Our results demonstrate that physical changes induced by mental stress can be observed by indices derived from FPG. Therefore, FPG can be one of the easiest methods by which to evaluate mental stress quantitatively. In particular, the largest Lyapunov exponent and the LF/HF ratio might be associated with acute mental change.
Validity of CWT as stress-inducing test
The CWT is known as a psychological stress test; several reports have described increased pulse rate and inhibited parasympathetic autonomic nerve activity during CWT [27, 35, 36]. As shown in Table 1, participants in this study showed significantly increased finger pulse rate during the test, by comparison with the rest periods. We also set three stages of difficulty for each task. In practice, a stepwise reduction in the correct response rate was found, affirming the gradual increase in difficulty. The values indicate that the finger pulse rate during testing was always higher than during the rest periods. Compared with other indices, SBP, DBP, and body temperature did not change after CWT. Finger pulse rate, respiratory rate, the LF/HF ratio, and the HF component were measured during CWT, but SBP and DBP were measured after not measure the blood pressure during CWT because the left arm was used for the measurement of FPG and the right hand was used for clicking the mouse. Although we conjectured that these were changed during CWT as well as other indices, we could not determine the change exactly in the present study. To overcome this problem, we should instead use a stress test in which a participant’s hand is not used.
The POMS also showed decreased ratings in the vigor subscale and increased ratings in the confusion subscale immediately after the test (Figure 4). Ratings in the depression-dejection and anger-hostility subscales may have decreased significantly after CWT as a side effect of the concentration workload. These results indicate that the CWT in this study was sufficient to cause acute confusion and reduce vigor. Although ratings in the tension-anxiety and fatigue subscales showed no significant changes, correlations between them and some physiological indices showed significant changes (Table 3). Physiological indices might reflect a slight mental change, but this point is unclear in this short-term study. In future studies, we should improve the mental test (for example, we could lengthen the periods of the task-time).
Acute mental stress affects peripheral circulation
Generally, sympathetic nerve activity is parallel to parasympathetic nerve activity. In our results, the HF component was decreased and the LF/HF ratio was increased during CWT, showing that mental stress activated sympathetic nerve activity and inhibited parasympathetic nerve activity. We speculate that during mental workload, peripheral circulation decreases with activation of sympathetic nerve balance. The FPG reportedly reflects sympathetic nerve activity , and alpha-adrenergic sympathetic nerve activity in the skin of the fingertips increases as a result of psychological stimulation [6, 38]. Increased sympathetic nerve activity due to psychological and physical strain is commonly known to cause constriction of peripheral arteries. This vasoconstriction reduces blood flow and subsequently leads to decreased amplitude of the finger pulse wave [39, 40]. Based on these reports, we believe that the finger pulse wave amplitude decreased due to increased peripheral sympathetic nerve activity.
Evaluation of mental stress by chaotic indices
In this study, the finger pulse wave amplitude decreased and the chaotic attractor shrank during CWT (Table 2 and Figure 5B). The results indicate that the decrease in peripheral circulation was induced by mental stress. Tsuda et al.  demonstrated that orbital unevenness of the chaotic attractor became small under treatment, and the attractors of patients with low ratings for all scores showed stiffening. Previous studies suggested that mental stress decreased peripheral circulation and dynamics [20, 41]. Since significant changes to the first Lyapunov exponent were not found in our study, we could not reveal any changes in peripheral dynamics. Moreover, we could not determine the clear shape of the chaotic attractor during testing. We therefore should improve the system of chaos analysis to show a clear shape even when the size of the attractor is small. It is difficult to determine the exact chaotic state, because of dependence on several physical and mental conditions, so more experiments are needed to investigate dynamics in the human body from different perspectives.
Relation between physiological indices and POMS
It seems that the LF/HF ratio tends to reflect scores on the fatigue subscale sensitively, whereas the largest Lyapunov exponent tends to reflect scores on the tension-anxiety subscale (Table 4). Interestingly, there were no correlations between them during rest periods, but strong correlations appeared during CWT. This result suggests that the LF/HF ratio and the largest Lyapunov exponent could be valuable indices to reflect emotional changes instantly. Imanishi et al.  reported that the largest Lyapunov exponent had positive correlations with anxiety and fear. They also found no relationship between heart rate and emotions. Because they used task-induced feelings of anxiety, fear and relief, our results did not coincide with their results. Correlations between the Lyapunov exponent and mental stress may change depending on the situation. Further investigations are needed to confirm the association between other emotional states and physiological indices. One difficulty in testing accurate correlations between subjective and objective data was the difference in timing of the measurements. The POMS questionnaire was completed before the rest period and after the task, whereas hand psychological indices were measured during the rest period and CWT. Although we are not able to measure both types of data at the same time, they should be measured simultaneously as much as possible to overcome this problem.
Effect of menstrual cycle on autonomic nervous system
The follicular phase is the period when concentration of estrogen is gradually increasing. Previous studies had indicated that parasympathetic activity presented by the HF component was influenced by estrogen . Thus, we conjectured that a higher level of estrogen was observed in the follicular phase than in the luteal phase during the pre-task, CWT-1 and CWT-2. However, such differences disappeared in the CWT-3 and post-task periods. This may have been because the strong mental stress might cause the large individual differences in these periods. We simply divided the menstrual cycle into two phases depending on participants’ responses. Future investigations should consider monitoring hormone levels and measuring the basal body temperatures before the experiment to more reliably identify the menstrual phases.
Clinical implications of FPG
Since FPG is noninvasive, the device can be made portable with simple modifications, and measurements can be performed anywhere. For example, desktop measurements could be performed in the workplace or at the hospital bedside in patients who are unable to express feelings, thus facilitating the identification of mental stress. Fujimoto and Yamaguchi  suggested that the chaotic attractor could be used to monitor the vital condition and support older people. To add to their idea, we suggest that it is better to calculate not only the chaotic attractor but also other indices, such as the autonomic nervous system activity and the largest Lyapunov exponent. Moreover, it can be used for self-checks by workers who perform heavy labor to assess their own stress levels visually and increase the awareness of stress-related disease prevention. Since FPG provides an estimate of artery stiffness , it is currently used for indices of blood vessel age . The measurement environment is still limited because of its sharp sensitivity, but indices derived from FPG may reflect not only the blood vessel status, but also mental stress.
Chaos indices such as the chaotic attractor have not been used to date in the medical field, because of the complex nature of the calculations involved. However, improvements in analytical techniques due to new computer technology have facilitated the use of chaos indices, which are expected to be developed for clinical application. If chaos indices are analyzed in greater detail with relation to other factors, such as clinical condition or lifestyle habits, larger amounts of health-related information can easily be acquired. One study noted that analyzing nonlinear characteristics plays a role in predicting some illnesses of the cardiovascular system . Further studies aimed at achieving clinical applications are necessary to investigate not only how chaos indices change in response to stress, but also to emotional patterns of restlessness, anger, and happiness.
Specific comments: study limitations
Our sample size was small and only comprised young, healthy women, so the results may not be generalizable to other populations (for example, men or older individuals). Furthermore, CWT is a validated and widely used test for inducing stress, but is not the same as stress in real life, of which chronically high levels are associated with negative effects on health. Thus, it is necessary to examine the use of FPG in more real-life situations.
In addition, the indices derived from FPG are strongly dependent on peripheral circulation, so we cannot use it for patients who have cardiovascular disease. We also cannot use it when the peripheral vessel is too contracted to measure.