Mental rotation performance after acute stress

Abstract

Background

The need to mentally rotate objects is common in everyday activities (e.g., driving, loading suitcases in a car, playing 3D video games) and in specific professions (e.g., air traffic controllers, mechanics, pilots, surgeons). Surprisingly, whether stress facilitates or impairs spatial reasoning in these situations remains an open question. Previous research has shown that spatial performance (e.g. navigation, visuospatial memory, attention) can be influenced by acute stress. However, few studies have investigated the effects of stress on mental rotation and these have produced mixed results. In the present study, we examined the influence of cardiovascular indicators of Sympathetic Adrenal Medulla (SAM) activation from an acute laboratory stressor (Paced Serial Addition Task [PASAT]) on mental rotation performance (MRT Accuracy and Response Time).

Methods

We assigned young, healthy, volunteer participants (39 men, 41 women; 18–23 years of age) to either a Stress (PASAT) or a Non-Stress (Control) group. All participants completed a Mental Rotation Task (MRT) before (Pretest) and after (Posttest) the PASAT or Control task. Cardiovascular measurements (blood pressure, heart rate) and self-reported affect were obtained to quantify values during a resting state, stress manipulation, MRT Pretest, MRT Posttest, and a Recovery period.

Results

The Stress group reported greater negative affect, less positive affect, and had higher systolic blood pressure than our Non-Stress Control group after our stress manipulation, thus demonstrating that our PASAT stressor was effective at increasing cardiovascular indicators of SAM activation. All participants became faster and more accurate from the MRT Pretest to the Posttest demonstrating a practice effect. Men were more accurate than women in MRT, but contrary to expectations, this sex difference did not change with stress induction. Overall, PASAT stress did not have an influence on MRT speed or accuracy. A regression analysis did not find a relationship between self-reported affect, cardiovascular indicators of SAM activation, and MRT performance.

Conclusions

Our results suggest that being in the Stress group (PASAT) had no influence on mental rotation performance. Future investigations might confirm this in professions performing specific rotation tasks under pressure and should also investigate the effect of physical stressors or stress-induced hypothalamic-pituitary (HPA) axis activation prior to MRT performance.

Background

Mental rotation is the cognitive ability to manipulate two or three dimensional objects to predict how the object would look if it were rotated in space. It is one of the most used measures of spatial ability and tends to show large individual differences [1–3]. Studies consistently report large effect sizes favoring males [4–6]. Notably, mental rotation ability is associated with occupational skills, such as laparoscopic surgery [7] and dental performance [8]. It is also related to success in aviation training [9] and achievement in chemistry courses and tasks [10]. In some professions, individuals will need to quickly and accurately perform spatial reasoning tasks during periods of acute stress. After a bird strike disabled both engines of his aircraft, pilot Chesley Sullenberger had moments to consider how changes in his speed, altitude, and heading might result in a safe landing, which he ultimately achieved on the Hudson River [11]. Surprisingly, whether stress facilitates or impairs spatial reasoning in these situations remains an open question, and this may be critical in professions where even minor errors can have significant, sometimes life-threatening consequences.

The central nervous system activates two main physiological systems to respond to stress: (1) the fast acting sympathetic adrenal medulla (SAM) axis, which results in the release of the action molecules epinephrine and norepinephrine that increase heart rate and blood pressure, and (2) the slower hypothalamic pituitary adrenal (HPA) axis, which causes the subsequent release of cortisol leading to gluconeogenesis. Activation of these biological stress systems may become influential during activities that require attention, learning, and memory [12, 13] and researchers have systematically investigated the complex interaction between psychological stress and these processes. These stress systems might be related to performance as they may activate or inhibit brain regions involved in spatial reasoning. For example, virtual navigation tasks are associated with increased activation in the right hippocampus and prefrontal cortex (PFC), and high levels of cortisol (HPA axis) have been linked to reduced hippocampal activation. With respect to SAM activation, moderate levels of epinephrine and norepinephrine released during stress can improve PFC functioning, whereas high levels can result in impairment [14, 15]. Mental rotation tasks specifically have been shown to involve the parietal lobe, motor areas, and PFC [16, 17]; however, these activation patterns also differ between men and women [16–18]. Sex differences may also exist in how men and women respond to stress, both with respect to neural activation patterns [19] and cortisol response [20, 21].

Indeed, there may be sex differences in how a stressor modulates processing during mental rotation tasks. For example, SAM activation after cold pressor stress has been shown to increase virtual navigation performance in males [13]. However, contradictory findings were reported [22] using a timed spatial stressor (Star Mirror Trace Task). In our research [22], we showed that SAM activation was associated with slower performance in men and women in both virtual navigation and in perspective taking tasks, which required individuals to make directional judgments from imagined orientations. Virtual navigation studies employing stronger socially evaluated stressors, such as the Trier Social Stress Test (TSST) or Socially Evaluated Cold Pressor Test (SECPT), result in HPA activation and subsequent cortisol release. These studies tend to show no performance changes after a stress manipulation [23–25]. Studies examining the influence of acute stress on visuospatial memory performance are also mixed. Stress induced increases in SAM activation leads to impaired performance [26–28], while stress studies resulting in HPA activation are associated with improvements [29–31] or sometimes no change in performance [32–34].

To date, only two studies have examined the relationship between acute stress and mental rotation performance. Gabriel et al. [35] did not include physiological measures (e.g., HR, BP, skin conductance) to confirm SAM activation in their stress condition, and they found no relationship between cold pressor test stress and MRT accuracy. However, in their study, females were faster on the MRT after stress. In contrast, Cohen et al. [36] showed that TSST improved accuracy in both men and women, though response time was not assessed.

The goal of our present study was to examine the effect of an acute psychosocial stressor (PASAT) on mental rotation accuracy and response time measures. Cardiovascular measurements were collected to assess SAM activation, and self-report measures of affect were collected throughout the study. Based on previous research we predict that stress-induced SAM activation will increase performance, and that this effect might be limited to women. We also predict a practice effect in that MRT performance will increase between the Pretest and Posttest and that men will have better overall MRT performance than women.

Methods

Participants

Participants were recruited through flyers and email announcements across campus. Sample size was determined with G*Power 3.1 [37] to detect a medium-large effect (f = 0.35), with power of 80%, and α = 0.05 two-tailed; a total sample of at least 67 participants was required to detect main effects and a two-way interaction. Expected effect size was based on previous meta-analyses on sex differences in mental rotation [4, 5] and the single stress and mental rotation study from Cohen et al. [36]. Eighty-two participants completed the study session, however two were excluded due to technical difficulties with the computer administration of our Mental Rotation task, leaving us with a total sample size of 80 participants (39 men, 41 women). Participants were healthy college student volunteers between 18 and 23 years of age (M = 19.66, SD = 1.24) with an average BMI of 23.74 (SD = 3.14). They reported no chronic health problems, no use of prescribed medications that should affect physiological or mood responses, and self-reported that they were not pregnant. Course extra credit and/or entry into a drawing for a $75 Amazon gift card was provided. The study was approved by the College’s Institutional Review Board (ID#: IRB-22-2) and procedures were performed according to their guidelines and regulations, including participants providing written informed consent.

Psychological measures

Perceived stress scale (PSS-10)

To assess the degree to which an individual perceives situations in their life as uncontrollable, overburdening, and unpredictable over the past month and to establish a baseline level of stress, the 10-item PSS questionnaire was administered [38]. Participants responded to each question using a 5-point Likert scale ranging from 0 (never) to 4 (very often). Scores on the PSS-10 can range from 0 to 40, with higher levels indicating greater perceived stress (e.g., 0–13 indicate low, 14–26 moderate, and 27–40 indicate high levels). The questionnaire demonstrates strong internal consistency reliability (Cronbach’s alpha reliability coefficients ranging from 0.78 to 0.89) [39, 40] and strong convergent validity (scores were highly associated with the State-Trait Anxiety Inventory Total Score) [40].

Positive and negative affect

A modified version of Monk’s Global Vigor and Affect scale [41] was used to measure participants’ affect after each period of the study. Participants indicated their affect (i.e., Happy, Bored, Anxious, Satisfied, Depressed, Interested, Angry, Frustrated, Irritated), by drawing a perpendicular mark through a 10 cm long scale line with anchors of “Not at all” and “Extremely” at each end. Similar assessments for the degree of challenge and difficulty of the task periods also occurred. Stress is positively related to negative affect (i.e., Anxious, Angry, Frustration, and Irritation) [42] and negatively related to positive affect (i.e., Happy, Satisfied) [43], so only the negative and positive affect variables were included in our subsequent analyses.

Physiological measures

Blood pressure and heart rate

Systolic blood pressure (SBP), diastolic blood pressure (DBP), and heart rate (HR) were measured using an automated digital blood pressure monitor (Omron digital blood pressure monitor HEM 907XL, Omron Healthcare, Inc., Bannockburn, IL). An appropriately sized cuff was attached to the participants’ non-dominant upper arm. All measurements were initiated by the experimenter in a separate room. To quantify the resting state, blood pressure and heart rate were obtained at minutes 7, 8, and 9 of the 10-minute Baseline period. During the 6-minute stress manipulation or Control task, cardiovascular measurements were obtained every minute beginning at the start of the tasks (i.e., minutes 0, 1, 2, 3, 4, and 5). Measurements were initiated every other minute during the 6-minute Mental Rotation Task (MRT) Pretest and MRT Posttest periods (i.e., 0, 2, 4, and 6) and during the 5-minute Recovery period (i.e., minutes 0, 2, and 4). Blood pressure and heart rate values within each time period were averaged, so that there was one value representing the Baseline period, Stress manipulation (PASAT or Control task), MRT Pretest, MRT Posttest, and the Recovery period.

Task descriptions

Stress manipulation - paced serial addition task (PASAT)

The PASAT, originally used as a measure of recovery from concussion [44], is a mental arithmetic task that has been adapted for use in cardiovascular reactivity studies by Willemson et al. [45]. It was modified for psychological stress-induction in laboratory-based cardiovascular reactivity research as it reliably increases sympathetic responses, such as increases in blood pressure and heart rate [46–48], skin conductance [49], and negative affect [50, 51]. The PASAT has also been shown to influence cognition in that it impairs visual attention [52] and reduces executive functioning performance such as task management, planning, monitoring, and inhibition [53]. Anecdotally, participants have stated that the PASAT is quite aversive, and some indicated that they would rather have a lumbar puncture than complete it [54].

The PASAT was presented from a digital audio file, which provided participants with a continuous list of numbers varying from one to nine. Each participant continuously added each new number to the previous digit. For example, if the numbers “3, 7, 9” are presented, the participant would add (3 + 7) and respond with “10”, followed by (7 + 9) and respond with “16”, and so on. The task is divided into three 2-minute periods with the digits presenting at varying speeds for a total of 6 min. The first period presents at one digit per 2.4 s, the second period at 2.0 s, and the third and final period at 1.6 s. The experimenter remained in the room in clear view of the participant and documented all incorrect verbalized answers. To maintain task engagement, participants were told that they should be able to answer 80% of their responses correctly. A brief practice session of ten numbers was repeated until the participant understood the task.

In the Non-Stress/Control group, participants read the same list of digits aloud at their own pace for 6 min without performing addition. The experimenter was not in the room and there was no time constraint.

Mental rotation task (MRT)

The MRT is a computerized version of the original test devised by Shepard and Metzler [1]. For each trial, a pair of 3D figures were presented on screen (Fig. 1). Participants determined whether the second figure was an identical but rotated version or a mirror image of the first figure and responded by clicking a computer key as quickly as possible. In each trial block, participants were presented with 40 pairs of figures that varied in rotation by 5, 45, 90, 135, or 180 degrees around the horizontal axis. Participants were presented with trials randomly with each rotation presented 8 times across a trial block. Each participant was given 6 min to complete the trial block and were instructed to wait quietly if they finished early. Participants completed both an MRT Pretest (before stress manipulation or control) and a MRT Posttest (after stress manipulation or control) in which the same 40 pairs were presented but in a different (random) order. The dependent measures were Accuracy (proportion correct) and average Response Time (seconds) to complete each item. During the experimental trial blocks, no Accuracy or Response Time feedback was provided to the participants.

Fig. 1.

Example MRT trial item with 45-degree rotation

Procedure

Prior to data collection participants were randomly assigned to either a Stress (PASAT; n = 38; 19 men, 19 women) or Non-Stress (Control; n = 42; 20 men, 22 women) group. Testing occurred between 12pm and 5pm to control for circadian changes in biological responses. All participants were instructed to refrain from alcohol, nicotine, and cannabis for 24 h; caffeine and strenuous physical activity for five hours, and food or drink (other than water) for two hours prior to the study session to control for short-term dietary or post-prandial effects [55, 56] and verbal affirmation of these requirements was confirmed at the start of the experimental session. Before data collection began participants signed an informed consent document.

Next, participants were seated and an inflatable cuff was applied to their non-dominant arm, connected to an automated blood pressure monitor for the confirmation of normotensive status, and the PSS was administered. Participants were then provided with four practice trials of the MRT with feedback to ensure that the task was clearly understood. Participants then relaxed and watched and listened to a YouTube video of Lofi hip hop music for a ten-minute Baseline period. They were then given the first block of 40 MRT trials to establish baseline performance (MRT Pretest).

Participants in the Stress group then underwent the 6-minute PASAT procedure while those in the Non-Stress group read the digits aloud for 6 min (Control task). Next, both groups immediately conducted a second block of 40 MRT trials (MRT Posttest). The experiment concluded with a 5-minute resting recovery period where participants once again watched and listened to the Lofi hip hop music video.

Data reduction and statistical analysis

To assess our stress manipulation, we conducted a series of 5 × 2 × 2 mixed design ANOVAs with Time (Baseline, MRT Pretest, PASAT or Control task, MRT Posttest, Recovery) as the within-subjects factor, and both Group (PASAT or Control) and Sex (men, women) as the between-subjects factors with each Cardiovascular (SBP, DBP, HR), Negative Affect (Anxiety, Anger, Frustration, Irritation), Positive Affect (Happy, Satisfied), and stress manipulation assessment (Challenging, Difficulty) variable as the dependent measures. To examine our main questions of interest, a series of 2 × 2 × 2 mixed design ANOVAs with Session (MRT pretest, MRT posttest), Group (Stress, Non-Stress) and Sex (men, women) as between-subjects factors were conducted with Mental Rotation Response Time and Accuracy as the dependent measures. When sphericity assumptions for ANOVA were violated, Greenhouse-Geisser corrected p values, along with the respective epsilon and uncorrected degrees of freedom, are reported [57]. Simple main effect analyses were conducted to explain significant (p < .05) interactions. Within-group (Stress, Non-Stress) differences from Baseline to task (PASAT, Control) were assessed through Bonferroni-corrected post hoc tests. This strategy allowed us to differentiate the magnitude of change from a calm to a stress state and thus examine the efficacy of our stress manipulation. To determine if self-reported perceived stress, cardiovascular measurements, and affect measures would continue to predict MRT performance after controlling for Sex and Stress Group, a regression analysis was also conducted. Since no previous studies have examined stress and cardiovascular reactivity in predicting MRT, this analysis was considered exploratory. Data were analyzed using JASP (version 0.19) open source software.

Results

Perceived stress

Internal consistency reliability was found to be robust for our administration of the Perceived Stress Scale (PSS; Cronbach’s alpha reliability coefficient = 0.89). Overall, participants self-reported moderate levels of perceived stress (PSS) at the beginning of the study, M = 14.7, SD = 6.40, as consistent with previous normative data [58]. The Stress (M = 15.6, SD = 6.07) and Non-Stress (M = 13.8, SD = 6.65) groups did not significantly differ in their levels of perceived stress, nor did levels between men (M = 13.3, SD = 6.67) and women (M = 16.0, SD = 5.92; all p’s > 0.05).

Cardiovascular measurements

Sex main effects indicated that women had overall greater HR as compared to men, while men had overall greater SBP than women (see Table 1). Consistent with our prediction that the PASAT task would elicit greater increases in cardiovascular indicators of SAM activation than the Control task, significant Time X Group interactions were demonstrated for SBP, DBP, and HR (see Table 1; Fig. 2). Bonferroni corrected post hoc testing (see Fig. 2) showed that participants in the Stress group had greater SBP, (p < .001, d = 1.00), DBP, (p < .001, d = 1.14), and HR (p < .001, d = 1.05) during the PASAT as compared to Baseline. In contrast to our prediction, participants in the Non-Stress group also had greater SBP (p = .003, d = 0.47), DBP, (p < .001, d = 0.72), and HR, (p < .001, d = 0.57) during the Control task as compared to Baseline.

Table 1.

Main effects and interactions for cardiovascular measures

Cardiovascular	F	df	ε	p	ηp2
Systolic Blood Pressure
Sex	31.87	1, 76		< 0.001	0.30
Time	42.82	4, 304	0.82	< 0.001	0.36
Group	0.54	1, 76		0.47	0.01
Time X Group	5.17	4, 304	0.82	0.001	0.06
Diastolic Blood Pressure
Sex	2.71	1,76		0.10	0.03
Time	60.84	4, 304	0.86	0.001	0.45
Group	0.87	1, 76		0.35	0.01
Time X Group	3.33	4, 304	0.86	0.02	0.04
Heart Rate
Sex	9.26	1, 76		0.003	0.11
Time	99.49	4, 304	0.75	< 0.001	0.57
Group	0.05	1, 76		0.82	0.00
Time X Group	8.69	4, 304	0.75	< 0.001	0.01

Fig. 2.

Mean (± SEM) for (a) systolic blood pressure (SBP), (b) diastolic blood pressure (DBP), and (c) heart rate (HR) values for participants in the Stress (PASAT) and Non-Stress (Control) groups at periods throughout the experiment. The Stress group had significantly greater SBP than participants in the Non-Stress group during the Task period (Stress or Control)

Although both Stress and Control groups showed increases in cardiovascular measures from Baseline to Task (PASAT, Control), simple main effects analysis indicated that participants in the Stress group had greater SBP than participants in the Non-Stress group during the Task period, F(1, 76) = 11.14, p = .001 (see Fig. 2a), thus demonstrating that the PASAT stress manipulation produced greater sympathetic activation than the Control task. For DBP, the difference between the Stress and Non-Stress groups during the task period did not reach statistical significance, F(1, 76) = 3.26, p = .074 (see Fig. 2b). No significant difference in HR during the Task period, F(1, 76) = 2.76, p = .17 (see Fig. 2c) was demonstrated. As expected, there were no differences in SBP, DBP, or HR between the Stress and Non-Stress groups at Baseline (all p’s > 0.05), demonstrating that both Groups began the experiment in similar cardiovascular states.

Positive and negative affect

Sex main effects indicated that women had overall greater Anxiety and rated the tasks to be more Challenging than men. Consistent with our prediction that the PASAT task would elicit greater negative affect increases than the Control task, significant Time X Group interactions were demonstrated for Anxiety, Anger, Frustration, Irritation, Happy, Satisfied, and also for the degree of Challenging, and Difficulty (see Table 2; Figs. 3 and 4). Bonferroni corrected post hoc testing showed that participants in the Stress group reported more Anxiety (p < .001, d = 1.18), more Anger (p < .001, d = 1.48), more Frustration (p < .001, d = 2.08), and more Irritation (p < .001, d = 1.43), were less Happy (p < .001, d = 1.30) and less Satisfied (p < .001, d = 1.59), and self-reported both greater Challenge (p < .001, d = 3.63) and Difficulty (p < .001, d = 3.82) during the PASAT as compared to Baseline. Participants in the Non-Stress group did not report significant increases in negative affect, Challenging, or Difficulty (all p’s > 0.05); however, participants were less Happy, (p = .005, d = 0.60) during the Control task as compared to Baseline.

Table 2.

Main effects and interactions for affect measures

Affect	F	df	ε	p	ηp2
Anxiety
Sex	4.33	1, 76		0.04	0.05
Time	24.14	4, 304	0.75	< 0.001	0.24
Group	5.66	1, 76		0.02	0.07
Time X Group	12.38	4, 304	0.75	< 0.001	0.14
Anger
Sex	0.00	1, 76		0.99	0.00
Time	15.67	4, 304	0.57	< 0.001	0.17
Group	5.69	1, 76		0.02	0.07
Time X Group	12.19	4, 304	0.57	< 0.001	0.14
Frustration
Sex	0.01	1, 76		0.94	0.00
Time	32.10	4, 304	0.76	< 0.001	0.30
Group	6.75	1,76		0.001	0.08
Time X Group	17.30	4, 304	0.76	< 0.001	0.19
Irritation
Sex	1.09	1, 74		0.30	0.01
Time	20.29	4, 304	0.66	< 0.001	0.22
Group	1.77	1, 74		0.19	0.02
Time X Group	7.19	4, 304	0.66	< 0.001	0.09
Happy
Sex	0.92	1, 75		0.34	0.01
Time	27.32	4, 300	0.80	< 0.001	0.27
Group	0.20	1, 75		0.66	0.00
Time X Group	6.19	4, 300	0.80	< 0.001	0.08
Satisfied
Sex	0.32	1, 75		0.58	0.00
Time	93.25	4, 304	0.82	< 0.001	0.55
Group	0.30	1, 75		0.59	0.00
Time X Group	4.97	4, 300	0.85	< 0.001	0.06
Challenge
Sex	6.25	1, 76		0.02	0.08
Time	95.71	4, 304	0.78	< 0.001	0.56
Group	39.79	1,76		< 0.001	0.34
Time X Group	67.58	4, 304	0.78	< 0.001	0.47
Difficulty
Sex	2.80	1, 76		0.10	0.04
Time	93.25	4, 304	0.82	< 0.001	0.55
Group	29.36	1, 76		< 0.001	0.28
Time X Group	73.10	4, 304	0.82	< 0.001	0.49

Fig. 3.

Mean (± SEM) for Anxious (a), Angry (b), Frustrated (c), Irritated (d), Happy (e), and Satisfied (f) values for participants in the Stress (PASAT) and Non-Stress (Control) groups at periods throughout the experiment. The Stress group had significantly greater negative affect (i.e., Anxious, Angry, Frustrated, Irritated) and less positive affect (i.e., Happy, Satisfied) than the Non-Stress group during the Task period (Stress or Control)

Fig. 4.

Mean (± SEM) for Challenging (a) and Difficulty (b) values for participants in the Stress (PASAT) and Non-Stress (Control) groups at periods throughout the experiment. The Stress group rated the Task Period (Stress or Control) as being more Challenging and Difficult than the Non-Stress group

As predicted, Simple main effects analysis indicated significant differences in self-reported Anxiety, F(1, 76) = 27.55, p < .001, Anger, F(1, 76) = 15.90, p < .001, Frustration, F(1, 76) = 31.60, p < .001, and Irritation, F(1, 74) = 10.46, p = .002, such that participants in the Stress group reported more negative affect during the task than the Non-Stress group (see Fig. 3). For positive affect, participants in the Stress group reported being less Happy, F(1, 75) = 6.60, p = .01, and less Satisfied F(1, 75) = 9.96, p = .002, during the task than the Non-Stress group. The Stress group also reported the PASAT to be more Challenging, F(1, 76) = 306.46, p < .001, and Difficult, F(1, 76) = 289.12, p < .001, than the Non-Stress group (see Fig. 4). As expected, there were no differences in negative or positive affect between the Stress and Non-Stress groups at Baseline (all p’s > 0.05), demonstrating that both Groups began the experiment in similar affective states.

Mental rotation performance

Response Time was log-transformed to correct for positive skew from long response times [59]. For MRTlog Response Time Session, participants were faster during the MRT Posttest (M = 0.52, 95% CI [0.48, 0.56]) as compared to the MRT Pretest (M = 0.68, 95% CI [0.63, 0.72]), F(1, 76) = 177.87, p < .001, η_p² = 0.70 (see Fig. 5). There were no differences in Response Time between men (M = 0.61, 95% CI [0.55, 0.67]) and women (M = 0.58, 95% CI [0.53, 0.64]), F(1, 76) = 0.53, p = .47, η_p² = 0.01, or between the Stress (M = 0.56, 95% CI [0.50, 0.62]) and Non-Stress (M = 0.64, 95% CI [0.58, 0.69]) Groups, F(1, 76) = 3.46, p = .07, η_p² = 0.04. Our hypothesized Session X Group interaction, F(1, 76) = 2.92, p = .09, η_p² = 0.04, was not significant, nor was our Session X Group X Sex interaction, F(1, 76) = 1.47, p = .23, η_p² = 0.02; thus contrary to expectation, our stress manipulation did not have a significant influence on MRT Response Time. We repeated the analysis with trimmed MRT Response Times (discarding scores > 2 Standard Deviations from the mean) to check if these results were sensitive to the log-transformation of Response Time scores. The Trimmed MRT Response Time analysis produced an identical pattern of significance as the results of the log-transformed Response Time scores.

Fig. 5.

Mean (± SEM) Pretest and Posttest Mental Rotation Response Time (seconds) for participants in the Stress (PASAT) and Non-Stress (Control) groups. Data analysis was conducted on log RT values, however untransformed RT values are presented for clarity. Overall MRT Posttest response time was significantly faster than Pretest response time

Participants were more accurate during the MRT Posttest (M = 0.77, 95% CI [0.73, 0.81]), than during the MRT Pretest, (M = 0.73, 95% CI [0.70, 0.77]), F(1, 76) = 13.43, p < .001, η_p² = 0.15; and men (M = 0.79, 95% CI [0.74, 0.84]) were more accurate than women, (M = 0.71, 95% CI [0.66, 0.76]), F(1, 76) = 4.31, p = .041, η_p² = 0.05 (see Fig. 6). Participants in the Stress group (M = 0.72, 95% CI [0.67, 0.77]) did not differ from the Non-Stress group, (M = 0.78, 95% CI [0.73, 0.83]), F(1, 76) = 2.74, p = .10, η_p² = 0.04. Our hypothesized Session X Group interaction, F(1, 76) = 0.01, p = .94, η_p² = 0.00, was not significant, nor was our Session X Group X Sex interaction, F(1, 76) = 0.01, p = .92, η_p² = 0.00; thus contrary to expectation, our stress manipulation had no influence on MRT performance.

Fig. 6.

Mean (± SEM) Pretest and Posttest Mental Rotation Accuracy (proportion correct) for participants in the Stress (PASAT) and Non-Stress (Control) groups. Overall MRT Posttest scores were significantly more accurate than Pretest scores, and overall Men were more accurate than Women

Correlations between self-reported perceived stress (as measured by the PSS at the beginning of the experiment) were not significantly related to performance measures: MRT Pretest Response Time, r(78) = − 0.01, p = .91, MRT Pretest Accuracy r(78) = − 0.05, p = .66, MRT Posttest Response Time, r(78) = − 0.02, p = .89, and MRT Posttest Accuracy r(78) = − 0.13, p = .24.

Regression analysis on mental rotation performance

To determine if self-reported perceived stress, cardiovascular measurements, and affect measures would continue to predict MRT performance after controlling for Sex and Stress Group an exploratory regression analysis was conducted. The outcome variables were changes in MRT accuracy and MRT logResponse Time (posttest-pretest); and the predictor variables were self-reported perceived stress (PSS), cardiovascular reactivity (i.e., Stress or Non-Stress task minus Baseline), and changes in negative and positive affect. To avoid collinearity (r’s > 0.70), we removed highly correlated predictor variables (i.e., DBP, Irritated). Using hierarchical regression, we entered Sex and Stress conditions in step 1, PSS in step 2, HR and SBP in step 3, negative affect measures (i.e., Anger, Anxious, Frustrated) in step 4, and positive affect measures (i.e., Happy, Satisfied) in step 5. For MRT Accuracy, no predictor variables were significant at any step, and the overall model was not significant, F(10, 77) = 0.63, p = .78, R² = 0.09, adjusted R² = − 0.05. For MRT log Response Time no predictor variables were significant at any step, and the overall model was not significant, F(10, 77) = 0.76, p = .66, R² = 0.10, adjusted R² = − 0.03. The regression analysis was also repeated with trimmed Response Time values instead of log-transformed values, and the model remained nonsignificant.

Discussion

In the present study, we found no difference in mental rotation speed and accuracy before and after exposure to an acute stressor (PASAT). To our knowledge this is only the third study to investigate the influence of a laboratory stressor on mental rotation performance [35, 36]. However, it is the most comprehensive study to date as it assesses: physiological measures of SAM activation; both mental rotation speed and accuracy; several measures of affect; and initial assessments of perceived stress.

Similar to previous research [46–48, 50, 51, 60], administration of the PASAT led to within-group increases in blood pressure, heart rate, self-reported negative affect (Anxiety, Anger, Frustration, Irritation), and both Challenge and Difficulty, with large effect sizes, thus demonstrating that our PASAT manipulation was an effective stressor. As expected, women had overall greater HR, and men had overall greater SBP throughout the study [61, 62]. Surprisingly, our Control task also resulted in a medium effect increase in blood pressure and heart rate; however, participants exposed to the PASAT stressor showed significantly greater reactivity in systolic BP, negative affect (Anxiety, Anger, Frustration, Irritation), and positive affect (Happy, Satisfied) than our Non-Stress group. Given these findings, our stress manipulation was more effective than our Control task in producing negative affect and cardiovascular SAM activation.

There are two possibilities to explain the unexpected increases in cardiovascular measures for the Non-Stress group. One explanation could be that for participants in the Non-Stress group simply reading numbers was in fact stressful. However, we reject this idea as the PASAT produced large increases in negative affect (e.g., Anxiety, Anger, Frustration, Irritation) in our Stress group, which is also observed in previous studies [50, 51]. These increases in negative affect were wholly absent in our Non-Stress group. Furthermore, the Non-Stress group did not self-report the tasks to be Difficult or Challenging. A second explanation for the increases in heart rate and blood pressure within the Non-Stress group could be due to vocalization from reading numbers aloud rather than stress-induced SAM activation. Previous studies have shown that blood pressure and heart rate increase when participants are simply talking [63, 64]. Specifically, talking has been found to increase blood pressure due to SAM activation [65, 66] and increases in peripheral vasoconstriction [67]. Zheng et al. [64] reported a 5mmHg increase in blood pressure from counting numbers aloud, similar to our own findings within our Non-Stress group. A more neutral task such as silent reading on a historical topic [36], watching a nature video [22] or sitting silently [53] may have mitigated these increases from vocalizations.

We found no overall effect of stress group (PASAT, Control) on MRT Response Time or Accuracy, or any Group or Sex interactions with MRT session (Pretest, Posttest). This is partially consistent with Gabriel et al. [35], in that our Stress and Non-Stress groups did not differ in MRT Accuracy. Inconsistent with our results, Gabriel et al. found that women became faster than men after stress. This inconsistency might be attributed to the difference in stressors used in the two studies. Stress researchers have identified that acute stressors can produce SAM activation, HPA activity, or both [68]. Furthermore, stressors may be categorized in many ways including, chronicity, predictability, controllability, or evaluative threat [69]; physical versus psychological [24]; and reactive versus anticipatory [70]. Gabriel et al. [35] used a physical stressor (cold pressor test) to induce stress, which involved participants’ immersing their hand in ice water for three minutes and encompassing elements of pain (reactive). Our present study used the PASAT, which is a SAM activating psychological challenge with reactive components.

While the second MRT stress intervention study by Cohen et al. [36] showed an overall improvement in MRT accuracy after stress, their experimental methodology differed substantially from Gabriel et al. [35] and our current study. Firstly, Cohen et al. only assessed MRT accuracy, which precludes comparison to the reaction time measures in Gabriel et al. and our present study. Second, Cohen et al. used a stronger social evaluative stressor (TSST), which resulted in HPA activation and a subsequent cortisol response. Since they were interested in the influence of HPA activation on performance, they waited 20 min post-stressor before administering the mental rotation test to assess the influence of peak cortisol during testing. Other stress studies resulting in HPA activation show that cortisol release can result in improvements in visuospatial tasks, such as map memory [29], visual working memory for faces and scenes [30], and visuospatial change detection [31]. It may be that stronger amounts of stress increase mental rotation performance or follow an inverted-U pattern in that extreme stress leads to subsequent impaired performance [71, 72]. The discrepancy between these stress and mental rotation studies may be that each experiment examined stress at different points along the inverted U-curve. Consequently, more research is needed to work towards an explanatory framework with respect to stress and mental rotation.

We also noted in our study that men were more accurate than women at Mental Rotation, with a medium effect size consistent with previous research [4, 5]. There were no Sex differences in Mental Rotation Response Time, a finding also commonly reported [73, 74]. All participants did however demonstrate a practice effect in that they became faster and more accurate in the MRT Posttest compared to the MRT Pretest, which is also consistent with the medium-large effect sizes found in previous research [75, 76].

To examine whether individual differences in cardiovascular measurements and self-reported affect would predict MRT performance we conducted a regression analysis. Our findings showed no relationship between cardiovascular reactivity (blood pressure and heart rate) and MRT performance, which is consistent with the lack of relationship between SAM activation (e.g. alpha-amylase) and mental rotation performance as reported by Cohen et al. [36]. Additionally, we found that neither self-reported perceived stress nor any of the self-reported negative or positive affect measures were related to MRT performance. Some studies show that inducing negative affect can impair cognition, including executive function [77], verbal working memory [78], and spatial working memory [26, 27]. In contrast, Kaltner and Jansen [79] reported that the presentation of fearful stimuli was related to improved mental rotation speed on body-based egocentric rotations, but not on object-based rotations (similar to the stimuli in this study). Similarly inconsistent findings have been demonstrated between positive affect and cognition, with both facilitation [80] and impairment [81] associated with spatial working memory performance. More research is needed to elucidate the influence of SAM activation and/or induced negative or positive affect on mental rotation performance.

The influence of stress on another spatial reasoning task where participants imagine how a scene looks from different viewpoints requiring imagined body rotations is also worthy of mention. Perspective taking is a spatial orientation task that requires participants to imagine how a scene looks from different viewpoints, requiring an imagined body rotation instead of imagined object rotation as in the MRT. One of our previous studies [22] showed that perspective taking performance slowed after acute stress. Although perspective taking and mental rotation are similar in that both require mental transformations of spatial information, and they are correlated to one another, confirmatory factor analysis showed that they are also dissociable factors [82] and involve different neural mechanisms [83]. It may be that acute stress only influences tasks including imagined body rotations.

Limitations and future directions

Previous research has shown that reactivity to laboratory challenges is affected by menstrual cycle hormonal fluctuation [84]. Oral contraceptive use and phase of the menstrual cycle were not controlled for with the women (n = 41; 51%) in our study. Studies investigating the effects of hormones on mental rotation without the presence of stress have shown differences across the menstrual cycle. Specifically, higher scores on mental rotation tasks during the menstrual phase and lower scores in the mid-luteal phase have been reported [85]. Greater accuracy and speed were also reported amongst women solving 3D mental rotation tasks in the early follicular phase when sex hormones are low, as compared during the mid-luteal phase, when progesterone and estrogen levels are higher [86]. These sex hormones may also interact with the stress response [36].

Two final limitations concern our participant sample. First, our group was homogeneous being composed exclusively of college students between 18 and 23 years of age. Second, the sample size was moderate, which may have been insufficient to detect small interaction effects. Expected effect sizes in the power analysis were based on known sex differences in mental rotation and on the expectation of a medium-large effect size, as detected by Cohen et al. [36], which may have been overestimated. It is also possible that our Non-Stress group may have produced an unexpected increase in cardiovascular indicators or SAM activation, reducing differences between our groups. Since our initial power analysis focused on the detection of two-way ANOVA interactions, the nonsignificant three-way interaction (Session X Group X Sex) should be viewed as exploratory rather than confirmatory. Consequently, future studies should employ more conservative effect size estimates when conducting power analyses to ensure adequate sample size.

Future studies should include a multi-dimensional approach for direct measurements of SAM activation, such as the collection of salivary biomarkers [87], as our study only included cardiovascular reactivity as a measure of SAM activation. One such bio-marker to consider is the collection and quantification of salivary alpha-amylase [88], which has been found to be a non-invasive, robust, and reliable metric to assess autonomic nervous system activity. Measuring the activation of neuroendocrine HPA activation and the subsequent effects of cortisol on mental rotation speed and accuracy should be another bio-marker assessed in future studies. Quantifying the effects of cortisol on MRT pretest and posttest performance would also be insightful. Designing future studies after stress tasks that are known to produce HPA activation, such as the TSST or SECPT, may further elucidate sex differences.

The influence of strategy selection after stress should also be considered. In some studies, acute stress induction results in no overall performance differences but can lead to changes in strategy. For example, Schwabe et al. [89] found that TSST stress led to a switch from a spatial to stimulus response strategy, and van Gerven et al. [48] showed that PASAT stress resulted in a shift from an egocentric to allocentric strategy in a virtual Morris water maze task. Individuals performing mental rotation have been shown to utilize different strategies in accomplishing the task (e.g., piecemeal vs. holistic rotation) [90, 91]. Further explorations might explore whether acute stress leads to changes in mental rotation strategy selection.

Conclusion

Our study is the most comprehensive investigation into the effect of acute stress on mental rotation performance to date. We included both physiological measurements of stress and self-reported measures of positive and negative affect, and perceived stress. Contrary to expectations, PASAT stress had no influence on speed or accuracy of the MRT, and Sex did not interact with our stress manipulation. We did demonstrate a practice effect between the Pretest and Posttest MRT administrations, however these increases in speed and accuracy between trials also did not interact with PASAT stress. As expected, men were more accurate than women at mental rotation. In summary, our results suggest that acute stress from a psychological stressor (PASAT) had no influence on mental rotation performance. Future investigations might confirm that acute stress has no influence on professions performing specific rotation tasks under pressure (e.g., during aviation or surgery simulation) and should also investigate the effect of physical stressors (e.g., cold pressor) or HPA-activation (e.g., TSST or SECPT) on mental rotation performance.

Abbreviations

SAM

Sympathetic Adrenal Medulla axis

HPA

Hypothalamic Pituitary Adrenal axis

SBP

Systolic Blood Pressure

DBP

Diastolic Blood Pressure

HR

Heart Rate

PASAT

Paced Auditory Serial Addition Task

MRT

Mental Rotation Task

TSST

Trier Social Stress Task

3D

3-dimensional

SECPT

Socially Evaluated Cold Pressor Test

PSS

Perceived Stress Scale

Authors’ contributions

A.E.R. and M.M.V.T. contributed equally to this manuscript and study; including conceptualization, study design, data collection and implementation, data analysis, writing the manuscript text, reviewing and reading the manuscript, preparing figures, and approving the final manuscript.

Funding

No internal or external funding was provided for this project.

Data availability

The dataset analyzed for the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

This study was approved by the Saint Michael’s College Institutional Review Board (ID#: IRB-22-2). All participants provided written informed consent at the beginning of the testing session.

Consent for publication

All participants provided written consent for their data to be aggregated and published.

Competing interests

The authors declare no competing interests.