Manuscript: Journal of Cognitive Neuroscience



THE EFFECTS OF TMS OVER DORSOLATERAL PREFRONTAL

CORTEX ON TRANS-SACCADIC MEMORY OF MULTIPLE OBJECTS

Tanaka, L.L.1, 2, Dessing, J.C.1,3, Malik, P. 1,2, Prime S.L.4, & Crawford, J. D.1,2

1 York Centre for Vision Research and Canadian Action and Perception Network

2 Neuroscience Graduate Diploma Program and

Departments of Psychology, Biology, and Kinesiology and Health Sciences, York University, Canada,

3 School of Psychology, Queen’s University Belfast, Northern Ireland,

4 School of Psychology, Victoria University of Wellington, New Zealand.

Correspondence to:

Dr. J. Douglas Crawford

Centre for Vision Research

York University

4700 Keele Street

Toronto

M3J1P3

CANADA

Phone: 416-736-2100 X 88621

E-mail: jdc@yorku.ca

Abstract

Humans typically make several rapid eye movements (saccades) per second. It is thought that visual working memory can retain and spatially integrate three to four objects or features across each saccade but little is known about this neural mechanism. Previously we showed that transcranial magnetic stimulation (TMS) to the posterior parietal cortex and frontal eye fields degrade trans-saccadic memory of multiple object features (Prime, Vesia, & Crawford, 2008, 2010). Here, we used a similar protocol to investigate whether dorsolateral prefrontal cortex (DLPFC), an area involved in spatial working memory, is also involved in trans-saccadic memory. Subjects were required to report changes in stimulus orientation with (saccade task) or without (fixation task) an eye movement in the intervening memory interval. We applied single-pulse TMS to left and right DLPFC during the memory delay, timed at three intervals to arrive approximately 100ms before, 100ms after, or at saccade onset. In the fixation task, left DLPFC TMS produced inconsistent results, whereas right DLPFC TMS disrupted performance at all three intervals (significantly for presaccadic TMS). In contrast, in the saccade task, TMS consistently facilitated performance (significantly for left DLPFC / perisaccadic TMS and right DLPFC / postsaccadic TMS) suggesting a dis-inhibition of trans-saccadic processing. These results are consistent with a neural circuit of trans-saccadic memory that overlaps and interacts with, but is partially separate from the circuit for visual working memory during sustained fixation.

Keywords: transcranial magnetic stimulation (TMS); dorsolateral prefrontal cortex (DLPFC); visual working memory; saccades; trans-saccadic perception

1. Introduction

Humans typically make three saccades per second (Rayner, 1998), so continuous representation of the external world must be reconstructed from a succession of discrete visual fixations (Matin, 1974). To do this, some information about object features and their spatial position must be retained and integrated across saccades, a process referred to here as trans-saccadic perception (Irwin, 1996; Melcher & Colby, 2008; Prime, Vesia, & Crawford, 2011). Traditionally, trans-saccadic memory has been viewed as a form of visual working memory, as both exhibit capacity limits of 3-4 items (Luck & Vogel, 1997; Irwin, 1992; Irwin & Andrews 1996; Prime, Tsotsos, Keith, & Crawford, 2007; Bays & Husain, 2008). However, as each saccade changes the spatial correspondence between world-fixed visual stimuli and the retina, trans-saccadic memory requires additional mechanisms that account for eye displacement (Hayhoe, Lachter, & Feldman, 1991; Prime, Niemeier, & Crawford, 2006).

A distributed saccade network, including the superior colliculus (SC), frontal eye fields (FEF), and posterior parietal cortex (PPC) has been shown to support the spatial updating of target locations across saccades in both the monkey (Walker, Fitzgibbon, & Goldberg, 1995; Umeno & Goldberg, 1997; Duhamel, Colby, & Goldberg, 1992) and the human (Ostendorf, Kilias, & Ploner, 2012; Medendorp, Goltz, Vilis, & Crawford, 2003; Merriam, Genovese, & Colby, 2003; Morris, Chambers, & Mattingley, 2007; van Donkelaar & Muri, 2002). It is thought that internal representations of point locations are spatially updated as a function of saccade size and direction, to align with final gaze direction at the end of the saccade (for recent reviews, see Sommer and Wurtz, 2008; Klier & Angelaki, 2008; Crawford, Henriques, & Medendorp, 2011). It has been proposed that such mechanisms could also support trans-saccadic memory of object features, either by linking spatially updated information with object identity (Crawford 1997; Cavanagh, Hunt, Afraz, & Rolfs, 2010), or directly remapping visual features relative to gaze (Prime et al. 2006, 2007, 2011; Melcher & Colby, 2008; Golomb & Kanwisher, 2011). Consistent with either of these views, it has been shown that transcranial magnetic stimulation (TMS) over both PPC and FEF disrupts trans-saccadic memory of multiple object features (Prime, Vesia, & Crawford, 2008, 2010).

It remains to be seen how visual working memory might interact with the saccade-specific updating processes required for trans-saccadic feature integration. Prime et al. (2006, 2007, 2011) proposed that this might occur through feedback to early visual centres (Nakamura & Colby, 2002) or through feedforward projections to higher-level cognitive structures, such as the dorsolateral prefrontal cortex (DLPFC). DLPFC is recognized to support working memory processes (Owen, Evans, & Petrides, 1996; Barbey, Koenigs, & Grafman, 2013), as well as saccadic eye movements (Pierrot-Deseilligny, Muri, Ploner, Gaymard, & Rivaud-Pechous, 2003; Pierrot-Deseilligny, Muri, Nyffeler, & Milea, 2005), however, the role of DLPFC in trans-saccadic memory has not been specifically tested. We chose to examine this by applying TMS over DLPFC while subjects performed the trans-saccadic memory task outlined in Figure 1A.

DLPFC has consistently been implicated in the cognitive control of behaviour, though the precise contribution is unclear (Cieslik et al., 2013; Duncan & Owen, 2000).

Difficulty defining DLPFC function stem in part from the wide range of cognitive processes associated with the region, and complexity in discerning the core nature of such processes. DLPFC is often synonymous with spatial working memory, as neurons show sustained activity during the retention interval of memory guided, delayed-response tasks from both neurophysiological recordings (Fuster & Alexander, 1971; Goldman-Rakic, 1987; Funahashi, Bruce, & Goldman-Rakic, 1989; Miller, Erickson & Desimone, 1996), and neuroimaging studies (Owen, Evans & Petrides, 1996; McCarthy et al., 1994; Sakai et al., 2002, Curtis & D’Esposito, 2003). Here, an appropriate motor response is generated on the basis of remembered stimuli. While persistent delay-period activity is an attractive contender for some form of memory signal representation, the nature of this activity is poorly understood (Ikkai & Curtis, 2011). A distributed prefrontal and parietal network exhibit sustained delay-period activity (Friedman & Goldman-Rakic, 1994; Chafee & Goldman-Rakic, 1998; Curtis, 2006); the same regions show similar patterns of activation during tasks associated with other cognitive functions including spatial attention and motor planning (Ikkai & Curtis, 2011; Bisley & Goldberg, 2010; Constantinidis, 2006; Awh & Jonides, 2001). Insofar as delay-period activity supports short-term spatial memory, it may also be a means to support flexible, goal-directed behaviour more generally (Curtis & D’Esposito, 2003; Fuster, 2001). It is important to note that these may not be mutually exclusive concepts, that the underlying nature of working memory representation likely extends beyond the capacities in which they are often tested. The role of DLPFC may be part of a larger distributed network to support spatial cognition and adaptive behaviour in different contexts (Cieslik et al., 2012; Mars & Grol, 2007; Rowe, Stephan, Friston, Frackowiak & Passingham, 2005).

DLPFC shows profuse connections with the saccade system and is involved in oculomotor paradigms that move beyond simple visually-guided saccades, to more complex behaviours (Sweeny et al., 1996). For example, DLPFC is important for maintaining spatial information used for forthcoming memory-guided saccades, inhibiting reflexive eye movements in anti-saccade paradigms and organizing saccade sequences (Pierrot-Deseilligny et al., 2005; DeSouza, Menon, & Everling, 2003; Munoz & Everling, 2004; Fujii & Graybiel, 2003). More broadly, DLPFC may be characterized as an area that modulates decisional processes related to current task demands (Pierrot-Deseilligny, Milea, & Muri, 2004; Johnson & Everling, 2006). Given the anatomical location of DLPFC and functional influence on spatial cognition, this region could play a role in both visual working memory during fixation and trans-saccadic memory across saccades.

Here, we hypothesized that if visual working memory and trans-saccadic memory are subserved by analogous prefrontal mechanisms, TMS over DLPFC should similarly disrupt visual memory during fixation and saccades. We tested this by applying single-pulse TMS to left and right DLPFC while subjects performed a trans-saccadic memory task similar to that used by Prime et al. (2008, 2010). In two previous studies, TMS to PPC and FEF was found to disrupt trans-saccadic memory of multiple objects (Prime et al. 2008, 2010). The present results were more complex than either our previous TMS results or the simple prediction posed above. We found both performance impairment and facilitation, suggesting that the notion of a unified visual working memory / trans-saccadic memory system needs to be replaced by one that includes dynamically interacting storage and updating mechanisms.

2. Methods

2.1 Subjects

Eight subjects (5 males, 3 females, median age 25.6) participated in this study after providing written informed consent. Another three potential subjects were tested and eliminated as they failed to meet criteria described below. All subjects had normal or corrected-to-normal visual acuity, with no known contraindications to TMS. All experimental procedures adhered to standards outlined by the York Human Participants Review Subcommittee.

2.2 Apparatus

Subjects were seated in a dark room in a customized experimental set-up with their head immobilized by a personalized dental impression bar. Three microprocessor personal computers were used for stimulus presentation and data recording. Visual stimuli were presented using a video projector that back-projected onto a display screen (1.9 X 1.4m, spanning 124.5° visual angle horizontally by 108.9° visual angle vertically). Subjects sat 50cm in front of the screen, at 114cm eye-level. Two-dimensional eye position was recorded at a sampling rate of 500 Hz using Eyelink (SR Research, Mississauga, Ontario, Canada). Eye movements were examined off-line throughout the entire trial. Three subjects were excluded due to excessive blinks and microsaccades during TMS trials. In the remaining eight subjects, we excluded trials in which subjects either failed to maintain fixation (due to blinks, microsaccades or erroneous eye movements), or failed to make a saccade of at least 2° following the go-signal in the saccade trials. This amounted to less than 5% trials omitted in each subject.

Custom software triggered magnetic pulses at 100ms, 200ms, and 300ms following the reappearance of the fixation cross immediately after mask offset (saccade go-signal) for the TMS trials, and at equivalent times in the fixation task (see Figure 1A). With respect to the normal latency distribution of saccades, this places the TMS pulse timing just before, during or after the eye movement in the saccade condition (see Figure 1C). Henceforth, we will also refer to these stimulation intervals (when they occur in the saccade task) as presaccadic TMS, perisaccadic TMS, and postsaccadic TMS. These stimulation times were chosen on the basis of previous findings that isolated the effect of TMS at different times in different brain regions in a similar experimental design (Prime et al., 2008, 2010).

2.3 Localization of brain sites

Each subject underwent anatomical MRI scanning at Sherman Health Science Research Centre, York University, prior to participation in the study. Left and right DLPFC was defined as the anterior third of the middle frontal gyrus, and was localized based on each subject’s structural MRI (Rajji et al., 2013; Fitzgerald et al., 2006). This area is thought to correspond with BA 9/46 (Oliveri et al., 2001; Hamidt et al., 2009). Online neuro-navigation was conducted using Brainsight (Rogue Research, Montreal, Canada).

Figure 1B illustrates the stimulation sites for a representative subject. Single-pulse TMS was administered at 60% fixed stimulation intensity. We chose a fixed stimulation output rather than tune the stimulation intensity to subjects’ own motor threshold, as there is little evidence that cortical excitability correlates with motor thresholds outside of the motor cortex (Robertson, Theoret, & Pascual-Leone, 2003; but see Deblieck, Thompson, Lacobini, & Wu, 2009). However, for comparison resting motor thresholds (RMT) were obtained from each subject, by stimulating left motor area (M1). This region was anatomically localized, and confirmed when a visible contraction of the first finger of the right hand was observed. The RMT was defined as the lowest stimulation output required to illicit this movement. The average RMT for all subjects was (M = 52% stimulator output, SD =+/- 2.6). For each session, the TMS coil was placed tangentially against the scalp with the handle pointed backwards, in an antero-lateral position. This coil orientation was deemed the most effective in inducing a TMS effect in the prefrontal cortex (Hill, Davey & Kennard, 2000). A control site was included to yield estimates of non-specific effects of TMS, the vertex of the head, Cz according to the 10-20 EEG (electroencephalogram) coordinate system. Cz is defined as the apex of the subject’s head – where the halfway point between the preauricular points, and halfway point between the nasion and inion, meet.

2.4 Experimental procedure

The task was similar to the one used by Prime et al. (2007, 2008, 2010) illustrated in Figure 1A. In the fixation task, subjects fixated on a cross randomly presented at one of 29 possible spatial locations (diamond shaped grid with points 3° apart) within the center of the display area spanning 18° X 18°. The target-display, consisting of Gabor patches 2.8° in diameter, with alternating black and white bars, was then briefly presented (100ms). Orientation of these patches was randomly selected from twelve possibilities (± 25°, 35°, 40, 45°, 55°, 65°). These were randomly presented in the display area, without overlapping with one another or the fixation cross. Originally, 1-7 patches were presented, but the higher numbers (5-7) were eliminated after preliminary tests suggested that they yielded close to chance (50%) performance and that our DLPFC-TMS results would be too complex for the statistical model fitting technique used in Prime et al. (2008, 2010). One possible reason why we observed lower performance for higher set-sizes in the current study, is that we doubled the number of orientation positions from six (± 25°, 35°, 45°) to 12. This was done to reduce the possibility that subjects might discern the correct response from final orientations at the most extreme angles, without comparing them to the remembered stimulus.

Thus, we used 1-4 patches, where one was the target and the rest were considered ‘non-probed’ memory items. Note that subjects did not know which item was the target, and therefore were required to remember the details (location and orientation) of as many items as possible.

Next, a mask (a uniform white field) was flashed (150ms) to reduce the possibility of visual after effects following the target-display. The fixation cross was then re-presented at the same initial location. As our experimental design sometimes required delivery of TMS pulses in the absence of saccades, or before saccade onset, we time-locked our TMS pulses to come 100ms, 300ms, or 300ms after the re-appearance of the fixation cross; which served as the go-signal in saccade trials. TMS trials (75% of total trials) were randomly intermingled with No TMS trials (25% of total trials). Figure 1C shows the TMS timing with respect to saccade latency for one representative subject, illustrating that the 100ms pulse arrived before the majority of saccades began (presaccadic TMS), the 200ms pulse arrived just before the peak distribution of saccade onset latencies (perisaccadic TMS), and the 300ms pulse arrived after the majority of saccades (postsaccadic TMS).

Following the re-appearance of the fixation cross and a brief delay (750ms), a probe was presented (100ms) at the same spatial location as the target. The probe resembled the target, except that it had made a 10° rotation either clockwise or counter-clockwise from its original position. The 10° rotation was chosen based on a previous experiment that obtained an average performance of 80% accuracy in a very similar task, using a set-size of two (Prime et al., 2007).

Subjects were required to indicate with their dominant hand whether the probe orientation had rotated clockwise (right mouse click), or counter-clockwise (left mouse click) relative to the target. In the saccade task, subjects were required to make an eye movement at the saccade go-signal, when the fixation cross re-appeared following the offset of the first mask. The location of the second fixation cross was randomly selected, among the possible initial fixation point locations. Each point was utilized before ‘cycling through’ the 29 grid points again. Subjects were required to compare probe orientation to the original, presaccadic target. This required subjects to update target position, relative to retinal position. During fixation, stimulus location remained stable on the retina. However, each saccade shifted the retinal position of the stimulus. This necessitates a mechanism to internally ‘remap’ the original stimulus relative to the new retinal position, so that the remembered positions of the first and second stimuli remains stable relative to each other and the visual field. Before commencing experimental sessions, subjects underwent behavioural pilot testing to ensure they were able to achieve a minimum of 75% performance with a solitary target in the fixation task.

Each experimental session began with a calibration sequence. Subjects performed the task in three experimental sessions at least one week apart, to minimize the fatigue effects of TMS. Each trial type (set-size of 1-4, No-TMS, TMS at 100ms, 200ms, and 300ms) was repeated 16 times for left DLPFC, right DLPFC, and Cz. Trials were performed a minimum of three seconds apart. A total of at least 2,688 trials were performed per subject.

2.5 Statistical analysis

To test our hypothesis that TMS would equally reduce performance in the fixation and saccade tasks, and for the sake of consistency, we utilized similar statistical analyses as reported by Prime et al. (2010). We conducted separate repeated measures analysis of variance (ANOVAs) comparing each time interval to its corresponding No-TMS baseline value, for each site and task. As these planned comparisons were not based on any resulting performance values, we chose to use the per comparison error rate of a = 0.05. To test for other unexpected effects we also did an overall ANOVA with condition (2 levels: fixation, saccade), TMS time (4 levels: No-TMS, TMS at 100ms, 200ms, 300ms), site (3 levels: left DLPFC, right DLPFC, Cz) and set-size (4 levels: 1, 2, 3, 4) as factors, using Bonferroni corrections for multiple comparisons,

[pic]

3. Results

3.1 Baseline performance: No-TMS   

Baseline performance in the No-TMS trials in both the fixation task and saccade task was pooled across experimental sites and averaged across all subjects (n = 8). (See the black data points in Figure 2 for examples of the unpooled No-TMS data intermingled with TMS trials.) The percentage of correct responses was identified for each set-size, from one (the target presented alone) to four (target plus three ‘non-probed memory items). In general, performance decreased with set-size and was slightly lower in the saccade condition. Performance with a solitary target was 86.5% in the fixation task and 79.5% in the saccade task. No-TMS performance was evaluated using a two-way repeated measures ANOVA (task X set-size). This analysis revealed a significant difference in performance for task (F1,7 = 12.48; p = 0.01) and set-size (F3,7 = 27.07; p = 0.001); the interaction between these factors was not significant (F3,7 = 2.59; p = 0.166).

3.2 TMS to Left and Right DLPFC and Cz

Figure 2 shows the main results for mean percentage correct responses, during the fixation task and saccade task, for left and right DLPFC and Cz. The three different TMS time intervals (100ms, 200ms, 300ms) are shown in grey, blue and red respectively, again with No-TMS shown in black. The most visually striking feature of this plot is that the Cz TMS results overlap closely with the No-TMS controls, whereas the TMS results for left and right DLPFC exhibit more variable patterns.

Separate repeated measures ANOVAs were conducted to compare the baseline No-TMS condition with each TMS time interval (100ms, 200ms, 300ms). In the fixation task, no significant differences were found for left DLPFC comparing No-TMS to TMS at 100ms (F1,7 = 0.87; p = 0.38), 200ms (F1,7 = 1.70; p = 0.23), or 300ms (F1,7 = 0.54; p = 0.49). For right DLPFC, TMS at 100ms significantly reduced performance compared to No-TMS (F1,7 = 20.01; p = 0.003); no differences were found for TMS at 200ms (F1,7 = 2.48; p = 0.16), or 300ms (F1,7 = 0.43; p = 0.53). In the saccade task during left DLPFC TMS, a significant facilitation of performance was observed for the 200ms interval (F1,7 = 7.66; p = 0.03) compared to No-TMS, but not 100ms (F1,7 = 0.95; p = 0.36), or 300ms (F1,7 = 2.44; p = 0.16). For right DLPFC TMS, a significant improvement in performance was found for the 300ms interval (F1,7 = 6.64; p = 0.04, but not 100ms (F1,7 = 1.66; p = 0.24), or 200ms (F 1,7 = 0.39; p = 0.55). No significant differences in performance were observed for stimulation during either the fixation or saccade task for Cz (all F1,7 ≤0.45; p ≥ 0.53). Separate repeated measures ANOVAs showed a significant difference between Cz and right DLPFC performance during the 100ms interval of the fixation task (F1,7 = 4.2; p = 0.04), but failed to show any other differences. Overall, these results suggest differences in performance both between left and right DLPFC, and between tasks, namely, impairment in the fixation task and facilitation in the saccade task.

Performance during the trans-saccadic memory task was assessed using an overall ANOVA with four factors (condition, TMS time, site, set-size) with (2, 4, 3, 4) levels. The main effects of TMS time (F3,8 = 1.50; p = 0.24) and site (F2,8 = 0.64; p = 0.54) were not significant. However, the main effect of condition (F1,8 = 41.20; p ................
................

In order to avoid copyright disputes, this page is only a partial summary.

Google Online Preview   Download