Effects of Thirty Days Isolation on Attention Networks: A Behavioral and Event-related Potential Study

Dear Editor,

Individuals in isolated, confined, and extreme environments, such as astronauts, researchers wintering over Antarctica, submarine soldiers, and participants in simulated isolation environment experiments, commonly report difficulty in focusing their attention [1, 2]. However, current experimental findings from studies conducted in such environments are insufficient to fully support these self-reports of reduced attention. Certain attention-demanding tasks such as simple reaction time (RT) tasks and dual tasks, including primary pursuit tracking and secondary RT, were not affected in space flight experiments [2, 3]. In standardized cognitive tests conducted on Antarctic winterers, some studies have reported reductions in accuracy and increases in RT, while other studies have not observed any decline in cognitive performance [4].

Attention refers to the ability to direct and focus psychological activities, such as perception or thinking on an objective object, and accompanies the individual in all cognitive activities. Selective attention is considered to be composed of three functionally independent subnetworks, including the alerting network, orienting network, and executive control network [5]. So far, few studies have examined the impacts of confined and isolated environments on individual attention networks. Furthermore, to the best of our knowledge, no studies have been conducted on the alterations in cognitive activity that occur due to such environments using event-related potential (ERP) technology. Only a few electroencephalogram (EEG) studies have been conducted for isolated environments, while they just examined resting-state EEG data. In the MARS500 study organized by the Russian Space Agency [6, 7] and the SIRIUS-19 mission organized by the National Aeronautics and Space Administration (NASA) [8], researchers found a decrease in overall brain cortical activity (alpha and beta activity) accompanied by elevated salivary cortisol levels as the duration of isolation increased. These results suggested that confined isolation is a chronic stressor and can lead to neurophysiological changes in the brain, which may be caused by the monotonous environment, monotonous daily activities, and less social communication in long-term isolation [8]. Continuing research on the effects of confined and isolated environments on attentional networks and the cognitive neural mechanisms underlying them is crucial. This significance is particularly apparent in missions such as long-term space flights where astronauts perform complex and hazardous tasks with very little tolerance for error and an extreme reliance on attention [9]. Therefore, the purpose of the current study is to examine the effects of confined and isolated environments on individual attentional networks and the associated changes in brain activity.

Six crew members completed a 30-day confined isolation experiment in a simulated spacecraft experiment capsule. Three Attention Network Tasks (ANTs) were conducted on days 2, 14, and 26 while EEG data were recorded. The number of participants and the principles of statistical analysis are consistent with two classical simulated space missions (MARS500 Mars Simulation Mission [6, 7], and SIRIUS-19 Mission [10]). A non-isolated control group of ten participants completed identical tests at corresponding time intervals. The inclusion and exclusion criteria for participant recruitment, experimental procedures, data acquisition, data analysis, and statistical analysis methods are detailed in the Supplementary Materials.

The primary finding of the ANT is that the interaction between group and isolation time had a significant effect on accuracy for incongruent targets [F_{(2, 45)} = 6.136, P = 0.027, η² = 0.305]. Further analysis using post hoc comparisons revealed that the accuracy of incongruent targets was significantly lower on the third test in the isolation group compared to the first test (P = 0.042), while no such change occurred in the control group (P = 0.576). Importantly, there was no significant change in the accuracy of congruent targets between the two groups (P > 0.9). These findings suggest that the ability of the isolation group to effectively respond to incongruent targets is negatively impacted as the isolation time increases, whereas the control group does not exhibit a similar decline in performance. The detailed behavioral results of the ANT as well as the detailed ERP results for N1, N2, and P3 are provided in the Supplementary Materials.

The target-locked N1 amplitude of the isolation group was found to be larger than that of the control group in both the alerting network [F_{(1, 94)} = 22.411, P < 0.001, η² = 0.615] and the orienting network [F_{(1, 94)} = 10.547, P = 0.006, η² = 0.430]. However, no discernible trend in N1 amplitude was observed with increasing isolation time. Grand averaged ERP waveforms are presented in Fig. 1.

Fig. 1.

Grand averaged ERP waveforms of each electrode position, cue, and target condition. From left to right are the first test, the second test, and the third test. The target (Tar) start time is 0.5 s (dashed line). Negative amplitude values are plotted upwards. A The isolation group. B The control group.

We observed a larger N2 amplitude in the isolation group compared to the control group [F_{(1, 94)} = 5.653, P = 0.032, η² = 0.288]. An interaction effect between group and target was also significant [F_{(1, 94)} = 6.204, P = 0.026, η² = 0.307]. Moreover, a trend towards an interaction effect [F_{(2, 45)} = 3.784, P = 0.060, η² = 0.431] between target type and isolation time in the isolation group was observed, but not in the control group [F_{(2, 45)} = 0.107, P = 0.899, η² = 0.012]. This suggests that the difference in N2 amplitude between the two targets in the isolation group was decreasing with increasing isolation time, whereas no similar trend was observed in the control group. The topographies of the execution control effect in the N2 in each test and the difference of topographies between the third and first tests were displayed in Fig. 2 Notably, the isolation group showed a decrease in N2 for incongruent targets and an increase in N2 for congruent targets in the anterior part of the topography. The difference between the third and first tests was evident in the anterior scalp of the isolation group, while no such changes and trends were observed in the topographic maps of the control group.

Fig. 2.

The topographies of the execution control effect in the N2 in each test and the difference of topographies between the third and first tests. A The topographical of the execution control effect in isolation group. B The topographical of the execution control effect in the control group. C The difference in topographies between the third and first tests in the isolation group. D The difference in topographies between the third and first tests in the control group.

The analysis of P3 revealed several significant interaction effects. Specifically, the target × group interaction was significant [F_{(1, 94)} = 12.735, P = 0.003, η² = 0.476], the time × target interaction was significant [F_{(2, 93)} = 9.200, P = 0.007, η² = 0.297], as well as the time × target × group interaction was significant [F_{(2, 93)} = 6.279, P = 0.006, η² = 0.310]. The interaction between targets and groups can be explained by the larger differences in P3 amplitude between different target types in the isolation group compared to the control group. Additional analysis of the other interaction effects revealed two primary trends. First, a gradual diminishing of the P3 amplitude difference between incongruent and congruent targets was observed with increasing isolation time in the isolation group, while no such trend was observed in the control group. Second, another trend observed was the overall decrease in P3 amplitude of the isolation group. Brain topography maps of N1, N2, and P3 are included in the supplementary materials (Fig. S2), which do not distinguish between various cues and target conditions as we aimed to highlight isolation time.

Prior to and following the isolation experiment, participants from the isolation group were interviewed twice by a psychologist. During the pre-mission interview, the participants expressed a general sense of stress due to their inexperience with extended isolation and their concerns about completing the 30-day isolation experiment successfully. After the experiment, the majority of participants (5 out of 6) reported difficulties in concentration during the latter half of the isolation period, which is similar to reports from astronauts and members overwintering in the Antarctic. Moreover, the interviews revealed that the participants experienced consistently high levels of stress throughout the experiment. The oppressive and lonely environment contributed to feelings of depression that were difficult to alleviate. The feelings of oppression, loneliness, and stress persisted throughout the 30-day isolation period.

The primary objective of this study was to assess the effects of confined and isolated environments on the alerting, orienting, and executive control functions of the attention network. In terms of behavioral results, both the isolation and control groups performed significantly faster and more accurately during congruent flanker conditions than incongruent flanker conditions, indicating a typical flanker congruency effect. These results were consistent with previous findings [11]. Moreover, we found that the executive control network function of the participants from the isolation group deteriorated, and this could be explained by the interaction of the group and the test time of incongruent targets. As the duration of isolation increased, the isolation group experienced greater difficulty with the flanker conflict effect, whereas the control group exhibited no significant changes. For the other attention networks, no statistically significant effects of isolation time and group were found behaviorally. Previously, NASA assessed changes in visual perceptual attention during a 30-day isolation study, which also found no behavioral evidence of altered visual perceptual attention [12]. It is consistent with our behavioral results in the alerting network.

Previous research has demonstrated that stress significantly impacts the amplitude of N1 and N2. Liu et al. utilized the same ANT paradigm to investigate the effects of chronic stress on attention networks, and revealed that the target N1 and N2 amplitudes were significantly higher in the chronic stress group than in the control group [11]. Similarly, a study using the Eriksen flanker task yielded comparable results, indicating that stress elevates N1 and N2 amplitudes [13]. We posit that the increase in N1 and N2 amplitudes observed in the isolation group in our study is also linked to this mechanism. While we did not quantitatively assess the stress levels in the isolation group, the two interviews conducted before and after the experiment provided some insight. Unlike the experienced astronaut crew, the participants lacked prior exposure to similar isolation experiments, increasing their stress levels due to the unknown challenges they may face during the 30-day isolation period. Moreover, the disconnection from the outside world made it challenging for the participants to alleviate negative emotions, further exacerbating their stress levels throughout the experiment. The N1 component is known to be closely associated with spatial attention and is typically enhanced with increasing selective attention resources allocated to target stimuli [14]. No additional outcomes were found in the N1 analyses of the alerting and orienting networks, and no evidence was found to support the impact of isolation duration on these networks.

Two primary findings were discovered concerning N2. First, the amplitude of the incongruent target-locked N2 was found to be greater than that of the congruent target, consistent with previous research on flanker tasks [15]. N2 in the anterior scalp has been linked to competing response conflicts, and has commonalities with the error-related negativity (ERN). Greater brain activation is observed when participants respond incorrectly to the target. Similar effects exist in the Stroop paradigm [16], and although latencies are later in this paradigm, they still reflect the same cognitive processes. Secondly, the difference between the two target-locked N2 amplitudes decreased with increasing isolation time in the isolation group, while no similar changes were found in the control group. This finding is noteworthy, as it corresponds to changes in executive control function observed in the behavioral performance of the isolation group. One study proposed that variations in N2 amplitude could potentially act as an indicator of the efficacy of conflict detection [17]. From this perspective, the decrease in amplitude difference between N2 of the incongruent target and the congruent target represents a decrease in the ability to monitor conflicts and resist interference. As a result, we suggest that confined and isolated environments have a negative impact on the function of the executive control network. This impact might be related to the chronic lack of corresponding external stimulus input due to isolation [6].

The P3 amplitude is also affected by stress. A recent study utilizing the Go/Nogo paradigm explored the effects of academic stress on response inhibition, and discovered a significant interaction of the P3 component between group and response type [18]. The P3 amplitude differences between the Nogo and Go trials were more significant in the stress group than in the control group. This is in line with our finding that larger differences in P3 amplitude between different target types were observed in the isolation group compared to the control group. By combining EEG and functional magnetic resonance imaging (fMRI), the researchers propose that N2 and P3 functions are separated, with N2 found to be associated with conflict detection and P3 with response inhibition [19]. The P3 activity in the flanker task reflects the allocation of cognitive resources, and the decrease in P3 for flanker stimuli indicates a reduction in attentional allocation towards task-relevant stimuli [18]. So incongruent target-locked P3 is suggested positively correlated with the ability of executive control [11]. In the later stages of isolation, a reduction in the difference between P3 amplitudes elicited by the two targets was observed, suggesting that there are fewer cognitive resources available for resolving conflicting information and controlling behavior. Furthermore, the isolation group exhibited an overall decline trend in P3 amplitude. Previous studies showed that isolated environments lead to a decrease in overall brain activity [8], and a change in P3 amplitude during the late stage of isolation might be related to this mechanism.

The distribution pattern of attentional processing regions throughout the brain indicates their extensive involvement in this crucial function. Notably, the executive control network is an important aspect of attentional processing as it involves higher-level cognitive processing and is primarily governed by the frontal cortex. Prior research has established a more pronounced impact of isolation on the frontal and parietal regions [12], which are the brain areas most critically implicated in the executive control function. This may account for the greater degree of impairment observed in the executive control network compared to the other two networks.

Despite our best efforts to improve the study, certain limitations still remained. Firstly, the spaceflight simulation experiment was hindered by limited space and high costs, resulting in a relatively low sample size. Our sample size was comparable to that of MARS500 and SIRIUS-19. To minimize the potential impact of anomalous data on our findings, a thorough examination of each data was conducted prior to statistical analysis. Moreover, we cannot fully discount the influence of other psychological factors, such as alterations in emotion, stress levels, and sleep rhythms. These factors are also associated with the isolation effect, and the changes observed in the attention network are likely to be the outcome of the multifaceted interplay of several factors. As a result, it is difficult to comprehensively disentangle these factors for further discourse.

The current findings suggest that isolated and confined environments lead to changes in attention control function. Moreover, we found evidence of the effects of isolation on attention networks through neurophysiological measurements. These results support the subjective perception of reduced attention reported by astronauts and are instructive to the development of more effective training and response programs for astronauts. Overall, the findings highlight the value of ERPs as electrophysiological markers of attentional changes in isolated environments. Future research should undertake a detailed exploration of the specific neural basis behind altered attentional processes in confined and isolated environments. This can lead to the development of more effective methods for enhancing attentional control in isolated settings and mitigating the adverse effects of isolation on cognitive function and mental health.

Supplementary Information

Below is the link to the electronic supplementary material.

Conflict of interest

The authors claim that there are no conflicts of interest.