- Email: [email protected]
Human cortical responses during one-bit delayed-response tasks: An fMRI study
Brain Research Bulletin 65 (2005) 383–390
Human cortical responses during one-bit delayed-response tasks: An fMRI study Claudio Babiloni a,b,c,∗ , Antonio Ferretti d,e , Cosimo Del Gratta d,e , Filippo Carducci a,c,d , Fabrizio Vecchio a,c , Gian Luca Romani d,e , Paolo Maria Rossini b,c,f a
Dipartimento di Fisiologia Umana e Farmacologia, Sezione di EEG ad Alta Risoluzione, Universit`a degli Studi di Roma “La Sapienza”, P.le Aldo Moro 5, 00185 Rome, Italy b IRCCS S Giovanni di Dio, Via Pilastroni, Brescia, Italy c AFaR.- Dip. di Neuroscienze, S. Giovanni Calibita, Fatebenefratelli Isola Tiberina, Rome, Italy d Dipartimento di Scienze Cliniche e Bioimmagini, Universit` a G. D’Annunzio, Chieti, Italy e ITAB Fondazione “Universit` a G. DAnnunzio”, Chieti, Italy f Clinica Neurologica, Campus Biomedico, Universit` a di Roma, Italy Received 25 August 2004; received in revised form 26 October 2004; accepted 31 January 2005 Available online 19 March 2005
Abstract Neuroimaging study of cognition across aging requires simple tasks ensuring: (i) high rate of correct performances in neurophysiological settings; and (ii) significant modulation of cortical activity. As a preliminary step, the present functional magnetic resonance imaging (fMRI) study tested the hypothesis that very simple delayed-response tasks fit these requirements in normal young adults. The short-term memory (STM) variant included a sequence of cue stimulus (two vertical bars), delay period (blank screen for only 5 s), go stimulus, and motor response compatible with the taller vertical bar. Noteworthy, the retention (only one bit) could be based on visuo-spatial, phonological, and somatomotor coding. In the control variant (no STM, NSTM), the cue stimulus was present during the delay period. Results showed high rate of correct performances in both tasks (about 95%). Compared to the NSTM task (delay period), the STM task enhanced cortical responses in bilateral dorsolateral prefrontal (Brodmann area 8–9 (BA 8–9)), lateral premotor (BA 6L), medial premotor (BA 6M), inferior parietal (BA 40), and superior parietal (BA 7) areas. In the STM task, cortical responses were stronger in right than left BA 8–9 and BA 6L. These results indicate that, in normal young adults, a simple STM variant of delayed-response tasks (one bit to be retained) is correctly performed and enhances bilateral fronto-parietal responses. Therefore, it may be used for future cognitive neuroimaging studies on aging. © 2005 Elsevier Inc. All rights reserved. Keywords: Short-term memory; Delayed responses; Functional magnetic resonance imaging (fMRI); Normal subjects; Frontal lobe
1. Introduction Short-term memory (STM) can be easily investigated by so called delayed-response tasks. In delayed-response tasks, subject has to retain a cue stimulus and a related motor response across a delay period up to an imperative go stimulus. In the STM variant of the task, the cue stimulus rapidly disappears. In the control variant (no STM,
∗
Corresponding author. Tel.: +39 06 49910989; fax: +39 06 49910917. E-mail address: [email protected] (C. Babiloni).
0361-9230/$ – see front matter © 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.brainresbull.2005.01.013
NSTM), the cue stimulus is delivered during the delay period. With respect to “delayed match-to-sample” and “nback” STM paradigms, the delayed-response tasks include no probe stimulus to be matched with the memorized cue stimulus. The delayed-response tasks have been successfully used for the study of cognitive processes in schizophrenia and aging [25,27,37,41]. Neurobiological correlates of the delayed-response tasks have been extensively investigated in the nonhuman primates [16–20,33]. Fronto-parietal systems are involved in the maintenance of the “representational memory” during delayedresponse tasks [5,6,16,17,19–21,24,29,33].
384
C. Babiloni et al. / Brain Research Bulletin 65 (2005) 383–390
In humans, studies using functional magnetic resonance imaging (fMRI) have confirmed the crucial role of the frontoparietal cortical system for a successful performance in the STM variant of delayed-response tasks. This system included dorsolateral prefrontal (Brodmann area 8–9 (BA 8–9)), lateral and medial premotor (BA 6L and BA 6M), and posterior parietal (BA 7, 40) areas. The right dorsolateral prefrontal area was especially involved when visuo-spatial contents had to be memorized [8,10,11,34–36]. Furthermore, recent fMRI studies have experimentally dissociated two parallel processes related to STM tasks, namely “maintenance” of motor responses during the delay period and “selection” of an item from memory to guide the motor response [38,39]. “Selection”, but not “maintenance”, was associated with the activation of BA 46 of the dorsolateral prefrontal cortex. In contrast, “maintenance” was associated with the activation of prefrontal BA 8 and the intra-parietal cortex. Functional neuroimaging studies have revealed abnormal cortical activity associated with losses in STM with age (for a review see [26]). However, this research field is characterized by great methodological difficulties, due to the relatively poor cognitive performances of aged people to standard STM tasks. Indeed, neuroimaging study of cognition across aging requires simple tasks ensuring: (i) high rate of correct performances in stressful neurophysiological settings; and (ii) significant modulation of the cortical activity. Neuroimaging data accompanying correct performances are at the basis of comparisons between normal young versus elderly subjects and between normal elderly subjects and elderly subjects with cognitive impairments. As a preliminary step, the present functional magnetic resonance imaging study tested the hypothesis that very simple delayed-response tasks fit these requirements in normal young adults. The STM variant included a sequence of cue stimulus (two vertical bars of unequal height), delay period (blank screen for only 5 s), go stimulus, and motor response compatible with the taller vertical bar. Noteworthy, the retention (only one bit) could be based on visuospatial, phonological, and somatomotor coding. In the control variant (no STM), the cue stimulus was present during the delay period. In previous studies, the same tasks have been able to modulate electroencephalographic rhythms in frontal and parietal areas of normal young [1,2] and elderly [3] adults. However, it should be stressed that electroencephalographic techniques prevented a fine topographical analysis of these cortical responses.
2. Materials and methods 2.1. Subjects Subjects were 15 healthy, right-handed (Edinburgh Inventory) adult volunteers (mean age of 27 years, range from 19 to 40 years), who gave their informed written consent according to the declaration of Helsinki. General procedures were approved by the local institutional ethics committee.
2.2. Experimental tasks The experiment included two recording blocks (i.e. STM, NSTM). Each recording block comprised 23 consecutive trials during which fMRI machine runs in continuous mode (inter-block interval of about 5 min). Subjects were told in advance if the block was NSTM or STM (pseudo-randomized across subjects). Examples of trial for both NSTM and STM tasks are illustrated in Fig. 1. For the STM task, sequence of events was as follows: (i) a cross in a red circle at the center of the monitor (about 0.7◦ for side) as a visual warning stimulus lasting 1 s; (ii) a couple of vertical bars (about 2◦ large and 2.5–7◦ high) as a visual cue stimulus lasting 2 s; (iii) blank screen as a delay period lasting 5 s; (iv) a cross in a green circle at the center of the monitor (about 0.7◦ of diameter) as a go stimulus lasting 1 s; and (v) right finger movement to press the proper button. Subjects had to click either left mouse button if the taller bar (cue stimulus) was at the left monitor side or right mouse button if the taller bar was at the right monitor side. A cross at the center of the monitor (16 s inter-trial interval) allowed the hemodynamic parameters to return to baseline level. Compared to the STM task, the NSTM task had the cue stimulus lasting across the delay period up to the go stimulus triggering the motor response. Before the recording session, a training of about 10 min made the subjects familiar with the experimental apparatus and the tasks. Of note, it should be remarked that the two delayedresponse tasks used in the present study were chosen with a view to future applications in aging research. This motivated the choice of minimal STM load (one bit for very few seconds) and the possibility of using any memorization strategy including visuo-spatial imagery, somato-motor preparation, and mental “phonological” coding and rehearsal. 2.3. fMRI recordings For the fMRI experiments, the subjects were positioned in the scanner with the head immobilized by foam support cushions (1.5 T MRI system; Siemens Magnetom Vision). They kept their arms resting, with the right index finger resting between two buttons spaced 6 cm apart. The visual stimuli were generated by a personal computer and projected by a LCD video projector on a screen behind the scanner. A mirror attached to the head coil, at 45◦ with the line of sight, allowed the subjects to see the image on the screen. fMRIs were recorded in an event-related mode. Blood oxygen level dependent (BOLD) contrast images were obtained by means of Echo Planar Imaging (EPI) sequences with the following features: TR 1189 ms, TE 60 ms, matrix size 64 × 64, FOV 220 mm, flip angle 90◦ , slice thickness 7 mm and no gap. Functional volumes consisted of nine bicommissural transaxial slices. In addition, high-resolution anatomical images of the brain were obtained with a 3D magnetization prepared rapid gradient echo (MP-RAGE) sequence with the following sequence parameters: TR 9.7 ms;
C. Babiloni et al. / Brain Research Bulletin 65 (2005) 383–390
385
Fig. 1. Sequence of the events for the two delayed-response tasks. In the short-term memory (STM) condition, these events were as follows: (i) a cross in a red circle at the center of the monitor as a visual warning stimulus; (ii) a couple of vertical bars as a visual cue stimulus; (iii) blank screen as a delay period; (iv) a cross in a green circle at the center of the monitor as a go stimulus; and (v) right finger movement to press the proper button. Subjects had to click either left mouse button if the taller bar (cue stimulus) was at the left monitor side or right mouse button if the taller bar was at the right monitor side. Compared to the STM task, the no STM (NSTM) condition had the cue stimulus lasting across the delay period up to the go stimulus triggering the motor response.
TE 4 ms, flip angle 12◦ ; FoV 256 mm; matrix size 256 × 256, 256 axial slices with a thickness of 1 mm (no gap). 2.4. Data analysis Analysis of the fMRIs was performed using Brain Voyager 4.4 software package (Brain Innovation, The Netherlands). After slice time alignment, individual fMRIs were registered with the respective structural data set from the same session. The co-registration transformation was determined using the slice position parameters of the fMRIs and the position parameters of the structural volume. Structural and functional MRIs were normalized to the Talairach space [44] using a piecewise affine and continuous transformation. Pre-processing of fMRIs included motion correction and linear detrending. Due to the relatively large voxel size (about 83 mm3 ), no spatial smoothing was performed even for the group analysis. The extreme simplicity of both NSTM and STM tasks in terms of general cognitive demands motivated a data analysis design aimed at stressing the expected modest differences of the cortical activity related to the two tasks. To minimize the statistical correction due to multiple comparisons, the final
statistical analysis was performed on a small number of regions of interest (ROIs). These ROIs were selected based on previous studies [8,10,11,34–36]. We considered the following bilateral ROIs: dorsolateral prefrontal (BA 8–9 and BA 46), inferior parietal (BA 40), superior parietal (BA 7), lateral premotor (BA 6L), medial premotor (BA 6M, i.e. supplementary motor area), and primary motor (BA 4) areas. The ROI analysis allowed statistical inter-hemispherical comparisons within task. The mentioned ROIs were defined manually using each subject’s structural MRIs coded into Talairach space. This was done for all subjects by two experimenters before the analysis of the fMRI data. BA 8 primarily was located in the superior frontal gyrus extending from the cingulate sulcus on the medial surface over the margin of the hemisphere to the middle frontal gyrus. BA 9 was situated in a portion of the superior frontal gyrus and the middle frontal gyrus. Its approximate boundary on the medial aspect of the hemisphere was the cingulate sulcus and, on the lateral aspect, the inferior frontal sulcus. BA 46 was approximately circumscribed to the middle third of the middle frontal gyrus and the most rostral portion of the inferior frontal gyrus. BA 40 lays primarily in the supramarginal gyrus surrounding the
386
C. Babiloni et al. / Brain Research Bulletin 65 (2005) 383–390
posterior ascending limb of lateral sulcus. BA 7 was placed in the superior parietal lobule and in precuneus. It was approximately bounded by the superior postcentral sulcus rostrally, the intraparietal sulcus laterally, the parieto-occipital sulcus caudally, and the subparietal sulcus on the medial bank of the hemisphere. BA 6M and L were situated primarily in the caudal portions of the superior frontal gyrus, the middle frontal gyrus, and the rostral portions of the precentral gyrus not occupied by the gigantopyramidal area 4. They extended from the cingulate sulcus on the medial aspect of the hemisphere to the lateral sulcus on the lateral aspect. BA 4 was located in the caudal part of the precentral gyrus. Noteworthy, the two experimenters systematically verified the coordinate position of the ROI borders into the Talairach atlas. In particular, they probed borders of each ROI with the cursor and verified that corresponding Talairach coordinates belonged to the proper ROI. For each task (STM, NSTM), the voxels within the above ROIs showing statistically significant differences of the fMRI responses during the delay period compared to the rest period (baseline) were identified by a means of a group analysis merging data from all subjects. First, the time courses of all subjects were concatenated and z-transformed to form a single time course for each task. Then, these data were analyzed using the general linear model (GLM) [15]. Delayed predic-
tors were defined for the delay period of the two tasks. To account for the hemodynamical inertia (about 4 s), the statistical analysis of the delay period was carried out in a time interval from 4 s after the cue stimulus offset to 3 s after the go stimulus onset. A 3D group statistical map was generated for the delay condition by associating to each voxel the F value corresponding to the related predictor. This map was thresholded at voxel level at p < 0.000003 (corresponding to p < 0.05 corrected for multiple comparisons by the Bonferroni method; the correction was performed considering the number of brain voxels in Talairach space, about 15,000) and superimposed on the anatomical scan of one of the subjects. After the activation areas were defined from group data, the comparison of activity in these areas was performed on the basis of individual subject activation. Therefore, individual subject analysis was also undertaken. The general linear model was computed for each subject using the same predictor for the delay period as in the group analysis. To avoid the severe loss of statistical power due to the Bonferroni correction, the final significance was evaluated in the individual subject case using a cluster-size algorithm [13]. Individual statistical maps of the delay condition were thresholded at p < 0.001 (per-voxel) and a minimum cluster-size of three voxels was required, yelding a false positive rate of 0.023 on
Fig. 2. Group analysis statistical maps of BOLD cortical activity relative to delay compared to baseline period for both NSTM and STM tasks. The statistical maps were superimposed on the anatomical magnetic resonance imaging (MRI) scan of one representative subject. Statistical threshold of these statistical maps was posed at voxel level of p < 0.001, which corresponded to p < 0.05 corrected for multiple comparisons with clusters of three voxels as evaluated by Monte Carlo simulation. Color scale: maximum (p = 0.05 corrected) and minimum (p < 0.05 corrected) are coded in yellow and red, respectively. Right hemisphere is on the right side of the figure.
C. Babiloni et al. / Brain Research Bulletin 65 (2005) 383–390
the entire 3D image, as evaluated by means of a Monte Carlo simulation (AlphaSim routine of AFNI package, [9]). Finally, the relative BOLD signal increase with respect to baseline (i.e. (delay signal − baseline signal)/(baseline signal)) was averaged across trials and across voxels in a ROI, yielding for each ROI a measure of activation. These mean relative BOLD signal increases were used as an input for an ANOVA analysis for repeated measures. The ANOVA analysis evaluated the fronto-parietal cortical responses during the delay period of the present one-bit STM task. The factors of the analysis were Condition (STM and NSTM), Hemisphere (left, right), and the aforementioned ROIs. Mauchley’s test evaluated the sphericity assumption, which took into account the level of auto-correlation among the variables to be compared. The correction of the degrees of freedom was made by Greenhouse–Geisser procedure, which took into account the complexity of the statistical design formed by factors and levels to adjust the univariate results of repeated measures. Duncan test was used for post-hoc comparisons (p < 0.05).
387
The ANOVA analysis of the BOLD percent values relative to delay compared to baseline period showed a statistical interaction (F(5,70) = 2.48; Ms error = 0.009; p < 0.04) among the three factors considered, namely Condition (NSTM, STM), Hemisphere (left, right), and ROI (BA 8–9, BA 40, BA 7, BA 6L, BA 6M, and BA 4). Fig. 3 illustrates the means (± standard error, S.E.) of the mentioned BOLD percent values for BA 8–9, BA 40, BA 7, BA 6L, BA 6M, and BA 4 ROIs. Duncan post-hoc testing indicated that, compared to the control NSTM task, the STM task induced stronger BOLD activations in bilateral dorsolateral prefrontal (BA 8–9, p < 0.002), lateral premotor (BA 6L, p < 0.002), medial premotor (BA 6M, p < 0.01), inferior parietal (BA 40, p < 0.006), and superior parietal (BA 7, p < 0.003) areas. Remarkably, the STM task induced stronger BOLD activations in right than left dorsolateral prefrontal (BA 8–9, p < 0.03) and lateral premotor (BA 6L, p < 0.02) areas. Both NSTM and STM tasks provoked stronger BOLD activations in left (contralateral) than right primary motor area (p < 0.00002).
4. Discussion 3. Results As expected, high rate of correct performances were observed in both NSTM (about 95%) and STM (about 93%) tasks (no statistical difference was found at p < 0.05). The fine statistical mapping of the BOLD cortical responses accompanying the delay period of these tasks compared to the baseline period is illustrated across subjects over an individual MRI (Fig. 2). Note that statistical mapping of the BOLD cortical activity appeared to be quite similar in the two tasks (p < 0.05 corrected). Statistically (p < 0.05 corrected) activated voxels in dorsolateral prefrontal BA 46 were observed only in five subjects during the STM task and in six subjects during the NSTM task. Therefore, this area was not further considered. The epicenters of the cortical activity within the ROIs are reported in Table 1 (Talairach coordinates). Of note, BOLD percent values (delay compared to baseline period) relative to the statistically activated voxels (p < 0.05 corrected) were further considered. These values for each ROI were averaged for the final ANOVA statistics at the ROIs level.
The present delayed-response tasks demonstrated frontoparietal cortical responses in young adults associated with a minimal STM load (one bit for a few seconds; the low load tasks may be appropriate for future research with elderly and/or impaired subjects). There was a high rate of correct performances for both the NSTM and STM tasks (about 95%). In the two tasks (delay period), there were significant responses in bilateral dorsolateral prefrontal (BA 8–9), lateral premotor (BA 6L), medial premotor (BA 6M), inferior parietal (BA 40), and superior parietal (BA 7) areas. Compared to the control task (delay period), the STM task induced bilaterally stronger responses in these areas. These results confirm and extend in fine spatial details previous electroencephalograhic experiments using the same tasks [1,2]. Furthermore, the present results suggest that fronto-parietal systems are involved with a similar functional topography during the delay period of the two tasks. In this framework, the retention processes of the STM task would be related to a further increase of the activation of these systems [12,14,43]. Remarkably, the lesion of single parts of these networks (i.e. prefrontal)
Table 1 Talairach coordinates of the epicenters of the regions of interest (ROIs) showing a BOLD activation (%) during the delay period of the two present delayedresponse tasks (short-term memory, STM; no STM, NSTM) Regions of interest (ROIs)
Right hemisphere
Left hemisphere
x
y
z
x
y
z
Frontal lobe: middle frontal gyrus dorsolateral prefrontal BA 8–9 Frontal lobe: precentral gyrus lateral premotor BA 6L Frontal lobe: medial frontal gyrus medial premotor BA 6M Frontal lobe: precentral gyrus primary motor BA 4 Parietal lobe: inferior parietal lobule inferior parietal BA 40 Parietal lobe: superior parietal lobule superior parietal BA 7
32 40 2 35 37 26
29 −1 −7 −26 −41 −62
38 36 54 54 39 43
−33 −43 −3 −36 −37 −26
24 −8 −8 −28 −43 −62
39 38 54 54 41 44
388
C. Babiloni et al. / Brain Research Bulletin 65 (2005) 383–390
Fig. 3. Mean activation (± standard error) across subjects of BOLD percent values relative to delay compared to baseline period for both NSTM and STM tasks. These values refer to the ROIs including bilateral dorsolateral prefrontal (Brodmann area 8–9, BA 8–9), lateral premotor (BA 6L), medial premotor (BA 6M), inferior parietal (BA 40), superior parietal (BA 7) areas, and primary motor (BA 4) areas. The ANOVA for repeated measures of these data showed a statistically significant interaction among the factors Condition (NSTM, STM), Hemisphere or Side (left, right), ROIs (the mentioned areas). The results of Duncan post-hoc testing are indicated by asterisks.
might not impair the behavioural performance due to a possible multiple coding of the information to be memorized, namely visuo-spatial, phonological, and sensorimotor [35]. On the whole, the present experimental design is of interest for future clinical applications since it is relatively quick and easy for aged people. This may ensure good rate of correct performances necessary for neurophysiologic comparisons. However, as a methodological limitation, that design does not allow the dissociation of three different strategies for cue stimulus retention. Namely, the strategies that involve visual representation of target stimulus during the delay period, or motor representation of previously selected response during the delay period, or linguistic representation of either stimulus or anticipated motor response during the delay period. In that sense, the use of the present paradigm should be limited
to the cases in which a global neurophysiological index of short-term retention capabilities is of interest. In the present experiments, the prominent activation of the right dorsolateral prefrontal and lateral premotor areas would reflect the storage and rehearsal of the visuo-spatial information (i.e. position and highness of the vertical bars), according to the explicit task demands. This explanation agrees with previous neuroimaging studies showing that coding and rehearsal of visuo-spatial representations mainly involve the right hemisphere [43]. It also agrees with previous EEG studies showing an activation in prefrontal areas, especially of the “non-dominant” right hemisphere, during visuo-spatial STM tasks [4,22,23,30,31,42]. However, it should be stressed that this is just a tentative explanation of the present results. As aforementioned, we cannot
C. Babiloni et al. / Brain Research Bulletin 65 (2005) 383–390
exclude that the mentioned cortical activity is related also to sensorimotor (less probably linguistic) representations, due to the lack of specific experimental manipulations. In the framework of the bilateral responses to the present STM demands, the activation of the left hemisphere might be due to “executive processes” other than to the encoding, store, and sub-vocal rehearsal of the cue stimulus representation. The “executive” processes are defined as meta-processes (switching attention, checking features, inhibiting responses, etc.) that regulate the manipulation of both spatial and verbal contents in STM [7,32]. In this regard, the present STM task was characterized by simple cue stimuli, short delay period, no distractor, and competition between only two possible motor responses. Hence, it seems reasonable that the cortical responses over the left hemisphere mainly indicated the strength of the encoding, storage, and rehearsal with respect to the “executive” processes. Future studies should address the specific relationships of “executive functions” and activation of left hemisphere during delayed-response tasks, by means of experiments manipulating visual-spatial, sensorimotor, and linguistic aspects of these tasks. The present delayed-response tasks were designed to be extremely simple. The subjects knew the side of the target movement during the retention period, so that phonological and sensorimotor coding could contribute to the task goal. This means that, during the delay period of the STM task, the right fronto-parietal cortex might be recruited for the motor planning [28] other than for the retention of visuo-spatial representations. In this regard, previous fMRI studies [38,39] have demonstrated that the “maintenance” of visuo-spatial representations in delayed tasks is associated with the activation of dorsolateral prefrontal (BA 8) and posterior parietal areas. In parallel, the motor preparation would involve the dorsolateral prefrontal BA 46, marginally activated in the present study. Noteworthy, the activation of BA 46 has been demonstrated to be sensitive to aging effects in working memory tasks [40]. Here, its minor activation may be due to the young age of the experimental subjects as well as the minimal load and repetitive nature of the present STM task. The present results may be considered as a basis for future functional neuroimaging studies aimed at extending actual neurophysiologic knowledge on poor cognitive performances of aged people to standard STM tasks (for a review see [26]). Indeed, the present methodological approach ensured many correct performances in fMRI setting and a significant modulation of cortical activity in the STM task variant. Of course, the experimental design should be refined if scientific issues of interest are focused on specific representations and functions involved in the stimulus retention (i.e. visuo-spatial, sensorimotor, linguistic).
5. Conclusions The present functional magnetic resonance imaging study tested the hypothesis that a very low STM load (one bit to
389
be retained) is able to modulate cortical activity in young adults involved in delayed-response tasks. Results showed high rate of correct performances in both NSTM and STM tasks (about 95%). Compared to NSTM task (delay period), STM task enhanced the cortical responses in bilateral dorsolateral prefrontal (BA 8–9), lateral premotor (BA 6L), medial premotor (BA 6M), inferior parietal (BA 40), and superior parietal (BA 7) areas. In STM task, cortical responses were stronger in right than left BA 8–9 and BA 6L. These results indicated that, in normal young adults, a simple STM variant of delayed-response tasks is correctly performed and enhances bilateral fronto-parietal responses. Therefore, it may be used for future cognitive neuroimaging studies on aging.
Acknowledgments The authors thank Prof. Fabrizio Eusebi, Chairman of the Biophysics Group of Interest of Rome I University, for his continuous support. The research was granted by Telethon Onlus Foundation (“Progetto E.C0985”). Dr. Fabrizio Vecchio participated to this study in the framework of his Ph.D. program at the Doctoral School in Neurophysiology, University of Rome “La Sapienza”.
References [1] C. Babiloni, F. Babiloni, F. Carducci, S.F. Cappa, F. Cincotti, C. Del Percio, C. Miniussi, D.V. Moretti, S. Rossi, K. Sosta, P.M. Rossini, Human cortical responses during one-bit short-term memory. A highresolution EEG study on delayed choice reaction time tasks, Clin. Neurophysiol. 115 (2004) 161–170. [2] C. Babiloni, F. Babiloni, F. Carducci, F. Cincotti, F. Vecchio, B. Cola, S. Rossi, C. Miniussi, P.M. Rossini, Functional frontoparietal connectivity during short-term memory as revealed by high-resolution EEG coherence analysis, Behav. Neurosci. 118 (2004) 687–697. [3] C. Babiloni, F. Babiloni, F. Carducci, S.F. Cappa, F. Cincotti, C. Del Percio, C. Miniussi, D.V. Moretti, S. Rossi, K. Sosta, P.M. Rossini, Human cortical rhythms during visual delayed choice reaction time tasks. A high-resolution EEG study on normal aging, Behav. Brain Res. 153 (2004) 261–271. [4] A. Begleiter, T. Allison, G. McCarthy, E. Roessler, A.C. Nobre, D.D. Spencer, Electrophysiological studies of color processing in human visual cortex, Electroencephalogr. Clin. Neurophysiol. 88 (1993) 343–355. [5] N. Butters, D. Pandya, Retention of delayed-alternation: effects of selective lesions of sulcus principalis, Science 165 (1969) 1271–1273. [6] M.V. Chafee, P.S. Goldman-Rakic, Matching patterns of activity in primate prefrontal area 8a and parietal area 7ip neurons during a spatial working memory task, J. Neurophysiol. 79 (1998) 2919–2940. [7] C.B. Cohen, Avoiding “Cloudcuckooland” in ethics committee case review: matching models to issues and concerns, Law Med. Health Care 20 (1992) 294–299. [8] J.D. Cohen, W.M. Perlstein, T.S. Braver, L.E. Nystrom, D.C. Noll, J. Jonides, E.E. Smith, Temporal dynamics of brain activation during a working memory task, Nature 386 (1997) 604–608. [9] R.W. Cox, AFNI: Software for analysis and visualization of functional magnetic resonance neuroimages, Comput. Biomed. Res. 29 (1996) 162–173.
390
C. Babiloni et al. / Brain Research Bulletin 65 (2005) 383–390
[10] M. D’Esposito, D. Ballard, G.K. Aguirre, E. Zarahn, Human prefrontal cortex is not specific for working memory: a functional MRI study, Neuroimage 8 (1998) 274–282. [11] M. D’Esposito, B.R. Postle, B. Rypma, Prefrontal cortical contributions to working memory: evidence from event-related fMRI studies, Exp. Brain Res. 133 (2000) 3–11. [12] J.A. Fiez, S.E. Petersen, Neuroimaging studies of word reading, Proc. Natl. Acad. Sci. U.S.A. 95 (1998) 914–921. [13] S.D. Forman, J.D. Cohen, M. Fitzgerald, W.F. Eddy, M.A. Mintun, D.C. Noll, Improved assessment of significant activation in functional magnetic resonance imaging (fMRI): use of a cluster-size threshold, Magn. Reson. Med. 33 (1995) 636–647. [14] D. Friedman, W. Ritter, J.G. Snodgrass, ERPs during study as a function of subsequent direct and indirect memory testing in young and old adults, Brain Res. Cogn. Brain Res. 4 (1996) 1–13. [15] K.J. Friston, A.P. Holmes, K.J. Worsley, J.P. Poline, C.D. Frith, R.S.J. Frackowiak, Statistical parametric maps in functional imaging: a general linear approach, Hum. Brain Mapp. 2 (1995) 173–181. [16] S. Funahashi, C.J. Bruce, P.S. Goldman-Rakic, Mnemonic coding of visual space in the monkey’s prefrontal cortex, J. Neurophysiol. 61 (1989) 331–349. [17] S. Funahashi, C.J. Bruce, P.S. Goldman-Rakic, Visuo-spatial coding in primate prefrontal neurons revealed by oculomotor paradigms, J. Neurophysiol. 63 (1990) 814–831. [18] S. Funahashi, C.J. Bruce, P.S. Goldman-Rakic, Neuronal activity related to saccadic eye movements in the monkey’s dorsolateral prefrontal cortex, J. Neurophysiol. 65 (1991) 1464–1483. [19] S. Funahashi, C.J. Bruce, P.S. Goldman-Rakic, Dorsolateral prefrontal lesions and oculomotor delayed-response performance: evidence for mnemonic ’scotomas’, J. Neurosci. 13 (1993) 1479–1497. [20] S. Funahashi, M.V. Chafee, P.S. Goldman-Rakic, Prefrontal neuronal activity in rhesus monkeys performing a delayed anti-saccade task, Nature 365n (1993) 753–756. [21] J.M. Fuster, Temporal processing, Ann. N. Y. Acad. Sci. 769 (1995) 173–181. [22] A. Gevins, B. Cutillo, Spatiotemporal dynamics of component processes in human working memory, Electroencephalogr. Clin. Neurophysiol. 87 (1993) 128–143. [23] A. Gevins, M.E. Smith, Neurophysiological measures of working memory and individual differences in cognitive ability and cognitive style, Cereb. Cortex. 10 (2000) 829–839. [24] P.S. Goldman-Rakic, Topography of cognition: parallel distributed networks in primate association cortex, Annu. Rev. Neurosci. 11 (1988) 137–156. [25] P.S. Goldman-Rakic, Prefrontal cortical dysfunctions in schizophrenia: the relevance of working memory, in: Psychopatholology and the Brain, Raven Press, New York, 1991. [26] C.L. Grady, F.I. Craik, Changes in memory processing with age, Curr. Opin. Neurobiol. 10 (2000) 224–231. [27] F. Harrington, B.K. Saxby, I.G. McKeith, K. Wesnes, G.A. Ford, Cognitive performance in hypertensive and normotensive older subjects, Hypertension 36 (2000) 1079–1082.
[28] M. Jahanshahi, I.H. Jenkins, R.G. Brown, C.D. Marsden, R.E. Passingham, D.J. Brooks, Self-initiated versus externally triggered movements. I. An investigation using measurement of regional cerebral blood flow with PET and movement-related potentials in normal and Parkinson’s disease subjects, Brain 118 (1995) 913–933. [29] R.P. Kessels, A. Postma, E.M. Wijnalda, E.H. de Haan, Frontallobe involvement in spatial memory: evidence from PET, fMRI, and lesion studies, Neuropsychol. Rev. 10 (2000) 101–113. [30] L.K. McEvoy, E. Pellouchoud, M.E. Smith, A. Gevins, Neurophysiological signals of working memory in normal aging, Brain Res. Cogn. Brain Res. 11 (2001) 363–376. [31] L.K. McEvoy, M.E. Smith, A. Gevins, Dynamic cortical networks of verbal and spatial working memory: effects of memory load and task practice, Cereb. Cortex. 8 (1998) 563–574. [32] D.E. Meyer, D.E. Kieras, A computational theory of executive cognitive processes and multiple-task performance: Part 1. Basic mechanisms, Psychol. Rev. 104 (1997) 3–65. [33] M. Mishkin, Effects of small frontal lesions on delayed alternation in monkeys, J. Neurophysiol. 20 (1957) 615–622. [34] S. Pollmann, D.Y. von Cramon, Object working memory and visuospatial processing: functional neuroanatomy analyzed by eventrelated fMRI, Exp. Brain Res. 133 (2000) 12–22. [35] B.R. Postle, M. D’Esposito, “What–Then–Where” in visual working memory: an event-related fMRI study, J. Cogn. Neurosci. 11 (1999) 585–597. [36] B.R. Postle, E. Zarahn, M. D’Esposito, Using event-related fMRI to assess delay-period activity during performance of spatial and nonspatial working memory tasks, Brain Res. Brain Res. Protoc. 5 (2000) 57–66. [37] S. Roux, I. Hubert, A. Lenegre, D. Milinkevitch, R.D. Porsolt, Effects of piracetam on indices of cognitive function in a delayed alternation task in young and aged rats, Pharmacol. Biochem. Behav. 49 (1994) 683–688. [38] J.B. Rowe, R.E. Passingham, Working memory for location and time: activity in prefrontal area 46 relates to selection rather than maintenance in memory, Neuroimage 14 (2001) 77–86. [39] J.B. Rowe, I. Toni, O. Josephs, R.S. Frackowiak, R.E. Passingham, The prefrontal cortex: response selection or maintenance within working memory, Science 288 (2000) 1656–1660. [40] B. Rypma, M. D’Esposito, Isolating the neural mechanisms of agerelated changes in human working memory, Nat. Neurosci. 3 (2000) 509–515. [41] D. Servan-Schreiber, J.D. Cohen, S. Steingard, Schizophrenic deficits in the processing of context. A test of a theoretical model, Arch. Gen. Psychiatry 53 (1996) 1105–1112. [42] R.B. Silberstein, J. Ciorciari, A. Pipingas, Steady-state visually evoked potential topography during the Wisconsin card sorting test, Electroencephalogr. Clin. Neurophysiol. 96 (1995) 24–35. [43] E.E. Smith, J. Jonides, Working memory: a view from neuroimaging, Cognit. Psychol. 33 (1997) 5–42. [44] J. Talairach, P. Tournoux, Coplanar Stereotaxic Atlas of the Human Brain, Thieme Medical Publishers, New York, 1988.
Human cortical responses during one-bit delayed-response tasks: An fMRI study Claudio Babiloni a,b,c,∗ , Antonio Ferretti d,e , Cosimo Del Gratta d,e , Filippo Carducci a,c,d , Fabrizio Vecchio a,c , Gian Luca Romani d,e , Paolo Maria Rossini b,c,f a
Dipartimento di Fisiologia Umana e Farmacologia, Sezione di EEG ad Alta Risoluzione, Universit`a degli Studi di Roma “La Sapienza”, P.le Aldo Moro 5, 00185 Rome, Italy b IRCCS S Giovanni di Dio, Via Pilastroni, Brescia, Italy c AFaR.- Dip. di Neuroscienze, S. Giovanni Calibita, Fatebenefratelli Isola Tiberina, Rome, Italy d Dipartimento di Scienze Cliniche e Bioimmagini, Universit` a G. D’Annunzio, Chieti, Italy e ITAB Fondazione “Universit` a G. DAnnunzio”, Chieti, Italy f Clinica Neurologica, Campus Biomedico, Universit` a di Roma, Italy Received 25 August 2004; received in revised form 26 October 2004; accepted 31 January 2005 Available online 19 March 2005
Abstract Neuroimaging study of cognition across aging requires simple tasks ensuring: (i) high rate of correct performances in neurophysiological settings; and (ii) significant modulation of cortical activity. As a preliminary step, the present functional magnetic resonance imaging (fMRI) study tested the hypothesis that very simple delayed-response tasks fit these requirements in normal young adults. The short-term memory (STM) variant included a sequence of cue stimulus (two vertical bars), delay period (blank screen for only 5 s), go stimulus, and motor response compatible with the taller vertical bar. Noteworthy, the retention (only one bit) could be based on visuo-spatial, phonological, and somatomotor coding. In the control variant (no STM, NSTM), the cue stimulus was present during the delay period. Results showed high rate of correct performances in both tasks (about 95%). Compared to the NSTM task (delay period), the STM task enhanced cortical responses in bilateral dorsolateral prefrontal (Brodmann area 8–9 (BA 8–9)), lateral premotor (BA 6L), medial premotor (BA 6M), inferior parietal (BA 40), and superior parietal (BA 7) areas. In the STM task, cortical responses were stronger in right than left BA 8–9 and BA 6L. These results indicate that, in normal young adults, a simple STM variant of delayed-response tasks (one bit to be retained) is correctly performed and enhances bilateral fronto-parietal responses. Therefore, it may be used for future cognitive neuroimaging studies on aging. © 2005 Elsevier Inc. All rights reserved. Keywords: Short-term memory; Delayed responses; Functional magnetic resonance imaging (fMRI); Normal subjects; Frontal lobe
1. Introduction Short-term memory (STM) can be easily investigated by so called delayed-response tasks. In delayed-response tasks, subject has to retain a cue stimulus and a related motor response across a delay period up to an imperative go stimulus. In the STM variant of the task, the cue stimulus rapidly disappears. In the control variant (no STM,
∗
Corresponding author. Tel.: +39 06 49910989; fax: +39 06 49910917. E-mail address: [email protected] (C. Babiloni).
0361-9230/$ – see front matter © 2005 Elsevier Inc. All rights reserved. doi:10.1016/j.brainresbull.2005.01.013
NSTM), the cue stimulus is delivered during the delay period. With respect to “delayed match-to-sample” and “nback” STM paradigms, the delayed-response tasks include no probe stimulus to be matched with the memorized cue stimulus. The delayed-response tasks have been successfully used for the study of cognitive processes in schizophrenia and aging [25,27,37,41]. Neurobiological correlates of the delayed-response tasks have been extensively investigated in the nonhuman primates [16–20,33]. Fronto-parietal systems are involved in the maintenance of the “representational memory” during delayedresponse tasks [5,6,16,17,19–21,24,29,33].
384
C. Babiloni et al. / Brain Research Bulletin 65 (2005) 383–390
In humans, studies using functional magnetic resonance imaging (fMRI) have confirmed the crucial role of the frontoparietal cortical system for a successful performance in the STM variant of delayed-response tasks. This system included dorsolateral prefrontal (Brodmann area 8–9 (BA 8–9)), lateral and medial premotor (BA 6L and BA 6M), and posterior parietal (BA 7, 40) areas. The right dorsolateral prefrontal area was especially involved when visuo-spatial contents had to be memorized [8,10,11,34–36]. Furthermore, recent fMRI studies have experimentally dissociated two parallel processes related to STM tasks, namely “maintenance” of motor responses during the delay period and “selection” of an item from memory to guide the motor response [38,39]. “Selection”, but not “maintenance”, was associated with the activation of BA 46 of the dorsolateral prefrontal cortex. In contrast, “maintenance” was associated with the activation of prefrontal BA 8 and the intra-parietal cortex. Functional neuroimaging studies have revealed abnormal cortical activity associated with losses in STM with age (for a review see [26]). However, this research field is characterized by great methodological difficulties, due to the relatively poor cognitive performances of aged people to standard STM tasks. Indeed, neuroimaging study of cognition across aging requires simple tasks ensuring: (i) high rate of correct performances in stressful neurophysiological settings; and (ii) significant modulation of the cortical activity. Neuroimaging data accompanying correct performances are at the basis of comparisons between normal young versus elderly subjects and between normal elderly subjects and elderly subjects with cognitive impairments. As a preliminary step, the present functional magnetic resonance imaging study tested the hypothesis that very simple delayed-response tasks fit these requirements in normal young adults. The STM variant included a sequence of cue stimulus (two vertical bars of unequal height), delay period (blank screen for only 5 s), go stimulus, and motor response compatible with the taller vertical bar. Noteworthy, the retention (only one bit) could be based on visuospatial, phonological, and somatomotor coding. In the control variant (no STM), the cue stimulus was present during the delay period. In previous studies, the same tasks have been able to modulate electroencephalographic rhythms in frontal and parietal areas of normal young [1,2] and elderly [3] adults. However, it should be stressed that electroencephalographic techniques prevented a fine topographical analysis of these cortical responses.
2. Materials and methods 2.1. Subjects Subjects were 15 healthy, right-handed (Edinburgh Inventory) adult volunteers (mean age of 27 years, range from 19 to 40 years), who gave their informed written consent according to the declaration of Helsinki. General procedures were approved by the local institutional ethics committee.
2.2. Experimental tasks The experiment included two recording blocks (i.e. STM, NSTM). Each recording block comprised 23 consecutive trials during which fMRI machine runs in continuous mode (inter-block interval of about 5 min). Subjects were told in advance if the block was NSTM or STM (pseudo-randomized across subjects). Examples of trial for both NSTM and STM tasks are illustrated in Fig. 1. For the STM task, sequence of events was as follows: (i) a cross in a red circle at the center of the monitor (about 0.7◦ for side) as a visual warning stimulus lasting 1 s; (ii) a couple of vertical bars (about 2◦ large and 2.5–7◦ high) as a visual cue stimulus lasting 2 s; (iii) blank screen as a delay period lasting 5 s; (iv) a cross in a green circle at the center of the monitor (about 0.7◦ of diameter) as a go stimulus lasting 1 s; and (v) right finger movement to press the proper button. Subjects had to click either left mouse button if the taller bar (cue stimulus) was at the left monitor side or right mouse button if the taller bar was at the right monitor side. A cross at the center of the monitor (16 s inter-trial interval) allowed the hemodynamic parameters to return to baseline level. Compared to the STM task, the NSTM task had the cue stimulus lasting across the delay period up to the go stimulus triggering the motor response. Before the recording session, a training of about 10 min made the subjects familiar with the experimental apparatus and the tasks. Of note, it should be remarked that the two delayedresponse tasks used in the present study were chosen with a view to future applications in aging research. This motivated the choice of minimal STM load (one bit for very few seconds) and the possibility of using any memorization strategy including visuo-spatial imagery, somato-motor preparation, and mental “phonological” coding and rehearsal. 2.3. fMRI recordings For the fMRI experiments, the subjects were positioned in the scanner with the head immobilized by foam support cushions (1.5 T MRI system; Siemens Magnetom Vision). They kept their arms resting, with the right index finger resting between two buttons spaced 6 cm apart. The visual stimuli were generated by a personal computer and projected by a LCD video projector on a screen behind the scanner. A mirror attached to the head coil, at 45◦ with the line of sight, allowed the subjects to see the image on the screen. fMRIs were recorded in an event-related mode. Blood oxygen level dependent (BOLD) contrast images were obtained by means of Echo Planar Imaging (EPI) sequences with the following features: TR 1189 ms, TE 60 ms, matrix size 64 × 64, FOV 220 mm, flip angle 90◦ , slice thickness 7 mm and no gap. Functional volumes consisted of nine bicommissural transaxial slices. In addition, high-resolution anatomical images of the brain were obtained with a 3D magnetization prepared rapid gradient echo (MP-RAGE) sequence with the following sequence parameters: TR 9.7 ms;
C. Babiloni et al. / Brain Research Bulletin 65 (2005) 383–390
385
Fig. 1. Sequence of the events for the two delayed-response tasks. In the short-term memory (STM) condition, these events were as follows: (i) a cross in a red circle at the center of the monitor as a visual warning stimulus; (ii) a couple of vertical bars as a visual cue stimulus; (iii) blank screen as a delay period; (iv) a cross in a green circle at the center of the monitor as a go stimulus; and (v) right finger movement to press the proper button. Subjects had to click either left mouse button if the taller bar (cue stimulus) was at the left monitor side or right mouse button if the taller bar was at the right monitor side. Compared to the STM task, the no STM (NSTM) condition had the cue stimulus lasting across the delay period up to the go stimulus triggering the motor response.
TE 4 ms, flip angle 12◦ ; FoV 256 mm; matrix size 256 × 256, 256 axial slices with a thickness of 1 mm (no gap). 2.4. Data analysis Analysis of the fMRIs was performed using Brain Voyager 4.4 software package (Brain Innovation, The Netherlands). After slice time alignment, individual fMRIs were registered with the respective structural data set from the same session. The co-registration transformation was determined using the slice position parameters of the fMRIs and the position parameters of the structural volume. Structural and functional MRIs were normalized to the Talairach space [44] using a piecewise affine and continuous transformation. Pre-processing of fMRIs included motion correction and linear detrending. Due to the relatively large voxel size (about 83 mm3 ), no spatial smoothing was performed even for the group analysis. The extreme simplicity of both NSTM and STM tasks in terms of general cognitive demands motivated a data analysis design aimed at stressing the expected modest differences of the cortical activity related to the two tasks. To minimize the statistical correction due to multiple comparisons, the final
statistical analysis was performed on a small number of regions of interest (ROIs). These ROIs were selected based on previous studies [8,10,11,34–36]. We considered the following bilateral ROIs: dorsolateral prefrontal (BA 8–9 and BA 46), inferior parietal (BA 40), superior parietal (BA 7), lateral premotor (BA 6L), medial premotor (BA 6M, i.e. supplementary motor area), and primary motor (BA 4) areas. The ROI analysis allowed statistical inter-hemispherical comparisons within task. The mentioned ROIs were defined manually using each subject’s structural MRIs coded into Talairach space. This was done for all subjects by two experimenters before the analysis of the fMRI data. BA 8 primarily was located in the superior frontal gyrus extending from the cingulate sulcus on the medial surface over the margin of the hemisphere to the middle frontal gyrus. BA 9 was situated in a portion of the superior frontal gyrus and the middle frontal gyrus. Its approximate boundary on the medial aspect of the hemisphere was the cingulate sulcus and, on the lateral aspect, the inferior frontal sulcus. BA 46 was approximately circumscribed to the middle third of the middle frontal gyrus and the most rostral portion of the inferior frontal gyrus. BA 40 lays primarily in the supramarginal gyrus surrounding the
386
C. Babiloni et al. / Brain Research Bulletin 65 (2005) 383–390
posterior ascending limb of lateral sulcus. BA 7 was placed in the superior parietal lobule and in precuneus. It was approximately bounded by the superior postcentral sulcus rostrally, the intraparietal sulcus laterally, the parieto-occipital sulcus caudally, and the subparietal sulcus on the medial bank of the hemisphere. BA 6M and L were situated primarily in the caudal portions of the superior frontal gyrus, the middle frontal gyrus, and the rostral portions of the precentral gyrus not occupied by the gigantopyramidal area 4. They extended from the cingulate sulcus on the medial aspect of the hemisphere to the lateral sulcus on the lateral aspect. BA 4 was located in the caudal part of the precentral gyrus. Noteworthy, the two experimenters systematically verified the coordinate position of the ROI borders into the Talairach atlas. In particular, they probed borders of each ROI with the cursor and verified that corresponding Talairach coordinates belonged to the proper ROI. For each task (STM, NSTM), the voxels within the above ROIs showing statistically significant differences of the fMRI responses during the delay period compared to the rest period (baseline) were identified by a means of a group analysis merging data from all subjects. First, the time courses of all subjects were concatenated and z-transformed to form a single time course for each task. Then, these data were analyzed using the general linear model (GLM) [15]. Delayed predic-
tors were defined for the delay period of the two tasks. To account for the hemodynamical inertia (about 4 s), the statistical analysis of the delay period was carried out in a time interval from 4 s after the cue stimulus offset to 3 s after the go stimulus onset. A 3D group statistical map was generated for the delay condition by associating to each voxel the F value corresponding to the related predictor. This map was thresholded at voxel level at p < 0.000003 (corresponding to p < 0.05 corrected for multiple comparisons by the Bonferroni method; the correction was performed considering the number of brain voxels in Talairach space, about 15,000) and superimposed on the anatomical scan of one of the subjects. After the activation areas were defined from group data, the comparison of activity in these areas was performed on the basis of individual subject activation. Therefore, individual subject analysis was also undertaken. The general linear model was computed for each subject using the same predictor for the delay period as in the group analysis. To avoid the severe loss of statistical power due to the Bonferroni correction, the final significance was evaluated in the individual subject case using a cluster-size algorithm [13]. Individual statistical maps of the delay condition were thresholded at p < 0.001 (per-voxel) and a minimum cluster-size of three voxels was required, yelding a false positive rate of 0.023 on
Fig. 2. Group analysis statistical maps of BOLD cortical activity relative to delay compared to baseline period for both NSTM and STM tasks. The statistical maps were superimposed on the anatomical magnetic resonance imaging (MRI) scan of one representative subject. Statistical threshold of these statistical maps was posed at voxel level of p < 0.001, which corresponded to p < 0.05 corrected for multiple comparisons with clusters of three voxels as evaluated by Monte Carlo simulation. Color scale: maximum (p = 0.05 corrected) and minimum (p < 0.05 corrected) are coded in yellow and red, respectively. Right hemisphere is on the right side of the figure.
C. Babiloni et al. / Brain Research Bulletin 65 (2005) 383–390
the entire 3D image, as evaluated by means of a Monte Carlo simulation (AlphaSim routine of AFNI package, [9]). Finally, the relative BOLD signal increase with respect to baseline (i.e. (delay signal − baseline signal)/(baseline signal)) was averaged across trials and across voxels in a ROI, yielding for each ROI a measure of activation. These mean relative BOLD signal increases were used as an input for an ANOVA analysis for repeated measures. The ANOVA analysis evaluated the fronto-parietal cortical responses during the delay period of the present one-bit STM task. The factors of the analysis were Condition (STM and NSTM), Hemisphere (left, right), and the aforementioned ROIs. Mauchley’s test evaluated the sphericity assumption, which took into account the level of auto-correlation among the variables to be compared. The correction of the degrees of freedom was made by Greenhouse–Geisser procedure, which took into account the complexity of the statistical design formed by factors and levels to adjust the univariate results of repeated measures. Duncan test was used for post-hoc comparisons (p < 0.05).
387
The ANOVA analysis of the BOLD percent values relative to delay compared to baseline period showed a statistical interaction (F(5,70) = 2.48; Ms error = 0.009; p < 0.04) among the three factors considered, namely Condition (NSTM, STM), Hemisphere (left, right), and ROI (BA 8–9, BA 40, BA 7, BA 6L, BA 6M, and BA 4). Fig. 3 illustrates the means (± standard error, S.E.) of the mentioned BOLD percent values for BA 8–9, BA 40, BA 7, BA 6L, BA 6M, and BA 4 ROIs. Duncan post-hoc testing indicated that, compared to the control NSTM task, the STM task induced stronger BOLD activations in bilateral dorsolateral prefrontal (BA 8–9, p < 0.002), lateral premotor (BA 6L, p < 0.002), medial premotor (BA 6M, p < 0.01), inferior parietal (BA 40, p < 0.006), and superior parietal (BA 7, p < 0.003) areas. Remarkably, the STM task induced stronger BOLD activations in right than left dorsolateral prefrontal (BA 8–9, p < 0.03) and lateral premotor (BA 6L, p < 0.02) areas. Both NSTM and STM tasks provoked stronger BOLD activations in left (contralateral) than right primary motor area (p < 0.00002).
4. Discussion 3. Results As expected, high rate of correct performances were observed in both NSTM (about 95%) and STM (about 93%) tasks (no statistical difference was found at p < 0.05). The fine statistical mapping of the BOLD cortical responses accompanying the delay period of these tasks compared to the baseline period is illustrated across subjects over an individual MRI (Fig. 2). Note that statistical mapping of the BOLD cortical activity appeared to be quite similar in the two tasks (p < 0.05 corrected). Statistically (p < 0.05 corrected) activated voxels in dorsolateral prefrontal BA 46 were observed only in five subjects during the STM task and in six subjects during the NSTM task. Therefore, this area was not further considered. The epicenters of the cortical activity within the ROIs are reported in Table 1 (Talairach coordinates). Of note, BOLD percent values (delay compared to baseline period) relative to the statistically activated voxels (p < 0.05 corrected) were further considered. These values for each ROI were averaged for the final ANOVA statistics at the ROIs level.
The present delayed-response tasks demonstrated frontoparietal cortical responses in young adults associated with a minimal STM load (one bit for a few seconds; the low load tasks may be appropriate for future research with elderly and/or impaired subjects). There was a high rate of correct performances for both the NSTM and STM tasks (about 95%). In the two tasks (delay period), there were significant responses in bilateral dorsolateral prefrontal (BA 8–9), lateral premotor (BA 6L), medial premotor (BA 6M), inferior parietal (BA 40), and superior parietal (BA 7) areas. Compared to the control task (delay period), the STM task induced bilaterally stronger responses in these areas. These results confirm and extend in fine spatial details previous electroencephalograhic experiments using the same tasks [1,2]. Furthermore, the present results suggest that fronto-parietal systems are involved with a similar functional topography during the delay period of the two tasks. In this framework, the retention processes of the STM task would be related to a further increase of the activation of these systems [12,14,43]. Remarkably, the lesion of single parts of these networks (i.e. prefrontal)
Table 1 Talairach coordinates of the epicenters of the regions of interest (ROIs) showing a BOLD activation (%) during the delay period of the two present delayedresponse tasks (short-term memory, STM; no STM, NSTM) Regions of interest (ROIs)
Right hemisphere
Left hemisphere
x
y
z
x
y
z
Frontal lobe: middle frontal gyrus dorsolateral prefrontal BA 8–9 Frontal lobe: precentral gyrus lateral premotor BA 6L Frontal lobe: medial frontal gyrus medial premotor BA 6M Frontal lobe: precentral gyrus primary motor BA 4 Parietal lobe: inferior parietal lobule inferior parietal BA 40 Parietal lobe: superior parietal lobule superior parietal BA 7
32 40 2 35 37 26
29 −1 −7 −26 −41 −62
38 36 54 54 39 43
−33 −43 −3 −36 −37 −26
24 −8 −8 −28 −43 −62
39 38 54 54 41 44
388
C. Babiloni et al. / Brain Research Bulletin 65 (2005) 383–390
Fig. 3. Mean activation (± standard error) across subjects of BOLD percent values relative to delay compared to baseline period for both NSTM and STM tasks. These values refer to the ROIs including bilateral dorsolateral prefrontal (Brodmann area 8–9, BA 8–9), lateral premotor (BA 6L), medial premotor (BA 6M), inferior parietal (BA 40), superior parietal (BA 7) areas, and primary motor (BA 4) areas. The ANOVA for repeated measures of these data showed a statistically significant interaction among the factors Condition (NSTM, STM), Hemisphere or Side (left, right), ROIs (the mentioned areas). The results of Duncan post-hoc testing are indicated by asterisks.
might not impair the behavioural performance due to a possible multiple coding of the information to be memorized, namely visuo-spatial, phonological, and sensorimotor [35]. On the whole, the present experimental design is of interest for future clinical applications since it is relatively quick and easy for aged people. This may ensure good rate of correct performances necessary for neurophysiologic comparisons. However, as a methodological limitation, that design does not allow the dissociation of three different strategies for cue stimulus retention. Namely, the strategies that involve visual representation of target stimulus during the delay period, or motor representation of previously selected response during the delay period, or linguistic representation of either stimulus or anticipated motor response during the delay period. In that sense, the use of the present paradigm should be limited
to the cases in which a global neurophysiological index of short-term retention capabilities is of interest. In the present experiments, the prominent activation of the right dorsolateral prefrontal and lateral premotor areas would reflect the storage and rehearsal of the visuo-spatial information (i.e. position and highness of the vertical bars), according to the explicit task demands. This explanation agrees with previous neuroimaging studies showing that coding and rehearsal of visuo-spatial representations mainly involve the right hemisphere [43]. It also agrees with previous EEG studies showing an activation in prefrontal areas, especially of the “non-dominant” right hemisphere, during visuo-spatial STM tasks [4,22,23,30,31,42]. However, it should be stressed that this is just a tentative explanation of the present results. As aforementioned, we cannot
C. Babiloni et al. / Brain Research Bulletin 65 (2005) 383–390
exclude that the mentioned cortical activity is related also to sensorimotor (less probably linguistic) representations, due to the lack of specific experimental manipulations. In the framework of the bilateral responses to the present STM demands, the activation of the left hemisphere might be due to “executive processes” other than to the encoding, store, and sub-vocal rehearsal of the cue stimulus representation. The “executive” processes are defined as meta-processes (switching attention, checking features, inhibiting responses, etc.) that regulate the manipulation of both spatial and verbal contents in STM [7,32]. In this regard, the present STM task was characterized by simple cue stimuli, short delay period, no distractor, and competition between only two possible motor responses. Hence, it seems reasonable that the cortical responses over the left hemisphere mainly indicated the strength of the encoding, storage, and rehearsal with respect to the “executive” processes. Future studies should address the specific relationships of “executive functions” and activation of left hemisphere during delayed-response tasks, by means of experiments manipulating visual-spatial, sensorimotor, and linguistic aspects of these tasks. The present delayed-response tasks were designed to be extremely simple. The subjects knew the side of the target movement during the retention period, so that phonological and sensorimotor coding could contribute to the task goal. This means that, during the delay period of the STM task, the right fronto-parietal cortex might be recruited for the motor planning [28] other than for the retention of visuo-spatial representations. In this regard, previous fMRI studies [38,39] have demonstrated that the “maintenance” of visuo-spatial representations in delayed tasks is associated with the activation of dorsolateral prefrontal (BA 8) and posterior parietal areas. In parallel, the motor preparation would involve the dorsolateral prefrontal BA 46, marginally activated in the present study. Noteworthy, the activation of BA 46 has been demonstrated to be sensitive to aging effects in working memory tasks [40]. Here, its minor activation may be due to the young age of the experimental subjects as well as the minimal load and repetitive nature of the present STM task. The present results may be considered as a basis for future functional neuroimaging studies aimed at extending actual neurophysiologic knowledge on poor cognitive performances of aged people to standard STM tasks (for a review see [26]). Indeed, the present methodological approach ensured many correct performances in fMRI setting and a significant modulation of cortical activity in the STM task variant. Of course, the experimental design should be refined if scientific issues of interest are focused on specific representations and functions involved in the stimulus retention (i.e. visuo-spatial, sensorimotor, linguistic).
5. Conclusions The present functional magnetic resonance imaging study tested the hypothesis that a very low STM load (one bit to
389
be retained) is able to modulate cortical activity in young adults involved in delayed-response tasks. Results showed high rate of correct performances in both NSTM and STM tasks (about 95%). Compared to NSTM task (delay period), STM task enhanced the cortical responses in bilateral dorsolateral prefrontal (BA 8–9), lateral premotor (BA 6L), medial premotor (BA 6M), inferior parietal (BA 40), and superior parietal (BA 7) areas. In STM task, cortical responses were stronger in right than left BA 8–9 and BA 6L. These results indicated that, in normal young adults, a simple STM variant of delayed-response tasks is correctly performed and enhances bilateral fronto-parietal responses. Therefore, it may be used for future cognitive neuroimaging studies on aging.
Acknowledgments The authors thank Prof. Fabrizio Eusebi, Chairman of the Biophysics Group of Interest of Rome I University, for his continuous support. The research was granted by Telethon Onlus Foundation (“Progetto E.C0985”). Dr. Fabrizio Vecchio participated to this study in the framework of his Ph.D. program at the Doctoral School in Neurophysiology, University of Rome “La Sapienza”.
References [1] C. Babiloni, F. Babiloni, F. Carducci, S.F. Cappa, F. Cincotti, C. Del Percio, C. Miniussi, D.V. Moretti, S. Rossi, K. Sosta, P.M. Rossini, Human cortical responses during one-bit short-term memory. A highresolution EEG study on delayed choice reaction time tasks, Clin. Neurophysiol. 115 (2004) 161–170. [2] C. Babiloni, F. Babiloni, F. Carducci, F. Cincotti, F. Vecchio, B. Cola, S. Rossi, C. Miniussi, P.M. Rossini, Functional frontoparietal connectivity during short-term memory as revealed by high-resolution EEG coherence analysis, Behav. Neurosci. 118 (2004) 687–697. [3] C. Babiloni, F. Babiloni, F. Carducci, S.F. Cappa, F. Cincotti, C. Del Percio, C. Miniussi, D.V. Moretti, S. Rossi, K. Sosta, P.M. Rossini, Human cortical rhythms during visual delayed choice reaction time tasks. A high-resolution EEG study on normal aging, Behav. Brain Res. 153 (2004) 261–271. [4] A. Begleiter, T. Allison, G. McCarthy, E. Roessler, A.C. Nobre, D.D. Spencer, Electrophysiological studies of color processing in human visual cortex, Electroencephalogr. Clin. Neurophysiol. 88 (1993) 343–355. [5] N. Butters, D. Pandya, Retention of delayed-alternation: effects of selective lesions of sulcus principalis, Science 165 (1969) 1271–1273. [6] M.V. Chafee, P.S. Goldman-Rakic, Matching patterns of activity in primate prefrontal area 8a and parietal area 7ip neurons during a spatial working memory task, J. Neurophysiol. 79 (1998) 2919–2940. [7] C.B. Cohen, Avoiding “Cloudcuckooland” in ethics committee case review: matching models to issues and concerns, Law Med. Health Care 20 (1992) 294–299. [8] J.D. Cohen, W.M. Perlstein, T.S. Braver, L.E. Nystrom, D.C. Noll, J. Jonides, E.E. Smith, Temporal dynamics of brain activation during a working memory task, Nature 386 (1997) 604–608. [9] R.W. Cox, AFNI: Software for analysis and visualization of functional magnetic resonance neuroimages, Comput. Biomed. Res. 29 (1996) 162–173.
390
C. Babiloni et al. / Brain Research Bulletin 65 (2005) 383–390
[10] M. D’Esposito, D. Ballard, G.K. Aguirre, E. Zarahn, Human prefrontal cortex is not specific for working memory: a functional MRI study, Neuroimage 8 (1998) 274–282. [11] M. D’Esposito, B.R. Postle, B. Rypma, Prefrontal cortical contributions to working memory: evidence from event-related fMRI studies, Exp. Brain Res. 133 (2000) 3–11. [12] J.A. Fiez, S.E. Petersen, Neuroimaging studies of word reading, Proc. Natl. Acad. Sci. U.S.A. 95 (1998) 914–921. [13] S.D. Forman, J.D. Cohen, M. Fitzgerald, W.F. Eddy, M.A. Mintun, D.C. Noll, Improved assessment of significant activation in functional magnetic resonance imaging (fMRI): use of a cluster-size threshold, Magn. Reson. Med. 33 (1995) 636–647. [14] D. Friedman, W. Ritter, J.G. Snodgrass, ERPs during study as a function of subsequent direct and indirect memory testing in young and old adults, Brain Res. Cogn. Brain Res. 4 (1996) 1–13. [15] K.J. Friston, A.P. Holmes, K.J. Worsley, J.P. Poline, C.D. Frith, R.S.J. Frackowiak, Statistical parametric maps in functional imaging: a general linear approach, Hum. Brain Mapp. 2 (1995) 173–181. [16] S. Funahashi, C.J. Bruce, P.S. Goldman-Rakic, Mnemonic coding of visual space in the monkey’s prefrontal cortex, J. Neurophysiol. 61 (1989) 331–349. [17] S. Funahashi, C.J. Bruce, P.S. Goldman-Rakic, Visuo-spatial coding in primate prefrontal neurons revealed by oculomotor paradigms, J. Neurophysiol. 63 (1990) 814–831. [18] S. Funahashi, C.J. Bruce, P.S. Goldman-Rakic, Neuronal activity related to saccadic eye movements in the monkey’s dorsolateral prefrontal cortex, J. Neurophysiol. 65 (1991) 1464–1483. [19] S. Funahashi, C.J. Bruce, P.S. Goldman-Rakic, Dorsolateral prefrontal lesions and oculomotor delayed-response performance: evidence for mnemonic ’scotomas’, J. Neurosci. 13 (1993) 1479–1497. [20] S. Funahashi, M.V. Chafee, P.S. Goldman-Rakic, Prefrontal neuronal activity in rhesus monkeys performing a delayed anti-saccade task, Nature 365n (1993) 753–756. [21] J.M. Fuster, Temporal processing, Ann. N. Y. Acad. Sci. 769 (1995) 173–181. [22] A. Gevins, B. Cutillo, Spatiotemporal dynamics of component processes in human working memory, Electroencephalogr. Clin. Neurophysiol. 87 (1993) 128–143. [23] A. Gevins, M.E. Smith, Neurophysiological measures of working memory and individual differences in cognitive ability and cognitive style, Cereb. Cortex. 10 (2000) 829–839. [24] P.S. Goldman-Rakic, Topography of cognition: parallel distributed networks in primate association cortex, Annu. Rev. Neurosci. 11 (1988) 137–156. [25] P.S. Goldman-Rakic, Prefrontal cortical dysfunctions in schizophrenia: the relevance of working memory, in: Psychopatholology and the Brain, Raven Press, New York, 1991. [26] C.L. Grady, F.I. Craik, Changes in memory processing with age, Curr. Opin. Neurobiol. 10 (2000) 224–231. [27] F. Harrington, B.K. Saxby, I.G. McKeith, K. Wesnes, G.A. Ford, Cognitive performance in hypertensive and normotensive older subjects, Hypertension 36 (2000) 1079–1082.
[28] M. Jahanshahi, I.H. Jenkins, R.G. Brown, C.D. Marsden, R.E. Passingham, D.J. Brooks, Self-initiated versus externally triggered movements. I. An investigation using measurement of regional cerebral blood flow with PET and movement-related potentials in normal and Parkinson’s disease subjects, Brain 118 (1995) 913–933. [29] R.P. Kessels, A. Postma, E.M. Wijnalda, E.H. de Haan, Frontallobe involvement in spatial memory: evidence from PET, fMRI, and lesion studies, Neuropsychol. Rev. 10 (2000) 101–113. [30] L.K. McEvoy, E. Pellouchoud, M.E. Smith, A. Gevins, Neurophysiological signals of working memory in normal aging, Brain Res. Cogn. Brain Res. 11 (2001) 363–376. [31] L.K. McEvoy, M.E. Smith, A. Gevins, Dynamic cortical networks of verbal and spatial working memory: effects of memory load and task practice, Cereb. Cortex. 8 (1998) 563–574. [32] D.E. Meyer, D.E. Kieras, A computational theory of executive cognitive processes and multiple-task performance: Part 1. Basic mechanisms, Psychol. Rev. 104 (1997) 3–65. [33] M. Mishkin, Effects of small frontal lesions on delayed alternation in monkeys, J. Neurophysiol. 20 (1957) 615–622. [34] S. Pollmann, D.Y. von Cramon, Object working memory and visuospatial processing: functional neuroanatomy analyzed by eventrelated fMRI, Exp. Brain Res. 133 (2000) 12–22. [35] B.R. Postle, M. D’Esposito, “What–Then–Where” in visual working memory: an event-related fMRI study, J. Cogn. Neurosci. 11 (1999) 585–597. [36] B.R. Postle, E. Zarahn, M. D’Esposito, Using event-related fMRI to assess delay-period activity during performance of spatial and nonspatial working memory tasks, Brain Res. Brain Res. Protoc. 5 (2000) 57–66. [37] S. Roux, I. Hubert, A. Lenegre, D. Milinkevitch, R.D. Porsolt, Effects of piracetam on indices of cognitive function in a delayed alternation task in young and aged rats, Pharmacol. Biochem. Behav. 49 (1994) 683–688. [38] J.B. Rowe, R.E. Passingham, Working memory for location and time: activity in prefrontal area 46 relates to selection rather than maintenance in memory, Neuroimage 14 (2001) 77–86. [39] J.B. Rowe, I. Toni, O. Josephs, R.S. Frackowiak, R.E. Passingham, The prefrontal cortex: response selection or maintenance within working memory, Science 288 (2000) 1656–1660. [40] B. Rypma, M. D’Esposito, Isolating the neural mechanisms of agerelated changes in human working memory, Nat. Neurosci. 3 (2000) 509–515. [41] D. Servan-Schreiber, J.D. Cohen, S. Steingard, Schizophrenic deficits in the processing of context. A test of a theoretical model, Arch. Gen. Psychiatry 53 (1996) 1105–1112. [42] R.B. Silberstein, J. Ciorciari, A. Pipingas, Steady-state visually evoked potential topography during the Wisconsin card sorting test, Electroencephalogr. Clin. Neurophysiol. 96 (1995) 24–35. [43] E.E. Smith, J. Jonides, Working memory: a view from neuroimaging, Cognit. Psychol. 33 (1997) 5–42. [44] J. Talairach, P. Tournoux, Coplanar Stereotaxic Atlas of the Human Brain, Thieme Medical Publishers, New York, 1988.