Migración de máquinas virtuales

Introduction

In this brief chapter I will describe the results of an experiment that used fMRI to map visual and tactile peripheral stimuli. Many imaging studies have been performed to study retinotopic maps in the occipital lobe (DeYoe et al. 1996; Tootell et al. 1997), and functional imaging has also been used to map the somatosensory representation of different body parts (Servos et al. 1998; Maldjian et al. 1999). However, to my knowledge nobody has identified modality specific and supramodal spatial maps concerning vision and touch.

Experiment 6 aimed at distinguishing areas that showed differential responses to left and right stimulation, and areas that responded to stimuli delivered to only one sensory modality versus areas that respond independently of the modality stimulated. The design was equivalent to the one used in Experiment 3, but here instead o f having a bimodal-bilateral stimulation, with subjects attending to one of the four streams, 1 simply delivered stimuli in one hemifield and one modalitv at a time. As in Experiment 3, a baseline condition highlighted areas that responded to both left and right stimulation (bilateral representations) and areas that responded to both visual and tactile stimulation (bimodal representations).

E. M acaluso Cross-m odal Spatial Attention

A second aim of this study was to pilot the feasibility of studying cross-modal spatial representation in a MR environment. In a PET scanner, the subject can directly see his own hands (see Fig. n.1.1). Hence, visual and tactile stimulations can be placed in the same spatial position, irrespectively of the frame of reference one wants to consider. On the other hand, in a MR scanner the subject is totally enclosed in a tight tube. Space is very restricted, and a direct view of the hands is impossible. To allow bimodal

stimulation with visual stimuli delivered near to the hands, I used a system of mirrors. However, because of the reflection, visual input (LED + hands + fixation point) appeared to be rotated 90 degree in respect of the real position. Subjects had their arms lying along the body, but viewed them as stretching in front of their eyes (see Fig. 111.(5.1). In this situation, visual and tactile input were in the same spatial position relatively to the world, but visual and proprieceptual sensory input did not agree, so that in a ‘receptor centred’ co-ordinate system the stimuli were actually in different positions.

MRI setup

M irror box RF-coil

Figure III.6.1. T his figure show s a schem atic representation o f the subject set up in the M RI scanner. T he su b ject’s arm s lie along the body. A box with two m irrors sits on top o f the R F-coil. Through the m irror system the subject can see his ow n hands, the LED and the fixation point. H ow ever because of the reflection, the visual scene appears rotated 90 degrees, appearing in front o f the subject looking upw ards (in red).

D esign

Subjects

Eight volunteers participated in this study. All were right-handed men. Mean age was 28 years old (range 21-33). None of them had psychiatric or neurological history, or was taking any drugs.

Paradigm

Functional data were acquired using a blocked fM RI protocol. Four conditions were organised in a 2x2 factorial design. One factor was the side of the stim ulation (left or right). The second factor was the modality of the stim ulation (visual or tactile). I will refer to the four conditions as: sLV (left-visual stim ulation), sRV (right-visual

E. M acaluso Cross-m odal Spatial Attention

between each block there was a period without any stimulation. This served as a baseline condition.

Stimuli and task

The data of Experiment 6 were acquired with the same experimental setup (and same subjects) as Experiment 8. The following description of the apparatus used for the stimulation will seem unnecessarily complicated for the present experiment. This is because the apparatus was designed for the complex task used in Experiment 8.

Subjects lay in the scanner with both hands resting on two separate plastic supports. Two LEDs were placed over each thumb: high and low position. The distance between the high and the low LED was 3 cm. The two LEDs were not exactly one above the other, so that the high LED was at 9 cm from the central fixation point, while the low LED was at 11 cm from the fixation point (see Fig. III.8.1). Under each thumb, in

correspondence to the LED, small tubes were connected to the plastic supports. These tubes were used to deliver the tactile stimulation, which consisted of air-puffs. Subjects viewed the LEDs, both hands and the central fixation point through a mirror system. The system consisted of two mirrors placed on top of the whole-head RE coil. The double reflection ensured that the volunteers saw their hands in the correct orientation, and not upside down. During the scanning period, subjects were asked to maintain central fixation. No task was performed with the stimuli. Stimulus presentation consisted of passive stimulation of the left or right hemifield, in either vision or touch. Each o f the

four conditions was repeated twice, with the order of the blocks counterbalanced within and between subjects.

Image acquisition

Functional images were acquired using EPI imaging. For each subject I acquired 80 volumes. Each stimulation block consisted of 5 volumes (14.5 seconds). Baseline data were acquired between each stimulation block, also in blocks of 14.5 sec. The acquisition was in a transverse orientation with 32 slices and an effective repetition time (TR) of 2.89 sec. The voxel size was 3 x 3 x 3 mm.

Data Analvsis

Data were analysed with SPM99b. For each subject, the 80 volumes were realigned using the first volume as reference. Images were normalised to the Montreal Neurological Institute (MNl) standard space, with the normalisation parameters estimated using the mean of the 80 functional images. Finally, all images were smoothed using an isotropic Gaussian kernel (FHWM of 10 mm).

Experiment 6 was analysed using a fix effect model. The analysis was set up in order to separate effects of side that were specific to one modality from the ones common to both modalities. Additionally, 1 wanted to test for areas, showing bimodal-bilateral responses. In order to do that, the 8 baseline blocks were modelled as 4 separate conditions. This produced independent baselines for each of the four stimulation

E. M acaluso Cross-m odal Spatial Attention

conditions. The advantage of having independent baselines is that it allowed me to test for main effect of stimulation above baseline, with the additional orthogonal constraint (masking) that each stimulation condition had to activate above baseline. Because of the complexity of the pattern of activation I wanted to isolate, I used exclusive masking. This approach consists in creating SPM maps that are defined by a given effect being present and other effects being absent. Table III.6.1 illustrates the criteria that I used to isolate the expected pattern of activation. Generally the exclusion criteria were set so that in

unimodal-contralateral areas no effect of side was to be observed in the second modality, and that in bimodal-contralateral areas no effect of modality was to be found. In

uni modal-bilateral areas, the effect of side within the modality had to be excluded, and in the bimodal-bilateral areas, all effects of side and modality were excluded. Additional criteria were also used to ensure that only the relevant conditions activated above baseline.