The experiment included a pretest (session 1), 4 learning sessions and a post-test (session 6) as summarized in Table I. During the first session, the students watched an instructional video on PVC insertion, then performed the technique twice under the supervision of 2 trainers that were naive in relation to the objectives of the experiment. Each student inserted the PVC on a double skin placed on the forearm of another student. It was an elastomeric venous system into which simulated blood was introduced through a syringe. The experimenters evaluated the quality of the PVC insertion through the respect of each stage using a common rating scale where they indicated whether the participants respected the afordmentioned steps or not. We also recorded the timing of the PVC insertion.
At the end of the first session, we evaluated the students MI ability with the Motor Imagery Questionnaire (MIQ-3),  and a mental chronometry test (Figure 1). We used the most recent version of the Movement Imagery Questionnaire (MIQ-3), early proposed by Hall and Pongrac (1983)  and later revised by Hall and Martin (1997).  This test consists of 12 elements describing real motor situations to be mentally reproduced. The participants assessed each item through a 7-level Likert scale (maximal score = 84 points) and then performed the mental chronometry test on a A4 printed sheet. The test requested to point towards 8 targets placed on two circles in a predetermined order, at free self-pace. The first 4 pointing towards the targets of the small circle were clockwise, and the next four, on the large circle, counterclockwise. The participants started from the center of the circles and went back to this position, after pointing towards each target (Figure 1). Both served as landmarks for each participant to start and stop the timer during MI. We recorded the timing of the 8 pointing during actual execution. Then, each participant mentally performed the same task. The comparison of real and imagined timing was an index of MI quality. The more MI duration matched actual duration (i.e. isochrony), the better the quality of MI, relative to speed preservation. As both the MIQ-3 and pointing towards targets tests addressed different components of MI quality, Williams et al. (2015) advised to use them together for a more comprehensive assessment of motor imagery ability. 
*** Please Insert Table 1 around here***
*** Please Insert Figure 1 around here***
Patient and public involvement
The participants were forty third-year medical students (23 women, 17 men, mean aged 20.6 ± 1.0) at Claude Bernard University Lyon 1, France. We also requested twenty professional nurses (aged 40.4 ± 4.0) to form the reference group (ref). The experiment did not involve any patient. All participants signed an informed-consent form before starting the study. The management and ethics committee of the Faculty of Medicine (Lyon-Est) approved the experimental design after the experimenters presented the objectives and procedures to the scientific board council. The study also obtained the scientific support of the Center for Education through Simulation in Health (CLESS), hosted by Lyon1 University. We randomly divided the 40 students into two homogeneous groups, the experimental (exp - 13 women and 7 men, mean age = 20.5, ± 1) and the control (ctrl - 10 women and 10 men, mean age = 20.7 ± 0.7), according to their gender, MI abilities and early performance in PVC insertion. We based the exp group training on the association of actual practice (AR) and motor imagery (MI). We requested the ctrl group to perform actual practice with a neutral task (looking at videos about medical care and articles related to ethics and palliative care) for equivalent duration as to the exp group. Although we kept a link with clinical care, the neutral task avoided any relation with MI. Thus, the exp and the ctrl groups trained during 4 successive sessions after the pre-test, performed the post-test and then, the retention test. The overall design was spread over 8 weeks to distribute the learning sequences across time. The ref group only performed the pre-test and the post-test.
We requested the exp group to alternatively perform actual practice and MI according to the experimental design we scheduled, i.e. increasing the rate of MI trials across sessions. The whole workload in the exp group was 36 trials (18 actual and 18 MI).
We read the MI script once for the first MI trial when starting each session as follow: i) Isolate yourself from the environment by closing your eyes and disregard the possible sound environment; ii) Represent all steps of the PVC insertion as an actor, i.e. as if you had to perform the movement yourself; iii) Use sensory information as a support for the construction of MI, mainly tactile (contact with patient's skin and all materials), visual (vein tracking, angle control between the needle and the skin) and proprioceptive (muscle effort, joint position, resistance offered by the skin to the needle insertion ...); iv) Stay motionless during MI, without mimicking the action; v) Start the sequence when spreading and stretching the skin; vi) Stop the action when you place the mandrel into the needle collector.
The MI script clearly described the main steps of the sequence, with the key information for a skilled execution.We thus gave several advices as follow: “When you represent the action, visualize all movements as parts of the action. To do this, close your eyes and try to clearly perceive all the steps of the PVC insertion, as if you were actually performing the movement. You should not move or mimic the movement. Place yourself in relation to the patient, as if you were going to actually make the needle insertion.
- Spread and stretch the skin with your left hand. Locate the exact place where you need to insert the needle.
- Check the needle-skin angle the before insertion.
- Insert the needle into the vein by exerting the proper force, try to perceive the resistance feedback of the tissues the needle crosses.
- When in the vein lumen, stabilize the needle.
- Insert the catheter, bevel up. Check that the angle is correct, apply the appropriate force, perceive the resistance that the tissues oppose to you.
- Force the passage of the catheter into the vein lumen.
- Monitor the arrival of blood in the mandrel.
- Slowly slide the catheter over the needle to position it in the vein.
- Loosen the tourniquet with one hand and hold the catheter with the other.
- Place the mandrel into the needle collector.
Since PVC insertion is a complex procedure, we determined two accurate boundaries separating starting (“spreading and stretching the skin …”) from ending (“placing the mandrel into the needle collector”). We therefore studied a specific part of the whole PVC insertion so that the MI script can be easily understood and memorized, with both accurate timing during actual execution and MI.
The duration of actual PVC insertion was an index of skilled action. We thus timed the last actual trial of each participant, at the end of each session in the exp and ctrl groups. We also timed MI accuracy from sessions 2 to 5, by comparing actual duration to that of the imagined action. The exp group triggered the timer when starting the MI sequence and stopped it at the end. We also assessed the vividness of MI on a 7-level Likert scale, from very blurred image (level 1) to a ultra-high definition image (level 7), intermediate levels being used for intermediate vividness.
Two expert instructors evaluated the quality of the PVC insertion in the exp and the ctrl groups. They were naive with respect to the objectives of the study and uninformed of the participants’ home group. They used a specific evaluation grid from the CLESS pedagogical team, made of 5 items that should be considered “reached” or “not reached” and completed at the end of each session. We assessed the quality of the PVC insertion during the last physical trial of each session, and gave a feedback to each participant to help them better memorizing the procedure. This grid served to guide the participants with a formative objective only and was not included into the statistical design. Only the actual duration was the dependent variable in the ref group. We timed a real trial to be compared with that of the exp and ctrl groups.
The participants completed the retention test one month after the post-test. Each participant only performed one real trial and, additionnaly, those of the exp group performed one MI trial. We evaluated the quality of the PVC insertion similarly as during previous tests (actual and MI duration).
We processed data of the last real trial of each session with a repeated-measures ANOVA. We then compared performance evolution of the exp group with that of both the ctrl and the ref groups. We also performed a repeated-measures ANOVA comparing the actual duration to that of MI in the exp group. We used Tukey's post-hoc tests for two-by-two comparison, provided that the analysis of variance reached the statistical threshold, set at .05.