The relation between physical and mental load, and the course of physiological functions and cognitive performance

Abstract

The objective of our study was to investigate differences in cognitive performance connected with physical load of varying intensities. One half of 88 examined persons sat on office chairs, and the other half sat on chairs with the added modification of the gymnastic (Swiss) ball called the dynamic directional seat pad (pad). The first rest phase was followed by the load phase, in which the subjects were administered a 20-minute sustained attention test. The number of correct answers and errors was evaluated. A BIOPAC apparatus continually recorded thoracic respiration, electrodermal activity, finger temperature, heart rate, and heart rate variability. Females on pads made 58% fewer errors than females on chairs; the number of errors was closely related to the depth of their breathing (tidal volume). It was found out that the use of the pad, in addition to the already known health benefits, also brings an increase in the precision of cognitive performance.

LIST OF ABBREVIATIONS

Keywords:

• Cognitive performance
• respiration
• electrodermal activity
• finger temperature
• heart rate variability

Relevance to human factors

It was found out that the use of the pad (slightly modified iteration of the Swiss ball), in addition to the already known health benefits, also brings an increase in the precision of cognitive performance (the number of errors was closely related to the depth of breathing). This fact could be helpful wherever the increased frequency of cognitive errors complicates the normal functioning of people in areas of their daily life. The fact that physical load dominates over mental stress in the behavior of physiological functions and that the number of errors is another criterion of cognitive performance besides the speed of work, could be also useful in the field of human factors.

Introduction

Understanding the most elementary components of mental and physical makeup and their relations is a necessary condition for the optimum adjustment of physiological functions that contribute to cognitive performance. However, research in this field still faces various methodological limitations (Boucsein 2012a). The fundamental fact of an inverted U relationship between physical activation (arousal) and cognitive performance was for example identified while walking on a treadmill (Levitt and Gutin 1971) or riding a bicycle ergometer (Salmela and Ndoye 1986; Sjöberg 1975; Basahel 2012). Although such research studies allow the measurement of heart rate, respiration or reaction time tasks during a physical load, the measurement of electrodermal activity, body temperature, heart rate variability or more complex cognitive processes (Gutin 1973) requires another methodological background (Brookhuis, et al. 2005; Mulder 1992). This methodology makes it possible to better detect the cause of changes in the course of physiological functions, e.g. to detect the influence of the autonomic nervous system (ANS) on the course of cognitive load and vice versa (Dawson, Schell, and Filion 2007; de Waard 1996; Boucsein 2005). Another limitation of such research studies is the separate measurement of physical or mental load (Antunes et al. 2006), or the measurement of mental load and cognitive performance before or after the end of physical load (Salmela and Ndoye 1986; Coles and Tomporowski 2008).

The aim of our study was therefore to investigate and further specify differences in cognitive performance (mental load) and the course of physiological functions connected with simultaneously occurring physical loads of varying intensities.

The findings of our study could be particularly helpful in teaching children with ADHD or ADD, in reducing the number of car accidents by eliminating momentary drowsiness, in curing anxiety and depression (Courtney 2009) or in the prevention of neurodegenerative diseases (Kramer and Colcombe 2018). In general, it could be helpful wherever the increased frequency of cognitive errors complicates the normal functioning of people in areas of their daily life (Boyer et al. 2015; Takeuchi 2000).

As a tool for increasing the physical load (Holinka, Gallo, and Zapletalová 2016, Holinka et al. 2017), we chose the dynamic directional seat pad (hereinafter “pad”) for our research. It is a smaller, and in shape, slightly modified iteration of the Swiss ball (Carriere 1998). Thanks to these modifications, it can be used without complications on a common office chair during everyday sedentary work. The effect of Swiss (gymnastic) balls in the exercise therapy of muscle strength and posture development or muscle relaxation and flexibility has been documented many times (Balakrishnan, Yazid, and Mahat 2016; Lee, Bang, and Ko 2003; Mills 1994; Marshall and Murphy 2005; Cosio-Lima et al. 2003). Similar effects on muscle strength and posture development were also confirmed for the pad (Holinka, Gallo, and Zapletalová 2016, Holinka et al. 2017).

Physiological functions associated with or affecting cognitive functions often mentioned in the literature (Brookhuis, et al. 2005; Mulder 1992) include respiration (R), electrodermal activity (EDA), heart rate (HR), heart rate variability (HRV), and finger temperature (FT).

It seems that there are quantitative and qualitative differences in the two R parameters (and in R patterns in general) depending on whether they are provoked mentally (through cognitive processes, volitional effort, emotions) or physically (physical work, exercise; Ganong 2005; Švancara 2003; Wientjes 1993; de Waard 1996). At light to moderate physical loads, higher minute ventilation (MV) is primarily induced by increased tidal volume (VT), while respiratory rate (RR) increases considerably only with a heavy physical load. In contrast, with increasing cognitive load – generally with a slight amount of mental arousal – RR accelerates but VT is shallower (Davies, Haldane, and Priestley 1919; Filo 2014; Veltman and Gaillard 1998; Wientjes 1993; Wilson, Fullenkamp, and Davis 1994). MV, which is a product of RR and VT over the course of one minute, increases as well. Hence, with mental effort RR accelerates greater than VT decreases. In summary, VT and MV in general are usually distinctly higher with physical load than with cognitive load; however, there is only a minimum difference in RR with the two types of loads (Wientjes 1993).

EDA can be considered a psycho-physiological indicator of arousal (Dawson, Schell, and Filion 2007; Boucsein et al., 2012b; Boucsein 2005; Kramer 1990; Rokyta 2000; de Waard 1996) in the subject, which must not be “contaminated” by thermoregulation processes (Rokyta 2000), stress hormones (Weiner 1982), and so forth (Boucsein 2012a; Kramer 1990; Schmidt et al. 2017). When the greater cognitive functioning associated with increased attention leads to higher arousal, EDA increases (Fröberg 1977; Boucsein 2005; Boucsein, Haarmann, and Schaefer 2007; Mulder et al. 2000; Reimer and Mehler 2011; Shimomura et al. 2008; Son et al. 2011). EDA is also affected by physical load and by irregularities in respiration (EDA grows with deeper breathing; see Boucsein 2012a).

There is no question about the effects of cognitive processes on HR and HRV (Brookhuis, et al. 2005; Kramer 1990; Schapkin et al. 2007; de Waard 1996; Weiner 1982). In the case of short-term tasks requiring sophisticated psychological (executive) operations relying on working memory or selective attention, however, the general cardiovascular pattern involves an increase in HR and blood pressure (increased blood pressure resulting in its reduced variability), a decrease in HRV (Kramer 1990; Mehler, Reimer, and Wang 2011; Miyake et al. 2007; Veltman and Gaillard 1998; de Waard 1996; Wilson, Fullenkamp, and Davis 1994), and a decrease in baroreflex sensitivity (Mulder et al. 2000). Although a loss in power occurs in all frequency bands (Brookhuis, et al. 2005), the loss in the mid-frequency band (0.10 Hz) seems to be most often demonstrated when there is an increased load of “resource-limited” (Kahneman 1973) cognitive processes (Kramer 1990; Mulder et al. 2000; Mulder et al. 2005; de Waard 1996). It also occurs when there is a constant mild physical load (de Waard 1996), which is the subject of our research. During physical activity HR increases and HRV decreases; nevertheless, intentionally slowing down HR results in ambiguous HRV behavior (Billman 2013; Billman et al. 2015). Apart from ANS, HR and HRV can be considerably influenced by breathing (Billman 2013; Ganong 2005; Kramer 1990; Mulder et al. 2000; Porges 1995; de Waard 1996) as well as by other mechanical effects leading to atrium compression and thoracic pressure change (Ganong 2005; Rokyta 2000; Van Roon et al. 2004), and by increased mental effort (increased cognitive activity). HRV decreases with increasing RR and increases with increasing VT. These phenomena have also been experimentally demonstrated to act in the opposite manner: HRV decreases with decreasing VT and increases with decreasing RR.

Body temperature values and their effects on the human body differ considerably (Boucsein 2012a; Ekman, Levenson, and Friesen 1983; Jirák et al. 1997; Kreibig et al. 2007; Pigeau et al. 1995; Švancara 2003; Takeyama et al. 2005; Vinkers et al. 2010; Weiner 1982; Wever 1979). It seems that mental-processing rates (including processing visual cues and other forms of cognitive processing causing mental load; see Calvin and Duffy 2007) depend on the degree of metabolic activity (or the arousal) of the brain’s cortex cells (Tůma 1989). Increased metabolic activity is related to increased core temperature (e.g., rectal, tympanic, etc.; see Folkard 1990; García-García and Drucker-Colín 1999; Kleitman 1963; Whang et al. 2007), while finger-skin-surface temperatures (i.e. peripheral body temperatures) generally decrease in such cases (Fröberg 1977; Vinkers et al. 2010).

We expect our research to confirm the inverted U relationship between activation and cognitive performance. Specifically, that the combination of a medium physical load (arousal) with a cognitive load will lead to higher cognitive performance (Levitt and Gutin 1971; Sjöberg 1975; Basahel 2012), and to higher values of physiological functions (Antunes et al. 2006; Fredericks et al. 2005) than when there is a cognitive load with a low physical load (arousal). Load intensity was determined by whether the examined person (EP) was sitting directly on a chair (hereinafter “chair”) or on a chair with a pad. Sitting on a chair can be considered a low physical load, while sitting on the pad can be considered a medium physical load (Holinka, Gallo, and Zapletalová 2016, Holinka et al. 2017; Wientjes 1993).

We shall further focus our attention on the course of cognitive performance. It is known that with a new task, performance improves at the beginning through conscious training and becomes stable at its optimum (the negative acceleration curve, Sternberg and Sternberg 2012). Nevertheless, it fluctuates according to the degree of automation (habituation) or according to physiologically (biologically or physically) given oscillations of attention (Senka, Kuruc, and Čečer 1992; Chmelař and Osecký 1974).

We also have to reckon with the assumption that females will have perceptual sensitivity lower than males in the visual test for sustained attention (Dittmar et al. 1993; Prinzel and Freeman 1997), which may reflect in greater changes of their cognitive functions (Kakizaki 1987).

Materials and methods

Research design and research sample

All measurements were carried out in an air-conditioned laboratory at a constant temperature of 23˚C (Boucsein 2012a; Dawson, Schell, and Filion 2007; Wever 1979), a humidity of 40–50%, a noise level lower than 45 dB (as measured by a calibrated noise meter), and illuminance of 1500–4000 lx (lux meter) in order to prevent possible influence on physiological functions (Kramer 1990; Wever 1979). Measurements were taken from October 2016 to January 2017. Altogether 88 EPs (29 female and 59 male aged 19–27) took part in our study. We tried to assemble a relatively homogeneous sample of healthy EPs without baseline differences between our study groups. The EPs were second-year forestry students with the same academic specialization. They were randomly divided by lot into two groups, a chair and a pad group, with an equal ratio of male and female EPs in each group. All subjects were strictly forbidden to drink alcohol for a minimum of two days before the measurements were taken (Mulder and Mulder 1987). Measurements began between 08:30 and 08:45 to ensure similar fatigue rates among EPs (for the effects of biorhythms and cognitive loads see Dawson, Schell, and Filion 2007; Wever 1979). None of the EPs had experienced sitting on the pad before. Short-term illnesses and long-term health problems (allergies, anemia, asthma, diabetes, etc.) were noted (Rokyta 2000; Billman 2013). EPs in the control group were seated on wooden, height-adjustable non-upholstered chairs with an unshaped seat so that they sat with their knees bent at a right angle. EPs seated on pads were forbidden to lean against the backrests of their chairs. Researchers monitored the movement of EPs on the chairs, which was supposed to be as limited as possible (Mulder 1992). The EPs were not informed of the duration or purpose of the experiment. They were forbidden to talk during the measurements because talking can considerably affect HR, HRV, and R behavior (Mulder and Mulder 1987; Mulder 1992; Wientjes 1993).

The measurement process involved the following steps:

1. EPs were seated on wooden chairs and received oral instructions about the sequence of individual measurement phases, including an explanation of the test of sustained, selected attention (Eysenck and Keane 2000; Bates and Lemay 2004; Brickenkamp and Zillmer 2000; Gabrhel 2014; Chmelař 1970a; Chmelař and Osecký 1970b, 1974; Mulder et al. 2000; Sternberg and Sternberg 2012). We used a version of the Bourdon test referred to as T-78, or BoPr (hereinafter, the Bo test), a type of cancelation test (Senka, Kuruc, and Čečer 1992). Test performance is evaluated by counting correctly cancelled symbols (BoC) and incorrectly cancelled symbols (errors; BoE). Practice was not permitted because it could influence the performance of EPs differently.

The test can ascertain the accuracy and rate of scanning, as well as the learning strategy and its development (Bates and Lemay 2004). Because of its duration and complexity, the capacity for shifting and holding attention (elements of permanent attention) and the capability to focus on selected stimuli (elements of selective attention) can be detected and discriminated between (Senka, Kuruc, and Čečer 1992). These “prototypical” capabilities (Mulder et al. 2000) are often assessed using this test (or by using a similar test named d2), e.g. in professional drivers (Gabrhel 2014), but also in testing performance in sport or in schools (Brickenkamp and Zillmer 2000). The Bo test was standardized on a sample of 1020 persons of different professions aged 17 − 36 years (Senka, Kuruc, and Čečer 1992). Results exhibited a normal distribution. Internal consistency (split-half) in BoE was r = 0.787 and in BoC: r = 0.989. Other validity and reliability indicators can be found in Senka, Kuruc, and Čečer (1992).

1. The control phase of the measurement lasted 2 minutes. In this phase, all EPs sat on chairs with no pads. Its purpose was to analyze the initial values of the EPs’ physiological functions.

2. The following resting phase of the measurement lasted 20 minutes. Control-group EPs continued sitting on chairs without pads, while EPs in the other group were seated on pads that had been added to their chairs. They would then sit on these pads until the end of the fourth phase. During the second phase, EPs were not to show any deliberate cognitive activity. Activities involving inadvertent attention, however, were allowed, for example, looking forward at the table or out of the window.

3. The load phase lasted 20 minutes, during which the EPs continuously filled out the Bo test.

4. The verification phase lasted 2 minutes. Following the end of the cognitive load, this phase also served to compare the development of physiological functions in the two groups of EPs.

A graphical visual illustration of our experiment setup is provided in Figure 1.

Figure 1. Graphical illustration of experiment setup.

Acquisition of biosignals

All four biosignals were measured simultaneously at the same sampling frequency of 1000 Hz and the same quantifying resolution of 24 bits. The biosignals were recorded using a Biopac MP35 (Biopac, Inc.) acquisition unit. All data were visually inspected once again, normal values were evaluated (Dawson, Schell, and Filion 2007), and deviations from standard oscillations (Göbel, et al. 2005) were examined.

Inspiration maxima (VT) were detected automatically by the PhysioCrate toolbox plug-in (Nejedly and Virgala 2016) and by SignalPlant software (Plesinger et al. 2016), and visually inspected at every step. Inspiration depth was measured as a relative chest volume change (VTR) with units in mV. To explore differences in mean VT values related to the use of the pad, it was necessary to transform the data. Absolute chest volume change (VTA) was subsequently calculated and expressed in liters from relative chest volume change (VTR), see Wientjes and Grossman 2005. We based this on the value of one inspiration in both males and females, which amounts to 0.5 l (Ganong 2005; Rokyta 2000) and on the minute ventilation (MV) at rest of healthy females aged 18–25 years, which is 10.5 liters. To ensure accuracy, we selected data from two time ranges during the control phase of the experiment (Bland and Altman 2010). The EDA curve carries information on skin conductance level (SCL) and skin conductance reflex (SCR). For the purposes of assessing the activity of sympathetic nerves, we used only skin conductance level (SCL) as a robust indicator of the sympathetic activity rate. We evaluated HR (1/1000 seconds), which was acquired from an electrocardiography (ECG) as the reverse value of the difference between consequent R-peaks detected by the QRS seeker program (Vítek and Kozumplík 2011). HRV parameters were analyzed on an R-R interval series (Barbieri et al. 2005). The heart rate variability analysis software (HRVAS) plug-in for Matlab software was used to compute HRV parameters (Ramshur 2010; Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology 1996).

Time domain parameters were evaluated, namely the mean of R-R interval (R-R mean) and its reciprocal value mean of heart rate (HR mean), standard deviation of R-R intervals (SDNN), root mean square of the successive differences (RMSSD), the number of successive intervals differing more than 50 ms (NN50) and its corresponding relative amount (pNN50). Geometric domain parameters included HRV triangular index (TI) and the baseline width of the R-R histogram evaluated through triangular interpolation (TINN). Frequency domain parameters were evaluated utilizing three different methods for spectra estimation: Fourier transform, Welch periodogram with window width of 256 samples and window overlap of 128 samples, and Burg autoregressive model with 16th order. All periodograms were divided into three standard bands: VLF 0 − 0.04 Hz, LF 0.04 − 0.15 Hz, HF 0.15 − 0.4 Hz. Frequency domain parameters were evaluated for results of each method: VLF, LF and HF band peak frequencies (VLF peak, LF peak, HF peak), absolute powers of VLF, LF, and HF bands (aVLF, aLF, aHF), relative power of VLF, LF, and HF bands (pVLF, pLF, pHF), powers of LF and HF bands in normalized units (nLF, nHF) and ratio between LF and HF band powers (LFHF). Non-linear HRV parameters assessed in this work included sample entropy (HRV Entropy), short term fluctuation slope (alpha 1) and long term fluctuation slope (alpha 2) as a result of detrended fluctuation analysis with a range from 4 to 64 samples and break point equal to 11, and Poincaré diagram descriptors SD1 and SD2 representing the standard deviation of diagram perpendicular to (SD1) and along (SD2) the line-of-identity (Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology 1996).

Standards for the evaluation of HRV unify the length of the assessed tachograph section to 5 minutes, so that both short and long-term HR changes can be optimally captured for a correct quantification of which a time section of at least 2 minutes is needed (Task Force of the European Society of Cardiology and the North American Society of Pacing and Electrophysiology 1996).

Results

Along with the description we are presenting an analysis of the dependent variable number of BoE in relation to the independent variables of gender and pad/chair use (ANOVA). We will also present the analysis of the dependent variables, namely physiological functions of females, in relation to the independent variable – pad/chair (t-test, the Mann-Whitney test). Finally, we will provide a description and statistical analysis of the contribution of physiological functions (predictors) to differences in average BoC and BoE values (dependent variables) in females seated on the pad or chair (standardized coefficients Beta and coefficient of multiple determination R²).

Cognitive functions

Firstly, mean values for BoC and BoE scores were calculated for each minute of the test separately. Then a total mean BoC and BoE score was calculated from the twenty partial minute average values.

In the set of males, no significant differences were found in connection with the use of the pad (see Table 1). As for the mean scores, differences were found only in the set of females and only in their BoE scores (Table 1). The number of BoE made by females sitting on the pad was statistically significantly lower than the number of BoE made by females sitting directly on the office chair (Table 2). The difference was approximately two errors, which was (with the mean number of errors being 1.77/min) a 58% difference in the number of errors in connection with the use of the pad. This overall contrast in BoE scores is primarily the result of significant differences between the beginning of the test and its last third (see Figure 2). The findings are exacerbated by the fact that higher (non-significant) BoE scores in females seated on the pads were found only in one minute (specifically in the sixth minute) of the 20 minutes measured. The power of independent samples in the t-test equals 0.79 (SD = 2.81; N = 15, 13; Es = 1.09). In the physiological functions, erroneous records of 5 EPs (1 female and 1 male seated on the chair, and 3 females seated on the pad) and 1 unfinished Bo test from a female seated on the chair were discarded. Thus, to compare the Bo test performance with the development of physiological functions, we worked with a sample of 24 females (chair = 12, pad = 12) and 58 males (chair = 28, pad = 30). Besides the subgroup of females seated on the pads (Kolmogorov-Smirnov = p < 0.008; Shapiro-Wilk = p < 0.001), the acceptable normality of distribution in subgroups was achieved by dividing gender by chair/pad (males on the chairs: Kolmogorov-Smirnov = p < 0.200; Shapiro-Wilk = p < 0.284; males on the pads: Kolmogorov-Smirnov = p < 0.114; Shapiro-Wilk = p < 0.033; females on the chairs: Kolmogorov-Smirnov = p < 0.200; Shapiro-Wilk = p < 0.980). The group of females on the pads exhibited right-skewed BoE data due to a relatively low median value. Due to this and the relatively small sample of females, both parametric and non-parametric statistical methods were used.

Figure 2. The number of BoE over the course of 20 minutes in males (A) and females (B) seated on chairs/pads (means, error bars of SD, simple moving average - SMA).

Table 1. Descriptive data on the number of BoE in all EPs and separately in males and females without outliers.

		Males (without outliers)
Mean of 20 minutes		N	Total Missing	Mean	Median	Std. Deviation	Std. Error Mean	Mean Rank	Sum of Ranks
BoC/minutes	Chair	28	0	30.16	30.30	3.69	0.70
	Pad	30	1	30.18	30.13	5.07	0.92
BoE/minutes	Chair	28	0	2.99	2.68	1.69	0.32
	Pad	30	1	3.13	2.80	1.65	0.30
		Females (without outliers)
BoC/minutes	Chair	13	1	30.42	30.30	5.29	1.47
	Pad	15	0	29.31	29.80	3.42	0.88
BoE/minutes	Chair	13	1	4.17	4.15	2.14	0.59	18.23	237.00
	Pad	15	0	2.40	1.85	1.10	0.28	11.27	169.00

Table 2. Significant differences in the number of BoE in females (without outliers) ascertained by t-test for independent samples, the Mann-Whitney test for two independent samples and the Kolmogorov-Smirnov test.

Females (without outliers)		t-test for Equality of Means
		t	df	Sig. (2-tailed)	Mean Difference	Std. Error Difference	95% Confidence Interval of the Difference
							Lower	Upper
BoE/minutes	Equal variances not assumed	2.82	26	0.009	1.78	0.63	0.48	3.07
Mann-Whitney U	Wilcoxon W	Z		Asymp. Sig. (2-tailed)		Kolmogorov-Smirnov Z	Asymp. Sig. (2-tailed)
49.00	169.00	−2.236		0.025		1.502	0.022

The significant BoE differences among females were identified by both the t-test for independent samples and its nonparametric analogues, the Mann-Whitney test for two independent samples and the Kolmogorov-Smirnov test.

Based on a univariate analysis of variance (ANOVA), we further explored differences in average BoE values in all EPs, in connection with both use of the pad and gender. Their potential interactions were also investigated. Table 3 shows the main effect of the pad, with the effect of differences between males and females being significant only in the interaction with the pad (in the number of errors, the gender effect was significant only in the fourth minute of the Bo test; T-test: P < 0.038; Mann-Whitney: P < 0.045). To sum up, BoE numbers were affected most strongly by the mutual interaction of the two factors; the more important role was undoubtedly played by the factor of the pad.

Table 3. Univariate analysis of variance (ANOVA) of the number of BoE in relation to gender and pad use.

Tests of Between-Subjects Effects
Source		df	Mean Square	F	Sig.
Corrected Model	BoE	3.00	7.586	2.724	0.050
Intercept	BoE	1.00	757.396	271.961	0.000
Pad, chair	BoE	1.00	12.504	4.490	0.037
Gender	BoE	1.00	0.948	0.340	0.561
Pad, chair * gender	BoE	1.00	17.377	6.240	0.014

Differences among the mean values of physiological functions in relation to the pad

Dissimilarities were found in VTR and RR and their derivatives VTA and MV, and also in FT, SCL, HR and HRV “sample entropy” (hereinafter HRV Entropy), which was calculated from the series of R-R intervals. A description of all physiological functions is presented in Table 4.

Table 4. Descriptive data on RR, VTR, VTA, MV, FT, SCL, HR and HRV Entropy of females. The table presents descriptive data on selected physiological functions capturing differences between females seated on chairs/pads during the load and rest phase.

	Females (without outliers)
	Load Phase 3							Rest Phase 2
		N		Median	Mean	Std. Deviation	Std. Error of Mean	Median	Mean	Std. Deviation
		Valid	Missing
RR	Chair	12	1	13.32	12.92	1.89	0.55	12.79	12.93	3.18
	Pad	12	3	14.43	14.36	1.39	0.40	13.90	14.10	2.33
VTR (mV)	Chair	12	1	0.79	1.41	1.27	0.37	0.50	1.17	1.12
	Pad	12	3	2.28	2.11	1.05	0.30	1.49	2.02	1.14
VTA (l)	Chair	12	1	0.76	0.87	0.47	0.13	0.58	0.60	0.18
	Pad	12	3	0.51	0.85	1.01	0.29	0.44	0.76	0.86
MV (l)	Chair	12	1	9.38	11.08	6.18	1.78	7.67	7.59	2.98
	Pad	12	3	6.88	11.67	13.78	3.98	6.17	10.11	10.59
FT (°C)	Chair	12	1	31.73	31.28	3.34	0.96	35.06	33.88	2.80
	Pad	12	3	31.93	31.88	2.55	0.74	34.36	33.50	2.13
SCL (µS)	Chair	11	2	17.70	18.52	7.64	2.30	15.12	15.51	8.02
	Pad	11	4	20.43	22.77	8.19	2.47	16.44	19.39	8.69
HR (bpm)	Chair	12	1	85.26	86.11	8.20	2.37	80.45	80.59	5.68
	Pad	12	3	85.96	86.76	9.11	2.63	83.89	82.98	8.97
HRV Entropy	Chair	12	1	1.56	1.58	0.49	0.14	1.59	1.70	0.35
	Pad	12	3	1.89	1.81	0.49	0.14	1.48	1.48	0.36

We have presented the results in Table 5, which were produced using categorized data, that is, from the averaged values of physiological functions for each minute of the 20-minute measurements. There are many reasons why we approached the criterion of significance with caution. One of the major ones is that means do not capture dynamic changes (oscillations) of functions in the course of measurements, which considerably impairs their informative value, especially when the function oscillations are distinctive and frequent. Apart from this criterion, we also assessed the individual findings according to the practical value of differences. For example, if a difference of 0.6 °C (between the examined groups) is actually important for real life situations or not.

Table 5. Significant differences (Sig. 2-tailed) between the physiological characteristics of females (without outliers) seated on the chairs/pads ascertained by means of t-test for independent samples and power calculation (power, standardized effect – Es) for N = 12, 12 and during the load (3) and rest (2) phase ascertained by means of paired samples t-test and its nonparametric analogues, the Wilcoxon signed ranks test for paired samples.

Females (without outliers)	Independent Samples T-Test: chair/pad phase 3
	t	df	Sig.	Mean Diff.	Std. Error Diff.	Power	Es
RR	−2.13	22.00	0.044	−1.44	0.68	0.539	0.878
	Paired Samples Test: pad phase 3/phase 2					Paired Samples Test: chair phase 3/phase 2
	t	df	Sig.	Mean	Power	t	df	Sig.	Mean	Power
VTA	–	–	–	–	–	2.39	11.00	0.036	0.28	0.419
FT	−2.43	11.00	0.033	−1.62	0.369	−3.77	11.00	0.003	−2.60	0.510
SCL	2.26	10.00	0.047	3.37	0.157	3.17	10.00	0.010	3.01	0.148
HR	2.45	11.00	0.032	3.78	0.167	2.41	11.00	0.034	5.51	0.451
HRV Entropy	2.36	11.00	0.038	0.33	0.438	–	–	–	–	–
	Wilcoxon Signed Ranks Test: pad phase 3/phase 2					Wilcoxon Signed Ranks Test: chair phase 3/phase 2
							Z	Sig.	Mean Rank	Sum of Ranks
VTR	–	–	–	–	–		2.040	0.041	7.22	65.00
	–	–	–	–	–				4.33	13.00
MV	–	–	–	–	–		2.118	0.034	6.60	66.00
	–	–	–	–	–				6.00	12.00

No significant differences were found in RR function in connection with sitting on the pad in the analysis using the Mann-Whitney test and the Kolmogorov-Smirnov test, which demonstrate its quite parametric distribution.

Differences in the development of physiological functions and cognitive performance in connection with the use of the pad

Figure 3 shows at least two basic trends in the oscillations (slight and conspicuous) and behaviors (ascending and descending tendencies) of the variables examined. Another assessment criterion could certainly be the degree of influence of “breaking points,” that is, the minutes in which distinctive changes occurred in the variables. Such breaking points are minutes 26 and 27, minutes 33 and 34, and to a lesser extent minutes 23 and 24, and minute 39. VTR, FT, SCL, and also BoE and HRV Entropy exhibited rather small oscillations with either ascending or descending tendencies, hinting at the possibility of introducing a linear function despite the partial influence of breaking points. The physiological functions mentioned increased in EPs sitting on pads during the test, while the number of BoE was continually decreasing. In contrast, the variables of MV and BoC exhibited generally strongly variable “behavior” that was more affected by the “breaking points.” Here too we observed increasing BoC performance in connection with sitting on the pad, although a noticeable slump occurred between minutes 27 and 33. The conspicuous inverted character of the behavior of both MV and BoC variables, particularly apparent in females seated only on the chairs, was also important. It must also not be neglected that the average values of the physiological functions of females were always higher when sitting on the pad during the whole measurement period (20 minutes) of the load phase than the values recorded when sitting just on the chair.

Figure 3. (C – H). Comparison of selected physiological functions with BoC (BoE; see Figure 2) in the two groups of females (means, error bars of SD, SMA). Figure 3 provides valuable data about frequent instances of the completely inverted course of the data mentioned (e.g., between MV and BoC).

A thorough investigation of the contribution of physiological functions to differences in average BoC and BoE values (standardized coefficients Beta) in females and the extent to which their share can be explained by means of these functions (coefficient of multiple determination R²) will be discussed in another paper. Current results show that the number of BoE was related most closely to VTR in both female groups. EPs seated on the chairs: R = 0.36; R² = 0.13; ANOVA: F = 35.96; P < 0.001; stand. Beta = −0.36; t = −6.00; P < 0.001. EPs seated on the pads: R = 0.31; R² = 0.09; ANOVA: F = 24.81; P < 0.001; stand. Beta = −0.31; t = −4.98; P < 0.001.

Discussion

The goal of this study was to search for differences in cognitive performance related to various intensities of physical load. Sitting on the chair can be considered a low physical load, while sitting on the pad can be considered a medium physical load (Levitt and Gutin 1971; Sjöberg 1975; Basahel 2012; Holinka, Gallo, and Zapletalová 2016, Holinka et al. 2017; Wientjes 1993).

Differences in cognitive performance were observed only in females. The assumption that females respond more sensitively to cognitive load than males appears to be valid (Kakizaki 1987). The explanation may lie in the lower perceptual sensitivity (Dittmar et al. 1993; Prinzel and Freeman 1997) which appears in females as compared with males in the tests of sustained attention to visual stimuli. It should be added that the EPs measured had no previous experience with the pad whereas the effects on muscle strength and posture development are at the strongest only after 3 months of everyday use of the pad (Holinka, Gallo, and Zapletalová 2016, Holinka et al. 2017). The question is, therefore, whether the results would not also be the same in males after that time, because of dissimilarly shaped pelvic floors, dissimilar attachment of muscles in this region, or dissimilar physical strength and amount of striated muscles present in males and females (Åstrand et al. 2003; Šedivý 2004; Šedivý 2006).

Females sitting on the pad made a significantly lower number of errors than those sitting on the chair. The basic assumption was therefore confirmed. When a medium physical load (20%–50% Wmax, cycling on a stationary bicycle fitted with an ergometer) and cognitive load are combined, cognitive performance increases as do the values of physiological functions (Basahel 2012; Antunes et al. 2006; Fredericks et al. 2005;) compared with when a low physical load (chair) and cognitive load are combined (Levitt and Gutin 1971; Sjöberg 1975). The findings can be interpreted by using the known function of inverted U in line with the Yerkes-Dodson law (Colquhoun 1971; Folkard 1975; Folkard 1990; Kahneman 1973). Nonetheless, Salmela and Ndoye (1986) or Basahel (2012) did not tackle the question of whether the number of errors is another criterion of cognitive performance besides the speed of work.

Surveys of Bo test validity indicate that errors are a measure of performance, which differs from the rate of processing that manifests itself in the number of correct answers (Břicháček, Brichcín, and Malotínová 1969; Senka and Daniel 1970). According to the factor analysis, BoC was part of the factor of “rate of operations” (Břicháček, Brichcín, and Malotínová 1969) or factors of perception, attention, imagination, quick-wittedness and memory (Senka and Daniel 1970). BoE then fell either in the category of “erroneous operations” (Břicháček, Brichcín, and Malotínová 1969), or – as claimed by Senka and Daniel (1970) – BoE created a special factor which is common with the errors of short-term numeric memory.

Our Bo test is similar to the well known d2 test of attention. According to Brickenkamp and Zillmer (2000), the rate of performance in the d2 test (in our case, BoC) reflects – in addition to the accuracy of visual perception – the degree of arousal (motivation side), while the number of errors (BoE) reflects the degree of self-control by means of conscious attention (Eysenck and Keane 2000; Gabrhel 2014; Chmelař 1970a; Chmelař and Osecký 1970b, 1974; Mulder et al. 2000; Sternberg and Sternberg 2012).

Bates and Lemay (2004) have further developed these findings, confirming the rather low correlation between the number of correct answers and the number of errors in the d2 test (Wientjes (1993) also arrived at similar results). It follows that the two outputs, search speed and accuracy, are in a way two independent cognitive dimensions. These authors have developed two concepts to evaluate these two independent factors: selective scanning accuracy, which is saturated by varying scores in the number of errors, and selective scanning deterioration or acceleration, which is saturated by scores in the distribution of the number of correct answers or errors in time during the test. The authors explain two possible mechanisms behind errors. Mistakes may be the result of incorrect motor reactions which may manifest themselves during a short-term drop in vigilance (in accordance with Brickenkamp and Zillmer 2000), or, from the viewpoint of the theory of signal detection, a result in a drop in the tendency to subordinate to the strict criterion of cognitive (mental) performance, which leads to interference with the implementation of the established goal.

In the following part of the discussion, we shall try to detect causes of the changes in cognitive and physiological functions according to literature results focused particularly on cognitive load (Boucsein 2012a). Compared with the results of research focused on physical load, these allow for the exploration of the effect of ANS on the course of cognitive and physiological functions directly (Dawson, Schell, and Filion 2007; de Waard 1996; Boucsein 2005) and offer a more precise explanation of their partial relationships (Antunes et al. 2006). Unfortunately, similar research for a rigorous comparison with our findings is lacking.

Supposing that the state of the EPs’ attention was continually monitored, the degree of the BoE can be derived from the degree of ANS activity. Since the degree of cognitive load was constant and hence the cognitive load can be categorized as mental effort devoted to the processing of information in the controlled mode (Mulder and Mulder 1987), it seems that changes in the output efficiency of EPs (number of errors) were actually related to the level of vigilance given by physiological arousal (Boucsein, Haarmann, and Schaefer 2007). According to the results of linear regression, we see a key function for maintaining vigilance (lucidity) in VTR (VTA) values. These are also reflected in RR values and in MV level. In females seated on the chairs, VTR decreased with increasing RR and HR, and the number of BoC increased with increased HR. Thus, we confirmed the frequently quoted relationship between physiological-function patterns and cognitive load (Kramer 1990; Mehler, Reimer, and Wang 2011; Miyake et al. 2007; Veltman and Gaillard 1998; de Waard 1996; Wilson, Fullenkamp, and Davis 1994). In this group of females, RR was so high during the load phase that the values of gross MV (l/min) increased despite the decreasing VTR values. Wientjes (1993) reported the same findings. Faster RR and shallower VT do not seem to provide the R pattern most appropriate for optimal cognitive activity. These findings were not observed in females seated on the pads. Their RR was not fast enough to contribute significantly to the resulting MV value. When RR eventually accelerated sporadically, it was reflected only in shallow breathing (VTA). Wientjes (1993) also found that under light to moderate physical load, higher R (MV) is primarily induced by increased VT. Deeper breathing can save more energy for cognitive performance than faster breathing because a further increase of RR apparently becomes inefficient due to the excessive use of respiratory muscles. Davies, Haldane, and Priestley (1919) explain this phenomenon as “secondary anoxemia” that arises during the “ordinary fatigue” of the respiratory center, which cannot even be suppressed by artificially adding O₂ into the inspired air (Davies, Haldane, and Priestley 1919). It is possible that increased muscular activity (deep stabilization of muscles of the back, abdomen, and buttocks; see Holinka, Gallo, and Zapletalová 2016, Holinka et al. 2017) impairs natural inspiration. In such cases, VT becomes deeper and RR slower (a decrease from approximately 14 to 7 inspirations per minute). Moreover, compared with the other group of EPs, females seated on the pads exhibited lower VTR values in the initial section of the load phase, which points to the same findings of Davies, Haldane, and Priestley (1919), when exploring the initial phase of respiration under conditions of increased resistance.

We further discovered that in females seated on the chairs, SCL increased with decreasing FT. This phenomenon was documented several times by Boucsein (2012a). By contrast, in females seated on the pads, SCL increased with increasing FT. It seems that due to physical load the dilatation of blood vessels and an increase in the number of open capillaries (Ganong 2005) are so prevalent in active muscles, and consequently in the entire body of females seated on the pads, that physical load trumps the effects of mental load. Thus, FT values determined in our study could also be the result of the ratio of vasoconstriction (mental load) and vasodilatation (physical load) of blood vessels (Boucsein 2012a).

The dominant influence of physical load over mental load in the behavior of physiological functions was also corroborated by Basahel (2012). He observed increased HRV values when the two types of load were combined in his research, despite the fact that HRV decreases if mental load alone increases (Kramer 1990; Mehler, Reimer, and Wang 2011; Miyake et al. 2007; Veltman and Gaillard 1998; de Waard 1996; Wilson, Fullenkamp, and Davis 1994). In females seated on the pads HRV Entropy increased with rising SCL. This phenomenon could be the result of respiration patterns (Veltman and Gaillard 1998) because HRV (in our case HRV Entropy) increases with rising VT. We do not dare present an unambiguous interpretation of this partial HRV indicator. HR and HRV can be distinctly influenced by breathing (Billman 2013), which is more than likely in our pilot research study.

Conclusion

We investigated differences in cognitive performance connected with physical loads of varying intensities. A combination of mild physical load (the use of the modified gymnastic – Swiss ball) and cognitive load leads to the increased precision of cognitive performance. Females exhibited greater changes in their cognitive functions than males. Females seated on pads (mild physical load) made 58% fewer errors than females on chairs; the number of errors was closely related to the depth of their breathing (tidal volume). The number of correct answers remained unchanged. Females on the pads also showed higher skin conductance levels and HRV Entropy values.

While breathing depth, skin conductance level, finger temperature, HRV Entropy, and the number of errors exhibited no great oscillations, variance in respiration frequency (minute ventilation), heart rate, and the number of correct answers of females was considerable in the load phase.

The fact that physical load dominates over mental load in the course of physiological functions particularizes the patterns for how one group of functions facilitates or inhibits the other one.

The finding that the number of errors is another criterion of sustained, selected attention besides the speed of work, can be applied for example in the area of traffic safety (air and road etc.).

Ethics approval and consent to participate

The study, filed as Project no. EK 04/2018, was approved by the Ethics Committee of the Department of Biomedical Engineering of the Faculty of Electrical Engineering and Communication at Brno University of Technology. Informed consent was obtained from each participant.

Availability of data and material

The datasets generated and/or analysed during the current study are available in the [https://www.mendeley.com] repository, [doi: 10.17632/7djsbz6prv.1].

ANS	=	autonomic nervous system
Bo test	=	Bourdon test T-78
BoC	=	correctly cancelled symbols of Bo test
BoE	=	incorrectly cancelled symbols (errors) of Bo test
FT	=	finger temperature
ECG	=	electrocardiography
EDA	=	electrodermal activity
HRV Entropy	=	sample entropy of HRV
EP	=	examined person
Es	=	standardized effect
HR	=	heart rate
HRV	=	heart rate variability
HRVAS	=	heart rate variability analysis software
MV	=	minute ventilation
pad	=	dynamic directional seat pad
R	=	respiration
RR	=	respiratory rate
SCL	=	skin conductance level
SCR	=	skin conductance reflex
SMA	=	simple moving average
VT	=	tidal volume
VTA	=	absolute tidal volume change
VTR	=	relative tidal volume change

Disclosure statement

No potential conflict of interest was reported by the author(s).