Shared and di erential neural substrates of copying versus drawing: a functional magnetic resonance imaging study

Transcription

1 BEHAVIOUR Shared and di erential neural substrates of copying versus drawing: a functional magnetic resonance imaging study Susanne Ferber a,e,richardmraz c,nicolebaker c and Simon J. Graham b,c,d Departments of a Psychology, b Medical Biophysics, University of Toronto, c Department of Imaging Research, Sunnybrook Health Sciences Centre, d Rotman Research Institute, Baycrest and e Heart and Stroke Foundation of Ontario, Centre for Stroke Recovery,Toronto, ON, Canada Correspondence to Richard Mraz, P. Eng, Sunnybrook Health Sciences Centre, 2075 Bayview Avenue,G432, Toronto, ON, Canada M4N 3M5 Tel: ext ; fax: ; richard.mraz@sunnybrook.ca Received 3 March 2007; accepted 21March 2007 Copying and drawing-from-memory tasks are popular clinical tests to assess visuo-motor skills in neurological patients. The tasks share some motor and visual processes; however, they di er substantially in their cognitive demands. We used functional magnetic resonance imaging to identify brain regions underlying processes involved in these tasks while avoiding confounds related to basic motor requirements, through use of a specially developed functional magnetic resonance imaging-compatible computer tablet. For the copying task, activation was observed in brain regions subserving visual processing and crossmodal attention (e.g. left lingual gyrus, cuneus). Drawing activated the anterior cingulate, an area associated with motor control and linking intention with action. These ndings suggest distinct neural networks subserving copying and drawing. NeuroReport 18:1089^1093 c 2007 Lippincott Williams & Wilkins. Keywords: constructional apraxia, copying, drawing, fmri Introduction Constructional apraxia refers to the inability of some braindamaged patients, who do not exhibit basic visual and motor impairments, to assemble the elements of a model object in the correct spatial relationship. Historically, the disorder was viewed as a left parietal deficit [1], but recent studies reveal that it is more frequent after right parietal damage [2,3]. Whereas the left parietal lobe subserves the planning and programming of purposeful movements [4 6], the right parietal lobe is involved in the spatial analysis of a given stimulus and of the required action toward that stimulus [7]. Typical tests of constructional apraxia such as copying or drawing-from-memory involve both aspects, planning of movements and spatial analysis. Accordingly, a recent study has shown that the nature of the deficit in these patients depends on which hemisphere is lesioned [8]. A functional magnetic resonance imaging (fmri) study of copying objects found activation in both parietal lobes, the dorsal and ventral premotor areas, the sensorimotor area and the right inferior temporal sulcus [9]. Given that the control condition did not include any hand movements, the observed motor and premotor activation patterns may be due to basic motor requirements. Further, the task involved copying a model object with the index finger and without any visual feedback, and cannot be considered a realistic copying scenario. Although the neural correlates of such a complex cognitive skill are intriguing, further investigation is needed. This study employed a custom-built fmri-compatible drawing tablet developed in our laboratory, enabling more realistic drawing tasks to be performed with both proprioceptive and visual feedback during fmri. The drawing and copying tasks were designed to be as comparable to the clinical bedside test of constructional apraxia as possible while also being highly similar to one another. The degree of semantic involvement in copying tasks was tested by asking the participants to copy nonsense shapes. A motor control task was designed with movements similar to those for the drawing and copying tasks, but with different cognitive demands. It was hypothesized that copying is a highly visual task requiring frequent shifts of spatial attention, whereas drawing-from-memory requires more constant attention and access to memory retrieval systems. Materials and methods Participants Twelve healthy, right-handed volunteers (age: 27.8 years, SD ¼ 3.9; eight female) gave informed consent to participate in this project, as approved by the Research Ethics Board at Sunnybrook Health Sciences Centre. Experimental apparatus and design A custom-built fmri compatible drawing tablet translated movements on an acrylic tablet into onscreen cursor movements [10]. The tablet used fiber-optic ShapeTape technology (Measurand Inc., Fredericton, NB, Canada) c Lippincott Williams & Wilkins Vol 18 No July

2 FERBER ETAL. to track the position and orientation of a plastic stylus. Data from the tape and stylus were relayed to a PC workstation [11] for visualization through fmri-compatible goggles (SV4021, Avotec Inc., Stuart, Florida, USA), which were calibrated for each participant before scanning. The experiment consisted of four different tasks: (A) DRAW: participants had to draw a picture of a given word (white on a black background) denoting a real, concrete object such as a house or fence (Fig. 1a, top image). Each word subtended B101 horizontally and B21 vertically and was presented in the upper panel of the display for the entire duration of the block. (B) COPY-R(eal): participants copied an image of each object (Fig. 1a, middle image) as accurately as possible. Each object subtended about and was presented in the upper panel of the display. (C) COPY-N(onsense): participants copied an image of a nonsense object in the upper panel of the display (Fig. 1a, bottom image), consisting of parts of the objects described above (same dimensions). (D) TRACE: participants watched a video playback of a previous drawing and traced over it as it evolved over time, by following a cursor. The playback was their performance during a previous DRAW or COPY task, presented at the same speed. The image from the previous task was again presented in the upper panel. In each run, tasks were pseudo-randomly alternated in 30 s blocks with interspersed fixation periods of 28 s (central white fixation cross on a black screen) followed by 2 s warning cues indicating the type of trial about to begin (Fig. 1b). Four fmri runs were executed, each containing six task blocks and six fixation blocks, for a total of four DRAW blocks, four COPY-R blocks, four COPY-R blocks and 12 TRACE blocks. Behavior was quantified by calculating the total line distance on the screen, recorded as screen pixels, summed across all trials for each task condition. The time spent performing each task was assessed by measuring the time point when the stylus last made contact with the tablet in a given 30 s trial. Image acquisition and data analysis Experiments were performed using a whole-body 3 T MRI system (GE Healthcare Inc., Waukesha, WI, USA) with a standard quadrature, bird-cage head coil. Functional images were obtained by spiral k-space readout [12] producing 26 axial slices and mm voxels (field of view: 20 cm, echo time: 30 ms, repetition time: 2000 ms, flip angle: 701). Functional data were aligned to neuroanatomy imaged using the 3D fast spoiled gradient echo technique with Draw Copy-N FENCE Copy-R Copy Fig. 1 Experimental setup. Example stimuli from the DRAW, COPY-R (real), and COPY-N (nonsense) tasks (not shown to scale). Example screenshot for the COPY task, including a representative drawing from one of the subjects. The cue indicating the trial type (here COPY) remained on the screen for the duration of the trial mm voxels (field of view: 22 cm, echo time: 4.2 ms, repetition time: 10.1 ms, flip angle: 151). The fmri data were analyzed using Analysis of Functional Neuroimages freeware (AFNI, release [13]). After the first 10 time points of each run were discarded, the remaining fmri time-series data were co-registered to the first remaining time sample to correct for the effects of small head motions during task performance. All runs were individually detrended and then concatenated, forming a single dataset per participant. Maps of brain activity were produced by fitting the general linear model (GLM) to the measured fmri time series for each voxel, without including a hemodynamic response function. Group analysis was performed by transforming the activation map for each participant to a standard brain atlas space [14], spatially smoothing the maps (6 mm full-width at half-maximum Gaussian), and conducting a two-factor, voxel wise, mixed effects analysis of variance with task as the fixed effect and participant as the random effect. Monte Carlo simulation was used to estimate type 1 error (Po0.05, corrected). Regions of interest (ROIs) of 30 voxels (810 ml) were selected at specific areas of peak activation to observed average signal time courses across participants and tasks. Results Behavior All participants displayed adequate control of the tablet after practice. The COPY-R condition (average total line distance m ¼ 2261 pixels, s ¼ 494 pixels) did not differ significantly from the COPY-N condition (m ¼ 2258 pixels, s ¼ 658 pixels, P ¼ 0.48). Both COPY conditions, however, had significantly (Po0.01) higher total line distances than the DRAW condition (m ¼ 2046 pixels, s ¼ 401 pixels). Participants took significantly (Po0.02) less time during the DRAW condition (m ¼ 28.5 s, s ¼ 1.4 s) versus the COPY- R condition (m ¼ 29.4 s, s ¼ 0.6 s) or the COPY-N condition (m ¼ 29.4 s, s ¼ 0.8 s). No significant differences were found between the two COPY conditions for total time spent drawing. Brain activation The fmri time-series data revealed anomalies in two individuals (less primary motor activation for the DRAW task when compared to the fixation baseline; average signal intensity more than two SDs away from the mean results of the other 10). These individuals were excluded from subsequent group analysis. The comparison DRAW versus TRACE revealed increased activation in the anterior cingulate and the medial frontal gyrus. TRACE compared to DRAW showed higher bilateral activation in the precuneus, cuneus, inferior and superior parietal lobes and the supramarginal gyrus (see Fig. 2a and Table 1). The higher activation for TRACE in bilateral parietal areas presumably reflects the increased praxic demands of the task. COPY-R versus TRACE (Fig. 2b, Table 1) again resulted in higher activation in the anterior cingulate and the medial frontal gyrus. In addition, higher activation was observed in the left middle occipital gyrus, cuneus and lingual gyrus. For TRACE versus COPY-R, we observed higher bilateral activation in the middle frontal gyri, the middle and superior temporal gyri, the left inferior parietal lobe and supramarginal gyrus Vol18 No1116July2007

3 NEURAL CORRELATES OF DRAWING AND COPYING (c) (d) DRAW vs TRACE R L z=0 z=15 z=30 z=45 COPY-R vs TRACE z=0 z=15 z=30 z=45 DRAW vs COPY-R z= 5 z=20 z=35 z=60 COPY-N vs COPY-R P<0.05 t=15 t=0 t= 15 Table 1 Brain regions identi ed for the contrasts DRAW versus TRACE and COPY-R versus TRACE in Talairach coordinate space (N ¼10) [t(1,9)42.26] Talairach coordinates Active area Side x Y z tscore Volume (ml) Precuneus R, L Inferior parietal lobule Superior parietal lobule Cingulate Supramarginal gyrus Precentral gyrus R Anterior cingulate R R Superior frontal gyrus Insula L Middle temporal gyrus R Superior temporalgyrus Superior occipital gyrus Inferior frontal gyrus R Cingulate L Middle occipital gyrus L Lingual gyrus Precentral gyrus L Anterior cingulate R Inferior parietal lobule L Supramarginal gyrus Superior temporalgyrus L Middle temporal gyrus z=0 z=20 z=40 z=55 Fig. 2 Representative statistical parametric maps of all relevant contrasts. Radiologic convention is employed (left hemisphere on the right). DRAW versustrace. We found increased activation for DRAW in the ACing and the MFG. TRACE versus DRAW resulted in higher bilateral activation in the PrCun, Cun, IPLs and SPLs and the SMG. COPY-R versus TRACE: we observed higher activation for COPY-R over TRACE in the ACing and MFG, and the MOG,Cun and LG of the left hemisphere.trace versus COPY-R showed higher bilateral activation in the MTG and STG and the MFGwith signi cant activation in the left IPL and SMG. (c) DRAW versus COPY-R. Copying concrete objects led to higher activation in the left LG, left Cun, left IOG and right STG and MTG. Drawing-from-memory activated the right ACing, MFG and the right SPL. (d) COPY-N versus COPY-R. Copying nonsense objects activated the left Cun, left LG, and right ACing and PCing. ACing, anterior cingulate; Cing, cingulate; Cun, cuneus; IFG, inferior frontal gyrus; IOG, inferior occipital gyrus; IPL, inferior parietal lobe; LG, lingual gyrus; MFG, middle frontal gyrus; MOG, middle occipital gyrus; MTG, middle temporal gyrus; PCing, posterior cingulate;prcg,precentralgyrus;prcun,precuneus;sfg,superiorfrontal gyrus; SMG, supramarginal gyrus; SPL, superior parietal lobe; STG, superior temporal gyri. Comparing DRAW to COPY-R (Fig. 2c, Table 2) resulted in higher activation in the left lingual gyrus, cuneus and inferior occipital gyrus and right superior and middle temporal gyrus for COPY-R, whereas the right anterior cingulate, middle and medial frontal gyri and the right superior parietal lobe showed stronger activation for DRAW. Finally, the comparison of COPY-N with COPY-R revealed increased activation in the cuneus, lingual gyrus, putamen and middle frontal gyrus of the left hemisphere and the right anterior and posterior cingulate (Fig. 2d). Po0.05, corrected. Table 2 Brain regions identi ed for the contrast DRAW versus COPY-R in Talairach coordinate space (N ¼10) [t(1,9)42.26] Talairach coordinates Active area Side x y z tscore Volume (ml) Inferior occipital gyrus L Lingual gyrus Anterior cingulate R Superior parietal lobule R Superior temporal gyrus R Middle temporal gyrus L Superior frontal gyrus Po0.05, corrected. Discussion Our study takes advantage of a newly developed fmricompatible tablet to explore the neural correlates of drawing and copying processes. The comparisons of the copying and drawing tasks to a low-level baseline showed reliable primary motor cortex activation in 10 out of 12 subjects indicating that our device and experimental setup allow for veridical assessment of neural structures involved in Vol18 No1116July

4 FERBER ETAL. z= 1 x= 11 y= 82 z= 17 x= 11 y= 85 Normalized intensity Normalized intensity DRAW COPY-R Time (s) DRAW COPY-R Time (s) Fig. 3 Time-course analysis for ROI in the left LG and left Cun. Plots depict the normalized intensity within the ROI averaged across the four task blocks and the 10 subjects. The error bars represent the standard error of the mean. Cun, cuneus; LG, lingual gyrus; ROI, region of interest. drawing and copying. The comparisons to the fixation baseline, however, do not eliminate the confounding effects of basic motor components of the tasks and thus, direct comparison between the drawing and copying tasks comprises the bulk of the discussion. Whereas copying requires constant visual feedback processes and crossmodal shifts of attention to compare one s own copy with the model object, drawing-frommemory requires maintenance of attention, access to memory systems to retrieve information about the stimulus, and internal monitoring of whether the performed action conforms to the original intention. Thus, activation was expected in brain areas associated with visual processing and crossmodal attention for the copying task, whereas activation was expected in brain areas subserving memory and monitoring attention and actions for the drawing task. Our results partially confirmed our hypotheses. Specifically, higher activation was observed in the left lingual gyrus and cuneus for copying concrete objects. A recent fmri study showed crossmodal (visual and tactile) attentional effects in the lingual gyrus [15] which is in line with the task demands of the copying task: the copy made by the participant had to be visually compared to the model object continuously available on the screen requiring crossmodal shifts of spatial attention. This idea is supported by ROI time courses of the left lingual gyrus (Fig. 3a) showing more sustained activity for COPY whereas activity for DRAW is more transient. Further support comes from our finding of higher activation in the lingual gyrus for COPY versus TRACE. The visual input in both conditions is similar but TRACE does not require frequent comparison processes to the model. Why did the cuneus show higher activity for COPY compared to DRAW? The left cuneus subserves the processing of spatial attributes in changing visual scenes [16]. Copying tasks require constant visual comparisons between the copy and the model object leading to frequent changes in the retinal image. Accordingly, the higher activation in the left cuneus for copying a model object versus drawing-from-memory can be explained by these differential task demands. This is reflected in more sustained activity in the cuneus for COPY versus DRAW as revealed by our ROI time courses (Fig. 3b). The comparison between COPY-R and COPY-N lends further support to our speculations regarding the roles of the lingual gyrus and the cuneus in copying tasks. Subjects cannot rely on any stored representations in COPY-N, requiring more visual comparison processes between the model object and the copy. Drawing an object from memory, however, elicited activation in the anterior cingulate and medial frontal gyrus. The anterior cingulate sits on the medial wall of each cerebral hemisphere and has dense connections with the motor and prefrontal cortex and parietal areas, pointing to its role in conflict monitoring and linking intention with action [17]. As such, the anterior cingulate is perfectly poised to subserve higher-order cognitive faculties such as drawing a concrete object. Our results differ from those reported by Makuuchi et al. [9] who found activity in the parietal lobes bilaterally, in sensorimotor and premotor areas, and in the posterior part of the right inferior temporal sulcus. Our comparisons of the COPY and DRAW tasks with their respective TRACE conditions did not reveal activation in those areas. These discrepancies can be explained by the lack of a motor control condition in the earlier study. In addition, the brain activations observed in our study are linked to use of the drawing tablet. Tablet behavior differed between the two copying tasks on the one hand and the drawing task on the other. Differences in task completion time, however, although statistically significant, were around 1 s and are unlikely to be observable in the intrinsically sluggish fmri signals. Further, this comparison revealed robust activation for the task undertaken in less time and with less line length. Corroborating results were also found when comparing the activation patterns of both copying tasks (with equal amounts drawn) with each other. Therefore, different tablet behavior across the tasks is unlikely to be a significant confound. As such, we believe that our results are likely more pertinent in ultimate relation to interpretation of constructional apraxia than are those of previous literature. Conclusion We used fmri to locate brain regions subserving copying and drawing tasks and found activation in areas associated with visual processing and crossmodal attention for our copying task; in contrast, the drawing task activated a brain area associated with motor control and linking intention with action. Our study shows that copying and drawing rely on distinct neural networks, and illustrates the utility of the fmri-compatible stylus and tablet for investigating the neural correlates of clinical tests. Acknowledgements This work has been funded by grants from the Canadian Institutes of Health Research awarded to SJG and SF and by 1092 Vol18 No1116July2007

5 NEURAL CORRELATES OF DRAWING AND COPYING a grant from the Heart and Stroke Foundation of Ontario awarded to SJG. References 1. Kleist K. Gehirnpathologie. Leipzig: Barth; Piercy M, Hecaen H, de Ajuriaguerra J. Constructional apraxia associated with unilateral cerebral lesions: left and right sided cases compared. Brain 1960; 83: De Renzi E. Visuospatial and constructional disorders. In: Feinberg TE, Farah MJ, editors. Behavioral neurology and neuropsychology. New York: McGraw-Hill; pp De Renzi E. Apraxia. In: Boller F, Grafman J, editors. Handbook of neuropsychology. Vol. 2. Amsterdam: Elsevier; pp Goodale MA. Brain asymmetries in the control of reaching. In: Goodale MA, editor. Vision and action: the control of grasping. Norwood, NJ: Ablex; pp Sirigu A, Daprati E, Pradat-Diehl P, Franck N, Jeannerod M. Perception of self-generated movement following left parietal lesion. Brain 1999; 122: Milner AD, Goodale MA. The visual brain in action. Oxford: Oxford University Press; Laeng B. Constructional apraxia after left or right unilateral stroke. Neuropsychologia 2006; 44: Makuuchi M, Kaminaga T, Sugishita M. Both parietal lobes are involved in drawing: a functional MRI study and implications for constructional apraxia. Cogn Brain Res 2003; 16: Zakzanis KK, Mraz R, Graham SJ. Cognitive set-shifting during the trail making test: an fmri study using an fmri-compatible writing device. Neuropsychologia 2005; 43: Mraz R, Hong J, Quintin G, Staines WR, McIlroy WE, Zakzanis KK, Graham SJ. A platform for combining virtual reality experiments with functional magnetic resonance imaging. CyberPsychol Behav 2003; 6: Glover GH, Lai S. Self-Navigated Spiral fmri: interleaved versus singleshot. Magnet Resonance Med 1998; 39: Cox RW. AFNI: software for analysis and visualization of functional magnetic resonance neuroimages. Comput Biomed Res 1996; 29: Talairach J, Tournoux P. Co-planer stereotaxic atlas of the human brain. New York: Thieme; Macaluso E, Frith CD, Driver J. Modulation of human visual cortex by crossmodal spatial attention. Science 2000; 289: Malouin F, Richards CL, Jackson PL, Dumas F, Doyon J. Brain activations during motor imagery of locomotor-related tasks: a pet study. Hum Brain Mapping 2003; 19: Paus T. Primate anterior cingulate cortex: where motor control, drive and cognition interface. Nat Rev Neurosci 2001; 2: Vol18 No1116July