Zoomed out: digital media use and depersonalization experiences during the COVID-19 lockdown

Depersonalisation is a common dissociative experience characterised by distressing feelings of being detached or ‘estranged’ from one’s self and body and/or the world. The COVID-19 pandemic forcing millions of people to socially distance themselves from others and to change their lifestyle habits. We have conducted an online study of 622 participants worldwide to investigate the relationship between digital media-based activities, distal social interactions and peoples’ sense of self during the lockdown as contrasted with before the pandemic. We found that increased use of digital media-based activities and online social e-meetings correlated with higher feelings of depersonalisation. We also found that the participants reporting higher experiences of depersonalisation, also reported enhanced vividness of negative emotions (as opposed to positive emotions). Finally, participants who reported that lockdown influenced their life to a greater extent had higher occurrences of depersonalisation experiences. Our findings may help to address key questions regarding well-being during a lockdown, in the general population. Our study points to potential risks related to overly sedentary, and hyper-digitalised lifestyle habits that may induce feelings of living in one’s ‘head’ (mind), disconnected from one’s body, self and the world.

To further investigate the pattern of lifestyle habits during the lockdown we conducted two principal components analyses (PCA) (see Table 1). The first PCA (A in Table 1) encompassed responses regarding frequencies of performing specific activities during the lockdown. Our analysis (model's Chi 2 (13) = 304.3, p < 0.001, Oblimin rotation, minimal correlation for factor loading r = 0.4) revealed that lifestyle habits investigated in our study comprised two independent factors (correlation between factors r = 0.08). The first factor consisted of items (all of them were positive factor loadings): (a) spending time outside home, (b) meeting people outside home, (c) physical activity, and (d) manual work. The second factor consisted of items: (e) playing computer games, (f) watching TV and audiovisual materials, and (g) participating in online meetings (also all responses were www.nature.com/scientificreports/ positive factor loadings). The results of the analysis revealed that interacting with people at one's home did not contribute to any of these factors. The second PCA (B in Table 1) was conducted on differences between lifestyle habits before and during the pandemic. The second PCA (model's Chi-squared(13) = 327.1, p < 0.001, Oblimin rotation, minimal correlation for factor loading r = 0.4) revealed a different factorial structure than the first one. The first extracted factor shows that people who were spending less time outdoors than before the pandemic were also (a) meeting relatively less people outside one's flat (than before the pandemic) and were (b) more likely to report playing more computer games, (c) watching more audiovisual materials and (d) participating more in online meetings. This factor indicates that the more people were restricting outdoors activities, the more they engaged in media consumption and digital social interactions. The second factor consisted of two items which both positively loaded the factor: (a) physical activity and (b) manual work, indicating that people who started performing more physical activity during the lockdown were also more likely to start performing more manual work. We also found the opposite: people who were doing less physical activity during the lockdown were also less likely to do manual work. The two extracted factors did not correlate with each other (r = 0.003). Once again, social interactions in person at one's home did not load any of the extracted factors.
Finally, we investigated how strongly each of the activities measured in our survey was related to the overall perception that lockdown influenced one's life. We have thus conducted a non-parametric correlation analysis between the estimated influence of lockdown on one's life and time spent weekly on the behaviours our survey measured. The results revealed ( Table 2) that the perceived influence of lockdown was significantly related to the change in performed activities before and during the lockdown for all activities included in our survey, except for manual work. The more people felt the influence of lockdown on their lives, (1) the less time they were spending being outside their flat/home, meeting people outside their flat/home, meeting people in their flat/home, and Relationship between depersonalization and lifestyle habits during lockdown. We conducted a correlational analysis (uncorrected for multiple comparisons) between the scores on the Cambridge Depersonalisation Scale (CDS-29) 56 (see "Methods" section below) and (a) responses to questions pertaining to participant's lifestyle habits now; (b) scores reflecting how participant's lifestyle habits have changed as a result of the pandemic. Table 3 presents the results of this analysis. Firstly, we found that the subjective estimation of how strongly lockdown is felt to have influenced one's life was correlated with the occurrence of DP experiences. Specifically, participants who reported that lockdown had strongly influenced their life also had higher scores on the CDS-29.
We also found that higher DP was positively related to digital media-based activities and distal social interactions. For example, participants who were spending more time during lockdown playing computer games, and who reported that they increased the time that they spent playing computer games during the lockdown (as compared to the pre-lockdown period) reported significantly stronger DP symptoms. Secondly, consumption of passive media such as watching movies, YouTube videos and television did not positively correlate with DP by itself, but people who reported that they were spending more time doing this during lockdown than before showed increased DP. Finally, frequent participation in online meetings was positively correlated with DP. Moreover, we found a weak significant negative correlation between the frequency of physical exercises during lockdown and the magnitude of DP showing that physical exercise during lockdown was weakly correlated to lower level of DP.
In contrast to digital media consumption, the amount of in person social contact outside of the home did not influence DP. Neither the absolute time spent meeting with other people indoors and outdoors, nor the difference in these activities before and after the lockdown, correlated with DP, with Bayesian analyses showing strong support for a null hypothesis.
Finally, with regards to the vividness of participants' emotions, we found that change in vividness of experienced emotions was significantly related to the occurrence of DP experiences. People with higher DP scores experienced positive emotions less vividly (Tau B = − 0.21, p < 0.001, BF10 = 2.2e + 11) and negative emotions more vividly (Tau B = 0.19, p < 0.001, BF10 = 4.1e + 9) than people with lower occurrences of DP experiences.

Discussion
This study examined the relationship between lockdown-related restrictions and experiences of DP during the lockdown as contrasted with before the pandemic. DP is a common dissociative phenomenon that makes people feel 'estranged' and detached from themselves and from their bodies 6 ,which in turn may lead to estrangement from others and consequent social isolation 9,10 .
Our study revealed several key findings. Firstly, we found that an increased use of digital media-based activities (especially playing computer games) correlated with higher feelings of DP. Strikingly, increased participation in online social e-meetings (as contrasted with the pre-lockdown period) was also positively correlated with DP Table 3. Correlations between depersonalization and lifestyle habits. *** = significant at the 0.001 level, ** = significant at the 0.01 level, and * = significant at the .05 level www.nature.com/scientificreports/ experiences. We also found that people who reported that they were spending more time during the lockdown engaged in sedentary digital media-consumption such as watching films, TV, YouTube videos (as contrasted with the pre-lockdown period), also showed DP experiences. Moreover, people reporting higher experiences of DP in our survey also reported enhanced vividness of negative emotions (as contrasted with positive emotions). Our study also reveals a weak significant negative correlation between the frequency of physical exercise during the lockdown and the occurrence of DP experiences. Intriguingly, the amount of in person social interactions, both indoors and outdoors, did not influence DP. Finally, our results suggest that the subjective estimation of how strongly lockdown restrictions influenced one's life was related to the occurrence of DP experiences. Specifically, participants who reported that lockdown influenced their life to a great extent had higher scores on the CDS-29.
Below we discuss these findings in detail. One of our key findings was that participants who (a) either spent a lot of time playing video games in general, or (b) reported that during the lockdown the time they spent playing video games had increased as compared to before the lockdown, showed higher depersonalisation experiences according to the CDS-29. While the direction of causation between these two phenomena is currently an open question, one may speculate about potential explanatory factors. For example, one may argue that computer-based gaming involves restricted and repetitive proprioceptive and proximal tactile stimulation (e.g. fingers on the keyboard, the mouse, or the telecommand, feet on the floor, sedentary posture, etc.). Indeed, videogaming, as the label suggests, rests heavily on enhanced visual and auditory sensory stimulation, to the detriment of proprioceptive, interoceptive, olfactory and tactile stimulation. This imbalance towards exteroceptive stimuli (being overstimulated by the outside or the 'virtual' world 'out there' may explain why people who spend more time videogaming also report higher feelings of selfdetachment and disembodiment, as well as lack of presence in the real world 'here and now' . Moreover, previous research has also shown the critical importance of dynamic, embodied and multisensory interactions with the social and physical environment for one's sense of self and mental health 40,46,52,57 . Another potential interpretation is that by disengaging from bodily signals, people may become more tempted to get more attentionally and emotionally involved in virtual realities than the real one. For example, playing computer games involves attentional detachment from both physical reality in favour of immersion in a virtual one, as well as detachment from one's body in favour of embodiment of one's digital body-an avatar 58,59 . Previous work outlined that the sense of self is critically dependent on bodily (i.e., interoceptive and proprioceptive) sensory processing 1,2,39,60,61 . If this is so, then distracting oneself from one's somatic feelings via digital audiovisual materials that puts the body in a more passive setting, should lead to higher DP. By contrast, higher dynamical and embodied engagements with the surroundings should lead to lower DP. Our findings seem to provide strong support for the former hypothesis and weak significant support for the latter hypothesis. Indeed, we found a weak significant negative correlation between the frequency of physical exercise during the lockdown and the occurrence of DP experiences. This suggests that physical exercise (i.e. dynamic activity engaging the entire moving body) may be related to lower DP. This idea is in line with previous work showing that DP is underpinned by disrupted bodily sensory processing as demonstrated by several studies [62][63][64][65][66] .
Another important finding is that increased participation in online and distal social e-meetings (as opposed to pre-lockdown period), positively correlated with higher DP experiences. The absolute time spent on e-meetings during the lockdown was positively related to increased DP, while the difference between lockdown and pre-lockdown did not reveal any effect. Again, while the direction of the causation between these phenomena remains to be further investigated, here we can speculate in favour of the impact of diminished dynamic and multisensory, proximal interactions with the physical and social environment on one's sense of self. During technology-mediated social interactions the body posture is primarily sedentary. Moreover, the communication is primarily audiovisual in a flat two-dimensional space, lacking the richness of the multisensory cues and affective reward that people typically gain from face-to-face in person interactions. Enhanced use of audiovisual signals situated 'out there' may prompt people to disengage, detach and pay less attention to bodily signals from 'here' , that is inside and closer to the physical body 39,40 .
Given that humans are highly social beings, one may find surprising our results suggesting that online social interactions enhance feelings of DP. For example, recent work exploring the link between loneliness, social isolation and social media provides a rich and more nuanced picture, with studies showing the importance of maintaining social media connectedness for people's well-being [67][68][69][70][71] . The relationship between social connectedness and social internet use is currently an open question, and further work is needed to carefully investigate the multifaceted aspects of these digital interactions. However, here we speculate in favour of the hypothesis that what matters for human wellbeing may not be just any type of social interactions, but close social interactions, that is, staying in touch (literally) with trusted and loved ones. This hypothesis seems to receive further support from recent studies exploring the relationship between social touch deprivation and psychological wellbeing during COVID-19 17 . Indeed, in early life and beyond, humans actively meet other peoples' bodies (via skin-toskin interactions) before they meet others' minds and interact with them from an visuo-distal viewpoint on a screen 39,40,46,48,52,72 . Yet, during the lockdown, and because of the social/physical distancing constraints, people spent a significantly higher amount of time in work-related distal e-meetings. Moreover, some people drastically reduced their face-to-face social interactions outside one's household, reducing thereby the richness of their social life. While some of those who experienced the lockdown with their close family may have benefited from the increase of amount of time spent together, by contrast, those who had conflictual family relationships may have benefited less from the indoors social interactions, putting them at risk. This may potentially explain our intriguing results highlighting a positive correlation between increased online e-meetings use and higher DP. Further work needs to disentangle these aspects in a systematic manner.
Another interesting result of our study is that people reporting higher experiences of DP in our survey also reported enhanced vividness of negative emotions (as opposed to positive emotions), which can be linked to higher anxiety and stress due to the uncertainty and threat posed by the COVID-19 pandemic. This is in line with www.nature.com/scientificreports/ previous work suggesting that a core feature of DP is atypical emotional processing: "I don't have any emotionsit makes me so unhappy" 12,62 . As Medford notes, this may seem self-contradictory, but on further questioning, the person explained that he experienced considerable inner turmoil, but felt reduced emotional response to external events or other people. Again, this is in line with the idea that DP involves feelings of being simultaneously 'trapped' in one's mind, and outside one's body, with enhanced focus on inner workings to the detriment of active and affective responsiveness to the external environment 48,73,74 . Intriguingly, the number of in person social interactions, both indoors and outdoors, did not influence depersonalization. This may be due however to a potential flooring/ceiling effect, as people were explicitly instructed to avoid in-person social interactions as much as possible. Indeed, given that (a) the largest decrease of self-reported activity during the lockdown (as opposed to pre-pandemic) was the time spent meeting people in person outside or inside one's home, and (b) we do not know to what extent these indoors and outdoors meetings involved close, touch-based social interactions (e.g. hugging vs saying 'hello' from safe distance), the relationship between social contact and DP remains inconclusive at this stage. Further work is needed to disentangle different aspects of social experiences (distal vs proximal) and DP.
It is likely that at the time of administering our questionnaire (23 April and 8 May 2021) all our participants had been significantly impacted by a lockdown in the last six months regardless of their place of residence (we have excluded those participants that reported not having experienced any lockdown in the past 6 months). However, different European countries imposed different lockdown measures (especially the Northern versus the Southern ones). In order to verify potential differences between participants' countries with regards to reported impact of lockdown on their lives we carried out an additional analysis (see Supplementary Material 1 and Supplementary Table 1) This analysis revealed no significant differences between any of the countries used in the analysis (we took the top six countries with respect to country of residence) with respect to the reported impact of lockdown on their lives. Thus, we are confident that our effects are down to the impact that the lockdown had on our participants on the individual level, rather than any between country differences with respect to the perceived impact of lockdown on everyday life.
Finally, our results suggest that the subjective estimation of how strongly lockdown restrictions influenced one's life was related to the occurrence of DP experiences. Specifically, participants that reported that lockdown influenced their life to greater extent, had higher scores in CDS-29. Although our study did not involve a systematic comparison of the same individuals before the pandemic and during its course (see "Limitations" section below), our findings strongly suggest that the experience of lockdown also led to increase in DP experiences. Indeed, lockdown-related restrictions led to a widespread increase in anxiety and stress 23,23,75 , which, combined with lack of proximal social interactions and increased digital-media consumption, may lead to DP experiences. On the other hand, since this study does not determine the direction of causation, it is also possible that people with higher DP experiences might be simply more likely to be sensitive to lockdown-related changes and hence report them as more strongly affecting this life. If the second hypothesis is correct, then on average participants tested during the lockdown should not differ in the occurrence of DP from people tested before the COVID-19 pandemic.
Although our study did not directly investigate this issue, our sample exhibited much higher DP scores compared to what has typically been observed in samples obtained before the COVID-19 lockdown. For example, Aponte-Soto et al. 76 published a study with 300 Italian participants where they reported an average general score in CDS-29 of M = 16.3 (SD = 18.3, range = 0-132). They also reported that only 2% (6 participants) obtained a score of ≥ 70 indicating, according to them, clinically significant DP symptoms. In contrast, in our study the average score was much higher and reached M = 49.3 (SD = 35.9, range = 0-244), and almost a quarter of participants (24%) obtained clinically significant scores ≥ 70. Other, smaller studies reported similar values to Aponte-Soto et al. 's. Sugiura et al. 77 in a Japanese version of CDS-29 tested on a Japanese population found that the average score in a control group that was not diagnosed with DP was M = 15.1 (SD = 16.5). Zingrone et al. 78 conducted another two large questionnaire studies with CDS-29 and found a relatively large average score on CDS-29 in their samples: Study 4: n = 256, M = 20.1, SD = 19.4, range = 0-93; Study 5: n = 591, M = 37.2, SD = 14.2, range = 0-279. However, while Study 4 was conducted on a random sample of a small city in central Virginia, USA, study 5 was conducted on subscribers of a para-psychology email newsletter. However, people interested in parapsychology might have the tendency to report more uncommon experiences than the rest of the population. Thus, the results from Zingrone and colleagues' study 5 might reflect elevated average DP experiences compared with the rest of the population.
More indirect evidence in support of the idea that lockdown-related experiences may lead to DP comes from the finding that in our sample the correlation between the CDS-29 subscales was higher than in previous studies. This suggests a more homogenous and global phenomenon, consistent with the fact that a significant amount of people worldwide experienced similar restrictions and ensuing negative states due to the pandemic. Overall, while our study does not provide direct ultimate evidence that COVID-19 lockdown caused an increase in the occurrences of experiences of DP, it provides indirect support for a causal link when our results are compared with the results of previous studies that used  Limitations and outlook. There are several important limitations of our study. Ideally, the same participants who took part in our online study should have been assessed for their DP experiences before the pandemic. Moreover, asking participants to recall their lifestyle habits from before the pandemic (probably six to nine months before the survey) is likely to be less accurate than if we had actually asked them about these habits before the pandemic (i.e., before March 2020). However, the occurrence of the COVID-19 pandemic was unforeseeable, hence our study was tailored 'on the fly' to assess lockdown-related experiences pre-vs during the www.nature.com/scientificreports/ pandemic as self-reported during the pandemic. Further work is needed to contrast DP experiences before and during the pandemic in the same large sample of participants. Second, our study was conducted between April and May 2021, and both the CDS-29 and our tailored lockdown survey were asking participants about their experiences in the previous 6 months. This means that participants were describing their lives during late autumn, winter, and early spring, as most of our participants were European-based. It is possible that life activities reported by participants were influenced by the fact that participants were asked about a period of the year in which they generally spend more time indoors (i.e., compared to summer). However, this should not affect correlations with CDS-29. Also, as we mentioned earlier, our study did not collect specific data about the type of in person interactions (distal versus proximal, romantic versus friends) that people had both indoors and outdoors during the pandemic.
Finally, we acknowledge that the use of an online recruitment tool (due to resource limitations) may limit the generalizability of our findings because demographic groups are unequally represented (e.g., an underrepresentation of those with limited computer use) due to coverage bias 79 . Future studies may employ probability sampling methods in order to address this topic in a more representative sample.
Our study may help to tackle key questions related to human well-being in the general population during a lockdown. Our results suggest that paradoxically, increasing online social interactions and digital activities may have negative effects in some people, such as inducing feelings of living more in one's 'head' (mind) and less in one's body. These findings point also to potential risks related to overly sedentary and hyper-digitalized lifestyle habits that may make people feel less 'real' and less connected with their close physical and social environment.

Methods
Participants. An a priori Power analysis conducted using G*Power 3.1.9.2 revealed that in order to detect correlations equal to or stronger than rho = 0.15 with alpha = 0.05 and beta = 0.95 the required sample size needs to be at least 571 participants. We estimated our target sample size to be approximately 640 people.
We recruited 691 people from the Prolific crowdsourcing platform (www. proli fic. co). Data collection took place between 23 April and 8 May 2021. For the analysis of effects of COVID-19 lockdown we excluded participants that reported that they hadn't experienced lockdown during the last 6 months for more than one month and those that left this question unanswered (69 people in total). Because we assumed that those who did not finish the survey (46 people) had also revoked their consent for us to use their data, their responses were not recorded.
Our final sample consisted of 622 participants. The mean age of all participants was 31.5 years (SD = 12.2, Median = 27, range . 257 of participants were female, 358 were male and 74 did not disclose their gender. All participants were required to be fluent in English (with respect to the participants' reported language proficiency on Prolific). Participants represented predominantly European nationalities: UK (220 participants), Poland (93), Portugal (65), USA (34), Italy (33), the Netherlands (22), Greece (18), and Germany (10). The remaining participants represented diverse nationalities from around the world. 73 participants did not disclose their nationality on Prolific. The places of residence of participants were predominantly European countries (550 participants), with the exception of Australia (1), Canada (7), Chile (1), Israel (3), Mexico (5), South Africa (18) and USA (33). 71 participants did not indicate their country of residence on Prolific. We did not place any constraints on the participants' place of residence or birth, therefore the distribution of our sample with regards to these was down to chance.
Informed consent was obtained from all participants before the start of the experiment according to procedures approved by the Monash University Human Research Ethics Committee (MUHREC). The experiment was conducted in accordance with the Declaration of Helsinki.
Measures. Cambridge Depersonalization Scale. All participants first completed the Cambridge Depersonalization Scale (CDS-29) 54 . CDS-29 is a 29 items standard questionnaire used to evaluate the severity of occurrence of depersonalization experiences by asking participants to estimate their frequency ("never", "rarely", "often", "very often", "all the time") and duration ("for a few seconds", "for a few minutes", "for a few hours", "for about a day", "more than a day", "more than a week") in the past six months. The total score (between 0 and 290) points was calculated by summing over all items. The CDS-29 has good statistical properties 56,76,77,80,81 with internal reliability for different language versions reported between 0.89 and 0.94 (Cronbach's alpha). Moreover, previous research has extracted four subscales from CDS-29 82 : Anomalous Body Experience, Emotional Numbing, Anomalous Subjective Recall, and Alienation from Surroundings. Given that in our sample all of these subscales correlated with each other more strongly (correlations between factors from r = 0.61 to 0.66) than in Sierra et al. 80 who reported correlations between r = 0.23 and 0.34, we decided to use only the full total score.
The average total score for the CDS-29 in our sample was 49.3 (SD = 35.9), with median at 41 points, mode at 32 points, and range between 0 and 244 points (with theoretical maximum being 290 points). The distribution of scores was strongly right-side skewed (Skewness = 1.29) and slightly leptokurtic (Kurtosis = 2.85) reflecting the fact that the majority of participants reported relatively low scores (the middle 50% of participants had scores between 23 and 68) (see Fig. 2). An investigation into the psychometric properties of our CDS-29 scores revealed that internal reliability was similarly high to previous studies, with Cronbach's alpha = 0.93. However, principal components analysis (PCA) revealed a different factorial structure than previous analyses. This was also reflected by relatively high correlations between subscales as determined by 80 , which varied between Pearson's r = 0.62 and 0.66. Due to this, all of our reported analyses were conducted on the total score rather than the subscales.

COVID-19 lifestyke habits survey.
(2) The second measure used in our study was a custom-designed lifestyle survey asking about life activities before and during the COVID-19 pandemic. Procedure. The study was programmed in Javascript using PsychoJS and hosted on the Pavlovia hosting service. Participants always started with completing the CDS-29 questionnaire, which was then followed with the survey asking about lifestyle before and during the COVID pandemic. Questions were displayed individually and there was no possibility to return to previous questions. Participants were informed that they could withdraw from the study at any time by pressing 'esc' . After completing the study, participants were debriefed via instructions displayed on the screen. They also received information about whether their score in CDS-29 was relatively high (if above 50) or low (below 20), and referred to a specific page for further information about depersonalization.
Data processing and analysis. Data processing and analysis was conducted using custom scripts written in Python. Bayesian analyses were conducted using JASP 0.9.0.1. For estimation of lifestyle habits before and during the lockdown we used (a) participants' responses about the frequency of each activity during the last 6 months as indices of their lifestyle habits during the lockdown, and (b) how much the pandemic had impacted their lifestyle habits, as indexed by the difference between the frequency of each activity before and during the pandemic measured as the frequency of the activity during the last 6 months minus the frequency of the activity before the pandemic (positive values represent an increase in the activity during the pandemic).
Due to the fact that frequencies were measured on an ordinal scale, all analyses involving them were conducted using non-parametric statistical methods. Wilcoxon's tests (W) were used in order to compare lifestyle behaviours before the pandemic with lifestyle behaviours during the pandemic. Kendall's tau B (Tau B) was used in order to investigate the relationship between the participant's estimations of the influence of lockdown on their life and the reported time they spent weekly on each life activity. We also used Kendall's tau B to estimate the  Table 4. Questionnaire measuring lifestyle habits before and during the lockdown.
Have you experienced COVID pandemic-related restrictions (lockdown) during the last 6 months for more than 1 month?

Yes No
To what extent have pandemic-related restrictions influenced your life? 1: I experience them less vividly than before the pandemic 2 3: I experience them similarly as before the pandemic 4 5: I experience them more vividly than before the pandemic How do you experience your NEGATIVE EMOTIONS (for example: sadness, anger, fear, disgust)?
1: I experience them less vividly than before the pandemic 2 3: I experience them similarly as before the pandemic 4 5: I experience them more vividly than before the pandemic