Averaging sleep spindle occurrence in dogs predicts learning performance better than single measures

Abstract

Although a positive link between sleep spindle occurrence and measures of post-sleep recall (learning success) is often reported for humans and replicated across species, the test-retest reliability of the effect is sometimes questioned. The largest to date study could not confirm the association, however methods for automatic spindle detection diverge in their estimates and vary between studies. Here we report that in dogs using the same detection method across different learning tasks is associated with observing a positive association between sleep spindle density (spindles/minute) and learning success. Our results suggest that reducing measurement error by averaging across measurements of density and learning can increase the visibility of this effect, implying that trait density (estimated through averaged occurrence) is a more reliable predictor of cognitive performance than estimates based on single measures.

Full text



Scientic Reports | (2020) 10:22461 | 
www.nature.com/scientificreports
Averaging sleep spindle occurrence
in dogs predicts learning
performance better than single
measures
Ivaylo Borislavov Iotchev1,4*, Vivien Reicher1,3,4, Enikő Kovács1,2,4, Tímea Kovács1,
Anna Kis2,5, Márta Gácsi1,3,5 & Enikő Kubinyi1,5
Although a positive link between sleep spindle occurrence and measures of post-sleep recall (learning
success) is often reported for humans and replicated across species, the test–retest reliability of the
eect is sometimes questioned. The largest to date study could not conrm the association, however
methods for automatic spindle detection diverge in their estimates and vary between studies. Here we
report that in dogs using the same detection method across dierent learning tasks is associated with
observing a positive association between sleep spindle density (spindles/minute) and learning success.
Our results suggest that reducing measurement error by averaging across measurements of density
and learning can increase the visibility of this eect, implying that trait density (estimated through
averaged occurrence) is a more reliable predictor of cognitive performance than estimates based on
single measures.
Sleep spindles are thalamocortical transmissions1,2 observed mostly in mammalian non-REM sleep3 as brief
(0.5–5 s4) trains of symmetric waves5 in the EEG signal. Dierent propositions for their dening frequency
(waves/second) overlap in the 9–16Hz band among humans6,7, mice8, and dogs9–11.
e most oen reported cognitive correlate of sleep spindles in humans is a positive relationship with post
sleep-recall (learning success). However, this has almost exclusively been reported in smaller samples12–23, and
thus the reliability of this eect requires a stronger conrmation. On one hand, the study using the largest to
date sample could not nd such anassociation24. Moreover, it is troublesome that dierent studies use dierent
algorithms2, since automatic spindle detection methods diverge in their estimates of spindle occurrence25. On
the other hand, invasive work in animal models has revealed putative mechanisms26 to explain how spindles
promote memory consolidation, as well as implicated causality27 where human data mostly allows only for cor-
relation. e issue thus remains controversial.
Here we report a replication analysis for the link between spindle occurrence and learning in dogs. e dog
(Canis familiaris) is a fairly new model species in sleep spindle research, but one advantage in addressing the
problem of replicability is that currently only one method for detecting canine spindles has been consolidated
across all published studies (from one single research group, Iotchev etal.9–11). e same method, adopted from
the human literature28 with minor alterations for use in dogs9, will also be used here to avoid the problem of
automatic detector divergence25. Moreover, the current analysis will include all unpublished data sets available
to us which are t for this analysis. is is crucial because in the human literature publication bias is suspected
to underlie eects reported for (fast) spindle density29. e goal of the analyses presented here will be to evaluate
if the relationship between sleep spindle occurrence and learning success is real or a type I error. To evaluate
this, we will look both at the prevalence of positive ndings and the conditions under which a positive or nega-
tive nding is observed. We will thereby also compare the prevalence of associations between (1) single and (2)
averaged measurements of sleep spindle density (spindles/minute) and learning, the latter being deemed more
likely to reect underlying traits and to be freer of measurement error.
OPEN
            
           
      
    
   
 *
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol:.(1234567890)
Scientic Reports | (2020) 10:22461 | 
www.nature.com/scientificreports/
Methods
Procedure. ree datasets were included in the analyses. In all datasets the basic learning paradigm (“novel
words paradigm”) was based on the one used to obtain the data for Iotchev etal.9 and Kis etal.30 (for details see
Supplementary Material). We included the dataset used in these studies for comparison (data set 0, N = 15). Data
set 1 (N = 19) originated from Reicher etal., in prep and data set 2 (N = 13) from Kovács etal. in prep. One dog
participated in data set 0 and 2, otherwise the samples did not overlap. Before sleep, dogs were required to learn
novel words (in English), and associate them with actions that they had been trained to perform to dierent ver-
bal commands before (in Hungarian). Aer sleep, the nal performance was measured as the percent of correct
trials (out of eighteen, on the re-test), and learning gain (% performance re-test minus test) was also calculated.
All three datasets were comprised of an adaptation sleep, followed by two counter-balanced, repeated-measures
conditions. Condition 1a used a supportive type of training (using both food and social reward in case of cor-
rect action and no scolding in case of incorrect action), while in condition 1b a controlling type of training was
conducted by a dierent experimenter (using only food reward without social reinforcement in the case of cor-
rect action and scolding in case of incorrect action). In condition 2a training was carried out by the owner in a
socially relevant manner (using both food reinforcement and social reward in case of correct action), in condi-
tion 2b training was carried out by an experimenter unknown to the dog in a socially irrelevant manner (using
food reinforcement but without social reward in case of correct action).
Subjects. A total of 46 dogs (23 females, age range 1–9years, 28 were purebred representatives from 16 dif-
ferent breeds) participated in the three studies (0, 1, 2). One dog was included in both study 0 and 2 (a female
golden retriever, aged 1year in study 0 and 2years in study 2). Because search for sleep spindles was restricted
to non-REM sleep as in Iotchev etal.9,10, dogs which did not reach this stage or had otherwise corrupted or miss-
ing les were assigned missing values for sleep spindle density. During adaptation (occasion 1), three dogs were
assigned missing values in study 0, ve dogs in study 1, and two dogs in study 2. Regarding data from experimen-
tal conditions (occasion 2 and 3), two missing values were assigned in condition 1a, and one in condition 2a (see
conditions below). Missing values were excluded from our analyses and the calculation of averages.
Ethical statement. According to the Hungarian regulations of animal experimentation, our non-invasive
polysomnography research does not qualify as an animal experiment. e Hungarian Scientic Ethical Commit-
tee of Animal Experiments issued a permission (under the number PE/EA/853-2/2016) approving of our non-
invasive protocol. All owners volunteered to participate in the study and were informed about the procedure
before beginning.
Electrode placement and EEG post-processing. Electrode placement (see Fig.1) followed the method
outlined by Kis etal.31 e polysomnographic recordings were manually categorized into sleep stages (wake,
drowsiness, non-REM, REM, see Supplementary for example images) according to standard criteria31 (validated
in Gergely etal.32). e traces identied as non-REM (descriptive statistics in the Supplementary) were scanned
for spindles using the frontal (Fz) and central (Cz) midline electrodes.
Spindle detection. Automatic detection was implemented as in Iotchev et al.9 on parts of the signal
marked as non-REM sleep and pre-ltered between 5 and 16Hz. Specics of the applied algorithm are detailed
in Iotchev etal.9,10 and based on similar criteria validated against visual experts on human EEG by Nonclercq
etal.28. Importantly, the algorithm invokes 2 steps, as initial detections are used to re-calculate boundaries for the
target amplitude and frequency of spindles for each dog and recording, in line with the assumption that these are
normally distributed within individuals. In the rst step the frequency is assumed to be 9–16Hz and a minimum
amplitude criterion of more than 1 standard deviation above the average of the searched signal is set. For the
individual adjustment in the second step the algorithm calculates maximum likelihood estimates for the means
and standard deviations of amplitude and frequency. e amplitudes and frequency of the nal detections have
to be within 2 standard deviations of the estimated means.
Figure1. Schematic drawing (by Vivien Reicher) of electrode placement in the dog identical in study 1 & 2
(study 0 used the same electrode placement, but without the F7 channel).
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol.:(0123456789)
Scientic Reports | (2020) 10:22461 | 
www.nature.com/scientificreports/
Analysis. Since only Fz was active in data set 0, we will focus below mainly on results obtained from Fz. We
will also refer to and discuss detections from across the whole 9–16Hz frequency range, considering that many
studies looking into spindle–learning correlations did not divide spindles into slow and fast ones12–14,33. We refer
to Supplementary TablesS1, S2, S4 and S5 for statistics obtained across both electrodes, and spindle sub-types
(slow and fast). All associations were tested with Pearson correlations using SPSS v25.
Results
By comparing single measurements (i.e. from one single attendance, same condition) of sleep spindle occur-
rence and learning success, we found an association between spindle density and learning gain for data sets 0
and condition 1a, but not for condition 1b and data set 2 (see Fig.2). Upon closer examination, both eects
were specic to the slow spindle type (TableS2) and for condition 1a a trend for the same eect was observed
on Cz, as well (TableS1).
Next, we compared averaged measurements (across attendances/conditions) of sleep spindle occurrence
and learning, to test how reducing measurement error could aect the relationship between spindle density and
learning in the three data sets, considering that a single measurement for each variable might be less reliable34.
Averaging across attendances/conditions was deemed valid, because one of the rationales for this analysis was to
approximate the underlying traits, rather than estimate a condition-specic expression of either spindle occur-
rence or learning performance. erefore, we averaged within each data set and for each dog spindle density
values obtained from all three recordings (in data set 0 these correspond to the adaptation, control and learning
conditions, while in data set 1 and 2 there was only an adaptation and two learning conditions; due to missing
values, for some dogs the averages were based on only two recordings, see Supplementary for further details).
We furthermore averaged learning performance variables (nal performance and learning gain) for the two
learning conditions in data sets 1 and 2 (this was not possible for data set 0 in which there was only one measure
for each). We tested associations with both (averaged) learning gain and (averaged) nal performance because
their distribution and range (TableS3) suggestedthat for each data set a dierent read-out variable might better
reect the underlying learning process.
Averaged density was positively associated with learning gain in data set 0 and average nal performance
in data set 2 (Fig.3). ese eects were also signicant for the slow sub-type and specic to Fz (TablesS4, S5).
Neither averaged learning performance variables were associated with averaged density in data set 1, but note
that in condition 1b, more than half of the dogs (57.9% or 11 out of 19 animals) worsened their performance
on the novel task aer sleep. In comparison, only 7 dogs (36.8%) did so in condition 1a. Together with a visual
inspection of the data (Fig.1) and the performance overview provided in TableS3 (lowest average values for
learning gain and nal performance), these numbers suggest a oor eect on learning success in condition 1b.
Discussion
A positive association between dogs’ spindle occurrence and learning success could be demonstrated in each
of the three data sets. As a note of caution,dierent transformations of the raw variables(like averaged scores)
and operationalization of learning (learning gain versus nal performance)were used between some of the
Figure2. Associations between spindle density (spindles/minute) and learning gain as measured over the
frontal midline electrode (Fz) in data sets 0, 1, and 2. For the already published nding in data set 09 we used
lighter colors. Two dogs did not sleep in condition 1a and one dog did not sleep in condition 2b, these animals
were excluded from the analysis. Schematic drawing (by the corresponding author) of how learning gain was
calculated during tests on the novel task.
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol:.(1234567890)
Scientic Reports | (2020) 10:22461 | 
www.nature.com/scientificreports/
comparisons,while all available samples arealsosmall (N < 20). An argument for treating the positive associations
as cumulativederives from usingthe same detection algorithm9,andthe same experimental operationalization
of learning (the “novel words paradigm” of Kis etal.30) across all experiments. Intriguingly, all signicant asso-
ciations were specic to slow and frontal spindles (see Supplementary TablesS2, S4, S5) which resembles what
is seen in humans when learning is tested with verbal material, like word-pairs13,16,23. In further support of the
spindle-learning association, Type II errors arecommonin small, and thereby likely underpowered samples35.
False negatives are also likely, considering that memory consolidation is not restricted to sleep alone in neither
humans nor dogs30,36. Also, many additional conditions are known to inuence if any eect is observed, e.g.
relative timing to ripples and slow-waves37, emotional arousal38 and exact stage of non-REM sleep14. However,
since in most animals it is hard to separate non-REM sleep stages from each other31,39 not all of these conditions
can be tested outside of humans.
Surprisingly, although sleep-dependent memory consolidation operates in the time-frame of a single
day40,41 and exposure to new information has been shown to result in direct increases in spindle occurrence
in humans15,23, rats33 and dogs9, our results for data sets 0 and 2 suggest that estimating trait density by averag-
ing across recordings might increase the visibility of spindle-learning associations. Other arguments for the
predictive utility of trait density come from reports of stable spindle occurrence across nights in humans42,43,
the heritability of sleep spindle density44, and the observation that dierent psychiatric conditions and natural
aging, each associated with memory problems, can measurably reduce spindle occurrence in humans45–49 and
specically the occurrence of slow spindles in dogs10.
We conclude thatthe here examined data-sets lendadditionalsupport to thepositive association between
sleep spindle occurrence and learningobservedin dogsearlier9, but the need for further evidence is not
exhausted. Even more and larger samples will berequiredtoestablish to what extendlow poweraccounts for the
proportion of null results. Moreover, since traitoccurrenceis also associated with general mental ability50more
studies with a control condition, in which sleep is not preceded by learning demand,will beneeded in dogs to
separateif these correlationsreect memory consolidationorgeneral learning potential.
Received: 19 June 2020; Accepted: 17 December 2020
References
1. Steriade, M. & Llinás, R. R. e functional states of the thalamus and the associated neuronal interplay. Physiol. Rev. 68, 649–742
(1988).
2. Fernandez, L. M. J. & Lüthi, A. Sleep spindles: Mechanisms and functions. Physiol. Rev. https ://doi.org/10.1152/physr ev.00042
.2018 (2020).
3. Kryger, M. H., Roth, T. & Dement, W. C. Principles and Practice of Sleep Medicine (Saunders/Elsevier, Amsterdam, 2011).
4. Rechtschaen, A. & Kales, A. A Manual of Standardized Techniques and Scoring System for Sleep Stages of Human Subjects (D.C.
U.S. Gov. Print. O. NIH Public, Washington, 1968).
5. Dutertre, F. Catalog of the Main EEG-Patterns .Handbook of Electroencephalography and Clinical Neurophysiology (Elsevier, Amster-
dam, 1977).
6. Bódizs, R., Körmendi, J., Rigó, P. & Lázár, A. S. e individual adjustment method of sleep spindle analysis: Methodological
improvements and roots in the ngerprint paradigm. J. Neurosci. Methods 178, 205–213 (2009).
7. De Gennaro, L. et al. e electroencephalographic ngerprint of sleep is genetically determined: A twin study. Ann. Neurol. 64,
455–460 (2008).
8. Kim, D., Hwang, E., Lee, M., Sung, H. & Choi, J. H. Characterization of topographically specic sleep spindles in mice. Sleep 38,
85–96 (2015).
9. Iotchev, I. B., Kis, A., Bódizs, R., van Luijtelaar, G. & Kubinyi, E. EEG transients in the sigma range during non-REM sleep predict
learning in dogs. Sci. Rep. 7, 12936 (2017).
Figure3. Associations between trait density (estimated here by averaging across recordings) and learning
success (also averaged for data set 2, based on nal performance) over the frontal midline electrode (Fz) in data
sets 09 and 2. Averaging learning gain was not possible in data set 0 as the experiment consisted of only one
learning condition. Data from the published data set 0 is presented in lighter colors.
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol.:(0123456789)
Scientic Reports | (2020) 10:22461 | 
www.nature.com/scientificreports/
10. Iotchev, I. B. et al. Age-related dierences and sexual dimorphism in canine sleep spindles. Sci. Rep. 9, 10092 (2019).
11. Iotchev, I. B., Szabó, D., Kis, A. & Kubinyi, E. Possible association between spindle frequency and reversal-learning in aged family
dogs. Sci. Rep. 10, 6505 (2020).
12. Clemens, Z., Fabó, D. & Halász, P. Twenty-four hours retention of visuospatial memory correlates with the number of parietal
sleep spindles. Neurosci. Lett. 403, 52–56 (2006).
13. Clemens, Z., Fabó, D. & Halász, P. Overnight verbal memory retention correlates with the number of sleep spindles. Neuroscience
132, 529–535 (2005).
14. Cox, R., Hofman, W. F. & Talamini, L. M. Involvement of spindles in memory consolidation is slow wave sleep-specic. Learn.
Mem. 19, 264–267 (2012).
15. Gais, S., Mölle, M., Helms, K. & Born, J. Learning-dependent increases in sleep spindle density. J. Neurosci. 22, 6830–6834 (2002).
16. Kuula, L. et al. Higher sleep spindle activity is associated with fewer false memories in adolescent girls. Neurobiol. Learn. Mem.
157, 96–105 (2019).
17. Seeck-Hirschner, M. et al. Declarative memory performance is associated with the number of sleep spindles in elderly women.
Am. J. Geriatr. Psychiatry 20, 782–788 (2012).
18. Lustenberger, C. et al. Feedback-controlled transcranial alternating current stimulation reveals a functional role of sleep spindles
in motor memory consolidation. Curr. Biol. 26, 2127–2136 (2016).
19. Yordanova, J., Kolev, V., Bruns, E., Kirov, R. & Verleger, R. Sleep spindles in the right hemisphere support awareness of regularities
and reect pre-sleep activations. Sleep 40, 1–13 (2017).
20. B arakat, M. et al. Fast and slow spindle involvement in the consolidation of a new motor sequence. Behav. Brain Res. 217, 117–121
(2011).
21. Tamaki, M., Matsuoka, T., Nittono, H. & Hori, T. Activation of fast sleep spindles at the premotor cortex and parietal areas con-
tributes to motor learning: A study using sLORETA. Clin. Neurophysiol. 120, 878–886 (2009).
22. Astill, R. G. et al. Sleep spindle and slow wave frequency reect motor skill performance in primary school-age children. Front.
Hum. Neurosci. https ://doi.org/10.3389/fnhum .2014.00910 (2014).
23. Schmidt, C. et al. Encoding diculty promotes postlearning changes in sleep spindle activity during napping. J. Neurosci. 26,
8976–8982 (2006).
24. Ackermann, S., Hartmann, F., Papassotiropoulos, A., de Quervain, D. J. F. & Rasch, B. No associations between interindividual
dierences in sleep parameters and episodic memory consolidation. Sleep 38, 951–959 (2015).
25. Warby, S. C. et al. Sleep-spindle detection: Crowdsourcing and evaluating performance of experts, non-experts and automated
methods. Nat. Methods 11, 385–392 (2014).
26. Rosanova, M. & Ulrich, D. Pattern-specic associative long-term potentiation induced by a sleep spindle-related spike train. J.
Neurosci. 25, 9398–9405 (2005).
27. Latchoumane, C. F. V., Ngo, H. V. V., Born, J. & Shin, H. S. alamic spindles promote memory formation during sleep through
triple phase-locking of cortical, thalamic, and hippocampal rhythms. Neuron 95, 424-435.e6 (2017).
28. Nonclercq, A. et al. Sleep spindle detection through amplitude-frequency normal modelling. J. Neurosci. Methods 214, 192–203
(2013).
29. Ujma, P. P. Sleep spindles and general cognitive ability—A meta-analysis. Sleep Spindl. Cortical Up States. https ://doi.
org/10.1556/2053.2.2018.01 (2018).
30. Kis, A. et al. e interrelated eect of sleep and learning in dogs (Canis familiaris); an EEG and behavioural study. Sci. Rep. 7,
41873 (2017).
31. Kis, A. et al. Development of a non-invasive polysomnography technique for dogs (Canis familiar is). Physiol. Behav. 130, 149–156
(2014).
32. Gergely, A. et al. Reliability of family dogs’ sleep structure scoring based on manual and automated sleep stage identication.
Animals. https ://doi.org/10.3390/ani10 06092 7 (2020).
33. Eschenko, O., Molle, M., Born, J. & Sara, S. J. Elevated sleep spindle density aer learning or aer retrieval in rats. J. Neurosci. 26,
12914–12920 (2006).
34. Reynolds, C. M., Gradisar, M. & Short, M. A. Reliability of sleep spindle measurements in adolescents: How many nights are
necessary?. J. Sleep Res. https ://doi.org/10.1111/jsr.12698 (2019).
35. Schmidt, F. L. & Hunter, J. E. Methods of Meta-Analysis: Correcting Error and Bias in Research Findings (SAGE Publications, ou-
sand Oaks, 2004).
36. Korman, M. et al. Daytime sleep condenses the time course of motor memory consolidation. Nat. Neurosci. 10, 1206–1213 (2007).
37. Clemens, Z. et al. Fine-tuned coupling between human parahippocampal ripples and sleep spindles. Eur. J. Neurosci. 33, 511–520
(2011).
38. Lehmann, M., Schreiner, T., Seifritz, E. & Rasch, B. Emotional arousal modulates oscillatory correlates of targeted memory reac-
tivation during NREM, but not REM sleep. Sci. Rep. https ://doi.org/10.1038/srep3 9229 (2016).
39. Genzel, L., Kroes, M. C. W., Dresler, M. & Battaglia, F. P. Light sleep versus slow wave sleep in memory consolidation: A question
of global versus local processes?. Trends Neurosci. 37, 10–19 (2014).
40. Stickgold, R., James, L. & Hobson, J. Visual discrimination learning requires sleep aer training. Nat. Neurosci. 3, 1237 (2000).
41. Tse, D. et al. Schemas and memory consolidation. Science 316, 76–82 (2007).
42. Silverstein, L. D. & Levy, C. M. e stability of the sigma sleep spindle. Electroencephalogr. Clin. Neurophysiol. 40, 666–670 (1976).
43. Gaillard, J. M. & Blois, R. Spindle density in sleep of normal subjects. Sleep 4, 385–391 (1981).
44. Hori, A. Sleep characteristics in twins. Psychiatry Clin. Neurosci. 40, 35–46 (1986).
45. Gorgoni, M. et al. Parietal fast sleep spindle density decrease in Alzheimer’s disease and amnesic mild cognitive impairment. Neural
Plast. 2016, 10 (2016).
46. Ferrarelli, F. et al. Reduced sleep spindle activity in schizophrenia patients. Am. J. Psychiatry 164, 483–492 (2007).
47. Merikanto, I. et al. ADHD symptoms are associated with decreased activity of fast sleep spindles and poorer procedural overnight
learning during adolescence. Neurobiol. Learn. Mem. 157, 106–113 (2019).
48. Guazzelli, N. et al. Sleep spindles in normal elderly: Comparison with young adult patterns and relation to nocturnal awakening,
cognitive function and brain atrophy. Electroencephalogr. Clin. Neurophysiol. 63, 526–539 (1986).
49. Smirne, S. et al. Sleep in presenile dementia. Electroencephalogr. Clin. Neurophysiol. 43, 4 (1977).
50. Fogel, S. M., Nader, R., Cote, K. A. & Smith, C. T. Sleep spindles and learning potential. Behav. Neurosci. 121, 1–10 (2007).
Acknowledgements
We thank our dog trainers Rita Báji and Barbara Csibra for their contributions to the work behind data set 1,
as well as all owner-dog pairswho participated in the discussed studies; Borbála Turcsan, Tamás Faragó for
useful comments to this manuscript, and Péter Ujma for inspiring debates on the reliability of spindle-learning
correlations.
Content courtesy of Springer Nature, terms of use apply. Rights reserved


Vol:.(1234567890)
Scientic Reports | (2020) 10:22461 | 
www.nature.com/scientificreports/
Author contributions
I.I. wrote the detection algorithm, analyzed the data and wrote the initial dra; R.V., A.K., E.Ko. and T.K. collected
data and participated in the writing, A.K., M.G. and E.Ku. supervised the research, participated in planning the
experiments and in the writing process.
Funding
is work has received funding from the European Research Council (ERC) under the European Union’s Horizon
2020 research and innovation programme (Grant Agreement No. 680040), the National Research Development
and Innovation Oce (OTKA FK128242, K132372), the BIAL Foundation (Grant No 169/16), Hungarian Acad-
emy of Sciences (F01/031) and the János Bolyai Research Scholarship of the Hungarian Academy of Sciences.
Competing interests
e authors declare no competing interests.
Additional information
Supplementary Information e online version contains supplementary material available at https ://doi.
org/10.1038/s4159 8-020-80417 -8.
Correspondence and requests for materials should be addressed to I.B.I.
Reprints and permissions information is available at www.nature.com/reprints.
Publisher’s note Springer Nature remains neutral with regard to jurisdictional claims in published maps and
institutional aliations.
Open Access is article is licensed under a Creative Commons Attribution 4.0 International
License, which permits use, sharing, adaptation, distribution and reproduction in any medium or
format, as long as you give appropriate credit to the original author(s) and the source, provide a link to the
Creative Commons licence, and indicate if changes were made. e images or other third party material in this
article are included in the article’s Creative Commons licence, unless indicated otherwise in a credit line to the
material. If material is not included in the article’s Creative Commons licence and your intended use is not
permitted by statutory regulation or exceeds the permitted use, you will need to obtain permission directly from
the copyright holder. To view a copy of this licence, visit http://creat iveco mmons .org/licen ses/by/4.0/.
© e Author(s) 2020
Content courtesy of Springer Nature, terms of use apply. Rights reserved

1.
2.
3.
4.
5.
6.
Terms and Conditions
Springer Nature journal content, brought to you courtesy of Springer Nature Customer Service Center GmbH (“Springer Nature”).
Springer Nature supports a reasonable amount of sharing of research papers by authors, subscribers and authorised users (“Users”), for small-
scale personal, non-commercial use provided that all copyright, trade and service marks and other proprietary notices are maintained. By
accessing, sharing, receiving or otherwise using the Springer Nature journal content you agree to these terms of use (“Terms”). For these
purposes, Springer Nature considers academic use (by researchers and students) to be non-commercial.
These Terms are supplementary and will apply in addition to any applicable website terms and conditions, a relevant site licence or a personal
subscription. These Terms will prevail over any conflict or ambiguity with regards to the relevant terms, a site licence or a personal subscription
(to the extent of the conflict or ambiguity only). For Creative Commons-licensed articles, the terms of the Creative Commons license used will
apply.
We collect and use personal data to provide access to the Springer Nature journal content. We may also use these personal data internally within
ResearchGate and Springer Nature and as agreed share it, in an anonymised way, for purposes of tracking, analysis and reporting. We will not
otherwise disclose your personal data outside the ResearchGate or the Springer Nature group of companies unless we have your permission as
detailed in the Privacy Policy.
While Users may use the Springer Nature journal content for small scale, personal non-commercial use, it is important to note that Users may
not:
use such content for the purpose of providing other users with access on a regular or large scale basis or as a means to circumvent access
control;
use such content where to do so would be considered a criminal or statutory offence in any jurisdiction, or gives rise to civil liability, or is
otherwise unlawful;
falsely or misleadingly imply or suggest endorsement, approval , sponsorship, or association unless explicitly agreed to by Springer Nature in
writing;
use bots or other automated methods to access the content or redirect messages
override any security feature or exclusionary protocol; or
share the content in order to create substitute for Springer Nature products or services or a systematic database of Springer Nature journal
content.
In line with the restriction against commercial use, Springer Nature does not permit the creation of a product or service that creates revenue,
royalties, rent or income from our content or its inclusion as part of a paid for service or for other commercial gain. Springer Nature journal
content cannot be used for inter-library loans and librarians may not upload Springer Nature journal content on a large scale into their, or any
other, institutional repository.
These terms of use are reviewed regularly and may be amended at any time. Springer Nature is not obligated to publish any information or
content on this website and may remove it or features or functionality at our sole discretion, at any time with or without notice. Springer Nature
may revoke this licence to you at any time and remove access to any copies of the Springer Nature journal content which have been saved.
To the fullest extent permitted by law, Springer Nature makes no warranties, representations or guarantees to Users, either express or implied
with respect to the Springer nature journal content and all parties disclaim and waive any implied warranties or warranties imposed by law,
including merchantability or fitness for any particular purpose.
Please note that these rights do not automatically extend to content, data or other material published by Springer Nature that may be licensed
from third parties.
If you would like to use or distribute our Springer Nature journal content to a wider audience or on a regular basis or in any other manner not
expressly permitted by these Terms, please contact Springer Nature at
onlineservice@springernature.com