It has often been proposed that there is a close link between representation of number and space. In the present work, single-pulse transcranial magnetic stimulation (TMS) was applied to the ventral intraparietal sulcus (VIPS) to determine effects on performance in motion detection and number comparison tasks. Participants' reaction times and thresholds for perception of laterally presented coherent motion in random dot kinematograms increased significantly when the contralateral VIPS was stimulated in contrast to the interhemispheric sulcus (Experiment 1) and to the ipsilateral VIPS (Experiment 2). In number comparison tasks, participants compared the magnitude of the laterally presented numbers 1–9 with the number 5. Again, reaction times significantly increased when TMS was applied to the contralateral VIPS in contrast to control sites. The finding that VIPS-directed TMS results in impaired efficiency in both motion perception and number comparison suggests that these processes share a common neural substrate.
There is increasing evidence on the connection between representations of number and space (for reviews, see De Hevia, Vallar, & Girelli, 2008; Hubbard, Piazza, Pinel, & Dehaene, 2005). Indeed, previous research has shown interference between number and space tasks known to be processed in the dorsal pathway (Fias, Lauwereyns, & Lammertyn, 2001; see also Izard, Dehaene-Lambertz, & Dehaene, 2008). However, besides object location, the dorsal pathway is also implied in motion perception (Ungerleider, Courtney, & Haxby, 1998). A predominance of attention to location processing over attention to other features such as color or motion has been shown (e.g., Hillyard & Anllo-Vento, 1998). Nonetheless, motion perception leads to the emergence of electrophysiological processes that are independent from the processing of location (Anllo-Vento & Hillyard, 1996).
In the research reported here, we focused on the relationship between motion and number processing. Because motion perception influences visuospatial imagery processes (Quinn & McConnell, 1996; Logie, 1995), it may influence numerical representations as well. Functionally, the spatial nature of number semantics in the form of a mental number line could have given rise to motion processes that direct search processes over this spatial representation. By this account, number comparison may require at least (a) a representation and (b) a process requiring active searching over that representation. Therefore, a motion component may be present in the process of searching along the mental number line in number comparison and, likely, in other numerical processes that imply the activation of core number semantics.
Various studies have shown that number comparison implies the access to numbers semantics and, therefore, some spatial-related indexes arise in the task. When performing a number comparison task, a distance effect (Moyer & Landauer, 1967) commonly appears, which is taken to indicate the access of the analogue magnitude representation. Comparison of numbers has even shown effects of hemisphere lateralization for high versus low numbers (Lavidor, Brinksman, & Göbel, 2004), and spatial neglect patients display difficulties in judgments of numbers that are lower than the reference (Vuilleumier, Ortigue, & Brugger, 2004), suggesting again that this task access to core number representation. On such tasks, fMRI studies have revealed activation in the horizontal intraparietal sulcus (HIPS) (Hubbard et al., 2005; Dehaene, Piazza, Pinel, & Cohen, 2003), an area hypothesized to code the abstract quantity meaning of numbers (Dehaene et al., 2003).
Our proposal about the presence of motion along the mental number line allocates these motion processes in the semantics of numbers, thus here we explore a neural link between motion and performance in this task. Indeed, McCrink, Dehaene, and Dehaene-Lambertz (2007) have recently proposed such a link between motion and approximate arithmetic, whereby approximate arithmetic operates according to quantitative rules, perhaps analogous to those characterizing movement on an internal continuum. If this is the case, areas known to be recruited in processing motion should also be activated in numerical processes which involve number semantics. These motion-sensitive areas would act in conjunction with the HIPS. We thus tested for the existence of interference on both motion and number processing in the ventral intraparietal sulcus (VIPS): A motion-sensitive area could have been recruited in order to operate over a spatial representation, a representation shown to be linked to HIPS (Dehaene et al., 2003).
Parietal cortex has been shown (Culham & Kanwisher, 2001) to be involved in spatial analysis (including attention, working memory, or representation) as well as other higher cognitive tasks (calculation and language-related processes). Besides the human homologue to the macaque in MT/V5, some parietal areas have been identified as motion-sensitive (for a recent review, see Orban et al., 2006). For example, motion imagery activates dorsal pathway areas but these activations become more important in more anterior areas such as posterior parietal cortex (PPC) (Podzebenko, Egan, & Watson, 2005; Goebel, Khorram-Sefat, Muckli, Hacker, & Singer, 1998; Farah, 1993). In particular, four motion-sensitive areas have been identified in PPC (Orban et al., 2003; Sunaert, Van Hecke, Marchal, & Orban, 1999). First, the VIPS is the most posterior region and is located in the occipital part of the IPS. The second area is the parieto-occipital IPS that resides at the confluence between the parietal and occipital portions of the IPS. The other two areas, medial dorsal IPS and anterior dorsal IPS, are located in the parietal or horizontal segment of the IPS. VIPS corresponds to human vIPS (Shulman et al., 1999) and to the junction of intraparietal and transverse occipital sulci (IPTO) (Wojciulik & Kanwisher, 1999), but it is different from the multimodal area responsive to motion in the monkey IPS (Bremmer et al., 2001), referred as the VIP area (Hubbard et al., 2005; Bremmer et al., 2001). The highly motion-sensitive area V3a—whose BOLD activations in front of motion stimuli normally follow in magnitude MT activations—is more posterior following VIPS. In humans, it has been hypothesized that motion information may reach the IPS through projections from V3a and that, in general, human IPS is more involved in the perception of motion than its monkey counterpart (Orban et al., 2006; Vanduffel et al., 2002).
This distribution of motion areas in human IPS and its proximity to other areas related to numbers, led us to select the VIPS as the focus area in our study. We hypothesized that an area implicated in motion processing could also overlap with the network processing numbers (Nieder, 2004), and would thus function as a neural locus of convergence between number representation and motion processes.
Other parietal areas are implied in number processing. According to the Triple Code Model of number processing (Dehaene & Cohen, 1995), revised more recently by Dehaene et al. (2003), numbers are coded in three representations. Each of these is located in a different cerebral area that is differentially activated depending on task requirements. The HIPS is involved in the core representation of quantity and likely takes the form of a mental number line. This representation is suggested to be domain-specific as it seems mostly activated in magnitude comparison and approximation tasks. The posterior superior parietal lobule supports the HIPS through attentional orientation to the mental number line, like on any other spatial dimension. Finally, the left angular gyrus complements the HIPS in language-mediated numerical tasks. The hypothesis that the human counterpart of VIP area could be associated to the number processing system, which has also been proposed as a multimodal motion responsive region in the monkey (Hubbard et al., 2005; Bremmer et al., 2001), is relevant for this study. Certainly, the human counterpart of the VIP area (Bremmer et al., 2001) has an anatomical localization similar to the HIPS of Dehaene et al. (2003); nonetheless, evidence is lacking of a possible relationship between numbers and motion in this or any neural substrate. We selected the VIPS (Orban et al., 2006) as our area of interest for the following reasons: it is not considered an area related to number processing by Hubbard et al. (2005) or Dehaene et al. (2003); it is close to these areas, yet far enough to allow its independent stimulation; it is indeed an area sensitive to motion stimuli and motion imagery, and thus, an implication of this area also in number comparison would suggest a link between motion and numbers.
Some TMS studies have shown parietal areas to be active in number comparison (Andres, Seron, & Olivier, 2005; Sandrini, Rossini, & Miniussi, 2004; Göbel, Walsh, & Rushworth, 2001). Although some left hemisphere predominance is proposed from these studies (Andres et al., 2005; Sandrini et al., 2004; Göbel et al., 2001), especially for close numbers (Andres et al., 2005), bilateral parietal implications in number processing have been repeatedly found in neuropsychological and neuroimaging literature (Eger, Sterzer, Russ, Giraud, & Kleinschmidt, 2003; Piazza, Mechelli, Price, & Butterworth, 2002; Simon, Cohen, Mangin, LeBihan, & Dehaene, 2002; Zorzi, Priftis, & Umiltà, 2002; Pinel, Dehaene, Riviere, & LeBihan, 2001; Chochon, Cohen, van de Moortele, & Dehaene, 1999; Dehaene & Cohen, 1997; Cipolotti, Butterworth, & Denes, 1991).
In the present investigation, we examined whether the application of TMS to the VIPS regions interferes with performance on both number comparison and motion perception tasks. Number magnitude comparison was investigated because it involves activation of the quantity system in terms of the semantic representation of numbers that recruits intraparietal areas (Dehaene et al., 2003). For motion perception, we used Random Dot Kinematograms (RDKs; Newsome & Pare, 1988). This type of stimulus consists of a pool of dots that move in a restricted area of space. The level of coherence of the RDK refers to the proportion of dots moving in the same direction, while the rest is moving randomly. It is important to notice that RDKs only imply a manipulation of motion, not of space or location. The observer cannot track the motion of single dots, and, as one dot disappears from the area of motion, another dot appears in the opposite side and the general pattern does not move.
Single-pulse TMS was delivered to one hemisphere at one time. Therefore, the RDK, numbers, and shape stimuli were presented as lateralized while subjects were looking at a central fixation point. Because mechanisms common to the two tasks may also relate to the preparation of saccades (Sereno, Pitzalis, & Martinez, 2001) or to general attentional load, a control task was added in which round corners must be detected from squared corners. This task requires visuospatial detection of the pattern, as well as attentional resources, allowing stimulation effects on motion and numbers processing to be dissociated from those on other more generic mechanisms. As event-related potentials data show that both number magnitude comparison and motion perception occur at around 200 msec from stimulus onset (Pinel et al., 2001; Niedeggen & Wist, 1999; Schwarz & Heinze, 1998; Temple & Posner, 1998), three different stimulus onset asynchronies (SOAs) from the presentation of stimuli (at 100, 150, and 200 msec) were chosen in order to capture the temporal evolution of the stimulus processing in the VIPS region. In a second experiment, we further tested the effectiveness of contralateral stimulation. This was achieved by comparing performance under contralateral and ipsilateral VIPS-directed TMS in both motion perception and number magnitude comparison tasks.
According to our hypothesis, TMS directed to the VIPS should lead to decreased efficiency of motion detection processes as compared to the stimulation of a central control site. Participants should thus show increased detection times in detecting coherent motion. Moreover, in order to detect coherent motion in the RDK pattern, they should require a higher signal-to-noise ratio. These effects would add new evidence to the implication of the VIPS in motion processing. Importantly, TMS directed to the VIPS was also expected to lead to inference in numerical comparison, resulting in increased reaction times (RT) when compared to the stimulation of the central control site. The absence of effects on the corner detection control task would exclude an alternative explanation in terms of general attentional mechanisms or preparation of saccades derived from the lateralized presentation of the stimuli.
Twelve right-handed healthy participants (3 men; mean age = 22.3 years) from the University of Pavia participated in the study. All of them provided informed consent before the experiment. The procedures were approved by the Ethics Committee of the Department of Psychology of the University of Pavia.
The experiment was performed in two sessions by the same 12 participants. In one of the sessions, comparisons on numbers higher than the reference were required and coherent motion to the right of a target had to be detected. In the other session, the response was made to numbers lower than the reference and coherent leftward motion had to be detected. In addition, the detection of corners control task was introduced in one of the sessions. The manipulation of visual field/side of stimulation (right visual field [RVF] stimulation of left the VIPS/left visual field [LVF] stimulation of the right VIPS), the conditions of central stimulation (RVF–central stimulation/LVF–central stimulation), and the three SOAs (100/150/200) resulted in a 2 (visual field: RVF vs. LVF) × 2 (site of stimulation: VIPS vs. central) × 3 (SOA) design. The same design was applied to the three tasks in a within-subjects design with the exception of the corner detection task, in which the visual field of presentation of the stimuli (paired with the hemisphere stimulated) was balanced between subjects. Half of the participants saw the shapes on their LVF while central or right VIPS stimulation was applied and the other half saw the stimuli in the RVF with central or left VIPS stimulation. Participants were randomly assigned to each of these conditions.
The dependent measure for the number magnitude comparison task was RT in a go/no–go paradigm. Because accuracy provided no information, no error rate data are reported. In the motion detection task, both RTs and proportion of detection for each level of coherence were measured. In the corner detection control task, the RTs were also examined.
In the number comparison task, numbers were presented in white on a black background and were 4° wide. For the motion detection task, RDKs were constructed using VisionEgg, generating a pool of 60 frames. Then jpg files were merged into mpeg video files of 15 frames/sec so that the duration of each movie was 4 sec. The total number of dots was 3000, sized at 1.5 pixels. The dot life span was 0.5 sec, and the velocity of dots displacement was 78 pixels/sec. The percentage of coherently moving dots was varied from 0 to 37%, generating nine different RDKs. For the rightward motion stimuli, the signal in the RDKs was set to 0°, in the leftward motion stimuli, the signal was set to 180°. The dots were white and the background black. The shapes of the control task were white over a black background (see Figure 1).
The experiment consisted of two sessions that were randomly assigned to each participant. In one of the sessions, participants were tested on two tasks administered in different blocks: (a) comparison on numbers (1 to 9) to a reference number 5, and (b) detection of coherent rightward motion on RDKs (Newsome & Pare, 1988) with nine levels of coherence from 0 up to 37%.
For the number comparison task, a go/no–go response was requested by pressing the spacebar when the number was bigger than 5. Each trial consisted in the presentation of a lateralized number until response or for 1300 msec. Trials were separated by a 3000-msec intertrial interval (ITI). A centered fixation cross was present during both trial and ITI. Comparison times were measured from the presentation of the number. Each block comprised four loops of the four numbers with go response times the three SOAs. Therefore, 16 data points were available for each condition and a total of 96 trials was included in each block.
For the motion detection task, a go/no–go response was also requested. The participant had to press the spacebar when coherent directional motion was perceived. Each trial consisted in the presentation of a lateralized RDK for 4000 msec or until response. Trials were also separated by a 3000-msec ITI. A centered fixation cross was also present thorough the task. Each block comprised four loops of the nine coherence levels times the three SOAs. Thus, four data points were available for each condition of coherence and a total of 108 trials were included in each block.
Single-pulse TMS was applied using a MagStim 200 stimulator: Monophasic posteriorly directed TMS pulses were delivered at the 110% of the phosphene threshold, using a 70-mm figure-of-eight coil with its center positioned over the cortical site tangentially to the scalp, oriented parallel to the mid-sagittal line, with the handle pointing upward. Experimental stimulation was delivered either to the right and left VIPS (±24, −76, 30 Talairach coordinates), at one out of three SOAs (100/150/200 msec) from the presentation of the number or the RDK. Stimuli were presented lateralized with respect to a central fixation cross (visual angle: 11°), in order to be projected only to a single hemisphere (the one stimulated) at one time. TMS was directed at one hemisphere at a time. The stimuli (RDKs, numbers, or shapes) were presented lateralized in the RVF or the LVF while TMS on the VIPS was applied to the contralateral hemisphere, where the stimuli were mainly projected. In a control-site condition, the same blocks were presented while stimulating the interhemispheric sulcus (0, −76, 30 Talairach coordinates) with the same parameters in time and intensity of the experimental stimulation.
The stimulation points on the subject's scalp were identified using the SoftTaxic Evolution Navigator system that works in the absence of radiological images. Based on both a set of digitized skull landmarks and an MRI-constructed stereotaxic template (accuracy >1 cm, Talairach space), the software can reconstruct the cerebral anatomy to guide accurate navigation. Although individual radiological head images (i.e., MRIs) were not available, Talairach coordinates of cortical sites underlying coil locations were automatically estimated for each subjects. Thus, stimulation sites were determined by entering the Talairach coordinates of the reference points (Figure 2), whereas the correspondence between coil positions and stimulation sites was continuously controlled through the FasTrak Polhemus neuronavigation system.
The other session 2 months later involved the same participants. In this session, number comparison go trials corresponded to judgments on numbers lower than 5, whereas in the motion task, leftward coherent motion had to be detected. In addition, a control task involving corner detection in shapes was included in order to rule out an explanation of the results in terms of a general attention mechanism of orienting to the hemifields. In this task, participants had to detect shapes with rounded corners, with respect to those with squared ones, among eight different irregular shapes (see Figure 1). The timing of the trials was exactly the same as that for the number comparison task: A fixation cross appeared for 3000 msec (ITI) and then the shape was presented for 1300 msec or until a response was made.
When the go response was related to numbers higher than 5, VIPS stimulation resulted in longer RTs with respect to the control central stimulation (Figure 3A). A 2 (visual field of the target: LVF vs. RVF) × 2 (stimulation conditions: experimental vs. control) × 3 (SOA: 100 vs. 150 vs. 200) ANOVA confirmed a main effect of stimulation [F(1, 11) = 17.48; MSE = 1873; p < .002]. Although the interaction between stimulation and visual field was not significant, we explored RTs through simple effects of the three-way interaction: Post hoc (DMS: Minimum Significant Difference) analysis showed a significant effect of stimulation at SOA = 100 for the RVF [t(1, 11) = 2.575, p < .03], for the LVF [t(1, 11) = 2.282, p < .04], and an effect of stimulation at 150 and 200 msec only for targets presented in the RVF [t(1, 11) = 2.662, p < .02 and t(1, 11) = 2.915, p < .01, respectively].
The 2 × 2 × 3 ANOVA showed a main effect of stimulation in the number comparison task to numbers lower than 5 [F(1, 11) = 14.95; MSE = 5377; p < .005; Figure 3B]. Conversely to the comparison of numbers higher than 5, a stronger LVF/right hemisphere involvement appeared. When the stimuli were presented to the LVF, all 100, 150, and 200 msec SOAs showed an effect of stimulation [respectively: t(1, 11) = 2.849, p < .02; t(1, 11) = 3.486, p < .005; t(1, 11) = 3.115, p < .01]. When stimuli were presented in the RVF, only a marginal effect of stimulation was found at SOA = 150 [t(1, 11) = 1.894, p = .08].
Motion Detection Task
For this analysis, the perception threshold was defined as the minimum percentage of coherence perceived in the 75% of the trials. A logistic fit between the different levels of coherence and the proportion of coherent motion detection (x/4) was calculated. The level of coherence over or equal to a proportion of 0.75 in this logistic fit was taken as the threshold for each condition and participant. Only those RTs corresponding to stimuli over the maximal threshold for all conditions and participant were considered.
In the rightward motion detection task, the same ANOVA 2 × 2 × 3 on the RTs confirmed a main effect of stimulation [F(1, 11) = 51.73; MSE = 34,901; p < .001; Figure 4A]. No other main effects or interactions were statistically significant [visual field: F(1, 11) = 1.89, ns; all other contrasts F < 1]. Regarding the threshold of motion detection, participants needed more dots moving coherently to perceive coherent motion when the VIPS was stimulated [23.7% to 26.3%: F(1, 11) = 19.09; MSE = 12.71; p < .001; Figure 4B].
When leftward coherent motion had to be detected, the 2 × 2 × 3 ANOVA again showed an effect of stimulation on the RTs [F(1, 11) = 8.35; MSE = 90,217; p < .01; Figure 4C]. No other main effects or interactions were statistically significant (visual field F = 1.03, ns; all other contrasts F < 1). Also in this condition, a significant displacement of the threshold was found [24.7 to 26.5: F(1, 11) = 13.25; MSE = 8.99; p < .004; Figure 4D].
Corner Detection Task
In the control task, no significant effects of stimulation appeared [F(1, 11) < 0.001, p = .9]. A 2 × 3 (Stimulation × SOA) ANOVA, including the field of the target presentation as between-subjects factor, showed neither an effect of stimulation nor an interaction between stimulation and visual field/hemisphere. The very low percentage of errors and the similar range of RT values suggested that the complexity of this task was the same as of the number comparison task in the control conditions (see Figure 5). The absence of effects of stimulation in the control task allows us to exclude possible interpretations of the previously reported stimulation effects as due to an interference with more general attentional processes.
A second experiment was carried out in order to evaluate the differential role of ipsilateral and contralateral areas. VIPS-directed TMS, in this case, was applied contralaterally and ipsilaterally to the visual field in which the stimuli were presented. Due to the null effects, the control task used in Experiment 1 was not included in Experiment 2. We hypothesized that TMS to the contralateral VIPS would result in disruption to performance both in number comparison and in the motion detection tasks. This disruption was expected on both RTs and thresholds depending on the stimulated hemisphere: An effect of contralateral stimulation with respect to ipsilateral stimulation would support the results obtained in the first experiment.
The 12 participants from the first study were contacted to participate again in the second experiment. The eight who responded participated in this experiment (2 men; mean age = 24.1 ± 1.5 years). All of them gave written consent.
Owing to the absence of appreciable differences between SOAs, single-pulse TMS in Experiment 2 was delivered only at 200 msec after the stimulus, which allowed to shorten the session aiming to include all the blocks while administering a limited number of pulses. Half the blocks in the number comparison task involved a response for numbers higher than 5 (the reference), whereas in the other blocks responses were required for numbers lower than the reference. In the motion detection task, the same RDKs of the first session of Experiment 1 were presented and again RTs and number of detections for the perception of coherent rightward motion were recorded. Therefore, the number comparison task consisted of a 2 (visual field: LVF vs. RVF) × 2 (site: ipsilateral vs. contralateral) × 2 (numerical size: numbers higher vs. lower than 5) within-subject design, whereas the motion detection task involved a 2 (visual field: LVF vs. RVF) × 2 (site: ipsilateral vs. contralateral) design.
The same digits and the same movies from the first session of Experiment 1 were used.
The trials procedures were the same as for Experiment 1, except for the procedure of localization of the VIPS. Anatomical MRI data were collected from the subjects of the sample with a Siemens Magnetom Vision scanner at 1.0 T with the following features: horizontal, matrix 256 × 256, FOV 230 mm, slice thickness 3 mm, no gap, in-plane voxel size 1 mm × 1 mm, TR = 9.0 msec, TE = 115 msec.
The bilateral VIPS was localized for each subject from the individual MR images, and the coil was positioned accordingly to the cerebral surface. Four points taken from the EEG 10–20 procedure (nasion, inion, A1, and A2) were identified on the head of the subjects (based on clinical procedures) and then on the tridimensional reconstruction of the heads (based on the MR images). The positions were registered, in the first case, by the application of the magnetic pen of the SofTaxic on each point, and in the second case, by moving the cursor on the position of the reconstructed image that visually corresponded to that of the participant's brain. The mean coordinates of the coil position were (standard deviations are into brackets): left VIPS x = −26.36 (3.75), y = −83.52 (4.4), z = 32.56 (6.46); right VIPS x = 26.4 (3.5), y = −83.44 (2.48), z = 34.28 (4.0). Because the VIPS is slightly internal with respect to the cerebral cortex, the representation on the cortex of the point ±24, −76, 30 has been calculated, resulting in: x = ±26.41, y = −83.64, z = 33.01. Moreover, the mean error for the coordinates of the several coil positions with respect to the projected point has been calculated: left 3.75, 4.4, 6.48; right 3.62, 3.11, 4.31. It can be seen how the mean position of the coil and the projection on the cortical surface of the VIPS are very similar and within the SD reported in previous imaging studies (e.g., Orban et al., 2003; Sunaert et al., 1999). Thus, it can be reasonably assumed that the points stimulated in the first experiment, through the SofTaxic reconstruction, were a precise estimation of the exact position of the VIPS.
Stimulation on the VIPS contralateral to the visual field in which the number was presented resulted in impaired performance in comparison to ipsilateral VIPS stimulation [F(1, 7) = 6.5; MSE = 1100; p = .038; see Figure 6]. This was the case regardless of the numerical size to which the response had to be made as the Site × Numerical size interaction was not significant (F < 1).
For this task, there was a significant main effect of site in the motion detection times [F(1, 7) = 10.2; MSE = 7218; p = .015]. Thus, stimulating the VIPS contralaterally to the visual field where the RDKs were presented led to higher detection times than when the ipsilateral VIPS was stimulated (see Figure 7A). Again a significant displacement of the threshold was found (from 20% to 23.5% coherence) [F(1, 7) = 6.3; MSE = 11.43; p = .04; see Figure 7B]. There were no other significant main effects or interaction effects.
Our investigation provides strong support for the involvement of the VIPS, both in visual motion processing and in performing numerical comparisons, which implies access to core numerical representations. The two experiments reported here show that TMS directed at the VIPS impairs both motion perception and number comparison processes. The differences in performance arising in the comparison between contralateral VIPS stimulation versus interhemispheric stimulation (Experiment 1) were replicated by the differences in performance found in the comparison between contralateral VIPS stimulation versus ipsilateral VIPS stimulation (Experiment 2). Therefore, the association between number comparison and motion perception seems to be robustly found in the VIPS. Although it has been previously suggested that the HIPS area, where core quantity processing is thought to be processed, corresponds to the polymodal motion area VIP (Hubbard et al., 2005; Bremmer et al., 2001), the present study is the first one providing a clear association between motion perception and number comparison within the VIPS, a motion-sensitive area signaled by Orban et al. (2003) and Sunaert et al. (1999).
A second finding is a tendency for hemispheric lateralization in the processing of numbers when they are presented as lateralized. This must be considered as a weak tendency because it did not arise in the second experiment. Specifically, Experiment 1 shows a tendency of lateralization pattern of TMS effects on the VIPS in inference of number comparison depending on the size of number. In the study of Lavidor et al. (2004), numbers of small magnitudes were detected faster in the LVF than in the RVF, whereas the opposite result was found for large magnitude numbers. Similarly, in the present study, when the numbers to detect are higher than the reference, the comparison is more impaired when numbers are presented in the RVF and therefore projected to the left VIPS, which is being stimulated. By contrast, the processing of numbers lower than the reference is impaired more when presented in the LVF and the right VIPS is stimulated. However, caution is warranted in considering this pattern of lateralization effects. According to the study of Vuilleumier et al. (2004), a temporary lateralization dependent on the task may be proposed instead as an evidence of the spatial nature of the mental number line. Like other spatial stimuli, numbers are thought to be represented from left to right depending on the interval chosen in the task and the number taken as reference. Thus, for example, different results would emerge for the number 4 if the reference number was 3, because in that case, the number 4 would have been higher than the reference. However, the temporary impact of the number size on the hemispheric preference indicates that VIPS stimulation altered the process of number comparison, hence, this interference is not due to another more global process. Moreover, this pattern of lateralization in number comparison suggests the allocation of the interference in the process of comparison and not simply in the access to numbers representation.
If number processing implies a motion component, there should be a motion-related area, whose disruption would alter number processing. The area that we targeted for stimulation has been recently described as sensitive to motion (Orban et al., 2003) and its stimulation was shown to affect also numerical processes. The mechanism behind the reported association may be one of functionally related networks that share some areas, one of them is the VIPS. As a part of a circuit, the VIPS may be one of the main areas involved in the process, but other areas (like MT/V5, V3a, or the group in PPC signaled by Orban et al., 2006) may perform part of the task. We tentatively suggest that motion-sensitive areas like the VIPS may have been recruited to complement the HIPS in the process of number comparison and likely also in other tasks calling to the core number representation. Exploring the mental number line (McCrink et al., 2007) may imply a motion-like process, similar to that necessary to compute motion stimuli (RDKs). In order to decide whether a number is higher/lower than the reference number, the focus of attention has to “move” along the mental number line in a similar way to that when sensory motion is computed. This process may be carried by motion-sensitive areas such as the VIPS. Further examination of performance differences following stimulation of the other areas functionally related to the VIPS is needed to determine the neural distribution of activation underlying motion processing and number comparison.
The present findings suggest that, at the neural level, there is a connection between motion processing and the mechanisms underlying number comparison (see also Walsh, 2003 for a proposal concerning the MT area as related to quantity processing). Although caution is warranted on the role of motion processes in mental number line search because they are supported by an association of processes in the VIPS, we suggest that these mechanisms may correspond to selective attention to motion (Beer & Röder, 2004, 2005; Lewis, Beauchamp, & De Yoe, 2000; Hillyard & Anllo-Vento, 1998). If this process devoted to motion is what was altered by TMS, then a mechanism implying motion may, in turn, be operating on the mental number line.
We thank Bibiana Bellotti, Mauro Frascaroli, and Guido Moro for their help and assistance in recruiting and testing participants.
Reprint requests should be sent to Elena Salillas, Department of Biology, University of Texas at San Antonio, One UTSA Circle, San Antonio, TX 78249, or via e-mail: Elena.Salillas@utsa.edu.