To the human kind, mental visualization and manipulation of an object is an essential ability to perceive and interact with our surrounding. Since the invention of the actual Transcranial Magnetic Simulation (TMS) method by Barker, Jalinous and Freestone in 1985, multiple studies have been conducted incorporating this method in order to demonstrate the causality and define the degree of involvement of specific brain regions when performing a task. According to Bode, Koeneke and Jäncke (2007), the laws followed by mental rotation of objects are similar to the one followed by real manipulation of objects. This suggest understanding how mental rotation works, and which areas of the brain are activated when performing mental rotation, is an important research topic in order to make discovery and progress related to the manipulation of real objects and overall operation of the brain.
Previous research has shown that multiple brain areas are activated when performing mental rotation of body parts and objects. Kosslyn, Ganis and Thompson (2001) demonstrated that the dorsal visual stream, parietal, and premotor cortex areas are all activated when mentally rotating body parts and objects. However, not only these areas were activated: they indicated that the primary motor cortex (M1) was involved, depending if the subject was mentally rotating hands or inanimate objects, by using functional magnetic resonance imagery (fMRI) and transcranial magnetic simulation (TMS) applied over the M1 region. They concluded that the involvement of M1 is “strategy-dependent” of what is being mentally rotated. Furthermore, Bode et al. (2007) expanded on these findings by showing that M1 is activated for general mental rotation even though no motor task is required. They displayed images of body parts (hands) and inanimate objects (houses and tools) and asked participants to evaluate if the image on the right is the same as the one on the left but either rotated or simply a “mirrored”. They recorded and averaged the motor evoked potentials (MEP) at the Abductor Pollicis Brevis (APB) muscle of the right hand of each participant while delivering TMS pulse every 6 seconds over a 24 trials period. When comparing these values with some baseline ones, they noticed a difference for every type of images except for hand.
Eisenegger, Herwig and Jäncke (2007) drawn the same conclusions as Bode et al. (2007) and Kosslyn et al. (2001) in their study. Using a comparable method as the one stated in Bode et al. (2007), Eisenegger et al. (2007) concluded that M1 is involved when mental rotation is performed, however, there is a slight difference in their conclusions. They deducted that M1 can either be: a) strictly involved (which was concluded by Georgopoulos, Lurito, Petrides, Schwartz and Massey, 1989), b) strategically involved (concluded by Kosslyn et al., 2001A) or c) the result of a “spill-over” effect from adjoining brain regions (concluded in Bode et al., 2007). In other words, Eisenegger et al. (2007) define the activation of M1 in c) being caused by the activation of the premotor cortex and areas surrounding M1.
However, the causal involvement of the M1 is not unanimous and opinions on the question diverge. In their research, Kosslyn, DiGiromalo, Thompson and Alpert (1998) suggested that two mechanisms underlie mental rotation with one process triggering the M1 region while the second one does not. Using positron emission tomography (PET), they compared the blood flow of participants while performing mental rotation of cubes figures as well as hands. As a result, they concluded that the M1 region was activated for rotation of hands, however, areas of the frontal motor cortex were not activated when rotating cubes figures. In their study, Tomasino, Budai, Mondani, Skrap and Rumiati (2005) also supported the idea that M1 is not necessarily involved in mental rotation. A motor cortex simulation device was implemented under the scalp of a 58 years old injured man from a motorcycle crash who had a lesion of the right inferior brachial plexus. He was then asked to perform a mental rotation of various figures such as tools, hands and cubes under different stimulation using the device. Results of the experiment showed that motor rotation (rotation of hands) involved the M1 region, however, activity of the M1 region was not recorded for visual rotation (rotation of cubes and tools).
The hypotheses of this study are similar to the ones stated in Bodel et al. (2007): first, the M1 region is causally involved when performing motor mental rotation (rotation of hands). Second, the M1 region is causally involved when performing visual rotation (rotation of Shepard figures). To demonstrate so, we predict an increase of reaction time (RT) when applying real TMS over the M1 region of participants when performing one or the other rotation compared to the RT of participants when sham (placebo) TMS is applied.
Method
Participants
Our study is a partial replication of the one of Bodel et al. (2007). A total of sixty right handed participants from The University of Melbourne were divided in two groups of 30. For the Shepard experimental condition, 15 females and 15 males were involved with a mean age of MD=23.6 and SD=2.87. For the hands experimental condition, 16 females and 14 males with a mean age of MD =24.03 and SD=3.33. They were all aged between 18 and 30 years old with normal vision and health and gave informed consent to perform the experiment under its approbation by the Melbourne Research Ethics Committee.
Stimuli and Apparatus
In order to interrupt processing of the M1 region, a Transcranial Magnetic Simulator was used combined to a figure-of-height coil (wing diameter: 70mm) to perform TMS and induce “neural noise”. The coil was placed tangentially over the hand area of the left M1 region with its handle facing backward at an angle of 45º from the midline of the scalp. The stimuli, consisting of images of Shepard figures and hands, were displayed on a Dell personal computer using the software Presentation Software Package. The object size was approximately kept to a maximum of 10cm in height or width in order to create a visual angle ranging from 7.6º and 8.4º (Bode et al., 2007).
Procedure
The participants were seated at a distance of approximately 70cm ( 4cm) from the screen on which the stimuli were displayed. They were asked to compare the images on the left and right and determined whether the right one is a rotated version of the left or mirrored. When a decision was made, the participant would press a response box with his right hand to indicate his choice. Reaction times (RT) were recorded over a number of 6 blocks, 24 trials per block, for a total number of 144 trials per participants. For every odd blocks, real TMS was applied, interrupting M1 processing while sham TMS was applied during even blocks. All stimuli were displayed twice to the participant where an actual match was possible in 50% of the trials while the conclusion that the right image is a rotated mirror of the left one represented the other 50%. Note that the degree of rotation necessary to conclude whether or not simple rotation or “rotation + mirroring” happened ranges from 45º and 315º. To avoid the experiment to become bias by accommodation of the participant to the stimulus, for each category of stimulus (hands or Shepard, 4 versions of 3 different objects were used (Bode et al., 2007).
Planned Statistical Analysis
During the experiment, RTs from the participants were retrieved for both conditions (hands and Shepard figures). We then compared the means RT between both conditions to validate or reject our hypotheses using a one-tailed paired sample t-test.
Results
Graph 1 displayed the mean RTs for real and sham TMS for both conditions. It reflects the results of this study explained below.
Hypothesis 1 predicted that participants subjected to real TMS rather than sham TMS would display higher mean RTs (would be slower) for motor rotation stimuli. In order to accept or reject this hypothesis, a one-tailed paired sample t-test was used comparing the mean RTs for real TMS and sham TMS. According to the conducted test, it was found that participants exposed to real TMS had higher response times than the ones exposed to sham TMS. There was a significant difference in the scores for real TMS (MD=1899.27 ms, SD=180.00) and sham TMS (MD=1774.09 ms, SD=143.51) conditions; t(29)=3.30, p=.0015, r2=.27.
Similarly, hypothesis 2 predicted that participants subjected to real TMS rather than sham TMS would display higher mean RTs (would be slower) for visual rotation stimuli. In order to accept or reject this hypothesis, a one-tailed paired sample t-test was used comparing the mean RTs for real TMS and sham TMS. According to the conducted test, it was found that participants exposed to real TMS had higher response times than the ones exposed to sham TMS. There was a significant difference in the scores for real TMS (MD =2115.19 ms, SD=173.42) and sham TMS (MD =1950.28 ms, SD=152.26) conditions; t(29)=3.84, p=.0005, r2=.34.
Discussion
In the results section, the effect size r2 had values greater than .25 for both of the conditions tested. According to Cohen (1988), these values are interpreted to be large effect size, indicating that for both conditions, the mean RTs were slower for real TMS applied compared to sham TMS. Therefore, these findings support both of our hypotheses, implying that the M1 region is causally involved for mental rotation or hands as well as Shepard figures.
The evidence supporting hypotheses 1 and 2 clearly align with the finding of Bode et al. (2007) as well as the ones of Eisenegger et al. (2007) who concluded the M1 region is substantially involved in mental rotation, regardless of the process (motor or visual). Moreover, the findings of our study support the hypotheses of Kosslyn et al. (1998) but not their conclusions: both motor and visual rotation trigger the M1 region compared to only the rotation of hands in their study. Similarly, Tomasino et al. (2005) found evidence supporting Kosslyn et al., but again going against our findings regarding the involvement of M1 for ration of objects.
In addition, the evidences found in relation with both of our hypotheses do not align with the findings of Sauner, Bestmann, Siebner and Rothwell (2006): they demonstrated in their study that M1 is not causally involved in motor and visual imagery but rather linked to it. They indeed demonstrated that M1 is not causally involved in the compulsory computations of mental rotation. In fact, they argued that the activity of M1 is altered by the result of the rotation being mapped onto the response and therefore, activate M1. Last but not least, they suggested that mental rotation is being processed in the premotor cortex and parietal brain regions, our findings diverging from this conclusion.
After reviewing this study, it is apparent that some limitations exist within these experiment. Similarly to those brought up by Eisenegger et al. (2007), the activation of the M1 region during the overall mental process could possibly be triggered by the fact that the participants pressed a button with their right hand. As stated before, M1 is activated when physically interacting with an object (Bode et al., 2007). Lastly, activation of the M1 region has been proven to potentially be resulting from a “spill-over” of other regions according to Bode et al., 2007 and Eisenegger et al., 2007. Future research will enable to demonstrate the validity of this statement.
To expand on the potential of future research, as stated in Kosslyn et al. (2001), the advancement of new technologies and the exponential progress of science will allow for a better understanding of the brain operation and causal involvement of the M1 region during mental rotation.