Student volunteers were recruited in Kobe Co-medical College for the present study. Fifteen healthy volunteers (12 female and 3 males; mean age, 28.1 ± 4.8 years) participated in the study after providing written informed consent. These participants were right handed and had no history of any psychopathological disorders or heat stroke. The participants completed the Japanese adaptation of the State-Trait Anxiety Inventory Form Y (STAI- JYZ) modified from the original form-Y before experiments in a calm environment with a room temperature of approximately 25˚C. The present study was approved by the Ethics Committee of the National Institute of Communication Technology.

The participants were exposed to a hot environment by wearing an adiabatic thermal suit (Figure 1). The suit was originally designed for preventing hypothermia in extreme cold; thus, it was made using heat-insulating materials. This suit can increase the microclimate temperature using no heat source except the body heat of the participant. A deep-body thermometer based on the zero-heat-flow method [45] was attached to the middle of the forehead or planta to continuously monitor the core temperature. Each participant achieved a high microclimate temperature (above 30˚C) before the experiments were initiated (Figure 2).

This study employed a wearable neck-cooling system (Figure 3) comprising 3 major parts: a cooling collar, pipe lines, and a radiator. The collar was equipped with an electronically-controlled Peltier cooling device. Waste heat from the cooling device was removed through water-circulating pipelines and an air-cooling radiator. The collar was attached to the lateral and posterior parts of the neck for direct cooling of the bilateral carotid arteries. The core temperature at the forehead was promptly decreased by the neck-cooling system. A typical temperature response curve is shown in Figure 4 shows typical changes in brain temperature during neck cooling in the heat for randomly selected subjects. The brain temperature was defined as a core temperature measured at the middle of the forehead. The prompt temperature reduction was attributed to the bilateral arterial blood streams. This indicated that the neck-cooling system realized a type of selective brain cooling [46] [47] .

Subjects underwent a battery of cognitive tasks comprising SRT-CPT and AX-CPT tasks that were performed on a computer following the Conners’ CPT. During the SRT-CPT task, subjects were required to press the button as quickly as possible every time a number was presented on the screen. Eighty stimuli comprising the same number “7” were presented at varying intervals. During the AX-CPT task, subjects were required to press the

button as quickly as possible only when the number “7” appeared after the presentation of the number “3”. Four-hundred number stimuli varying randomly from 1 to 9, including 40 targets (3 - 7), were presented at varying intervals.

Errors were categorized as omission and commission errors. Although omission errors were detected based on a latency of >1000 ms for the target stimuli, commission errors were detected based on a latency of <100 ms for the target in both the SRT- and AX-CPT tasks. In addition, inadequate responses to non-target stimuli were also taken into account in the AX-CPT task.

EEG was continuously recorded throughout the experiments. The recording was performed using a conventional international 10 - 20 system having 21 Ag/AgCl electrodes. Analog EEG signals acquired from the electrodes were digitized at 512 Hz with 24-bit ADC. Montage signals were produced from these digitized signals using the mastoid electrodes as a reference.

The EEG alpha-2 (10 - 13 Hz) rhythm was dominant in the occipital region (Figure 5(a)). Time-series data of the occipital (O1, O2) alpha-2 powers were calculated at every 31.25 ms from the fast Fourier transform using 2-s epoch EEG data at O1 and O2 electrodes. The average of these powers temporally fluctuated (Figure 5(b)). A slower component (≤0.04 Hz) of this fluctuation was defined as a DBA index.

Statistical data processing was performed using the Student’s t-test along the significant level set at p < 0.05 to examine significant changes in reaction time between normal and heat environments and neck-cooling effects. Correlation analyses were performed according to the conventional parametric method with Pearson’s correlation

coefficients and corresponding probability values. Nonlinear relationships between two items were derived from the polynomial regression analyses.

STAI-JYZ, as an advanced Japanese version [48] modified the original STAI-Y [49] precisely provides State? Trait anxiety measures using independent subscales related to positive and negative emotions. STAI-JYZ comprises 40 items, and subjects mark a rating from 1 to 4 for each item. The two anxiety measures are numerically evaluated in the range of 20 - 80 from their self-report answer sheets.

The experimental protocol is shown in Figure 6. After completing the STAI questionnaire, subjects were fitted with an EEG cap and a sensor head at the middle of the forehead for core-temperature measurements. Subjects took rest for 5 min with eyes closed. The first battery of cognitive tasks comprising the SRT- and AX-CPT tasks was then completed. Subjects then wore the thermal suit and stood by for approximately 30 min until the microclimate temperature rose to a threshold of 30˚C. After confirming the required increase in temperature, HT- group subjects promptly started the second battery of cognitive tasks, whereas the NC-group subjects turned on the neck-cooling system, rested for few minutes, and started the 20-min battery of tasks. EEG recording during cognitive tasks was performed with the eyes open. The experiment, requiring a total duration of approximately 2 h, was performed for each subject in the period between 3 - 6 pm.

All subjects made almost no errors during the simple reactive tasks (SRT-CPT), regardless of the ambient temperature being normal or hot. As shown in Figure 7(a), the error rate in the AX-CPT was higher in the lower DBA-index region for both omission and commission errors in a normal-temperature environment. In contrast, both omission and commission errors were lower in the NC-group subjects independent of the DBA index, whereas more errors appeared even in a higher DBA-index range for HT-group subjects in hot environments. These aspects were accentuated in the results of correlation analyses for logarithmic error rate vs. DBA index with excluding error-free data, as shown in Figure 7(b). The reaction time (RT) in SRT- and AX-CPT tasks did not change significantly between normal and hot environments, regardless of whether neck cooling was used (Figure 8).

As shown in Figure 9, neck cooling increased the average DBA index during cognitive tasks. The increase was

greater for subjects with higher average DBA indices at rest. This relationship was more significant for AX-CPT than SRT-CPT.

State- and traitanxiety were independently evaluated from self-reports of STAI. Subjects exhibited a linear relationship between these anxiety measures (Figure 10(a)). Trait anxiety was used as an index for characterizing

propensity in cognitive performance and behaviors. As shown in Figure 10(b), the average DBA index decreased in both higher and lower trait-anxiety regions, although it had been expected that the DBA would exhibit a simple negative correlation with trait anxiety.

Thermoregulatory responses of subjects to cognitive tasks were investigated during heat exposure by continuously monitoring the microclimate and core temperatures. Figure 11(a) shows a case with neck cooling. The core temperature promptly increased as soon as the task began, whereas the microclimate temperature decreased. These 2 temperature responses were reciprocal. The core temperature reached a peak and then decreased, whereas the microclimate temperature remained at the minimum. In contrast, as shown in Figure 11(b), a case with no neck cooling exhibited a much different profile, with the microclimate temperature slowly increasing and the core temperature rise lagging, independent of task performance.