Sideways Camera Rotations of 90° and 135° Result in Poorer Performance of Laparoscopic Tasks for Novices

Abstract

Objective:

The aim of this study was to determine the impact of sideways visuomotor rotations between 0° and 180° on novice performance in a laparoscopic simulator.

Background:

The laparoscopic surgical environment often involves visuomotor rotations because the laparoscope may be placed to the surgeon’s side. Basic research by Cunningham indicated that visuomotor rotations between 90° and 135° result in peak performance decrements. Research by Ames and colleagues failed to replicate Cunningham’s results in the laparoscopic environment, possibly due to (a) confounds from carryover effects or (b) use of an alternative laparoscopic training task rather than the straight-line pointing task used by Cunningham. Two experiments were conducted to determine if Cunningham’s results generalize to the laparoscopic environment when controlling for carryover effects for a three-dimensional “straight-line” pointing task (Experiment 1) and a laparoscopic training task (Experiment 2).

Method:

In Experiments 1 and 2, participants were assigned to one of five visuomotor rotations: 0°, 45°, 90°, 135°, or 180°. Utilizing a laparoscopic simulator, participants performed either a three-dimensional pointing task (Experiment 1) or a peg transfer task (Experiment 2).

Results:

In both experiments, visuomotor rotations of 90° or 135° resulted in the poorest performance.

Conclusion:

When controlling for carryover effects, Cunningham’s results generalize to novices’ performance of a pointing and a peg transfer task in the laparoscopic environment.

Applications:

The results indicate that 90° and 135° sideways laparoscope placements may result in worse performance for novices in the laparoscopic environment, indicating potentially longer learning curves for these conditions in the laparoscopic as well as other teleoperation environments.

Keywords

perceptual-motor adaptation laparoscopy laparoscopic training minimally invasive surgery visuomotor rotations

Introduction

Laparoscopic surgery is a minimally invasive surgical method whereby the surgical site is portrayed on a monitor via a laparoscopic camera that is partially inserted through a small opening in the patient. This technique requires that surgeons manipulate the target tissue using long, thin instruments that are partially inserted into the patient through small openings. This surgical approach provides advantages to patients compared to traditional open surgery, including shorter recovery and hospitalization times as well as lower costs, infection rates, and postoperative complications (Cuschierie, 1995; McGreevy et al., 2003; Varela, Wilson, & Ngygen, 2010). However, proficiency in laparoscopic surgeries is associated with long learning curves (Rattner, 1999; Schauer, Ikramuddin, Hamad, & Gourash, 2003). Long learning curves are a concern because limited proficiency in performing operations has been associated with surgical complications. Surgical issues, which include injuries caused by surgeons (e.g., organ damage), accounted for almost 25% of complications in laparoscopic cholecystectomies (Veen, Bik, Janssen-Heijnen, De Jongh, & Roukema, 2008).

A number of perceptual-motor distortions inherent in the laparoscopic environment may contribute to the long learning curves associated with acquiring laparoscopic skills, such as reduced depth information (Reinhardt-Rutland, Annett, & Gifford, 1999) and the fulcrum effect (Gallagher, McClure, McGuigan, Ritchie, & Sheehy, 1998; Kunde, Müssler, & Heuer, 2007). Most relevant to the present study is the problem of visuomotor rotations due to the laparoscope (camera) placement, which is often controlled by another member of the surgical team. It may be rotated around its long axis and is frequently placed to the left or the right of the surgeon, resulting in visuomotor rotations.

Cunningham (1989) conducted an influential study to assess the impact of visuomotor rotations ranging between 0° and 180° on performance of a “straight-line” pointing task. This task required participants to move an occluded stylus across a two-dimensional plane, which controlled a cursor on a monitor. The most substantial performance impairment occurred for visuomotor rotations between 90° and 135° compared to 0°, 45°, and 180° rotations. These results are consistent with studies using joystick tracking tasks (Kim, Ellis, Tyler, Hannaford, & Stark, 1987; Macedo, Kaber, Endsley, Powanusorn, & Myung, 1998) as well as a straight-line pointing task in a three-dimensional virtual reality environment (Ellis, Adelstein, & Yeom, 2012). Gradual rotation of an internal mapping/reference frame is assumed to account for performatory adaptation for rotations up to 90° (Bock, Abeele, & Eversheim, 2003). In contrast, the relatively good performance at 180° suggests that this particular rotation does not require a gradual rotation of the internal mapping/reference frame; rather, it requires only reversal of the movement (Cunningham, 1989), possibly by the inversion of the axes that are associated with left-right and forward-backward directions (Bock et al., 2003). Although this reversal is assumed to pose relatively low cost to performance, it is higher than performance costs observed for rotations of 0° (Cunningham, 1989). Further, adaptation to rotations ranging between 90° and 180° has been shown to be caused by a movement reversal and a gradual backward rotation to the prescribed distortion (Bock et al., 2003; Cunningham, 1989).

A considerable amount of visuomotor adaptation research has been conducted since the 1800s, but research on visuomotor rotations in the laparoscopic environment has been relatively sparse. Zhang and Cao (2012) assessed the impact of rotating the laparoscope around its long axis by 0°, 45°, 90°, 135°, or 180° at varying laparoscope locations. Notably, only the top-view laparoscope (camera) condition assessed by Zhang and Cao resulted in distortions consistent with those induced by Cunningham (1989). In this condition, the camera was vertically rotated to 90°, providing a top-down view. With the laparoscope in this position, the poorest performance tended to occur when the laparoscope was rotated 135° around its long axis.

Although it is important to assess the impact of laparoscopic camera rotations around its long axis, these distortions are qualitatively different from those induced by rotating the laparoscopic camera to the operator’s side, especially when a camera’s location does not provide a top-down view (camera vertically rotated 90°). A precise understanding of the effect of sideways camera rotations on novice laparoscopic performance is particularly relevant because some procedures require large side rotations of the laparoscope in order to adequately view the surgical field (Ferzli & Fingerhut, 2004). Prior research on novices’ performance in the laparoscopic training environment indicated that a 90° camera rotation to the operator’s side resulted in poorer performance compared to a 0° camera placement (directly in front of the participant) when performing a peg transfer task (Klein et al., 2008; Klein, Riley, Warm, & Matthews, 2005). Similarly, a 135° sideways camera rotation resulted in poorer performance compared to a top camera view in a similar study that also utilized a peg transfer task (Klein, DeLucia, & Olmstead, 2012).

To the authors’ knowledge, Ames, Frisella, Yan, Shulman, and Landman (2006) conducted the only study that systematically assessed performance in a laparoscopic trainer by inducing the five distortions assessed by Cunningham (1989; 0°, 45°, 90°, 135°, and 180°) by varying camera placements to the side of participants. These placements ranged between locations directly in front of the participant (0° rotation) to directly opposite the participant (180° rotation). Participants performed a pick-and-place task using a laparoscopic simulator and were exposed to all five levels of rotations. In contrast to Cunningham, Ames and colleagues did not find better performance for rotations of 180° compared to 90° and 135° rotations. Instead, performance deteriorated as the angle of rotation increased, with the poorest performance at 135° and 180° rotations.

Possible reasons for the discrepancy with Cunningham (1989) are primarily methodological in nature. First, although both studies involved a repeated-measures design whereby each participant was exposed to each of the five visuomotor rotations, Cunningham assessed each participant’s performance at the different rotations on different days. Trempe and Proteau (2010) indicated that aftereffects for visuomotor distortions were persistent after a 24-hr interval; thus carryover effects may potentially be present in Cunningham’s findings. However, this break between assessments may potentially reduce those effects. In comparison, Ames and colleagues’ (2006) description of the Method section implies that each participant’s performance on the different rotation conditions was assessed during the same experimental session, which may have produced significant carryover effects, as no information was provided regarding any breaks between exposure to the different rotations. Previous research indicates that prior exposure to a given visuomotor rotation may impact performance at a novel visuomotor rotation (Bock et al., 2003). Thus, Ames and colleagues’ discrepant results might be due to carryover effects.

Further, a small sample size in Ames and colleagues’ (2006) study did not allow the order of rotations to be fully counterbalanced to manage carryover effects, although this limitation was also an issue for Cunningham’s study. Another issue is that the aforementioned research replicated Cunningham’s (1989) findings using pointing as well as tracking tasks. Thus, the discrepancies in the results observed by Ames and colleagues may be due to a different task (pick-and-place task) being employed. Interestingly, Ames and colleagues’ results, which indicated worse performance as the degree of rotation increased, are akin to findings observed in the mental rotation paradigm.

Mental rotation research indicates increased reaction time with increased degree of rotation when participants determine if two images of three-dimensional line drawings are the same object (Shepard & Metzler, 1971). Thus, different processes may be at work when adapting to sideways camera rotations in a pick-and-place task relative to a straight-line pointing task. Initially, the processes involved in mental rotation appear to explain the results observed by Ames and colleagues (2006); however, it is important to note that motor performance while experiencing visuomotor rotations and judgments required during mental rotation tasks are thought to depend on qualitatively different processes (Cunningham, 1989). Specifically, performing movements while experiencing visuomotor rotations usually involves continuous feedback and allows for online error corrections (Cunningham, 1989); effective adaptation to such distortions appears to depend on implicit processes rather than explicit cognitive strategies (Mazzoni & Krakauer, 2006). In contrast, mental rotation tasks are cognitive in nature, requiring participants to explicitly judge whether two objects are the same (Shepard & Metzler, 1971). Thus, we expect that mental rotation research does not generalize to performance of a laparoscopic pick-and-place task.

The purpose of the current study was to resolve these discrepancies. Two experiments were conducted to assess whether Cunningham’s (1989) results, which indicated worst performance for rotations ranging between 90° and 135°, would generalize to sideways camera rotations in the laparoscopic environment by utilizing a between-groups design to control for carryover effects. Experiment 1 involved a three-dimensional straight-line pointing task similar to the two-dimensional pointing task used by Cunningham. Because the surgical field is three-dimensional, the research team favored a three-dimensional over a two-dimensional pointing task.

The second experiment was conducted to assess whether the discrepancy between Cunningham’s (1989) and Ames and colleagues’ (2006) findings was due to the nature of the task, while still controlling for carryover effects. Experiment 2 involved a peg transfer task, which is a standard laparoscopic training task (Fried et al., 2004). Like the pick-and-place task used by Ames and colleagues, the peg transfer task requires the ability to move the instrument tip from a start to an end position and to pick up and place down objects, whereas the peg transfer task also requires the coordination of tools with both hands. The peg transfer task utilized in the present study is similar to the peg transfer task of the Fundaments of Laparoscopic Surgery program that has been developed by the Society of American Gastrointestinal Endoscopic Surgeons and has been endorsed by the American College of Surgeons. Thus, given the differences between the mental rotation and visuomotor mapping/rotation paradigms, we hypothesize that Cunningham’s (1989) findings generalize to the laparoscopic environment when controlling for carryover effects, even when performing a laparoscopic peg transfer task. For both Experiment 1 and 2, we postulated the following hypotheses:

Hypothesis 1: The 90° and 135° sideways camera rotations as a whole will result in worse performance than each of the other sideways rotations when controlling for carryover effects.

Hypothesis 2: Because prior research has indicated that humans have the ability to adapt to visuomotor rotations (Cunningham, 1989), participants’ performance will improve over the course of the experiment, resulting in superior performance during the last experimental block.

Hypothesis 3: Consistent with Cunningham’s (1989) findings, the 90° and 135° sideways camera rotations as a whole will result in worse performance than each of the other sideways camera rotations during not only the first but also the last experimental block.

Experiment 1

Experiment 1 had been reported by Wheeler, Klein, and Craig (2012). However, the present manuscript reports a dependent variable (movement time) that was not previously published by Wheeler and colleagues.

Method

Participants

Forty students (20 males and 20 females) in an introductory psychology course took part in Experiment 1. Participants were between 18 and 27 years of age (M = 20.17, SD = 2.28). All reported being right-handed, having normal or corrected-to-normal vision, normal hearing, and no history of neuromuscular disorders.

Experimental design

Four males and four females were assigned to each of the five visuomotor rotations employed by Cunningham (1989): 0°, 45°, 90°, 135°, or 180°. These rotations were induced using a laparoscopic simulator. Participants performed eight experimental blocks of 32 straight-line movements. Performance was assessed using trial completion time averaged for each block, resulting in a 5 (rotations) × 8 (experimental blocks) mixed ANOVA design.

Apparatus

A low-fidelity laparoscopic simulator, with a camera that induces the visuomotor rotations typical in the laparoscopic environment, was employed (see Figure 1). The simulator was located on a table and was dome shaped (diameter of base = 30.4 cm, height = 16.0 cm) and had a 13.0 cm × 19.5 cm opening in its top. A rigid plastic strap (width = 2.5 cm) was placed across the opening. This strap had a 1-cm diameter opening in its center, which served as an entry port (trocar) for the grasper (see Figure 1). This configuration allowed a direct view of the target area located inside of the apparatus during the baseline blocks. The target area was located at the center of the apparatus. It consisted of a raised target that was located 35 mm above and parallel to the base. This raised target was created by extending a dowel from the side of the apparatus; its end reached the center of the target area and served as one of the targets. Four additional base targets were located on the base of the apparatus in a semicircle (35-mm radius). Each base target (6-mm diameter) was a different color, allowing the experimenter to easily describe them. The first base target was black and was located to the left of the dowel. The second was red and was located 45° counterclockwise from the black target. The third was yellow and was located 90° counterclockwise from the red target. The fourth was green and was located 45° counterclockwise from the yellow target. The larger opening between the red and yellow targets allowed all targets to be visible during all movements because the camera (LifeCam HD 5001) was located opposite to the target area, directly facing the dowel’s tip.

Figure 1.

Apparatus utilized in Experiment 1. (A) Top view of the exterior of the apparatus. (B) Top view of the interior of the apparatus. (C) Side view of the interior of the apparatus. (D) View of the interior of the apparatus as shown on the monitor.

An X mark, which was located 45° counterclockwise from the red target, served as the starting location for each block. The distance between each colored base target and the target area of the raised dowel was 49.5 mm. The camera was rotated 30° vertically from the base of the apparatus and projected the target area onto a 19-inch Karl Storz monitor (Tuttlingen, Germany). The monitor was located behind the apparatus, facing participants approximately at eye height. A model laparoscopic grasper was placed through the port located in the plastic strap that stretched across the opening in the top of the simulator. The model grasper had the same dimensions as a 5-mm Ethicon grasper (Ethicon Endo-Surgery, Inc., Somerville, NJ), with the only exception that the tip of the model grasper was pointed. Two micro-electromagnetic sensors (9.9 mm long × 2 mm diameter) were embedded in the center of the instrument shaft and connected to a motion tracker (Ascension Technology Corporation, Burlington, VT). These sensors provided two three-dimensional coordinates, allowing for the extrapolation of the three-dimensional coordinate of the grasper’s tip.

The five visuomotor rotations (0°, 45°, 90°, 135°, and 180°) were induced by rotating the entire apparatus along with the camera and target area counterclockwise, while the monitor and table remained stationary. A forward movement of the tip of the laparoscopic instrument inside of the apparatus in the 0°, 45°, 90°, 135°, and 180° rotations was viewed on the monitor as a forward, forward-right, left-right, backward-right, and backward movement, respectively. Rotating the camera and target area along with the apparatus allowed the five visuomotor rotation conditions to have an identical view of the target area on the monitor. This setup allowed for the control of potential confounds associated with the different views of the target area (DeLucia & Griswold, 2011).

Procedure

Prior to the arrival of the participant, the experimenter rotated the apparatus to 0°, 45°, 90°, 135°, or 180°, depending on the participant’s condition. Upon arrival, all participants signed a consent form, confirmed that they were right-handed, had normal hearing and normal or corrected-to-normal vision, and did not have a history of neuromuscular disorders. Then, participants were provided with verbal instructions, which were followed by participants completing two baseline (direct view) blocks (Blocks 1 and 2) and eight experimental blocks (Blocks 3 to 10; an opaque cover was placed over the central opening to prevent direct view of the target area). Each block consisted of 32 trials. At the beginning of each block, the grasper tip was placed on the X on the simulator’s base by the experimenter. The first trial of each block consisted of moving from the X to the raised target (the dowel’s target area). The subsequent trials consisted of movements between the dowel’s target area and one of the four base targets or vice versa. The experimenter said, “Dowel,” as soon as participants reached one of the base targets and said the color of one of the four base targets as soon as the participant reached the dowel. The order of the movements to the base targets was pseudorandom (participants visited each target four times during each block). Therefore, participants could not reliably predict which base target they were moving to during their next trial.

Results

Performance was analyzed using average trial movement time, which is reported in seconds and was measured at a resolution of 1/1000 of a second. Movement time was defined as the total time from when the grasper’s extrapolated tip left the starting target area’s coordinates until it reached the coordinates of the goal target area. Using Matlab (MathWorks, Inc.), we determined movement time post hoc using the grasper’s and targets’ three-dimensional coordinates as well as the time stamp recorded by the motion tracker.

A 5 (rotations) × 2 (baseline blocks) mixed ANOVA was conducted to determine if the different apparatus rotations impacted baseline performance. The five rotation conditions did not differ during the baseline (direct view) blocks, as indicated by a nonsignificant rotations main effect, F(4, 35) = 1.02, p = .41, as well as a nonsignificant Rotations × Baseline Blocks interaction, F(4, 35) = .71, p = .59.

To test our predictions, a 5 (rotations) × 8 (experimental blocks) mixed ANVOA was computed. Due to violations of the sphericity assumption, the ANOVA was analyzed using the Greenhouse-Geisser correction (Field, 2005). This analysis indicated a significant rotations main effect, F(4, 35) = 7.29, p < .001, η_p² = .45; a significant experimental blocks main effect, F(1.51, 52.91) = 38.76, p < .001, η_p² = .53; and a significant Rotations × Experimental Blocks interaction, F(6.05, 52.91) = 3.46, p = .006, η_p² = .28.

As seen in Figure 2, participants in the 90° and 135° rotations exhibited the longest average trial movement times. To test Hypothesis 1 (90° and 135° rotations as a whole result in worse performance than each of the other rotations), the rotations main effect was further analyzed using three Bonferroni corrected one-tailed linear contrasts (Keppel & Wickens, 2004). Type I error rate was set at α = .0167. These contrasts (see Table 1) indicated that the 90° and 135° rotations as a whole resulted in poorer performance compared to the 0°, 45°, and 180° rotations.

Figure 2.

Mean trial completion time collapsed across experimental blocks for each rotation condition. Error bars show ±1 standard error of the mean.

Table 1:

Linear Contrasts Comparing the 90° and 135° Rotations as a Whole (M = 6.30, SE = 0.45) With Each of the Other Rotation Conditions

Rotation	Contrast Value	t(35)	p Value (one tailed)
0° (M = 3.59, SE = 0.21)	2.71	4.78	<.001*
45° (M = 4.49, SE = 0.21)	1.80	3.18	.0015*
180° (M = 4.36, SE = 0.36)	1.94	3.42	<.001*

Note. Performance was averaged over the experimental blocks. One-tailed t-tests were computed; significance (using the Bonferroni correction) is indicated by asterisks. SE refers to the standard error of the mean.

As can be seen in Table 2, performance continuously improved from the first experimental block (Block 3) to the last experimental block (Block 10). To test Hypothesis 2 (performance is superior in the last experimental block, i.e., Block 10), the blocks main effect was further analyzed using one-tailed repeated-measures t-tests to assess if performance in Block 10 was superior compared to each of the other experimental blocks. To correct for Type I error inflation, these t-tests were analyzed using the Bonferroni correction (Type I error rate per comparison, α = .007). As can be seen in Table 2, performance observed in Block 10 was superior to that observed in each of the other experimental blocks, and the better performance observed over blocks appears to be primarily due to adaptation effects occurring in the 90°, 135°, and 180° rotations (see Figure 3).

Table 2:

Comparison of the Average Trial Completion Time at Block 10 (M = 3.54 s, SE = .12 s) With That Observed at the Other Seven Experimental Blocks

Block	t(39)	SE _Diff	p Value (one tailed)
Block 3 (M = 8.75 s, SE = .89 s)	−6.23	0.84	<.001*
Block 4 (M = 6.01 s, SE = .47 s)	−6.30	0.39	<.001*
Block 5 (M = 5.19 s, SE = .30 s)	−7.53	0.22	<.001*
Block 6 (M = 4.59 s, SE = .19 s)	−8.57	0.12	<.001*
Block 7 (M = 4.24 s, SE = .18 s)	−6.02	0.12	<.001*
Block 8 (M = 3.98 s, SE = .15 s)	−4.80	0.09	<.001*
Block 9 (M = 3.75 s, SE = .13 s)	−3.04	0.07	.002*

Note. Performance was averaged over rotations; one-tailed t-tests were computed. Significance (using the Bonferroni correction) is indicated with asterisks. SE_Diff refers to the standard error of the difference scores between Block 10 and the given experimental block, whereas SE refers to standard error of the mean.

Figure 3.

Performance observed for each rotation condition at every experimental block. Error bars are ±1 standard error of the mean, allowing to compare rotations within each block. Standard errors relevant for assessing repeated-measures effects (changes across blocks) are not presented, as such error bars would differ for each individual comparison of interest.

Performance at the different rotation conditions observed during the experimental blocks is shown in Figure 3. To test Hypothesis 3 (the 90° and 135° rotations as a whole result in the worst performance not only during the initial experimental block but also during the last experimental block), the Rotations × Experimental Blocks interaction was further analyzed using linear contrasts. Specifically, performance in the 90° and 135° rotations as a whole was compared to each of the other three rotation conditions at the first experimental block (Block 3) as well as the last experimental block (Block 10); see Table 3. The Bonferroni correction was employed to control for Type I error inflation (error rate per comparison, α = .0083). As can be seen in Table 3, performance in the 90° and 135° rotations as a whole was worse than the 0° and 45° rotations at Block 3. Further, tendency of better performance in the 180° rotation compared to the 90° and 135° rotations was observed at Block 3, although this tendency did not reach significance. At Block 10, the 90° and 135° rotations as a whole resulted in poorer performance than that observed at the 0° and 180° rotations. Taken together, performance in the 90° and 135° rotations also had a tendency toward worse performance compared to that observed at the 45° rotation, but this tendency did not reach significance.

Table 3:

Linear Contrasts Comparing the 90° and 135° Rotations as a Whole With Each of the Other Rotations at Block 3 and Block 10

Rotation	Contrast Value	t(35)	p Value (one tailed)
Block 3 (90° and 135° rotation as a whole, M = 12.50, SE = 1.81)
0° (M = 5.05, SE = 0.45)	7.44	3.63	.0005*
45° (M = 6.29, SE = 0.44)	6.21	3.03	.0023*
180° (M = 7.44, SE = 0.80)	5.06	2.46	.0094
Block 10 (90° and 135° rotation as a whole, M = 3.97, SE = 0.19)
0° (M = 2.97, SE = 0.14)	1.00	3.61	.0005*
45° (M = 3.66, SE = 0.20)	0.30	1.09	.1415
180° (M = 3.14, SE = 0.21)	0.83	2.98	.0026*

Note. Significance (using the Bonferroni correction) is indicated with asterisks. SE refers to the standard error of the mean.

Experiment 2

Method

Participants

Forty participants (20 males, 20 females) from an introductory psychology course took part in this study. Participants indicated via self-report that they were right-handed and had normal or corrected-to-normal vision and no history of neuromuscular disorders. Participants were between 18 and 26 years old (M = 19.88, SD = 2.10).

Experimental design

Eight participants (four males and four females) were each assigned to one of the five rotations previously utilized by Cunningham (1989): 0°, 45°, 90°, 135°, and 180°. Consistent with Experiment 1, these rotations were induced using a laparoscopic simulator. Participants performed eight blocks of two peg transfers. Performance was analyzed using the trial completion time and average number of times a transfer item (foam star) was dropped during each transfer. The average of these performance indices was calculated for each block. Therefore, Experiment 2 had a 5 (rotations) × 8 (experimental blocks) mixed ANOVA design. Rotations served as the between-subjects variable.

Apparatus

The apparatus (see Figure 4) was identical to the apparatus used in Experiment 1, with three exceptions. First, two 1-cm ports were located in the plastic strap, one 4 cm to the left and the other 4 cm to the right of its center. Second, rather than a laparoscopic model grasper, two 5-mm Clickline Reddick-Olsen dissecting grasping forceps (Karl Storz, Tuttlingen, Germany) were employed, with one grasper inserted through each of the two ports. Third, the target area consisted of a pegboard. The pegboard comprised a wooden base (8.0 mm high × 18.5 cm wide × 13.0 cm deep) with two pairs of pegs. Each pair of pegs consisted of two 1.5-cm-tall pegs, each with a diameter of 2.0 mm. Each pair of pegs was separated on the horizontal (x) axis by 3 cm. The pairs were arranged on the wooden base, separated by 2 cm in depth, resulting in two rows of two pegs. Two five-pointed foam stars, each with a 6-mm circular opening in the center (distance of each tip to the closest location of the circumference of the central opening was 9 mm, weight of each star = 0.11 g, height = 3 mm) were located each on either the right or the left two pegs. The pegboard was positioned at the center of the simulator so that the first pair of pegs faced toward the camera.

Figure 4.

Apparatus of Experiment 2. (A) Top view of the apparatus’ exterior. (B) Top view of the interior of the apparatus. (C) Side view of the apparatus’ interior.

Identical to Experiment 1, the entire apparatus was rotated counterclockwise to induce the five different visuomotor rotations. However, in Experiment 2, the plastic strap with the two port openings was always kept parallel to the participants irrespective of the given visuomotor rotation, ensuring that participants did not have to rotate their torsos and arms as a function of visuomotor rotation.

Procedure

Consistent with Experiment 1, prior to the participant’s arrival, the experimenter rotated the apparatus to induce the visuomotor rotation conditions to which the participant was randomly assigned. After arrival, participants signed the consent form and confirmed via self-report that they were right-handed and had normal hearing and normal or corrected-to-normal vision as well as no history of neuromuscular disorders. Participants then viewed an instructional video that explained the peg transfer task. At the beginning of each block, the two foam stars were located on either the two left or the two right pegs (this location was randomly determined). Participants picked up the two graspers from the side of the apparatus and inserted them through the two holes in the top of the apparatus. Depending on the experimenter’s instructions, the participant picked up the foam star closest to the camera or farthest from the camera with his or her right or left grasper (the initial star and grasper were randomized and indicated via the stars’ colors; e.g., “Please start with the green star first and use the grasper in your right hand”). Participants were instructed to pass the star midair to the other grasper and place it on the peg opposite its original position. They then completed the same movement with the other star, starting with the same grasper as during the preceding movement. Once they completed both transfers, they removed the graspers from the top of the simulator and placed them to the right and the left of the simulator. This constituted one block of two trials.

In Experiment 2, each block consistent of 2 trials rather than 32 trials because pilot testing indicated that completing two peg transfers took approximately as long as completing 32 trials of three-dimensional pointing movements utilized in Experiment 1. Participants were tested individually in a quiet room. The experimenter counted the number of times the foam stars were dropped during each trial and used a stopwatch to record the trial completion time. The first trial of a block began when the graspers appeared on the monitor. Each trial ended when the foam star was placed on a peg and touched the base of the apparatus.

Participants completed the first two blocks while viewing the pegboard directly through the opening in the top of the simulator. Then, they performed eight experimental blocks, which required the viewing of the pegboard on the monitor. During the experimental blocks, an opaque cover prevented participants from directly viewing the pegboard.

Results

Performance was analyzed using the average trial movement time and average drops per transfer observed in each block. A 5 (rotations) × 2 (baseline blocks) mixed ANOVA was computed for both performance indices with rotations serving as the between-groups variable and baseline blocks as the repeated-measures variable. The analysis of trial movement time indicated that baseline performance did not differ for participants assigned to the five different visuomotor rotations, as indicated by a nonsignificant rotations main effect, F(4, 35) = .82, p = .52, as well as a nonsignificant Rotations × Baseline Blocks interaction, F(4, 35) = .14, p = .97. Similarly, the analysis of the average drops per transfer also indicated that the baseline performance did not differ for participants assigned to the five rotations, as indicated by a nonsignificant rotations main effect, F(4, 35) = 0.80, p = .53, and a nonsignificant Rotations × Baseline Blocks interaction, F(4, 35) = 0.36, p = .84.

To tests our predictions for Experiment 2, a 5 (rotations) × 8 (experimental blocks) mixed ANOVA was calculated for average trial time and drops per transfer; rotations served as the between-groups variable and blocks as the repeated-measures variable. The Greenhouse-Geisser correction was used to correct for Type I error inflation caused by violations of the sphericity assumption.

Trial movement time

There were significant main effects for rotations, F(4, 35) = 18.25, p < .001, η_p² = .68, and experimental blocks, F(2.89, 101.22) = 27.41, p < .001, η_p² = .44, as well as a significant Rotations × Experimental Blocks interaction, F(11.57, 101.22) = 4.54, p < .001, η_p² = .34.

To test Hypothesis 1 (90° and 135° rotations as a whole result in worse performance than the other three rotation conditions), the rotations main effect (see Figure 5) was further analyzed using three Bonferroni-corrected one-tailed linear contrasts (error rate per comparison was set at α = .017). As can be seen in Table 4, performance in the 90° and 135° rotations as a whole was worse than that observed in the other rotations.

Figure 5.

Average trial completion time (averaged over experimental blocks). Error bars are ±1 standard error of the mean.

Table 4:

Linear Contrasts Comparing the 90° and 135° Rotations as a Whole (M = 57.35, SE = 3.95) With Each of the Other Rotations

Rotation	Contrast Value	t(35)	p Value (one tailed)
0° (M = 19.63, SE = 1.07)	37.72	7.51	<.001*
45° (M = 26.62, SE = 2.46)	30.73	6.12	<.001*
180° (M = 36.15, SE = 3.56)	21.19	4.22	<.001*

Note. Performance was averaged over the experimental blocks. Significance (using the Bonferroni correction) is indicated with asterisks. SE refers to the standard error of the mean.

To test Hypothesis 2 (performance improves over experimental blocks, resulting in superior performance in the last experimental block), the experimental blocks main effect was further analyzed using seven one-tailed Bonferroni-corrected repeated-measures t-tests to compare performance in Block 10 with that observed at each of the other experimental blocks. The error rate per comparison was set at α = .007. As can be seen in Table 5, performance in Block 10 was superior to that observed in Blocks 3 through 5, and this superiority is mostly due to adaptation occurring in the 90°, 135°, and 180° rotations (see Figure 6).

Table 5:

Comparison of the Average Trial Completion Time (sec) at Block 10 (M = 28.89, SE = 2.49) With Those Observed During the Other Experimental Blocks.

Block	t(39)	SE _Diff	p Value (one tailed)
Block 3 (M = 74.83, SE = 9.55)	−5.82	7.90	<.001*
Block 4 (M = 47.54, SE = 4.36)	−4.55	4.10	<.001*
Block 5 (M = 40.51, SE = 3.73)	−4.07	2.86	.001*
Block 6 (M = 34.92, SE = 3.28)	−2.07	2.91	.023
Block 7 (M = 33.02, SE = 3.58)	−1.25	3.31	.110
Block 8 (M = 27.95, SE = 1.79)	0.38	2.49	.354
Block 9 (M = 27.69, SE = 2.23)	0.46	2.59	.324

Note. Trial completion time was averaged across rotations; one-tailed t-tests were computed. Significance (using the Bonferroni correction) is indicated by asterisks. SE_Diff refers to the standard error of the difference scores between Block 10 and the given experimental block, whereas SE refers to the standard error of the mean.

Figure 6.

Average trial completion times for the different rotation conditions observed at each experimental block. Error bars are ±1 standard error of the mean, to compare rotations with each block. Standard errors relevant for assessing repeated-measures effects (changes across blocks) are not presented, as such error bars would differ for each individual comparison of interest.

The performance observed at the different rotation conditions during the experimental blocks is presented in Figure 6. To test Hypothesis 3 (the 90° and 135° rotations as a whole result in worse performance than the other rotation conditions in both the first and last experimental blocks), the Rotations × Experimental Blocks interaction was further analyzed using linear contrasts. Performance in the 90° and 135° rotations as a whole was compared to that observed in each of the other rotations during the initial experimental block (Block 3) and the last experimental block (Block 10). The Bonferroni correction was employed to control for Type I error inflation, resulting in a comparison-wise error rate of α = .0083. As can be seen in Table 6, at Block 3 the 90° and 135° rotations as a whole resulted in worse performance than that observed at the 0°, 45°, and 180° rotations. At Block 10, the 90° and 135° rotations as a whole resulted in worse performance than that observed at 0° and 45° rotations. However, although at Block 10 a tendency toward better performance was observed at the 180° rotation compared to the 90° and 135° rotations taken together, this tendency did not reach statistical significance.

Table 6:

Linear Contrasts Comparing the 90° and 135° Rotations as a Whole With Each of the Other Rotations at Block 3 and Block 10

Rotation	Contrast Value	t(35)	p Value (one tailed)
Block 3 (90° and 135° rotation as a whole, M = 126.33, SE = 14.80)
0° (M = 21.08, SE = 1.48 )	105.25	5.79	<.001*
45° (M = 35.46, SE = 5.00)	90.87	5.00	<.001*
180° (M = 64.98, SE = 13.07)	61.36	3.38	.001*
Block 10 (90° and 135° rotation as a whole, M = 39.34, SE = 4.55)
0° (M = 17.33, SE = 1.74)	22.01	3.97	<.001*
45° (M = 19.25, SE = 1.84)	20.09	3.63	<.001*
180° (M = 29.18, SE = 3.83)	10.16	1.83	.038

Note. Significance (using the Bonferroni correction) is indicated with asterisks. SE refers to the standard error of the mean.

Average drops per transfer

There was a significant rotations main effect, F(4, 35) = 6.77, p < .001, η_p² = .44 . Figure 7 displays the average drops per transfer observed at each rotation condition. To test our prediction that the 90° and 135° rotations as a whole result in worse performance than the other rotations, Bonferroni-corrected one-tailed linear contrasts were computed (error rate per comparison was set at α = .017). These contrasts indicated that the average drops per transfers at 90° and 135° rotation conditions as a whole (M = 0.36, SE = 0.10) were significantly higher than those observed at the 0° rotation (M = .08, SE = .03), t(35) = 2.89, p = .003, and the 180° rotation (M = 0.08, SE = 0.02), t(35) = 2.89, p = .003. However, even though there was a tendency for higher drops per transfer at the 90° and 135° rotations as a whole than at the 45° rotation (M = .16, SE = 0.05), this tendency did not reach statistical significance. As can be seen from Figure 7, worst performance was observed in the 90° rotation, with better performance in the 135° rotation. To assess if the described differences of the 90° and 135° rotations as a whole compared to the 0° and 180° rotations were due to mostly the 90° rotation, Tukey post hoc tests were computed. These tests indicated that the 90° rotation was significantly worse than the 0°, 45°, 135°, and 180° rotations, with no other group differences observed. Thus, the observed differences between the 90° and 135° rotations as a whole and the 0° and 180° rotations are driven primarily by the 90° rotation.

Figure 7.

Drops per transfer (averaged over experimental blocks) observed at different rotations. Error bars are ±1 standard error of the mean.

Discussion

The goal for the present project was to determine whether Cunningham’s (1989) results generalize to the laparoscopic environment when controlling for carryover effects, with Experiment 1 utilizing a three-dimensional straight-line pointing task and Experiment 2 employing a peg transfer task requiring coordinated movements of laparoscopic tools in both hands.

Rotations of 90° and 135° Result in Poorer Performance

The results of both Experiment 1 and Experiment 2 indicated, when controlling for carryover effects, poorer performance for novices in the laparoscopic environment when they were exposed to sideways camera rotations of 90° and 135° for both a pointing task and a peg transfer task. These results are consistent with those of previous visuomotor rotation research (Cunningham, 1989; Ellis et al., 2012; Kim et al., 1987). This pattern may be caused by two different processes. First, prior studies indicated that the gradual rotation of an internal mapping/reference frame accounts for performance adaptation for rotations up to 90° (Bock et al., 2003). In contrast, the relatively good performance at 180° is thought to be due to the reversal of the movement direction (Bock et al., 2003; Cunningham, 1989). Although this reversal is assumed to pose relatively low cost to performance, it is higher than that observed at rotations of 0°. Consistent with this observation, tendencies toward worse performance emerged in the 180° conditions when compared to the 0° conditions in Experiments 1 and 2 (see Figures 2, 3, 5, and 6). Furthermore, rotations greater than a certain threshold may lead to a combination of the two processes, with a movement reversal plus a rotation back to the prescribed angle. For example, Cunningham (1989) provided support that performance adaptation to 135° rotation can be explained by the reversal of movement direction to 180° of rotation, followed by 45° backward rotation to 135°.

Performance Improves Over Experimental Trials

Consistent with prior perceptual-motor adaption research (e.g., Dolezal, 1982), the two experiments showed performance improvement with practice (Tables 2 and 5). Specifically, performance observed for Experiment 1 was superior in the last experimental block (Block 10) when compared to all other experimental blocks (Blocks 3 through 9). The performance data observed in Experiment 2 also were consistent with the expectation that performance would improve over the course of adaptation. The last experimental block (Block 10) indicated superior performance compared to the first three experimental blocks (Blocks 3 through 5) for the average trial completion time. However, the average trial completion time observed in the last experimental block was not significantly lower than the average trial completion time observed in Blocks 6 through 9 (Table 5).

The achievement of asymptotic performance during the latter experimental blocks of Experiment 2 and not Experiment 1 is of particular interest. This difference might be due to the peg transfer task in Experiment 2 allowing for greater movement variability than the pointing task in Experiment 1. Prior research indicated that that greater variability during practice along multiple dimensions can result in enhanced performance under test conditions (Schöllhorn et al., 2006). This finding suggests that performing a task under different contexts allows for better perception of the relationship between internal and external constraints, which allows for better performance (Frank, Michelbrink, Beckmann, & Schöllhorn, 2008). Specifically, the pointing task in Experiment 1 required participants to reach target locations using their right hand, with perfect performance being reflected by a straight-line movement originating at the starting position and ending at the target. Thus, the movement variability might have been more limited for this pointing task than for the peg transfer task in Experiment 2. The peg transfer task required participants to use both hands, with a random starting hand. This task had the potential to be executed in different ways, allowing for considerable variability of movement execution. This increased movement variability might have allowed participants in Experiment 2 to more efficiently adapt to the perceptual-motor distortions, as indicated by asymptotic performance during the latter experimental blocks.

Is the Relative Difficulty of the Visuomotor Rotations Maintained Over the Course of Adaptation?

Prior research by Cunningham (1989) indicated that the relative difficulty of different rotations is preserved over time, in that performance in the 90° and 135° rotations is more difficult when compared to the 180° rotation during both the early and latter stages of adaptation. These results suggest that the functional characteristics (relative difficulty) of visuomotor performance are maintained over the course of adaptation. However, Cunningham’s sample size was very small, which might have resulted in insufficient power to detect a Rotations × Experimental Blocks interaction. In the present study, both Experiment 1 and Experiment 2 revealed a Rotations × Experimental Blocks interaction, initially suggesting that the results do not support such a “static” nature of visuomotor adaptation. However, when taking a closer look at Figures 3 and 6, it is apparent that the interaction is caused by greater adaptation taking place in the 90°, 135°, and 180° rotations than in the 0° and 45° rotations.

Interestingly, in both studies, the first experimental block (Blocks 3) and the last experimental block (Blocks 10) both demonstrated the worst performance with the 90° and the 135° rotations, with a tendency toward better performance in the 180° rotation. Notably, adaptation observed in the 180° rotation for the pointing task (Figure 3) rapidly approached performance observed in the 45° rotation during the second experimental block (Block 4) and resembled the performance observed in the 0° rotation already during the fourth experimental block (Block 6). In contrast, the performance profile of the peg transfer task indicated the 180° did not adapt as rapidly to resemble performance in the 0° rotation (see Figure 6). Therefore, straight-line pointing tasks might be especially amenable for the movement-inversion process.

Even though the follow-up analysis of the Rotations × Experimental Blocks interactions indicated different patterns regarding which rotations differ significantly from the 90° and 135° rotations as a whole for both Block 3 and Block 10 (see Tables 4 and 6), the results of these statistical tests alone are not sufficient to rule out Cunningham’s (1989) claim regarding the consistent relative difficulties of different rotations over the course of adaptation. However, the result of the present study raises some challenges, as differences between rotations appear to diminish over time.

Future Research

The results of the present study indicated poorer performance for visual rotations ranging between 90° and 135°, for both a “simple” pointing task and a more complex peg transfer task. Our results are consistent with prior perceptual-motor research that utilized repeated-measures designs (Cunningham, 1989; Ellis et al., 2012; Kim et al., 1987), but our findings are not consistent with those observed by Ames and colleagues (2006). Specifically, research by Ames and colleagues indicated that performance breakdown in a laparoscopic simulator increased as the sideways laparoscope rotation increased, with worst performance being observed at rotations of 135° and 180°. One possible reason for this discrepancy might be carryover effects, caused by each participant being exposed to multiple visuomotor rotations in Ames and colleagues’ study. No systematic research had been conducted to assess carryover effects and their underlying mechanisms in the laparoscopic training environment to the authors’ knowledge. Given that the present study showed worse performance for rotations ranging between 90° and 135° using a similar task as Ames et al. when employing a between-groups design, carryover effects may be the primary reasons for the observed differences between the current and Ames and colleagues’ study. Further, because performing laparoscopy might require the relocation of the laparoscope during surgery, potentially resulting in similar carryover effects, authors of future research need to assess the impact of such relocations on laparoscopic performance and if it varies as a function of surgical training.

Basic perceptual-motor research has demonstrated that if a visuomotor rotation has the same direction as a subsequent visuomotor rotation, performance during the latter rotation might be facilitated, particularly if both rotations are smaller or larger than a critical rotation angle. This critical rotation angle has been estimated to be between 105° and 150° when the second rotation is larger than the preceding rotation and between 75° and 120° when the second rotation is smaller than the initial rotation (Abeele & Bock, 2001). Thus, authors of future research need to assess if these critical rotation angles generalize to the laparoscopic environment and if they vary as a function of expertise.

Further, because expert surgeons have been exposed over an extended period to a multitude of different visuomotor rotations by differing laparoscope placements, the results of the present study may not generalize to expert surgeons. Authors of future studies should consider experts when delineating the impact of differing laparoscope placements, carryover effects when switching between different laparoscope positions and their underlying mechanisms, and if the findings of the present manuscript generalize to other environments that entail visuomotor rotations, such as teleoperation environments. Finally, it is important to note that in the current project, we assessed only performatory adaptation, which is not necessarily equivalent to perceptual adaption. Perceptual adaptation refers to adaptation to perceptual ambiguities and inconsistencies. Such perceptual experiences have previously been described, for example, during prism adaptation (Dolezal, 1982).

Key Points

The results of the present study indicated worst performance for novices performing pointing and peg transfer tasks in a laparoscopic simulator when the camera was rotated between 90° and 135° to the participant’s side.

The results of the present study indicated that novices’ task performance in the simulator improved with practice for all camera rotations, with greatest performatory adaption occurring in the 90°, 135°, and 180° rotations.

Footnotes

Martina I. Klein is an assistant professor of psychology at Texas Tech University. She received her PhD in human factors psychology in 2008 from the University of Cincinnati.

Noah J. Wheeler is a graduate human factors psychology student at Texas Tech. He received his MA in psychology in 2013 from Texas Tech University.

Curtis Craig is a graduate human factors student at Texas Tech University. He received his MA in psychology in 2008 from Texas Tech University.

References

Abeele

Bock

(2001). Sensorimotor adaptation to rotated visual input: Different mechanisms for small versus large rotations. Experimental Brain Research, 140, 407–410. doi:10.1007/s002210100846

Ames

Frisella

A. J.

Yan

Shulam

Landman

(2006). Evaluation of laparoscopic performance with alteration in angle of vision. Journal of Endourology, 20, 281–283.

Bock

Abeele

Eversheim

(2003). Human adaptation to rotated vision: Interplay of a continuous and a discrete process. Experimental Brain Research, 152, 528–532. doi:10.1007/s00221-003-1643-x

Cunningham

H. A.

(1989). Aiming error under transformed spatial mappings suggests a structure for visual-motor maps. Journal of Experimental Psychology: Human Perception and Performance, 15, 493–506. doi:10.1037/0096-1523.15.3.493

Cuschieri

(1995). Whither minimal access surgery: Tribulations and expectations. American Journal of Surgery, 169, 9–19.

DeLucia

P. R.

Griswold

J. A.

(2011). Effects of camera arrangement on perceptual-motor performance in minimally invasive surgery. Journal of Experimental Psychology: Applied, 17, 210–232. doi:10.1037/a0024041

Dolezal

(1982). Living in a world transformed: Perceptual and performatory adaptation to visual distortion. New York, NY: Academic Press.

Ellis

S. R.

Adelstein

B. D.

Yeom

(2012). Human control in rotated frames: Anisotropies in the misalignment disturbance function of pitch, roll, and yaw. In Proceedings of the Human Factors and Ergonomics Society 56th Annual Meeting (pp. 1336–1340). Santa Monica, CA: Human Factors and Ergonomics Society. doi:10.1177/1071181312561385

Ferzli

G. S.

Fingerhut

(2004). Trocar placement for laparoscopic abdominal procedures: A simple standardized method. Journal of the American College of Surgeons, 198, 163–173. doi:https://dx-doi-org.web.bisu.edu.cn/10.1016/j.jamcollsurg.2003.08.010

10.

Field

(2005). Discovering statistics using SPSS. Thousand Oaks, CA: Sage.

11.

Frank

Michelbrink

Beckmann

Schöllhorn

W. I.

(2008). A quantitative dynamical systems approach to differential learning: Self-organization principle and order parameter equations. Biological Cybernetics, 98, 19–31. doi:10.1007/s00422-007-0193-x

12.

Fried

G. M.

Feldman

L. S.

Vassiliou

M. C.

Fraser

S. A.

Stanbridge

Ghitulescu

Andrew

C. G.

(2004). Proving the value of simulation in laparoscopic surgery. Annals of Surgery, 240, 518–525. doi:10.1097/01.sla.0000136941.46529.56

13.

Gallagher

A. G.

McClure

McGuigan

Ritchie

Sheehy

N. P.

(1998). An ergonomic analysis of the fulcrum effect in the acquisition of endoscopic skills. Endoscopy, 30, 617–620.

14.

Keppel

Wickens

T. D.

(2004). Design and analysis: A researcher’s handbook (4th ed.). Upper Saddle River, NJ: Pearson Education.

15.

Kim

W. S.

Ellis

S. R.

Tyler

M. E.

Hannaford

Stark

L. W.

(1987). Quantitative evaluation of perspective and stereoscopic displays in three-axis manual tracking tasks. IEEE Transactions on Systems, Man and Cybernetics, 17, 61–72. doi:10.1109/tsmc.1987.289333

16.

Klein

M. I.

DeLucia

P. R.

Olmstead

(2013). The impact of visual scanning in the laparoscopic environment after engaging in strain coping. Human Factors, 55, 509–519. doi:10.1177/0018720812467458

17.

Klein

M. I.

Riley

M. A.

Warm

J. S.

Matthews

(2005). Perceived mental workload in an endoscopic surgery simulator. In Proceedings of the Human Factors and Ergonomics Society 49th Annual Meeting (pp. 1014–1018). doi:10.1177/154193120504901103

18.

Klein

M. I.

Warm

J. S.

Riley

M. A.

Matthews

Gaitonde

Donovan

J. F.

(2008). Perceptual distortions produce multidimensional stress profiles in novice users of an endoscopic surgery simulator. Human Factors, 50, 291–300. doi:10.1518/001872008x288312

19.

Kunde

Müsseler

Heuer

(2007). Spatial compatibility effects with tool use. Human Factors, 49, 661–670. doi:10.1518/001872007X215737

20.

Macedo

J. A.

Kaber

D. B.

Endsley

N. R.

Powanusorn

P. P.

Myung

(1998). The effect of automated compensation for incongruent axes on teleoperator performance. Human Factors, 40, 541–553. doi:10.1518/001872098779649256

21.

Mazzoni

Krakauer

(2006). An implicit plan overrides an explicit strategy during visuomotor adaptation. Journal of Neuroscience, 26, 3642–3645. doi:10.1523/JNEUROSCI.5317-05.2006

22.

McGreevy

Goodney

Birkmeyer

Finlayson

Laycock

Birkmeyer

(2003). A prospective study comparing the complication rates between laparoscopic and open ventral hernia repairs. Surgical Endoscopy, 17, 1778–1780. doi:10.1007/s00464-002-8851-5

23.

Rattner

(1999). Future directions in innovative minimally invasive surgery. Lancet, 353(Suppl. 1), 12–15.

24.

Reinhardt-Rutland

A. H.

Annett

J. M.

Gifford

(1999). Depth perception and indirect viewing: Reflections on minimally invasive surgery. International Journal of Cognitive Ergonomics, 3, 77.

25.

Schauer

Ikramuddin

Hamad

Gourash

(2003). The learning curve for laparoscopic Roux-en-Y gastric bypass is 100 cases. Surgical Endoscopy, 17, 212–215. doi:10.1007/s00464-002-8857-z

26.

Schöllhorn

Beckmann

Michelbrink

Trockel

Sechelmann

Davids

(2006). Does noise provide a basis for unifying different motor learning theories? International Journal of Sport Psychology, 2, 34–42.

27.

Shepard

R. N.

Metzler

(1971). Mental rotation of three-dimensional objects. Science, 171, 701–703. doi:10.1126/science.171.3972.701

28.

Trempe

Proteau

(2010). Distinct consolidation outcomes in a visuomotor adaptation task: Off-line leaning and persistent after-effect. Brain and Cognition, 73, 135–145. doi:10.1016/j.bandc.2010.04.005

29.

Varela

J. E.

Wilson

S. E.

Nguyen

N. T.

(2010). Laparoscopic surgery significantly reduces surgical-site infections compared with open surgery. Surgical Endoscopy, 24, 270–276. doi:10.1007/s00464-009-0569-1

30.

Veen

E. J.

Bik

Janssen-Heijnen

M. L. G.

De Jongh

Roukema

A. J.

(2008). Outcome measurement in laparoscopic cholecystectomy by using a prospective complication registry: Results of an audit. International Journal for Quality in Health Care, 20, 144–151. doi:10.1093/intqhc/mzm073

31.

Wheeler

N. K.

Klein

M. I.

Craig

(2012). Camera placement in simulated laparoscopic surgery influences performance. In Proceedings of the Human Factors and Ergonomics Society 56th Annual Meeting (pp. 1346–1350). Santa Monica, CA: Human Factors and Ergonomics Society. doi:10.1177/1071181312561387

32.

Zhang

Cao

C. L.

(2012). Effect of automatic image realignment on visuomotor coordination in simulated laparoscopic surgery. Applied Ergonomics, 43, 993–1001. doi:10.1016/j.apergo.2012.02.001