , 
2002
, vol. 
5
 (pg. 
272
-
276
)
Shallice
T
Specific impairments of planning
Philos Trans R Soc Lond B Biol Sci
 , 
1982
, vol. 
298
 (pg. 
199
-
209
)
Shallice
T
Burgess
PW
Deficits in strategy application following frontal lobe damage in man
Brain
 , 
1991
, vol. 
114
 
(Pt 2)
(pg. 
727
-
741
)
Sirigu
A
Zalla
T
Pillon
B
Grafman
J
Agid
Y
Dubois
B
Selective impairments in managerial knowledge following pre-frontal cortex damage
Cortex
 , 
1995
, vol. 
31
 (pg. 
301
-
316
)
Smith
EE
Jonides
J
Storage and executive processes in the frontal lobes
Science
 , 
1999
, vol. 
283
 (pg. 
1657
-
1661
)
Stuss
DT
Benson
DF
Neuropsychological studies of the frontal lobes
Psychol Bull
 , 
1984
, vol. 
95
 (pg. 
3
-
28
)
Stuss
DT
Benson
DF
The frontal lobes
1986
New York
Raven Press
Talairach
J
Tournoux
P
Co-planar stereotaxic atlas of the human brain
1988
New York
Thieme Medical Publishers
Wood
JN
Grafman
J
Human prefrontal cortex: processing and representational perspectives
Nat Rev Neurosci
 , 
2003
, vol. 
4
 (pg. 
139
-
147
)
Wood
JN
Knutson
KM
Grafman
J
Psychological structure and neural correlates of event knowledge
Cereb Cortex
 , 
2005
, vol. 
15
 (pg. 
1155
-
1161
)
Zalla
T
Koechlin
E
Pietrini
P
Basso
G
Aquino
P
Sirigu
A
Grafman
J
Differential amygdala responses to winning and losing: a functional magnetic resonance imaging study in humans
Eur J Neurosci
 , 
2000
, vol. 
12
 (pg. 
1764
-
1770
)