A total of 24 female (28 ± 3 years in age) students participated in the study, none of whom practiced QMT before (for the detailed procedure, see Dotan Ben-Soussan et al., 2013). In order to avoid gender-dependent differences in performance (Terlecki et al., 2008) and reduce intra-group variability we focused on female participants. Data were collected both before and after a single training session lasting 7 min in each of three training groups (the participants having been randomly allocated to these): (1) Quadrato Motor training (QMT—3 choices and whole-body response, n = 9); (2) Simple motor training (SMT—1 choice and whole-body response, n = 7); and (3) Verbal training (VT—3 choices and verbal response, n = 8). The collected data included the HFT, reported here. Other measures included EEG and an Alternative Uses task, reported elsewhere (Dotan Ben-Soussan et al., 2013).

Quadrato Motor Training (QMT). Briefly, the participant stood at one corner of a 0.5 m × 0.5 m square and made movements in response to verbal instructions given by an audio tape recording. There were three optional directions of movement. The instructions directed participants to keep the eyes focused straight ahead, hands loose at the side of the body. They were also told to immediately continue with the next instruction and not to stop due to mistakes. At each corner, there were three possible directions to move. The training thus consisted of 12 possible movements (Figure 1). Training consisted of a sequence of 69 commands, lasting 7 min.

Two variables that were addressed in other studies of motor learning are limb velocity and the decision regarding the responding limb (Criscimagna-Hemminger et al., 2002; Donchin et al., 2003). In order to control these parameters, we used a movement sequence paced at a rate of an average of 0.5 Hz (similar to a slow walking rate), and we instructed the participants to begin all movements with the leg closest to the center of the square (a detailed description can be found in Dotan Ben-Soussan et al., 2013).

Simple Motor Training (SMT). The SMT group provided similar motor performance as the QMT group but with reduced cognitive demands. This group moved from corner to corner on the square in exactly the same manner as the QMT group (pace, duration, auditory cue), but their movement was consistently 1-2-3-4-1 etc. The participants heard the same recordings as the QMT group. However, while the QMT group was told that each number represented a different corner of the square, the SMT group was told to simply begin at a certain corner and to continue to the next corner clockwise in response to the instructions. That is, regardless of the number specified on the tape, they always moved in the same sequence. This reduced the uncertainty and the cognitive demand, compared to the QMT group.

Verbal Training (VT). The VT group was designed to reduce motor load while keeping the same cognitive load and uncertainty. The participants, who were instructed to only make verbal responses, stood 1 m in front of the square, but did not move to the corners. Instead, they responded to the taped commands verbally by stating what direction of movement would be required in order to reach the corner specified by the command. For example, for a movement from corner 1 to corner 2, they were required to say “straight.” All other training parameters were kept identical to the QMT (pace, duration, auditory cue).

We employed a computerized version of the Hidden Figures Test (HFT, detailed in Glicksohn and Kinberg, 2009), generated using E-Prime 1.1 (Psychology Software Tools). The participants were required to locate a simple figure embedded within a complex figure, both of them appearing on the screen, side by side (Figure 2). The figures ranged in size between 10.5 × 4 cm and 12 × 6.5 cm, and were viewed from a comfortable viewing distance. The participants were allocated 30 s for each of 16 trials. Based on the data collected by Glicksohn and Kinberg (2009), we split the test into two even sets of 8 tasks, matched for degree of difficulty, defined as number of hits for item/(number of hits + number of false alarms for item) in that study (n = 80). The two sets were presented in a counterbalanced order across participants. Before the test, one practice trial was given. The participant attempted to locate the simple figure as quickly as possible, within the time allocated. On detecting the figure, the participant pressed a button, and the complex figure was presented again, so that the participant could indicate the embedded figure for the experimenter. Two scores were obtained for each trial: correct/incorrect detection, and reaction time (RT). All RTs were subsequently log-transformed to normalize the data, and mean log(RT) was computed for each participant.

To answer the question regarding the effects of QMT on HFT performance, we ran a Group (QMT, SMT, VT) × Training (pre, post) analysis of variance (ANOVA) for correct HFT responses, adopting the Greenhouse-Geisser criterion. Whenever needed, we added post-hoc t-tests.

A total of 37 volunteers (29 ± 4 years in age) participated in the study, which was conducted both in Israel (10 females—out of which nine participated in Study 1 in the QMT group, and 8 males) in the MEG unit at the Gonda Brain Research Center, and in Italy (7 females, and 12 males) in the cognitive neurophysiology laboratory of the Research Institute for Neuroscience Education and Didactics. All participants were right-handed with no medical history that might affect their EEG. The study was approved by the ethics committee of Bar-Ilan University. Upon entering the lab, the participant signed a written informed consent. Then, we recorded baseline EEG for 5 min (2.5 min eyes open and fixed and then 2.5 min eyes closed). Subsequently, the HFT Task (see section Hidden Figures Test) was presented. All data were collected both before and after a single QMT session lasting 7 min. In order to pool the female participants together and all male participants together, we conducted two t-tests (the first for comparing the participants who took part in Study 1 to the additional 8 females; and the second for comparing the Israeli and Italian male participants). The two independent t-tests revealed no differences in performance in the HFT within the female and male groups, allowing us to pool our participants.

The task was the same as in Study 1. In addition to the analysis conducted as in section Hidden Figures Test, based on Glicksohn and Kinberg (2009), four profiles were defined: (1) Field Dependent (FD) = low success + long RT; (2) Field Independent (FI) = high success + short RT; (3) Reflective (Ref) = high success + long RT; (4) Impulsive (Imp) = low success + short RT. High success was defined as an above-median number of correct detections at baseline (median = 4); long RT was defined as an above-median Log(RT) at baseline (median = 4.21).

EEG data were recorded using a 65-channel geodesic sensor net (Electrical Geodesics Inc., Eugene, USA), sampled at 500 Hz and referenced to the vertex (Cz) with analog 0.1–200 Hz band-pass filtering. Impedance was usually kept under 40 kΩ, lower than the customary 50 kΩ with this system (Ferree et al., 2001). EEG signals showing eye movements or muscular artifacts were manually excluded, and bad channels were replaced using spatial interpolation (Perrin et al., 1989). The data were referenced offline to average reference. The first 32 non-overlapping, artifact-free epochs of 2.048 s duration were extracted from each electrode for further analysis from the eyes-closed resting state period, as previously reported (Dotan Ben-Soussan et al., 2013). Coherence values were calculated from the multitapered pair-wise cross-spectra. The coherence values were normalized using Fisher's z transformation. We calculated coherence within the theta, alpha, and gamma frequencies (4–7, 8–13, 25–45 Hz, respectively). Notably, gamma effects are at risk of contamination by muscle activity from scalp and neck (Whitham et al., 2007) and saccade-related spike potentials (SP) due to eye movements (Yuval-Greenberg and Deouell, 2009). In order to minimize this risk, and as SP is elicited at the onset of small saccades, which occur during eyes-open fixation (Martinez-Conde et al., 2004), our report focuses on eyes-closed conditions. Furthermore, to eliminate muscle artifacts, we used three additional steps for caution. First, we carefully visually inspected the raw data, manually extracting artifact-free epochs. Second, as the EMG peaks at 70–80 Hz (Cacioppo et al., 1990), we used much lower frequencies of 25–45 Hz. And third, we excluded from statistical analyses all the circumference electrodes, closest to eyes, neck and face muscles (Berkovich-Ohana et al., 2012).

For the sake of data reduction and statistical comparisons, we chose one central electrode site in each region of interest. Since frontal, central, temporal and parietal areas are important for cognition and action, we chose to focus on bilateral frontal, central, temporal, and parietal electrode sites (F3, F4, C3, C4, T7, T8, P5, P6) (Figure 4). We defined electrode “pairs of interest” (POI), on the basis of prior knowledge concerning movement and meditation (Andres et al., 1999; Lutz et al., 2004, 2007).

In order to answer the question of gender-difference in HFT performance, we ran a Gender × Training (pre, post) analysis of variance (ANOVA) for the two measures of HFT: (1) number of correct detections; (2) mean log(RT). Whenever needed, we added post-hoc one-tailed t-tests. Change in number of correct detections was calculated by subtracting the number of correct figures detected before QMT from the number of correct detections following QMT. Change in log(RT) was calculated by subtracting pre from post QMT log(RT).

Then, in order to study the possible gender-dependent electrophysiological counterparts of the QMT effects, we ran six ANOVAs for the z-transformed theta, alpha and gamma coherences, for intra- and inter-hemispheric connections separately. For intra-hemispheric coherence, we ran a 4-way ANOVA: Gender × Training × Hemisphere × Electrode Pair (F-T, F-P, T-P). For inter-hemispheric coherence, we ran a Three-Way ANOVA: Gender × Training × Bilateral Electrode Pair (F-F, C-C, T-T, P-P). For all the ANOVAs we adopted the Greenhouse-Geisser criterion.

We found a significant Training effect [F_{(1, 35)} = 9.57, MSE = 1.66, p <.005] for the number of correct detections (Figure 5A), which extend the results of Study 1 to males; but not for mean log(RT). The Gender × Training interaction was not significant for either measure. No difference in performance between the male and female participants was found either at baseline, or following training, for either measure (Figures 5B,C). As can be seen from Figure 5, on average there is an increase in the detection of one more figure following QMT, and the test-retest correlation found here of 0.57 (n = 37, p < 0.0001) is relatively high, though is lower than the.78-.92 range found for the GEFT (Kepner and Neimark, 1984). This correlation is comparable to the correlation that we have computed between the two sets derived from the HFT of a previous study (Glicksohn and Kinberg, 2009), which is.67 (n = 80, p < 0.0001). An overall view of the relationship between HFT performance (see section Results) and RT is presented as a scatter plot (Figure 6). Of the 37 participants, 12 (6 females) exhibited a trend toward reflectivity or FI following QMT. The results show increased reflectivity in both genders.

A positive correlation was found between the change in score for the two change in number of correct detections and change in log(RT) measures (r = 0.36, p < 0.05, n = 37), indicating that better performance on the HFT following QMT is related to longer latency of correct response. Again, we can compare this to the respective difference scores computed for two sets derived from the HFT of the previous study, which is 0.22 (n = 76, p = 0.053). Thus, it is not simply the case that QMT leads to faster RT, as reported elsewhere for a simple reaction time task (Dotan Ben-Soussan et al., 2013), rather that performance on the HFT is indicative of reflectiveness (more correct detections coupled with longer RT; Glicksohn and Kinberg, 2009).

Of importance to the question related to training-induced gender-dependent effects, the first set of three ANOVAs yielded within the theta band a significant Gender × Training interaction [F_{(1, 35)} = 11.32, MSE = 0.03, p < 0.005]. As seen in Figure 7A, while the female group demonstrated significantly increased intra-hemispheric theta coherence [t₍₁₆₎ = −2.56, p < 0.05], this was significantly decreased in the male group [t₍₁₉₎ = 2.57, p < 0.05]. A similar pattern was found within the alpha band, the Gender × Training interaction being significant [F_{(1, 35)} = 7.22, MSE = 0.05, p < 0.05]. As seen in Figure 7B, while the female group demonstrated significantly increased intra-hemispheric alpha coherence [t₍₁₆₎ = −2.02, p < 0.05], this was significantly decreased in the male group [t₍₁₉₎ = 1.76, p < 0.05]. No such interaction was found for intra-hemispheric gamma coherence.

The second set of three ANOVAs yielded a significant Gender × Training interaction for both theta [F_{(1, 35)} = 11.05, MSE = 0.02, p < 0.005] and alpha coherence [F_{(1, 35)} = 5.92, MSE = 0.06, p < 0.05]. As seen in Figures 7C,D, while the female group demonstrated significantly increased inter-hemispheric theta coherence [t₍₁₆₎ = −2.46, p < 0.05], theta and alpha coherence significantly decreased in the male group [t₍₁₉₎ = 2.74, 1.81, respectively, p < 0.05]. Post-hoc analysis revealed a significant increase in temporal alpha coherence in the females who did not improve having low HFT performance at baseline [t₍₄₎ = −4.87, p < 0.01], while an opposite pattern was observed in males. More specifically, frontal alpha coherence significantly decreased in the males who improved [t₍₆₎ = 2.84, p < 0.05]. Although not significant, a similar trend occurred for intra-hemispheric alpha. In addition, a significant positive correlation was found between change in theta and alpha coherence, both for intra- and inter-hemispheric connections (r = 0.75, and 0.79, respectively, p < 0.0001, n = 37), in accordance with previous reports (Sauseng et al., 2002; van Albada and Robinson, 2013).

In addition, a significant Gender × Training × Electrode Pair interaction was found for inter-hemispheric gamma coherence [F_{(3, 105)} = 3.01, MSE = 0.01, p < 0.05]. As seen in Figure 8, only the female group demonstrated post-training increased inter-hemispheric bilateral temporal gamma coherence [t₍₁₆₎ = −2.16, p < 0.05].

Marginally significant differences were found between the female and male groups at baseline for inter- and intra-hemispheric gamma coherence [t₍₃₅₎ = 1.97, 1.96, p = 0.06], but not for theta or alpha. Following QMT, a significant difference was found for intra-hemispheric theta and alpha [t₍₃₅₎ = 2.80, 2.79, p = 0.008]. A similar difference was found for inter-hemispheric theta and alpha [t₍₃₅₎ = 2.36, 3.38, p < 0.05] but not for gamma coherence.

To summarize Study 2, following training, reflectivity improved for 32% of the participants. Importantly, while a similar trend of improved reflectivity was observed in both genders, there was a differential effect in EEG coherence for males and females. More specifically, temporal alpha increased in the females, while an opposite pattern was observed in males. In addition, the female group demonstrated post-training increased inter-hemispheric bilateral temporal gamma coherence

Showing a similar pattern as the differential gender trait differences, QMT-induced electrophysiological changes included significantly increased theta and alpha coherence in females, in support of our hypothesis, while the opposite was found for males. The increased theta and alpha coherence observed in the female group is consistent with studies examining electrophysiological changes following meditation. Meditative practices, irrespective of gender, have consistently been related to increased coherence, especially inter- and intra-hemispheric theta and alpha coherence (Alexander et al., 1987; Aftanas and Golocheikine, 2001; Travis, 2001; Travis et al., 2002, 2009; Cahn and Polich, 2006; Baijal and Srinivasan, 2010). Unfortunately, gender-related analyses within meditation reports are scarce, as either gender is not reported at all (e.g., Lutz et al., 2004), or there are no gender-specific baseline differences, hence these are not subsequently considered (Travis et al., 2009), or not mentioned (e.g., Aftanas and Golocheikine, 2001; Travis et al., 2002; Baijal and Srinivasan, 2010). Although gender differences have not yet been systematically investigated in previous electrophysiologically-based meditation studies, some gender differences in behavioral measures were found following meditation. For example, while Qigong meditation was found to contribute positively to addiction treatment outcomes in both genders, female participants reported significantly more reduction in anxiety and withdrawal symptoms than did the male group (Chen et al., 2010). An opposite pattern was observed for TM, for which reduced drinking rates was reported among male university students but not in female students (Haaga et al., 2011). Our finding of QMT-induced gender-related differences in coherence is in accord with another study reporting gender-related electrophysiological differences. In this study (Duregger et al., 2007), a motor related evoked potential (contingent negative variation—CNV) was shown to manifest differently in females, showing higher frontal activation, compared to males, showing higher temporoparietal activation. The more frontal brain area was related to motor preparation reflecting a higher level of cognitive involvement in females compared to males, while the higher male temporoparietal brain activity was interpreted as reflecting brain processes related to other sensory processing (Duregger et al., 2007). Related to this, it is important to keep in mind the gender-dependent differences frequently observed in motor and cognitive realms (Baron-Cohen and Hammer, 1997). More specifically, while women are better in fine-motor coordination (e.g., placing pegs in pegboard holes) men are better in target-directed motor skills, such as guiding or intercepting projectiles (Kimura, 1992). Finally, there are gender differences in brain morphology in relation to spatial cognition and mental rotation performance: while females have more gray matter in the parietal lobe than males, males have greater parietal surface area (Koscik et al., 2009). Koscik et al. (2009) further concluded that the structural gender difference could be a neurobiological substrate for the gender difference in mental rotation performance. In light of the above studies, showing gender-dependent strategies and neural structure, the current results of gender-dependent electrophysiological change following training could be interpreted as reflecting different strategies to achieve the same cognitive and motor goal.

The gender-dependent results may further be explained by the neuronal efficiency hypothesis, which claims that better performance is associated with more efficient brain functioning, that is to say, decreased cortical activation (Haier et al., 1988), indicated by lower event-related desynchronization (ERD). This hypothesis was examined in different tasks, such as spatial cognition (Neubauer et al., 2002; Grabner et al., 2006). One study examining gender-dependent changes in neuronal efficiency and the effects of training on a mental rotation task (Neubauer et al., 2010), reported that while a general increase in neural efficiency from pre to post-test was found for males, it was observed in females only in the 3D version and not in the 2D presentation of the task. In addition, previous studies have also shown that males and females display neuronal efficiency and an inverse IQ–activation relationship in just that domain in which they usually perform better, i.e., females in the verbal domain, and males in the visuospatial domain (Neubauer et al., 2005). Important to the current results, it was further suggested that neural efficiency in males is reflected in local cortical activation (as measured with the ERD-method), whereas neural efficiency in females is manifested in the functional coupling of several brain areas, as assessed by EEG coherence (Neubauer and Fink, 2003). Following this line of thought, training should lead to an increase in neural efficiency, namely, to a decrease in cortical activation from pre- to post-test (Kelly and Garavan, 2005; Neubauer and Fink, 2009; Neubauer et al., 2010) in a gender-dependent manner. Thus, the QMT-induced electrophysiological changes may amplify resting state gender trait differences.