Design

A between-subjects design was used to contrast synaesthetes and controls across four tasks; touch-colour consistency, tactile-texture discrimination, tactile-orientation discrimination and tactile-visual texture discrimination. For the consistency task, the dependent variable was the amount of test-retest colour error, with each synaesthete being assessed individually against the control group. For the discrimination tasks, the independent variable was the group (synaesthete, control), while the dependent variables were either the percentage correct score for the tactile and tactile-visual texture tasks or the worked out hypothetical grating gap width (in mm) that would produce a 75% correct response rate for the tactile-orientation task (smaller mm gaps indicate better tactile discrimination). These were analysed using a series of Mann Whitney U tests due to the low sample size and uncertainty about normal distribution for the synaesthetic group.

3.4.3. Materials and procedure

A custom testing booth was created to allow access to the experimental laptop and provide comfortable visual-obscuring of both the participant’s arms and the experimenter when they are in the testing position. Participants sat approximately 30 centimeters away from the testing screen,

and while horizontal viewing angle was restricted by the booth layout, vertical viewing angle was dependent on participant height. Commercially available auditory brown noise was loaded onto an mp3 player with ear plugs for use during the tactile-visual texture discrimination task. All participants first read an information sheet and signed the consent form before filing out a demographics form. Participants were tested in a quiet room with consistent lighting conditions, and asked which was their dominant hand (in the case of ambidextrous, right hand was considered the ‘dominant’ hand). The order of tasks consisted of the consistency test, tactile orientation task, tactile-vision texture task, tactile texture task and finally the consistency retest. The tasks took approximately one hour to complete and synaesthetes did an additional twenty minute photism- illustration task afterwards.

3.4.3.1. Touch-colour consistency test

The touch-colour consistency task used thirty novel items that were comprised of materials that lacked an implied real-world colour (e.g. wood-brown). The objects varied in weight, size, shape, texture and softness in a non-systematic manner (see fig. 3.2) to eliminate the creation of rule sets that may aid consistency for non-synaesthetes (Simner & Ludwig, 2012). A colour picker was created based on the Simner and Ludwig (2012) study, featuring a rotating colour wheel (to avoid spatial co-ordinates being used as a proxy for colour consistency), a brightness bar and a large colour preview window. Instructions were given on screen. No numeric colour information was provided to participants (i.e. RGB or HSL values), to avoid these being used to aid consistency. All aspects of the colour choice need to be actively selected by the participant to progress to the next stage (to avoid apathy resulting in inadvertent consistency). Consistent lighting conditions were provided in the testing environment to avoid inter-trial distortions in perceptual colour space.

Participants were sat down in the testing booth with the colour picker program running with instructions on screen. Participants were told they would be presented a series of objects out of sight in their non-dominant hand, and that their task was to pick the colour that best approximated either ‘how the object feels in your hand’ (for non-synaesthetes) or ‘that best reflects your synaesthetic colour experience’ (for synaesthetes). Objects were presented in a pseudo-random order and placed into their hand, palm-up for active exploration. The palm-up orientation of the participant’s hand allowed the participant to gather weight information alongside other tactile attributes. Knowing that participants were able to gauge weight allowed a weight-luminance correlation to be conducted (see section 3.5.2). The ordering was determined through a single randomised order for the objects’ starting points in the first presentation, and another randomised

order for their starting point in the second presentation. This first and second ordering was the same across all participants. This was done to facilitate the repeated presentation of objects to the participant in a simple to understand and timely fashion across the entire task. The purpose was to prevent presentation order errors made by the experimenter while keeping a steady pace for participants in their colour selections that might be introduced in a fully random presentation.

Figure 3.2. All thirty items used in the touch-colour consistency task. Items were picked in a manner to avoid implied real world colours and varied in a non-systematic manner across tactile dimensions to avoid participants utilising rulesets which might enhance their touch-colour consistency.

For the retest, participants were re-briefed on how they would select their colours both psychologically and on the computer. When they were ready they were subsequently presented with 30 items to pair with colours in a different pseudo-random order. The retesting within an hour was chosen as consistency across non-synaesthetes and synaesthetes within short timeframes have been found to be comparable to longer retest periods (Ward et al., 2006).

3.4.3.2. Tactile orientation and texture tasks

The tactile orientation and texture tasks used commercially available JVP spatial grating domes. These have been found to provide a more reliable threshold than alternative measures of tactile acuity (Craig, 1999; Johnson & Phillips, 1981). These are small plastic domes spanning 25mm in diameter with a square wave cut into them to create a series of raised ridges and gaps that are pressed against the skin to create evenly spaced distributions of pressure. The spacing of the ridge and subsequent gap are always the same as one another on a given dome, however there are multiple domes with different spacings. There are several domes and these consist of the following spatial distances for the ridge and gap sizes: 0.35, 0.5, 0.75, 1, 1.25, 1.5, 2 and 3mm.

For the tactile orientation task, participants were told ‘raised ridges’ would be applied to their fingertip and their task would be to respond with the direction the ridges are orientated on the fingertip. Participants were allowed to visually familiarise themselves with the domes before a quick practice on the tactile sensation of the domes orientated along and then across the fingertip. For the task, when their hand was visually obscured and the dome was applied, participants were asked whether the dome was orientated along or across the axis of the fingertip verbally. Twenty trials were given per dome spacing and the dome progression (either progressing onto more difficult narrower domes or easier wider dome spacings) depended on the correct response rate, 15 correct stopped the task, <15 correct proceeded to the next widest ridge spacing and >15 correct proceeded to the next narrowest ridge spacing. The reasoning for this is that the final score for the tactile acquity of the participants is based on working out what dome spacing participants would be expected to get 75% of the answers correct (this is the middle point between chance and perfect performance) where a lower grating width indicating higher tactile acuity. In order to work out this 75% grating spacing the following formula was used:

g75 = glow + (((0.75 – plow) / (phigh – plow))*(ghigh – glow)) g grating spacing

p trials correct / number of trials

high the grating spacing or probability of correct response on the lowest grating spacing on which the participant responded correctly more than 75% of the time

low the grating spacing or probability of correct response on the highest grating spacing on which the participant responded correctly less than 75% of the time

g75 the hypothetical grating spacing on which the subject would have scored 75% had it been present

No feedback on performance was given and the order of presentation was pseudo- randomised. The domes were applied manually until a response was given as performance is relatively unaffected by the force of application (Johnson & Phillips, 1981). To reduce time taken for the experiment, the 1mm dome was chosen as the starting point as it forms the middle ground between typical tactile acuity and enhanced mirror-touch synaesthetic acuity (Banissy et al., 2009).

For the tactile texture task, participants were shown two JVP domes of 1.5mm and 1.25mm width. These were described as the wide and narrow domes respectively, and that the participant’s task was to verbally identify which of the two is administered to them tactually across their non- dominant hand’s index fingertip. The comparison conditions included twenty trials of 1.5mm vs 1.25mm, 1.25mm vs 1mm and 1mm vs 0.75mm. The order of wide and narrow presentation was done in a pseudo-random order. Stimulation was manually applied until a response was given and a higher percentage correct denotes a higher tactile acuity.

3.4.3.3. Tactile-visual texture discrimination task

The tactile-visual discrimination task used commercially available aluminum oxide sandpaper for variations of tactile texture. The GRIT values (reflecting particles per square inch, with smaller numbers of particles denoting larger particles and a rougher texture) of 40 (the roughest), 60, 80, 120, 180 and 240 (the smoothest) were used. For the visual presentation of texture, photographs of each type of sandpaper were taken under consistent lighting conditions; the images were then converted to greyscale and several versions of these images were created through rotating the image to avoid identical visual repetition or cues to the participant. The pictures were analysed in terms of their mean luminance using GIMP image manipulation software. Analysis revealed that the pictures had minor random variations in mean luminance across all of the GRIT values. Since we did not want participants to mistake variations in average luminance as a cue to which visual texture they were seeing, a standardisation process was applied. All images were edited to have equal luminance ratings of 140 out of 200. Besides the overall luminance of an image there is also the amount of variation present between the lighter and darker elements of an image. When it comes to the variation of brightest and darkest individual parts of an image, since rougher GRIT values had larger particles, these naturally created larger and darker shadows than the smoother GRIT values. This variation in lightest and darkest elements of an image can be analysed in terms of the average standard deviation a pixel has from the mean luminance of the image as a whole. The roughest GRIT value had a standard deviation of 33 and the smoothest had a standard deviation of 15. This confirms that rougher GRIT values have darker and more luminance elements of an image (while at the same time the mean luminance is equal) relative to the smoother GRIT values that only show subtle variability from the mean luminance. Using GRIT values as a guide we adjusted the standard deviation from the mean luminance value to be linear between these points. This gives us the following standard deviations for each image: 40 GRIT (SD = 33), 60 GRIT (SD = 31), 80 GRIT (SD = 29), 120 GRIT (SD = 25), 180 GRIT (SD = 21), 240 GRIT (SD = 15). As such all pictures are equally luminant, however variations between the darker and lighter parts of an image are larger for lower GRIT ratings which visually indicates that the particles are larger and hence rougher than smoother GRIT values.

Participants were allowed to freely view pictures of the sandpaper for a minute that would later be used in the experiment. They were then explained the purpose of the experiment was would be to indicate whether sandpaper that is tactually felt is the same as or different to, the sandpaper that is presented visually on the computer. During the task, auditory cues as to perceived roughness were masked via the presentation of auditory brown noise on an mp3 player (Guest,

Catmur, Lloyd & Spence, 2002). The participant's dominant hand would tactually explore the sandpaper presented for five seconds before the visual sandpaper is presented, participants would then indicate whether the tactile and visual sandpapers were the same or different with the ‘S’ and ‘D’ keys respectively with their non-dominant hand. The visual presentation and recording of responses was done using e-prime 2.0 professional. The program featured six familiarisation, six practice (with feedback) and sixty trial phases (without feedback).

3.4.3.4. Photism-illustration task

In order to explore the psychophysics of how tactile stimulation relates to the visual photisms of the synaesthetes, eighteen photism-drawing tasks were devised. Synaesthetes were presented with answer sheets and commercially available drawing materials ranging across the colour spectrum. Synaesthetes were instructed to have their hand, body and head at specific orientations for each condition. They were then told that the experimenter would administer tactile stimulation three times on their palm while they had their eyes closed. The synaesthetes were then asked draw the photisms experienced from this stimulation. Any motion of the photism would be indicated by descriptions or arrows. The specific orientations of the hand and types of tactile stimulation are detailed in the results section.