Hyperthermia exposure impaired the early stage of face recognition: An ERP study

Abstract

We investigated the effect of hyperthermia exposure on the early stages of face processing by recording event-related potentials (ERPs) elicited by faces and non-face stimuli presented in upright and inverted orientations. Across all conditions, both the peak latencies of P1 and N170 components were earlier in the hyperthermia group than in the control participants. Although no effects of P1 amplitudes were influenced by hyperthermia, the face effect (larger amplitude for faces relative to other object categories) of the N170 was modulated by hyperthermia, whereas the face effect was significant in the control group, it was minimised in the hyperthermia group. The inversion effect of faces on N170 amplitudes, however, was not affected by hyperthermia. These data suggest that the detection of faces in the visual field and their initial streaming to face-specific structural encoding mechanisms are impaired by hyperthermia. However, subsequent face-specific configural processing revealed by the N170 inversion effect is not affected by hyperthermia. In addition, hyperthermia accelerates the early stage of visual perception, regardless of faces or non-face objects.

• Face perception
• hyperthermia
• inversion effect
• N170

Introduction

It has been known that hyperthermia negatively affects physical performance [1], such as circulatory, metabolic, muscular, nutritional, immunological and thermoregulatory factors [2–4]. While hyperthermia increases muscle metabolic rate and blood flow to the skin, there appears to be no effect on cardiac output, mean arterial pressure and muscle or skin blood flow or isometric contractions of muscle function, and the ability of muscle to extract oxygen seems to be well preserved [1]. As well as the above physiological changes, hyperthermia also induces changes in the central nervous system, i.e. central fatigue [3], [5].

Although several hypotheses considered that hyperthermia may have effects on the synthesis and metabolism of serotonin, dopamine and noradrenaline, interleukins, β-endorphins and ammonia [6–10], recently it was suggested that hyperthermia impaired cognitive function alterations [11], [12] and in particular, decreased memory capacity in both working memory and short-term memory, suggesting the dysfunction of frontal lobe activity in hyperthermia condition [13]. By recording and analysing electroencephalogram (EEG) signals, it was found that increasing body temperature resulted in faster α rhythms in resting subjects [14], There was also evidence that the α/β ratio increased with the increasing core temperature, and that these changes were generated in frontal motor areas, and the central motor cortex as well as the occipital cortex [15], [16]. In addition, a difference in brain temperature of only 1°C was sufficient to obtain a significant difference in peak θ frequency in the rapid-eye movement (REM) sleep EEG [17], indicating that body temperature can cause changes in the EEG power spectra. Moreover, there was evidence that an increase in oesophageal temperature from 37.14 ± 0.25°C to 39.05 ± 0.15°C produced a significant reduction of auditory evoked potentials, which depend on brain stem temperature [18], indicating that the low-level nervous system was affected by brain stem temperature. Another study relevant to finger shock stimulation revealed that P50 and P50-N70 amplitudes decreased regularly and significantly over the whole duration of the heating period, indicating that the hyperthermia could potentially disturb the somatosensory central nervous system [19]. In particular, a recent study revealed that passive heat exposure for a long time (1 h) impaired pre-attentive processing reflected by the decrease of mismatch negativity (MMN) amplitudes [20]. Particularly relevant to the current study, low-level visual functions are also impaired in heat-stress conditions. For example, there was evidence that the visual evoked potential (VEP) amplitude elicited by black and white checkerboard was reduced after heating, especially in patients with multiple sclerosis [21], [22], implying that conduction of more axons in the visual pathways was blocked. However, whether high-level visual perception abilities are affected by heat stress is less evident. To this end, in the present study we investigated possible heat effects on early stages of face processing.

Faces are ecologically important stimuli that provide essential cues relevant to a person's identity, mood, emotion, or intent; perceptual information that forms the basis of interpersonal communication. Ample evidence indicates that the processing of human faces relies both on local features and/or on the configural or holistic relations between these features [23–29]. As mentioned above, there have been several studies showing that the early stage of information processing (P50-N70 of finger shock stimulation, P2 components of VEP, auditory MMN) was impaired in hyperthermia [19], [21], [22]. It is possible, therefore, that heat stress may have a deleterious effect on the early stage of face processing that is independent of cognitive performance such as working memory [13]. To address this question, in the present study we explored the early stages of face perception using the N170, an electrophysiological index of early face processing [30]. In addition, the P1, which is an electrophysiological index related to the processing of basic visual stimuli, was also investigated.

The N170 is a robust negative deflection peaking normally between 160 and 180 ms at occipito-temporal sites with significantly larger amplitude for faces relative to other object categories [30]. Previous study showed that the N170 is not sensitive to face identity [31–33], is larger (and delayed) for face components (particularly eyes) than full faces [30], [34], [35], larger (and delayed) for inverted faces [30], [36] and equally large for scrambled and normally configured faces [37]. It was suggested that the N170 is triggered by the detection of global face structures as well as other face-related information in the visual field and initiates the perceptual processes that facilitate the individuation and identification of faces within category [30], [37], [38]. In particular, converging behavioural data suggested that the face-inversion effect [39], a disproportionate decrease in recognition performance when the face is inverted upside down, is due to the disruption of holistic/configural processing [34], [35]. Therefore, the N170 inversion effect (enhanced and delayed) has been the indicator of configural analysis of faces [40], [41].

In the present study we explored possible heat effects on face perception (as reflected by the N170 effect) and the application of configural processing strategies (reflected by the N170 inversion effect). An oddball paradigm was used, in which faces and non-face objects (tables) were randomly interspersed with butterfly pictures as visual target stimuli, and the participants kept a mental count of the occurrence of the target [30], [42]. A face-specific effect in early visual processing under heat-stress conditions should be reflected in the difference between the N170 elicited by faces and objects. Moreover, if heat stress does impair face configural processing, the N170 inversion effect should be decreased under heat conditions compared with normal temperature conditions.

Material and methods

Subjects

A total of 28 right-handed young male undergraduates (mean age 21.75 years, range 19–24 years) were divided into two groups, the control group (14 subjects) with 1 h exposure of 25°C, and the heat group with 1 h exposure of 50°C. All participants reported normal or corrected-to-normal vision and had no history of current or past neurological or psychiatric illness and took no medications known to affect the central nervous system. They signed an informed consent approved by the Jinan Ethical Committee and were paid for their participation.

Procedure

Subjects performed the task in a dimly lit room, whose indoor temperature could be controlled strictly within 1°C by four automatic heaters (one was located 1.2 m in front of the subject, one 1 m behind and one at each side at 0.5 m). Stimuli were presented on the computer screen (38.1 cm) at 60 cm from the subjects. The stimuli were 30 unfamiliar young faces, 30 tables, and 30 flowers. These stimuli were used to form five stimulus conditions: (1) upright faces, (2) inverted faces, (3) upright tables, (4) inverted tables and (5) targets (flowers) (Figure 1). All images were presented by grey-scale 11.5 × 12.0 cm photographs. Half of the faces were male and half female, and were presented without hair, glasses, or other accessories. All images were equated for luminance and root mean square (RMS) contrast (not including the grey background in calculation), using Adobe Photoshop CS4 software.

Figure 1. Examples of target flowers and non-target stimuli (faces and tables) in the upright and inverted orientations.

The stimuli were presented at fixation, seen from a distance of 60 cm and occupying a visual angle of 5.37°–6.25°. The subjects sat in a comfortable chair and were instructed to keep a mental count of flowers that occurred occasionally on the screen, while ignoring all other stimuli. This procedure has been frequently used in N170 research in order to assign similar task relevance to faces and non-face stimuli [42]. The 900 stimuli were presented one at a time in six blocks, each consisting of 30 upright faces, 30 inverted faces, 30 upright tables, 30 inverted tables, and 30 flowers (targets). The order of trials within each block was randomised. Stimulus exposure time was 300 ms and separated by an inter-trial interval of 1000 ms (randomised between 800–1200 ms). There was a 1-min break between blocks, when the subject numbered off the counted flowers to the experimenter.

During the experiment, the temperature inside the room was kept constant to simulate a passive heat environment. Two electronic heaters were placed inside and four automatic thermal controllers were placed in the four corners of the room, respectively. Once the temperature inside the room reached the designated temperature, 50°C, the heater would stop working or the heaters would quickly work again. Thus, the temperature inside the room was kept constant at the expected level of 50°C.

For each participant, to avoid weight loss 100 mL water was given before the start of the experiment. Body weights and body core temperature (a thermocouple (Hohour, ZWJ-2, Shanghai, China) inserted beyond the anal sphincter) were measured before and after experiments. During the experiment, participants were asked to refrain from drinking and urinating.

Electrophysiological recording and measures

EEG was recorded continuously by a set of 32 Ag/AgCl electrodes placed according to the 10/20 system. The EEG recording sites were: FP1, FP2, F7, F3, Fz, F4, F8, FT7, FC3, FCz, FC4, FT8, T7, C3, Cz, C4, T8, TP7, CP3, CPz, CP4, TP8, P7, P3, Pz, P4, P8, O1, Oz, and O2 as well as left (M1) and right (M2) mastoids. Electrooculograms (EOG) were recorded via electrodes placed on the bilateral external canthi and the left infraorbital and supra-orbital areas to monitor for eye movements and blinks. Both EEG and EOG were sampled at 1000 Hz, with a 0.1–100 Hz band pass using a NuAmps digital amplifiers system (Neuroscan Labs, El Paso, TX). The tip of the nose was used for reference during recording. Electrode impedances were kept below 5 kΩ.

EOG artefacts were corrected using a correlation method proposed by Semlitsch and colleagues [43] (Neuroscan Labs, El Paso, TX, USA). Then the recorded EEG data was segmented in epochs of 1000 ms beginning 200 ms prior to stimulus onset and averaged separately for each of the five conditions. Segments contaminated with artefacts exceeding an amplitude of ±100 µV were excluded from averaging. After this procedure, averaged event-related potentials (ERPs) included at least 55 trials for faces and tables and 30 trials for flowers. The averaged ERPs were low-pass filtered at 30 Hz (24 dB/octave).

Based on the visual inspection of grand-averaged ERPs and on the existing literature [44], we analysed the N170 components only at a subset of sites including the lateral sites P7 and P8, M1 and M2, as well as more medially at the O1 and O2 electrodes. In addition, since group differences were observed for the preceding P1 component, we have also analysed these differences at the same locations. The peak amplitudes and latencies were measured automatically between 70 ms and 120 ms and between 120 ms and 200 ms for the P1 and N170, respectively (Figure 2, Tables 1–8).

Figure 2. The diagrammatic drawing of peak amplitudes and latencies for P1 and N170, which were automatically measured between 70 ms and 120 ms and between 120 ms and 200 ms, respectively.

These measurements were analysed by using repeated ANOVA, with the Group (control and hyperthermia) as a between-subject factor, and Testing time (before heat, post-heat), Stimulus type (faces, tables), Orientation (upright, inverted), Hemispheres (left, right), and Sites (P7/8, M1/2, O1/2) as within-subject factors. Statistical probability from the ANOVAs was corrected using the Greenhouse-Geisser procedure. Post hoc tests (Bonferroni corrected for multiple comparisons) were performed when necessary.

Results

Body weight and temperature responses

For the control group there were no significant changes for body weight (68.8 kg and 68.8 kg, p = 0.9) and body temperature (37.2°C and 37.3°C, p = 0.8). For the hyperthermia group, compared to control conditions before the test, body temperature was significantly increased (37.3°C and 38.4°C for control and heat conditions, respectively; p < 0.001) and the body weight losses were less than 0.1% less than control (69.6 kg and 68.9 kg for control and heat conditions, respectively; p = 0.89).

ERP data

Target monitoring was equally good in both groups (97.9% for control participants and 96.8% for hyperthemia participants, respectively; F(1, 26) < 1). Non-target stimuli, i.e. faces and tables, elicited clear P1 and N170 components with significant occipital temporal distribution in both control and hyperthermia participants (Figures 3 and 4). Although there were some slight latency differences of P1 (F(1, 26) < 1) and N170 (F(1, 26) < 1) between control and hyperthermia groups, according to the grand-averaged waveforms the P1 and N170 peaked at similar latency time windows. As mentioned in the Methods section above, therefore, we measured the peak latencies and amplitudes in the same time windows for both groups, between 70 ms and 120 ms and between 120 ms and 200 ms for the P1 (seen in Tables 1, 2, 5, 6) and N170 (seen in Tables 3, 4, 7, 8), respectively.

Figure 3. (A) The grand-averaged waveforms elicited by face and table stimuli before and after 1 h exposure in the heat and control groups, respectively. (B) Grand-averaged ERP topography elicited by upright faces and tables at the analysed sites for control and hyperthermia participants after exposure.

Figure 4. The grand-averaged waveforms elicited by upright and inverted face stimuli before and after 1 h exposure in the heat and control groups, respectively.

P1 component

In the P1 amplitude analysis, no main effect of Group was found (F(1, 26) < 1) (as presented in Figure 4). The analysis of the amplitude yielded several significant effects of factors with repeated measures, none of which, however, interacted with Group. The P1 elicited by faces (3.6 µV) was bigger than that elicited by tables (2.7 µV; F(1, 26) = 21.9, p < 0.001) and the amplitude was larger at the right hemisphere sites (3.4 µV) than the left hemisphere sites (2.8 µV; F(1, 26) = 19.4, p < 0.001). At the more posterior sites, the P1 was larger than at the more anterior sites (4.2 µV, 3.3 µV, and 1.9 µV for O1/2, P7/8, and M1/2, respectively; F(1, 26) = 13.2, p < 0.001), with Bonferroni comparisons showing significant differences between O1/2 and P7/8 (p = 0.005), between O1/2 and M1/2 (p < 0.001) and between M1/2 and P7/8 (p < 0.001). Although the P1 for faces was larger than for tables at all sites (F(1, 26) = 13.8, p < 0.05), this effect was smaller at M1/2 (0.4 µV) than O1/2 (1.1 µV, p < 0.001) and P7/8 (1.2 µV, p < 0.001). The inversion effect (inverted stimuli minus upright stimuli) (Figure 4) was not significant at all sites (p > 0.1 for each). For the analysis of P1 latency, there was no significant difference between two groups (108 ms and 104 ms for control and heat groups, respectively; F(1, 26) < 1, p = 0.447) and, interestingly, P1 latency was shorter after 1 h exposure (104 ms) than that before 1 h exposure (108 ms; F(1, 26) = 12.260, p = 0.002). The main effect of Site was also significant (F(2, 52) = 7.45, p = 0.03), with further Bonferroni comparisons showing shorter latency at O1/2 (103.5 ms) than at P7/8 (107.0 ms, p < 0.010) and M1/2 (107.9 ms, p < 0.001), with no difference for the latter two sites (p = 0.790). Testing time × Site interaction (F(2, 52) = 11.817, p < 0.001) indicated that whereas the Testing time effect was evident at O1/2 (before minus after 1 h heat exposure, 6.0 ms; p < 0.01) and M1/2 (3.7 ms; p < 0.05), it was not significant at P7/8 (1.6 ms; p = 0.19). What should be stressed is the interaction of Testing time × Group (F(1, 26) = 13.54, p < 0.001), indicating that the Testing time effect only existed in the hyperthermia group (109 ms and 100 ms for before and after 1 h exposure at 50°C, p < 0.001), while the latency of P1 was not modulated by Testing time in control conditions (107 ms and 108 ms for before and after 1 h exposure at 25°C, p = 0.4).

N170 component

As presented in Figures 3 and 4, during 100–300 ms post-stimuli onset, faces elicited more negative N170 than did tables, i.e. N170 face effect, regardless of heat or orientation conditions. After heat exposure, the N170 face effect was more conspicuous in the controls than in the heat group. Although there was no difference before and after 1 h exposure in the control group, the N170 declined significantly in the heat group after 1 h exposure.

In the N170 amplitude analysis, overall the N170 was similar between the two groups (−6.5 µV and −5.7 µV for the control and heat groups, respectively; F(1, 26) < 1). The following effects were significant and did not interact with Group: Faces elicited larger N170 (−7.3 µV) than did tables (−4.9 µV; F(1, 26) = 45.121, p < 0.001) but this effect was qualified by the two-way interaction of Orientation × Stimulus type, F(1, 26) = 40.183, p < 0.001. Further analysis revealed that whereas inversion enhanced the N170 of faces (−7.8 µV and −6.8 µV for inverted and upright faces, respectively; p < 0.001), the N170 amplitude in response to inverted tables was decreased (−4.6 µV) over upright tables (−5.2µV; p = 0.005). The Site × Stimulus type interaction was significant, F(1, 26) = 69.731, p < 0.001. Further analysis revealed that, whereas for faces the N170 was significantly larger at P7/8 (−8.2 µV) than at either M1/2 (−6.8 µV) (p < 0.05) or O1/2 (−6.9 µV) (p < 0.05), with no significant difference between the latter two sites (p = 0.882), for tables, the N170 was decreased significantly at M1/2 (−2.7 µV) than at either P7/8 (−5.9 uV, p < 0.001) and O1/2 (−6.1 uV, p < 0.001), with no difference between the latter two sites (p = 0.639). Furthermore, although significant at all sites, the Stimulus type effect was largest at the P7/8 sites and smallest at the O1/2 sites (−4.1 µV, −2.3 µV and −0.8 µV for the P7/8, M1/2, and O1/2 sites, respectively). Finally, the N170 was larger at the right hemisphere site (−6.5 µV) than at the left hemisphere site (−5.7 µV; F(1, 26) = 13.142, p < 0.05).

More relevant to the scope of the present study, however, were the effects which were modulated by hyperthermia. For N170 amplitude, a significant two-way interaction of Testing time × Group, F(1, 26) = 5.911, p < 0.05, indicated that, while in the control group the N170 was maintained after 1 h exposure of 25°C (−6.3 µV and −6.6 µV for before and after 1 h exposure, respectively; p = 0.491), it was decreased by the heat stress (−6.4 µV and −5.0 µV for before and after 1 h heat exposure; p < 0.05). This two-way interaction was further qualified by two three-way interactions of Stimulus type × Testing time × Group (F(1, 26) = 6.127, p < 0.005) and Sites × Testing time × Group (F(2, 52) = 11.124, p < 0.005). Further analysis revealed that although the N170 face effect was not affected by 1 h exposure of 25°C (−3.2 and −3.0 µV for before and after 1 h exposure; p = 0.714), it was decreased significantly after 1 h exposure in heat stress (−2.5 and −0.9 µV for before and after 1 h exposure of 50°C, respectively; p < 0.05). However, for tables, N170 amplitude was not significantly changed before and after 1 h exposure of 25°C (F(1, 26) = , p = 0.530) or 50°C (F(1, 26)=, p = 0.411). Moreover, the decrease of N170 amplitude was significant at P7/8 (−7.5 µV and −5.8 µV for before and after 1 h exposure of 50°C, respectively; p < 0.05) and O1/2 (−7.3 µV and −5.2 µV for before and after 1 h exposure of 50°C, respectively; p < 0.05) not at M1/2 site (−4.3 µV and −4.1 µV for before and after 1 h exposure of 50°C, respectively; p = 0.89). Finally, the Hemisphere × Sites × Group interaction was also significant, F(2, 52) = 21.045, p < 0.001, showing that, while in the control group the N170 amplitudes were significantly increased on the right side at the M1/2 (right minus left: −1.0 µV, p < 0.05) and P7/8 sites (−2.0 µV, p < 0.05), this hemisphere effect, i.e. higher N170 over right hemisphere than left, was significant at P7/8(−1.1 µV, p < 0.05) and O1/2(−1.3 µV, p < 0.05) sites in the hyperthermia group.

Analysis of peak latency revealed that, overall, the N170 latency was similar in the control (167 ms) and hyperthermia (165 ms; F(1, 26) < 1) groups. Across groups, inversion delayed N170 latency significantly (165 ms and 167 ms for upright and inverted conditions, respectively; F(1, 26) = 8.159, p < 0.05) and the significant interaction of Orientation × Stimulus type (F(1, 26) = 4.696, p < 0.05) showed that the inversion effect was evident for faces (p = 0.002), not for tables (p = 0.38). Although faces elicited slightly but significantly shorter N170 latency (165 ms) than did tables (167 ms; F(1, 26) = 6.481, p < 0.05), this Stimulus type significantly interacted with Group (F(1, 26) = 9.242, p < 0.05), revealing that the Stimulus type effect was conspicuous in the control group (165 ms and 169 ms for faces and tables, respectively; p = 0.001), not in the hyperthermia group (166 ms and 165 ms for faces and tables, respectively; p = 0.73). The main effect of Testing time was also significant, F(1, 26) = 37.227, p < 0.001, indicating that compared to pre-test condition (170 ms), N170 peaked earlier after 1 h exposure (163 ms). Interestingly, this effect was qualified by the two-way interaction of Group × Testing time, F(1, 26) = 7.953, p < 0.05, revealing that the Testing time effect of N170 latency was larger in the hyperthermia group (11 ms; p < 0.001) than that in the control group (4 ms; p = 0.028), although the N170 latency was very similar for pre-test conditions between the two groups (169 ms and 170 ms for control and hyperthermia groups, respectively; p = 0.79). Furthermore, a significant Stimulus type × Orientation × Sites × Group interaction (F(2, 52) = 5.318, p < 0.05) followed by Bonferroni corrected pair wise comparison post hoc tests showed that, whereas in the control group inverted faces elicited a comparatively longer latency only at P7/8 (0.4 ms, 0.8 ms and 0.1 ms for M1/2, P7/8 and O1/2, respectively; p = 0.013 for P7/8, p > 0.05 for M1/2 and O1/2), in the hyperthermia group the delayed latency appeared at all sites (0.8 ms, 1.7 ms and 1.2 ms for M1/2, P7/8 and O1/2, respectively; p < 0.05 for M1/2, P7/8 and O1/2).

Discussion

In this study we investigated whether perceptual processes involved in the early stages of face processing were influenced by hyperthermia. Should such effects be found, we sought to determine their nature and time course. To achieve this goal, the hyperthermia effects on face detection and configural analysis were assessed by recording the early ERP components (P1 and N170) elicited by faces and objects (tables) presented in upright and upside down orientations.

The major findings could be summarised as follows: Compared with control conditions, body core temperature was significantly increased and the body weight loss was less than 0.1% lower for both groups. Across all experimental stimulus conditions, both the peak latencies of P1 and N170 components were earlier after 1 h exposure in both control group and hyperthermia group. In addition, we found a modulation of hyperthermia for P1 and N170 latency. Both the peak latencies of P1 and N170 components were earlier in the hyperthermia group than in the control participants. Furthermore, the hyperthermia effects on latency were modulated by stimulus type and orientation: the N170 latency difference between faces and tables disappeared after 1 h at 50°C heat exposure, and the scalp distribution of N170 inversion effect (delayed peak latency for inverted faces) was expanded.

Thermal environment can produce different effects such as dehydration, physical and fatigue stressors [45]. There was evidence that a 2% or more decrease in body weight could result in dehydration which could lead to deterioration of cognitive abilities [46], and, possibly due to adequate fluid replacement, cognitive performance was indeed unaffected when body weight changes did not exceed 0.5% [47]. In the present study, to avoid dehydration effects all participants ingested water before the experiment and, finally, body weight loss was less than 0.1% of control. Hence, the observed face detection changes were not caused by weight changes per se. Compared to the controls, in heat conditions the increment of body core temperature was significant and mean body temperature was increased 1.1°C by thermal stress in our experiment. In line with the current findings, previous studies reported when deep body temperature was disturbed with thermal stress exposures away from both normal and steady-state conditions, vigilance could be impaired (Hancock, 1986). In addition, a study evaluating psychological stress reactions during hyperthermia treatments showed that hyperthermia treatments are not excessively stressful, either systemically or psychologically [48], which implies that the underlying mechanism responsible for the measured ERP changes is hyperthermia, rather than the psychological changes elicited by hyperthermia exposure.

Across the two groups the scalp distribution of P1 and N170 demonstrated that these two components are dissociated: P1 was largest at the occipital sites (O1 and O2), whereas N170 was largest at the posterior lateral sites (P7 and P8). Moreover, whereas the N170 elicited was right lateralised only to faces, P1 was larger at the right hemisphere for both faces and tables. Nevertheless, the amplitude of P1 was sensitive to stimulus manipulations as well as N170. It was higher in response to faces than tables. Importantly, however, unlike N170, none of these effects of P1 amplitudes were influenced by hyperthermia, and the P1 amplitudes were equal for inverted and upright stimulus. The face effect (larger amplitude for faces relative to other object categories) of the N170, on the other hand, was modulated by hyperthermia: whereas the distinction between faces and non-face stimuli was significant in the control group, the face effect was minimised in the hyperthermia group. The inversion effect of faces on amplitude, however, was not affected by hyperthermia.

Although anticipating no P1 modulations, the present study focused primarily on the N170. One of the important findings is that the face effect decreased after 1 h heat exposure. Studies about the cognitive mechanisms of face processing have proven that faces are recognised in configural processing, while the tables were perceived in a featural manner [25], [49–52]. The visual N170 component peaks at occipito-temporal electrode sites are believed to be a marker to investigate non-invasively the time course of configural face recognition in the human brain and characterise their sensitivity to various stimulus manipulations. The larger amplitude of N170 for faces than for tables seems to lie in the recruitment of face-specific cortex [53]. Generally, there are three face-selective regions which respond to changes in different aspects of faces [54]: the fusiform face area (FFA) [55–57] found in the mid-fusiform gyrus, the occipital face area (OFA) [58] found in the lateral inferior occipital gyri, and a face-selective region in the posterior part of the superior temporal sulcus (fSTS) [59], [60]. Several studies revealed that the face-selective responses in FFA and fSTS were highly correlated with the face-selective N170 component [61]. Following this hypothesis, the decreased face effect of the N170 in hyperthermia conditions may be caused by the weakened activation of face-selective cortex, while the neurons for non-faces were not affected significantly.

In contrast to the N170 face effect, the inversion effect on the N170 was not modulated by high temperature atmosphere. Conclusive evidence has shown that faces presented upside-down are harder to perceive, memorise and recognise than upright faces, and the inversion effect primarily reflects the recruitment of object-selective neurons needed to process inverted faces, due to the disruption of the face configuration [34], [35]. Given that the featural processing of tables is not influenced by hyperthermia, it is not so surprising that the activation of object-selective neurons for inverted faces, like that of tables, is not modulated by hyperthermia. The maintained inversion effect in the hyperthermia group may be due to the proportionally decreased activation in FFA and STS for upright and inverted faces and minor effect on object-selective neurons, which leads to a maintained inversion effect of faces. Combining the decreased face effect and maintained inversion effect, we can conclude that hyperthermia mainly disrupts the face-selective neurons while the non-faces based activation is less affected.

P1 is an early index of endogenous processing of visual stimuli, with sources in the striate and extrastriate cortex [62]. It is thought to represent the psychophysical differences in the stimuli or top-down or attentional modulations [63] as well as to reflect physical variations in the stimuli used such as luminance, colour, contrast or spatial frequencies of the stimulus [64–66]. Although most ERP studies of face processing did not find categorical sensitivity to faces at P1 [67–71], other studies have reported larger P1 in response to faces than to objects [53], [72], [73]. At the same time, there is evidence that the P1 are sensitive to violation of the first-order configuration of a face, such as orientation and Mooney faces defined by shape from shading [42], [66], [73–75], which may barely be accounted for by low-level factors in vision. The present study is not the first to report higher P1 to faces. The inversion effect of P1 for faces, however, was not found. What is more, the face effects of P1, without any changes of luminance, colour, contrast or spatial frequencies of the stimulus were not modulated by heat stress. This phenomenon supports that P1 may be just a component reflecting physical variations in the stimuli, rather than a component representing the configural processing [64], [71], as it is not modulated in the same pattern as N170 during long exposure at high temperature. The face effect of P1 may be due to differences in the global or local low-level parameters between stimulus categories generally left uncontrolled in ERP studies; although in the present study all images used in this study were equated for luminance and contrast, the area subtended by tables was much smaller than the area occupied by faces [42]. Clearly determining the exact processing mechanism of P1 awaits additional research. Importantly, however, P1 is not modulated by hyperthermia, which may suggest that the low-level processing of faces is not influenced by heat stress.

Most of the above interpretations are based on the pattern of amplitude modulations. It is noteworthy that, overall, we found adequate evidence for hyperthermia-related latency modulation for either P1 or N170. Although the Test–time effect was not evident in the control group, the latencies of early components were significantly shortened after 1 h of 50°C heat stress, regardless of faces or objects. We considered that the speed perception for visual stimuli could be due to the increased conduction velocity of the optic nerve in hyperthermic conditions. Supporting this hypothesis, several studies on the effects of temperature on the conduction velocity for myelinated nerve fibres, including the phrenic nerve [76], auditory nerves [77], sural nerve [78], median motor and sensory nerves [79], and median nerve [80], concluded that the conduction velocity of nerves increased non-linearly with increase in skin temperature [79]. Although we did not directly test the conduction velocity of the optic nerve, it is still reasonable that hyperthermia can increase optic nerve conduction velocity significantly, which shortens the intervals between the onset of stimulus and early ERP components.

Before concluding, we should reiterate three procedural decisions that constrain the interpretation of the present findings. In the present study we investigated hyperthermia effects on humans under a degree of thermal stress with a designated 50°C, approximately 1 h heat exposure. First, the exposure timing and temperatures were fixed at the same values, i.e. 50°C, approximately 1 h heat exposure. It has been shown that the extent of cognitive performance deficits was associated with the extent of heat exposure, including the duration and the intensity of heat exposure [12]. Therefore, it is necessary to further explore the relationship between face perception and the extent of heat exposure. Second, all the participants in this study were young men. The effects of gender on face perception have been established, for instance a lesser degree of lateralisation of brain functions related to face coding in women than in men [81–83]. In addition, ample evidence indicates the aging modulation of the early stage of face perception. For example, the N170 inversion effect of faces, which was evident in younger people, was indeed absent in elderly people [42]. Hence, further investigation using gender and age subgroups may explicate the adverse effect of hyperthermia more clearly. Finally, we did not investigate the physiological basis, such as plasma osmolarity, hormonal concentrations, neurotransmitter or redistribution of fluids, of the influenced cognitive performance. Though previous studies showed that hyperthermia-mediated central fatigue may include cerebral perturbations such as reduced perfusion of the brain, induced hypoglycaemia, accumulation of ammonia or depletion of neuronal energy stores, with a clear deteriorating effect on motor performance and prolonged exercise [3], [84], the true mechanism was still unclear as this evidence is mainly based on the blood, rather than the true internal environment of the central nervous system. As the technique develops in the future, the impact of physiological changes of the central nervous system on cognition should be able to shed light on the hyperthermia-induced effect on cognition.

These limitations notwithstanding, our study findings suggested that the detection of faces in the visual field and their initial streaming to face-specific structural encoding mechanisms are affected by hyperthermia revealed by decreased N170 face effect. However, subsequent face-specific configural processing revealed by the N170 inversion effect is not affected by hyperthermia. In addition, hyperthermia accelerates the early stage of visual perception, regardless of faces or non-face objects.

Table 1.  The detailed latency (ms) and amplitude (uV) of P1 in the control group under pre-test conditions.

Upright face			Inverted face			Table
Channel	Latency	Amplitude	Channel	Latency	Amplitude	Channel	Latency	Amplitude
FP1	68	0.11145	FP1	56	−0.633756	FP1	54	0.351617
FP2	129	0.347481	FP2	58	−0.568055	FP2	68	0.223488
F7	68	0.085294	F7	67	−0.626483	F7	50	0.095097
F3	129	0.015819	F3	50	−1.055251	F3	70	−0.012767
FZ	67	0.103699	FZ	50	−0.995654	FZ	50	0.149993
F4	129	0.585555	F4	50	−0.930675	F4	50	0.184119
F8	129	0.418209	F8	66	−0.626638	F8	50	0.437108
FT7	64	−0.080781	FT7	62	−0.406632	FT7	50	0.133534
FC3	129	0.245498	FC3	50	−0.932818	FC3	50	0.110783
FCZ	129	0.071756	FCZ	50	−1.121502	FCZ	50	−0.006466
FC4	129	0.620057	FC4	50	−0.883657	FC4	50	0.37202
FT8	129	0.368049	FT8	80	−0.606903	FT8	50	0.43986
T3	66	0.001582	T3	70	−0.363468	T3	50	0.236693
C3	129	0.478946	C3	129	−1.033673	C3	50	0.338958
CZ	129	0.54852	CZ	50	−1.078017	CZ	50	−0.033235
C4	129	0.708796	C4	83	−0.704501	C4	50	0.46813
T4	96	0.942411	T4	95	0.163422	T4	50	0.663528
TP7	124	0.461621	TP7	92	0.530357	TP7	118	0.461635
CP3	129	0.590616	CP3	81	−0.610461	CP3	50	−0.029794
CPZ	129	0.985973	CPZ	82	−0.757532	CPZ	50	0.118107
CP4	129	0.875219	CP4	87	−0.084258	CP4	50	0.351029
TP8	92	2.049065	TP8	94	1.837923	TP8	50	0.585231
M1	117	0.906549	M1	116	1.62054	M1	118	0.932702
T5	118	1.222436	T5	97	1.563984	T5	119	0.843335
P3	129	0.761183	P3	88	0.015336	P3	50	0.033886
PZ	129	1.051226	PZ	89	0.216546	PZ	84	0.405325
P4	129	1.275537	P4	91	0.409821	P4	50	0.561051
T6	93	3.549444	T6	95	3.702532	T6	115	1.4331
M2	97	1.572422	M2	98	1.56435	M2	119	1.271487
O1	99	1.562231	O1	97	3.085518	O1	116	0.926144
OZ	94	1.469857	OZ	94	2.562554	OZ	92	0.615216
O2	93	1.985041	O2	94	2.615734	O2	100	0.797831

Table 2.  The detailed latency (ms) and amplitude (uV) of P1 in the control group under post-test conditions.

Upright face			Inverted face			Table
Channel	Latency	Amplitude	Channel	Latency	Amplitude	Channel	Latency	Amplitude
FP2	129	0.083843	FP2	129	−0.579095	FP2	65	−0.065041
F7	129	0.001226	F7	129	−0.213003	F7	68	0.340672
F3	129	0.0912	F3	129	−0.599873	F3	68	0.22275
FZ	129	0.277123	FZ	129	−0.649683	FZ	68	0.033644
F4	129	0.095873	F4	129	−0.368284	F4	67	−0.414613
F8	129	−0.294707	F8	129	−0.480549	F8	67	−0.289708
FT7	129	−0.252617	FT7	129	−0.075955	FT7	70	0.129329
FC3	129	−0.063724	FC3	129	−0.587662	FC3	68	0.094275
FCZ	129	0.127166	FCZ	129	−0.583489	FCZ	68	0.120699
FC4	129	−0.01801	FC4	129	−0.414641	FC4	68	−0.250197
FT8	89	−0.499897	FT8	92	−0.47814	FT8	90	0.511495
T3	106	−0.069317	T3	129	0.010569	T3	67	0.057133
C3	129	0.059296	C3	129	−0.082253	C3	68	−0.141917
CZ	129	0.14152	CZ	129	−0.325162	CZ	67	−0.060358
C4	129	−0.180087	C4	129	−0.088655	C4	70	−0.255981
T4	88	0.111153	T4	91	0.514699	T4	85	−0.115936
TP7	101	0.543356	TP7	128	0.297874	TP7	129	−0.19907
CP3	129	0.160846	CP3	129	0.089853	CP3	67	−0.241129
CPZ	129	0.057313	CPZ	129	0.139466	CPZ	67	−0.130795
CP4	129	−0.09966	CP4	91	0.459263	CP4	73	−0.060506
TP8	91	1.468908	TP8	92	2.337197	TP8	88	0.654504
M1	105	1.018549	M1	116	1.228049	M1	102	0.704426
T5	105	1.288931	T5	97	0.936072	T5	103	0.622045
P3	129	0.153134	P3	129	0.361742	P3	68	−0.237587
PZ	129	0.211918	PZ	89	0.298572	PZ	66	0.109396
P4	88	0.609309	P4	91	1.433254	P4	77	0.412554
T6	90	3.33709	T6	92	4.327697	T6	92	1.498915
M2	96	1.300626	M2	97	1.959669	M2	104	0.79182
O1	93	2.614334	O1	92	2.923918	O1	93	1.300369
OZ	92	2.073305	OZ	92	2.703707	OZ	94	0.886346
O2	91	2.674916	O2	91	3.0485	O2	93	1.074477

Table 3.  The detailed latency (ms) and amplitude (uV) of N170 in the control group under pre-test conditions.

Upright face			Inverted face			Table
Channel	Latency	Amplitude	Channel	Latency	Amplitude	Channel	Latency	Amplitude
FP2	199	−1.120307	FP2	120	−1.513732	FP2	123	−0.8482
F7	120	−0.651984	F7	120	−1.245943	F7	120	−0.740512
F3	199	−1.094055	F3	120	−2.132868	F3	122	−0.644608
FZ	199	−1.193632	FZ	120	−2.523083	FZ	130	−0.810543
F4	199	−1.036403	F4	120	−1.669348	F4	120	−0.763352
F8	199	−0.674912	F8	120	−1.235422	F8	126	−0.54671
FT7	120	−0.814758	FT7	120	−1.014685	FT7	199	−0.619688
FC3	120	−0.9184	FC3	120	−2.12444	FC3	120	−0.720883
FCZ	120	−1.293285	FCZ	120	−2.645169	FCZ	120	−1.217878
FC4	199	−0.739603	FC4	120	−1.717276	FC4	128	−0.797195
FT8	168	−1.108374	FT8	190	−1.145097	FT8	130	−0.352217
T3	173	−1.831382	T3	192	−1.203688	T3	175	−0.6857
C3	120	−0.603983	C3	120	−1.750115	C3	124	−0.67763
CZ	199	−0.986954	CZ	120	−2.514205	CZ	120	−1.365163
C4	199	−0.688127	C4	120	−1.402525	C4	127	−0.790778
T4	165	−2.691009	T4	177	−2.655888	T4	172	−1.046579
TP7	170	−4.19388	TP7	173	−4.136565	TP7	174	−2.028029
CP3	178	−0.905531	CP3	120	−1.482397	CP3	122	−0.655177
CPZ	199	−0.937622	CPZ	120	−2.20166	CPZ	120	−1.225391
CP4	174	−1.125084	CP4	120	−1.219739	CP4	124	−0.727273
TP8	167	−6.154478	TP8	172	−6.421705	TP8	168	−3.139988
M1	165	−5.276427	M1	170	−5.612003	M1	173	−1.457829
T5	169	−7.111151	T5	172	−7.468181	T5	170	−3.825438
P3	173	−2.304833	P3	185	−1.433088	P3	171	−1.69327
PZ	178	−0.919188	PZ	120	−1.503557	PZ	127	−0.780988
P4	172	−1.976134	P4	182	−1.207173	P4	167	−1.454117
T6	167	−8.228829	T6	172	−9.125854	T6	166	−4.530831
M2	165	−6.817051	M2	169	−7.268038	M2	168	−2.091681
O1	169	−5.75551	O1	169	−6.372889	O1	164	−5.109099
OZ	170	−3.9393	OZ	169	−3.761926	OZ	167	−3.275505
O2	171	−4.799818	O2	172	−4.69898	O2	169	−3.728536

Table 4.  The detailed latency (ms) and amplitude (uV) of N170 in the control group under post-test conditions.

Upright face			Inverted face			Table
Channel	Latency	Amplitude	Channel	Latency	Amplitude	Channel	Latency	Amplitude
FP2	199	−1.123312	FP2	120	−1.267998	FP2	199	−1.878913
F7	199	−1.124028	F7	120	−0.691864	F7	193	−1.19936
F3	199	−1.159102	F3	120	−1.261637	F3	199	−1.125349
FZ	199	−1.198408	FZ	120	−1.606083	FZ	120	−1.337775
F4	199	−1.407732	F4	120	−0.999027	F4	120	−1.226159
F8	199	−1.219798	F8	199	−0.994148	F8	199	−1.234408
FT7	199	−1.144141	FT7	199	−0.984667	FT7	188	−1.082654
FC3	120	−1.109519	FC3	120	−1.470648	FC3	120	−1.06383
FCZ	120	−1.22384	FCZ	120	−1.732124	FCZ	120	−1.376058
FC4	199	−0.981585	FC4	120	−1.177665	FC4	120	−1.105839
FT8	189	−1.258424	FT8	193	−1.327818	FT8	199	−0.875134
T3	175	−1.379318	T3	177	−1.935638	T3	183	−1.283731
C3	120	−0.872068	C3	120	−1.070839	C3	120	−1.247823
CZ	120	−1.259834	CZ	120	−1.664878	CZ	120	−1.578164
C4	120	−1.126794	C4	120	−0.864165	C4	120	−1.17126
T4	171	−2.779625	T4	171	−3.488233	T4	160	−1.867586
TP7	169	−4.516413	TP7	172	−5.324196	TP7	165	−2.534286
CP3	120	−0.694227	CP3	120	−0.914024	CP3	120	−1.285741
CPZ	120	−1.21652	CPZ	120	−1.237907	CPZ	120	−1.635232
CP4	189	−1.107457	CP4	120	−0.694357	CP4	152	−1.151275
TP8	165	−6.336318	TP8	168	−7.362976	TP8	162	−3.641725
M1	163	−5.872844	M1	166	−6.286015	M1	164	−1.98786
T5	166	−8.395267	T5	169	−9.200617	T5	167	−4.091932
P3	172	−2.092279	P3	175	−2.328827	P3	163	−2.428172
PZ	120	−0.867812	PZ	120	−1.19072	PZ	154	−1.456432
P4	174	−1.507923	P4	172	−1.857008	P4	155	−2.237167
T6	163	−8.667503	T6	167	−10.59754	T6	161	−5.487996
M2	161	−7.299803	M2	163	−7.854068	M2	162	−2.500842
O1	165	−6.857307	O1	167	−7.416913	O1	161	−6.136679
OZ	164	−4.060819	OZ	166	−4.521304	OZ	158	−4.077366
O2	165	−4.540834	O2	167	−5.52277	O2	159	−4.152922

Table 5.  The detailed latency (ms) and amplitude (uV) of P1 in the hyperthermia group under pre-test conditions.

Upright face			Inverted face			Table
Channel	Latency	Amplitude	Channel	Latency	Amplitude	Channel	Latency	Amplitude
FP1	129	0.855755	FP1	129	0.628154	FP1	50	0.34292
FP2	129	1.090228	FP2	129	0.780314	FP2	50	0.250408
F7	129	0.412244	F7	129	0.358284	F7	50	−0.14917
F3	129	0.561641	F3	129	−0.076524	F3	50	−0.19202
FZ	129	0.851626	FZ	129	0.414672	FZ	50	−0.26591
F4	129	1.361764	F4	129	0.772561	F4	61	−0.03989
F8	129	0.823174	F8	129	0.741831	F8	60	0.313715
FT7	129	0.603538	FT7	129	0.007067	FT7	61	−0.05388
FC3	129	0.587816	FC3	129	−0.123598	FC3	50	−0.42458
FCZ	129	0.689419	FCZ	129	0.094186	FCZ	58	−0.37463
FC4	129	1.182241	FC4	129	0.674043	FC4	59	−0.32187
FT8	129	0.322099	FT8	129	1.380087	FT8	60	0.066089
T3	129	−0.520172	T3	129	−0.385953	T3	50	−0.17094
C3	129	0.388098	C3	129	−0.288464	C3	50	−0.40007
CZ	129	0.76757	CZ	129	−0.069916	CZ	50	−0.59621
C4	129	1.072921	C4	129	0.341911	C4	60	−0.48358
T4	129	0.30827	T4	129	0.109244	T4	50	0.139809
TP7	113	−0.878257	TP7	84	−0.733188	TP7	59	−0.32115
CP3	129	−0.043475	CP3	129	−0.580596	CP3	50	−0.75718
CPZ	129	0.574958	CPZ	129	0.004843	CPZ	50	−0.74632
CP4	129	0.767216	CP4	129	−0.271945	CP4	58	−0.97363
TP8	109	0.387432	TP8	113	0.812518	TP8	116	0.134941
M1	106	0.709347	M1	111	0.467339	M1	119	0.279987
T5	96	−0.507125	T5	92	0.27376	T5	103	−0.50401
P3	129	−0.473217	P3	129	−0.59096	P3	50	−1.03493
PZ	129	0.317936	PZ	129	−0.140338	PZ	57	−0.92731
P4	129	0.069284	P4	129	−0.092813	P4	58	−1.05643
T6	106	1.402455	T6	108	1.953835	T6	115	1.05858
M2	110	0.848855	M2	110	1.011522	M2	118	0.970858
O1	95	1.113134	O1	93	1.734712	O1	107	0.36699
OZ	93	0.68355	OZ	92	1.708575	OZ	101	0.112764
O2	94	1.195245	O2	93	2.659928	O2	102	0.542967

Table 6.  The detailed latency (ms) and amplitude (uV) of P1 in the hyperthermia group under post-test conditions.

Upright face			Inverted face			Table
Channel	Latency	Amplitude	Channel	Latency	Amplitude	Channel	Latency	Amplitude
FP1	129	1.381023	FP1	129	1.58103	FP1	50	0.061601
FP2	129	3.501933	FP2	129	1.804696	FP2	50	0.046306
F7	129	1.593295	F7	129	0.91887	F7	73	−0.55872
F3	129	1.033588	F3	129	1.444736	F3	65	−0.3025
FZ	129	1.377002	FZ	129	1.56983	FZ	62	−0.65035
F4	129	1.812557	F4	129	1.525637	F4	63	−0.3489
F8	129	1.387757	F8	129	0.593386	F8	50	1.831239
FT7	129	1.592752	FT7	129	1.460496	FT7	67	−0.32491
FC3	129	1.359736	FC3	129	0.965715	FC3	66	−0.8093
FCZ	129	1.355992	FCZ	129	1.287162	FCZ	66	−1.01143
FC4	129	2.149549	FC4	129	1.39138	FC4	64	−0.9076
FT8	129	1.337564	FT8	129	0.750409	FT8	80	0.306528
T3	129	0.831529	T3	129	0.664345	T3	53	−0.55083
C3	129	1.223374	C3	129	0.526312	C3	67	−0.79175
CZ	129	1.179601	CZ	129	1.016546	CZ	65	−1.20011
C4	129	2.356754	C4	129	1.218382	C4	76	−0.23077
T4	129	1.02849	T4	129	0.582093	T4	81	0.306651
TP7	90	1.538719	TP7	120	1.598792	TP7	104	0.000075
CP3	129	1.674456	CP3	129	0.934627	CP3	78	−0.07471
CPZ	129	1.292121	CPZ	129	1.254231	CPZ	78	−0.58301
CP4	129	1.5278	CP4	129	0.95748	CP4	80	−0.24704
TP8	91	1.099975	TP8	93	0.557465	TP8	108	0.021709
M1	99	0.871903	M1	101	0.682677	M1	101	0.556417
T5	90	3.397281	T5	101	3.056698	T5	105	0.881294
P3	129	1.226224	P3	129	0.397901	P3	83	0.659548
PZ	129	1.403674	PZ	129	0.92382	PZ	80	−0.15356
P4	129	0.647918	P4	129	0.545603	P4	84	0.328141
T6	92	3.147679	T6	96	3.113531	T6	105	1.848269
M2	104	1.598176	M2	98	0.852457	M2	106	1.254443
O1	90	4.159509	O1	95	4.569549	O1	90	2.558592
OZ	89	3.666601	OZ	93	3.718858	OZ	88	2.362736
O2	90	4.073209	O2	93	4.522532	O2	90	2.84671

Table 7.  The detailed latency (ms) and amplitude (uV) of N170 in the hyperthermia group under pre-test conditions.

Upright face			Inverted face			Table
Channel	Latency	Amplitude	Channel	Latency	Amplitude	Channel	Latency	Amplitude
FP2	199	−0.87915	FP2	120	−0.43933	FP2	199	−0.959926
F7	199	−1.534	F7	199	−1.00034	F7	199	−0.963691
F3	199	−1.17306	F3	120	−1.23397	F3	120	−0.979649
FZ	199	−0.95315	FZ	120	−0.89873	FZ	120	−0.736613
F4	199	−0.60589	F4	120	−0.31304	F4	199	−0.303423
F8	199	−0.70265	F8	199	−0.25778	F8	193	−0.976546
FT7	199	−1.5058	FT7	198	−1.47391	FT7	125	−0.948161
FC3	199	−1.15392	FC3	120	−1.35621	FC3	120	−1.173501
FCZ	120	−1.16015	FCZ	120	−1.36083	FCZ	120	−1.289863
FC4	120	−0.40322	FC4	120	−0.40048	FC4	120	−0.762521
FT8	199	−1.4081	FT8	199	0.040242	FT8	189	−0.604709
T3	187	−2.82477	T3	193	−2.4465	T3	163	−1.468952
C3	199	−1.58086	C3	120	−1.40657	C3	120	−1.081764
CZ	199	−1.17764	CZ	120	−1.48767	CZ	120	−1.49521
C4	199	−1.0289	C4	120	−0.55447	C4	184	−1.100357
T4	177	−2.11485	T4	192	−1.96541	T4	175	−1.682238
TP7	175	−4.59146	TP7	188	−4.61774	TP7	171	−3.100198
CP3	199	−1.97978	CP3	199	−1.69994	CP3	166	−2.41071
CPZ	199	−1.68392	CPZ	120	−1.32468	CPZ	120	−1.518264
CP4	183	−2.01082	CP4	194	−1.68937	CP4	171	−2.684356
TP8	171	−5.44248	TP8	171	−6.032	TP8	172	−3.832097
M1	170	−4.39415	M1	175	−4.48459	M1	175	−1.888703
T5	169	−7.13802	T5	170	−7.19522	T5	165	−5.603069
P3	182	−2.78922	P3	195	−2.48892	P3	166	−3.819804
PZ	199	−2.14227	PZ	199	−1.81066	PZ	170	−2.791081
P4	179	−3.50881	P4	176	−2.81046	P4	168	−4.207095
T6	168	−8.01089	T6	170	−9.29685	T6	171	−5.676867
M2	167	−4.20823	M2	174	−5.02492	M2	180	−1.582956
O1	163	−4.99682	O1	166	−5.75858	O1	164	−5.970306
OZ	165	−3.88007	OZ	168	−4.17333	OZ	166	−4.860272
O2	166	−5.82395	O2	168	−6.29586	O2	167	−5.95998

Table 8.  The detailed latency (ms) and amplitude (uV) of N170 in the hyperthermia group under post-test conditions.

Upright face			Inverted face			Table
Channel	Latency	Amplitude	Channel	Latency	Amplitude	Channel	Latency	Amplitude
FP2	196	0.042607	FP2	120	−0.02514	FP2	199	−0.61846
F7	198	−0.12897	F7	199	−0.81014	F7	120	−1.121352
F3	196	−1.1487	F3	120	−0.43564	F3	199	−1.307215
FZ	199	−0.79594	FZ	120	−0.69573	FZ	199	−1.406518
F4	198	−0.23842	F4	120	−0.45616	F4	199	−0.732452
F8	120	0.231566	F8	120	−0.808	F8	199	−0.010268
FT7	195	−0.14217	FT7	199	−0.4961	FT7	199	−1.277028
FC3	195	−0.31199	FC3	120	−1.02454	FC3	128	−1.671536
FCZ	199	−0.95878	FCZ	120	−1.04843	FCZ	120	−1.417822
FC4	197	0.260197	FC4	120	−0.58608	FC4	129	−1.15769
FT8	120	0.433381	FT8	120	−0.40386	FT8	199	−1.12895
T3	179	−0.78208	T3	186	−1.46103	T3	158	−1.849522
C3	194	−0.42548	C3	120	−1.13015	C3	129	−1.623367
CZ	197	−1.10723	CZ	120	−1.28514	CZ	120	−1.761923
C4	194	0.803898	C4	120	−0.48216	C4	134	−0.585963
T4	120	0.648635	T4	176	−0.41817	T4	147	−1.223149
TP7	165	−1.77393	TP7	172	−2.05284	TP7	159	−3.283889
CP3	186	0.474503	CP3	186	−0.31092	CP3	147	−1.54313
CPZ	195	−0.70121	CPZ	120	−0.69345	CPZ	131	−1.422187
CP4	185	0.115144	CP4	120	−0.22703	CP4	143	−1.670177
TP8	164	−2.38317	TP8	164	−4.45403	TP8	159	−3.536466
M1	164	−3.56096	M1	165	−4.50522	M1	161	−2.096493
T5	160	−3.61426	T5	165	−5.36873	T5	158	−4.927924
P3	174	0.098476	P3	180	−1.05206	P3	151	−2.40813
PZ	186	0.022096	PZ	188	−0.43808	PZ	141	−2.199466
P4	174	−1.30037	P4	178	−1.1166	P4	150	−3.077027
T6	160	−5.26908	T6	163	−8.109	T6	160	−5.166918
M2	161	−2.88476	M2	160	−4.15812	M2	170	−1.502871
O1	155	−2.31329	O1	156	−4.24739	O1	151	−5.141047
OZ	155	−0.80333	OZ	156	−2.47271	OZ	150	−4.094002
O2	156	−3.29645	O2	157	−4.82874	O2	150	−5.282829

Declaration of interest: The authors report no conflicts of interest. The authors alone are responsible for the content and writing of the paper.