Following ethical approval and after having obtained informed written consent, 12 surgical residents agreed to participate (median age = 31.5 years, 4 females) (Table 1). All subjects were screened for handedness and neuropsychiatric illness (n = 0) and were asked to refrain from alcohol and caffeine intake for 24 h prior to participation.

Participants were asked to perform a laparoscopic suturing (LS) task using an intracorporeal technique on a laparoscopic box trainer (iSim2, iSurgicals, UK). The task involved inserting a 2–0 Vicryl^® suture (Ethicon, Somerville, NJ, USA) as close to pre-marked entry and exit points on either side of a defect in a Penrose drain. To tie a knot laparoscopically, participants were instructed to formulate one double throw followed by two single throws of the suture (Figure 1).

A block design experiment was conducted in which all participants initially performed the knot-tying task three times (baseline), with an inter-trial rest period of 30-s. Subjects were then randomized into intervention (mental rehearsal [MR]) or control (textbook reading [TR]) groups using a random number generator. Following either MR or TR, participants tied another three knots as described above (Figure 2).

Upon completion of the baseline task, participants in the control group read extracts from a paper which outlined the technical steps of the laparoscopic knot-tying task (Croce and Olmi, 2000). The subjects were asked to read the extracts for 90 s and repeated this exercise 5 times.

Each participant in the mental rehearsal group listened to a 90-s pre-recorded MR script (Supplementary Text S1). In order to create the script, instructional videos identified the key steps in performing LS as performed by consultant surgeons. Sensory and kinaesthetic sensory cues were identified and included in the script. The script contained not only a list of procedural steps for laparoscopic suturing, but also a vivid description of associated imagery cues to enhance the representation of the task in the subject’s mind and enable them to mentally experience the procedure. Having been reviewed by a consultant surgeon and three trainees, the participants listened to the script five times and were instructed to imagine the steps of the procedure and the associated feelings and sensations, whilst refraining from any physical movement.

One experimenter (RP) was responsible for allocating participants to the MR or TR groups and was therefore not blinded to group allocation. Those involved in data collection and analysis (HNM, MO, and HS) were blinded to group allocation. Furthermore, as participants in both groups were required to perform an activity during the training period (i.e., mentally rehearse a task or read a passage of text), they were unaware as to whether they were allocated to the experimental or control group.

Statistical analysis was performed using SPSS version 23.0 (IBM Corp., Armonk, NY, USA). A threshold p < 0.05 was set as the threshold for statistical significance.

For each group, pre- and post-intervention channel activation was confirmed by comparing the average baseline rest HbO₂ data sampled over 10 s before task onset (HbO_2Rest) with average task HbO₂ data sampled over 110 s starting 10 s after task onset (HbO_2Task) using the Wilcoxon Signed Ranks test. Channels displaying a statistically significant (p < 0.05) increase in HbO₂ were considered activated.

For each channel and each hemoglobin species, a variable ΔHbO₂ was computed as follows:

For each group, HbO₂ in each channel was compared pre- and post-intervention using the Wilcoxon Signed Ranks test. Similarly, ΔHbO₂ in each channel was compared between groups in both pre- and post-intervention suturing sessions using the Mann–Whitney U test.

In this study, mental rehearsal was found to significantly improve performance on a laparoscopic suturing task. This is congruent with findings from other studies which have also demonstrate enhanced surgical performance following mental rehearsal training (Arora et al., 2011; Cocks et al., 2014; Komesu et al., 2009; Bathalon et al., 2005; Sanders et al., 2008; Immenroth et al., 2007). For example, in study by Sanders et al. (2008) novice surgeons were randomized to receive either further mental rehearsal or engage in textbook reading after which they were asked to open and close a wound on a live anesthetised rabbit (Sanders et al., 2008). The authors found that mental rehearsal was associated with improved technical performance and appeared to facilitate skills transfer from practice to actual surgery better than textbook reading (Sanders et al., 2008). Others have investigated the effects of mental rehearsal on more advanced procedures. For example, Immenroth et al. (2007) randomized surgical trainees into ‘no training,’ ‘practical training,’ or ‘mental rehearsal’ groups (Immenroth et al., 2007). Assessment of technical performance on a simulated laparoscopic cholecystectomy after the assigned intervention showed significant performance gains in the mental rehearsal group, but not in the no training or practical training groups (Immenroth et al., 2007). Similarly, a randomized controlled trial by Arora et al. (2011) demonstrated that surgical novices who received mental rehearsal training in addition to physical practice obtained significantly higher Objective Structured Assessment of Technical Skills (OSATS) scores on a simulated laparoscopic cholecystectomy task compared with those who undertook physical practice alone (Arora et al., 2011). Improvements in technical performance with mental rehearsal have also been observed during cystoscopy (Komesu et al., 2009), crycothyroidotomy (Bathalon et al., 2005) and endovascular surgery (Patel et al., 2012).

Interestingly, certain studies fail to show performance benefit of mental rehearsal during a range of surgical procedures such as laparoscopic suturing and knot-tying (Donnon et al., 2005; Jungmann et al., 2011), pattern cutting (Mulla et al., 2012), carotid endarterectomy (Wetzel et al., 2011), and hysterectomy (Geoffrion et al., 2012). However, these studies have several limitations which may explain the apparent ineffectiveness of MR training interventions. Firstly, none of the studies used any form of ‘manipulation checks’ such as imagery diary exercises or post-intervention interviews that explore participants imagery experience and ensure that subjects are compliant with the mental imagery script (Sevdalis et al., 2013). Indeed, Geoffrion et al. (2012) admit that subjects in the textbook reading arm of the study may have unknowingly mentally rehearsed the procedure as they were reading the relevant textbook chapters (Geoffrion et al., 2012). Secondly, it appears that subjects’ imagery ability was not assessed in any of these studies. This is an important consideration in order to control for differences in imagery ability between control and intervention groups (Sevdalis et al., 2013), and can be achieved by using, for example, the mental imagery questionnaire (MIQ) which can determine individuals’ ability to generate and control images in their mind (Arora et al., 2011). Without manipulation checks or assessment of imagery ability, the internal validity of the results of any mental rehearsal study may be called into question. Finally, there was a prolonged time interval between mental rehearsal and task execution in some of the studies. Evidence suggests that mental rehearsal is most effective when carried out no more than 24 h prior to physical task performance (Sapien and Rogers, 2012). However, in some studies the time lag between mental rehearsal and physical task performance was up to 1 week (Donnon et al., 2005; Mulla et al., 2012), whereas in others the timing was unclear (Jungmann et al., 2011; Wetzel et al., 2011), thereby mitigating any positive effect mental rehearsal may have had on task execution. These potential methodological shortcomings were addressed in the current study. For example, MIQ scores in the mental rehearsal and textbook reading groups were found to be equal at baseline, and the intervention was received immediately prior to physical task execution.

There was no difference between groups in terms of subjective workload or stress in the current study. This finding is mirrored in other studies which also failed to show that mental rehearsal attenuates stress in surgeons. For example, Wetzel et al. (2011) demonstrated no difference in objective or subjective stress between mental rehearsal and control groups when using a simulated carotid endarterectomy model. Instead, subjects in the mental rehearsal group were found to be using more stress-coping strategies (Wetzel et al., 2011). Similarly, mental rehearsal had no effect on self-reported stress among novice surgeons during a basic surgical skills task (Sanders et al., 2008). The lack of effect of MR on workload and stress observed in the current study would suggest that gains in technical performance observed with MR may be due to its direct impact on surgeons’ learning curves and the underlying neural processes rather than an indirect consequence of reducing cognitive workload.

In the current study, mental rehearsal led to attenuated responses in the prefrontal cortex as well as improved technical performance during laparoscopic suturing, whereas no such changes in activation were observed with textbook reading.

There is wealth of neuroimaging literature highlighting the importance of the prefrontal cortex for novel skill acquisition in surgery (Leff et al., 2007; Leff et al., 2008; Leff et al., 2008; Nemani et al., 2015; Leff et al., 2008; Ohuchida et al., 2009; Shetty et al., 2016; Khoe et al., 2020). For example, Leff et al. (2008) used fNIRS to investigate the neurocognitive mechanisms of task-related expertise in 62 surgeons of varying experience performing five trials of a bimanual open knot-tying task. Technical skill was assessed objectively using time on task, number of movements and instrument pathlength. Whist experienced subjects demonstrated stable technical performance and minimal fluctuation in PFC activity across all five trials, novice surgeons showed significant performance improvement and prefrontal attenuation suggesting that practice-related acquisition of a novel task is associated with a decrease in prefrontal demands (Leff et al., 2008). Similar results have been observed with laparoscopic procedures. Khoe et al. (2020) used fNIRS to examine the variation in PFC activation before and after a laparoscopic training workshop. Novice medical students were randomized into trained and untrained groups. All subjects were shown a basic tutorial video, following which the trained group received additional one-on-one training. Activation responses in the left PFC among the trained group were significantly less than those observed in the untrained group (Khoe et al., 2020). Similarly, Nemani et al. (2018) sought to objectively differentiate surgical skill by assessing patterns of cortical activation during a laparoscopic pattern cutting exercise. Attending surgeons demonstrated decreased PFC activity compared with surgical residents when performing the task on a physical trainer (Nemani et al., 2018).

These studies suggest that in the early phases of motor learning when performance is unrefined and attention-demanding, there is greater recruitment of prefrontal regions (Geoffrion et al., 2012). However, as expertise develops and skills become more automated less demands are placed on executive centers and prefrontal activity diminishes (Leff et al., 2007; Leff et al., 2011). Therefore, the attenuated prefrontal responses that accompanied technical skill improvement in the mental rehearsal group in the current study, may suggest that MR accelerates skill acquisition and expertise development in surgeons at both a motor and cognitive level. Given that mental rehearsal has been shown to activate similar areas of the brain as physical task execution (Moran, 2009; Jeannerod, 2001; Jeannerod, 1995; Kuhtz-Buschbeck et al., 2003; Gerardin et al., 2000; Saiote et al., 2016; Hardwick et al., 2018), it could be hypothesized that MR helps encode a skill on a cognitive level, strengthens central representations of the skill, and facilitates automaticity the same way that physical practice would be expected to do (Sevdalis et al., 2013).

In addition to attenuated prefrontal responses, mental rehearsal led to a spatially more localized increase in activation in the left motor cortex compared with textbook reading—indicative of greater neural efficiency. Given that all subjects in this study were right hand dominant, left lateralisation of motor activity in the mental rehearsal group is likely to genuinely reflect more efficient motor representation during task execution.

There is evidence to suggest that similar neuroplastic changes occur during skills learning. For example, Nemani et al. (2018) used fNIRS to capture cortical haemodynamic responses over the PFC, SMA and M1 to classify surgical expertise during a laparoscopic pattern cutting task. Novice surgeons were shown to have significantly greater activation in the PFC and less activation in the left medial M1 and SMA compared with expert surgeons (Nemani et al., 2018). These findings have been confirmed by longitudinal studies which have delineated changes in motor cortical activity as expertise develops over time (Floyer-Lea and Matthews, 2005; Karni et al., 1995; Ma et al., 2010). Floyer-Lea and Matthews (2005) sought to identify changes in cortical activation that occur with short term (fast) learning during which performance improves rapidly, and long term (slow) learning during which performance gains are incremental. Using fMRI, the authors showed that during short-term learning of an isometric force task, activity in the PFC and primary motor cortex decreased, whereas with long term learning (3 weeks), increased activity was observed in the primary motor cortex (Floyer-Lea and Matthews, 2005). In line with these findings, Karni et al. (1995) trained subjects in a motor sequence task for 3 weeks and found a progressive increase in activation of the primary motor cortex. These findings suggest that over time, there is an enlargement of the motor cortical representation a learned skill, which may underlie long-term skill retention and enable finer motor control (Ungerleider et al., 2002; Dayan and Cohen, 2011).

Neural efficiency refers to patterns of more spatial localized or less intense brain activity when performing a motor skill (Neubauer and Fink, 2009). Many studies have shown that brain activation in expert athletes is more spatially localized compared with non-experts. For example, Chang et al. (2011) compared the activation maps of elite archers and non-archers during mental rehearsal using fMRI. In non-archers, a wide area of activation was observed and included premotor, SMA, inferior frontal region, basal ganglia and cerebellum. In contrast, activation was localized to only the SMA in expert archers (Chang et al., 2011). Similarly, Milton et al. (2007) used fMRI to compare brain activation during the pre-shot routine of novice and expert golfers and found a more focused and efficient organization of task-related neural networks among expert golfers. In contrast, novices exhibited a broader area of activation implying a difficulty in filtering out task-irrelevant information (Milton et al., 2007).

In sum, motor skill learning is associated with a redistribution of cortical activity from anterior to posterior regions (Neubauer and Fink, 2009; Kelly and Garavan, 2005). Specifically, as expertise develops prefrontal cortical activity attenuates, interpreted as less consumption of attentional resources as a skill is internalized and becomes automated (Leff et al., 2011). Concurrently, an increase in activity in motor regions is observed suggesting greater recruitment of additional motor units into a local network that represents that acquired skill (Ungerleider et al., 2002). Furthermore, this increase in activity in motor regions is more spatially localized to task-relevant cortical regions suggesting a greater degree of neural efficiency as learning progresses. In the current study, mental rehearsal led to similar neuroplastic changes in prefrontal (decreased activity) and motor regions (increased and spatially more localized activity), which suggests that mental rehearsal facilitates skill acquisition by accelerating the neuroplastic changes that accompany skills learning.

There are several limitations to the current study which need to be acknowledged. Firstly, the number of participants was relatively low which increases the risk of type 1 and/or type 2 errors. However, a sample size estimation was not feasible as there have been no previous studies comparing mental rehearsal and textbook reading in a surgical setting in order to perform a pre-hoc power calculation. Indeed, studies investigating the effect of mental rehearsal on cortical activation in other domains have not incorporated sample size calculations and the cohort size in our study is comparable to these studies (Nyberg et al., 2006; Moriya and Sakatani, 2017). Furthermore, the current work is a hypothesis-generating study, data from which could be used for a sample size calculation for a larger confirmatory study.

Secondly, there may be a selection bias in the recruitment of study participants. Although all general surgical residents within a postgraduate training region were invited to participate, it is possible that only residents who felt confident in their laparoscopic knot-tying ability agreed to enroll. The subspecialty interest of participants was not recorded, and there may have been a disproportionate number of residents with a specialist interest in surgical disciplines in which laparoscopic skills are a fundamental part of training (e.g., upper gastrointestinal surgery). Therefore, the neuroergonomic and performance benefits of mental rehearsal may not be representative of the wider surgical community. Furthermore, generalisability of the study findings to other surgical procedures cannot be assumed. Empirical data is required to ascertain whether mental rehearsal can have similar effects on performance and brain behavior during other types of surgical skill.

Thirdly, whilst mental imagery ability was found to be comparable between the two groups at baseline, manipulation checks to determine whether subjects are adhering to the MR script were not carried out. Future studies can address this shortcoming by conducting post-intervention interviews or collecting qualitative data from imagery diaries.

Fourthly, short separation channel data was not collected which would have provided an indication as to whether the measured cortical activations were contaminated by superficial haemodynamic signals. However, analysis of heart rate data demonstrated that there was no significant change in the systemic physiological response within or between groups which would suggest that the observed cortical haemodynamic signals reflect genuine underlying activation responses.

Finally, we did not investigate whether the effects of mental rehearsal on cortical activation are retained in the long-term. This would be the focus of future work to provide supportive evidence for incorporating mental skills training in the surgical curriculum. Nonetheless, the short-term neuroergonomic benefits of mental rehearsal which this study investigates are equally important to appreciate since, in the real-world setting, surgeons usually mentally visualize the critical steps of the procedure, often with the aid of pre-operative imaging, just prior to starting an operation.