Sleep restriction affects vigilant attention:

Behavioural and neural correlates

Deborah Apthorp & Lourdes Machin

University of New England, NSW

Talk information

• This talk was written in Quarto (R Markdown/R Studio) & is available online

bit.ly/ANS-Sleep

Stages of sleep

• Traditional hypnogram
• Fitbit

• Staging sleep involves classifying the EEG signal, often still by hand
• Modern wearable technology enables much easier sleep tracking
• Accuracy not as good as polysomnography, but improving
• Enables inexpensive in-home sleep assessments
• Big data?
• Also relatively cheap

The gold standard of sleep measurement is, of course, polysomnography, where a patient stays overnight in a fully-equippped sleep lab and has continuous EEG recorded. This is then analysed with a variety of techniques and sleep is divided into stages based on patterns in the EEG - as you no doubt already know, there’s REM seep and three stages of non-REM sleep, N1 and N2, known as light sleep, and N3 which is deep, delta or slow-wave sleep.

Most of us won’t have the time or resources to spend a night or two in a sleep lab, but nearly everyone now has a smart watch and these can do a reasonable job of measuring sleep stages based on physical signals such as movement and heart rate.

You’ll hear more about this from both Hannah and Laura, including some really exciting stuff about big data from Hannah, and measuring sleep stages in mice from Laura, and in flies from Bruno.

Sleep restriction vs. sleep deprivation

Sleep loss can cause daytime sleepiness (image credit: Jude Conning Photography)

• Much research looks at total sleep deprivation (no sleep overnight)
• Profound effects on cognition (memory, attention, etc)
• But sleep restriction (less sleep than usual) is much more common
• Effects on cognition not as well researched
• But this type of sleep loss is much more common!

Sleep loss can affect our daytime functioning outside of just sleepiness - it can cause lapses in concentration and microsleeps which are potentially fatal, for instance while driving. There’s also a continued reduction in the average hours people are sleeping. In spite of this, the effect of mild sleep loss are much less studied than total sleep deprivation.

Background

• These data were originally collected in 2015 as part of an Honours project by Lucienne Shenfield at ANU
• Now working as a clinical psychologist specialising in sleep
• Some of this work is now published
• What we’ll mainly talk about is the work of another Honours student Lourdes Machin (UNE, 2022)
• Lourdes analysed the during-task EEG data for the Psychomotor Vigilance Task

Background

• CNS centres regulating sleep overlap with those regulating attention & arousal
• Disrupting sleep disrupts attention
• But what is attention?
• Catch-all term for multiple processes
• Here we focus on sustained attention/vigilance

“Attention is psychology’s weapon of mass explanation” - David Burr, personal communication

Participants

• Inclusion criteria: 18 – 65 years; regular sleep 7-8 h/night

• Exclusion criteria:

  • Regular smokers
  • Shift work
  • High caffeine consumption (> 5 cups/day)
  • Diagnosed ADD or sleep problems
  • Recent head injury or history of seizures

• Final sample: 25 participants (15 females)

  • Age: 21 – 55 (M – 24.79 years, SD = 6.82)

Experimental design

• Within-subjects: Normal sleep (NS) and Sleep-restricted (SR)
• Sleep restriction: Delay bedtime by 3 hrs, set alarm for 5 hrs later
• Counterbalanced order
• Sleep monitoring: Sleep diary and FitBits
• Tasks: Attentional Blink (AB) and Psychomotor Vigilance Task (PVT)
• EEG: Resting state and ERP (during tasks)

Sleep restriction

Minutes of Sleep: Diary and FitBit

So this is just the data showing that people did, in fact, restrict their sleep, and that the sleep diaries largely agreed with the FitBit data. Because this was 2015, FitBits were still relatively new, so we don’t have data on people’s sleep stages in this study, although that would have been interesting. They did see it as an attraction to do the study, though! We also used the FitBit data to ensure that they were sleeping the regular amount on the non-sleep-restricted nights as per the exclusion criteria.

Task: Psychomotor Vigilance Task

• 10 minutes long
• Participants press a button as soon as they see red numbers appear
• Numbers appear at random intervals (2-10s apart)
• Reaction time (ms) is measured
• Also lapses (>500ms), false starts (<100ms)
• ~80 trials per session

EEG

• Compumedics NuAmps 32-channel EEG system
• Electrodes 10/20 system
• Electrode maximum impedance: 5 \(k\Omega\)
• Signal recorded at 1000 Hz
• NuAmps digital amplifier
• Curry 7.0.9 software
• Resting state and during-task recording

EEG preprocessing

flowchart TD
  A[1. Import data, event markers, channel locations into EEGLAB] --> B[2. Down-sample EEG data to 256 Hz]
  B --> C[3. Apply Finite Impulse Response filter: 1-40 Hz ]
  C --> D[4. Remove non-relevant channels - EOG, Mastoids]
  D --> E[5. Manually remove channels with uncommonly high or low power]
  E --> F[6. Re-reference data to common average]
 F --> G[7. Interpolate removed electrodes, using spherical interpolation]
  

flowchart TD
  A(8. Visual inspection of  channel plots. Removal of large artifacts ) --> B(9. Run independent components analysis )
  B --> C(10. Run ICLabel - Remove muscle and eye IC artifacts )
  C --> D(11. Remove remaining blink-like ICs with probability between 0.8 and 0.9 )
  D --> E(12. Extract epochs of 3s length for ERP and time frequency analysis )
  E --> F[13. Locked to stimulus onset: -1s, +2s ]
  E --> G[14. Locked to stimulus onset: -1.5s, +1.5s ]

Behavioural data

Mean reaction times by sleep condition

Subjective sleepiness

Mean Karolinska Sleepiness Scale scores by sleep condition

ERP data

ERP amplitude differences: ERP Topographic Maps 250-500 ms

ERP data - differences - Pz

Grey bar shows statistically significant differences in amplitude at the start of P3

ERP data - differences - CPz

Grey bar shows statistically significant differences in amplitude at the start of P3

Source localisation - NS - SR

eLORETA Source Localisation of Current Source Density Differences (from ERP) between NS and SR

Source localisation - NS - SR

328 ms

332 ms

336 ms

340 ms

• Best Match - Brodmann area 5, postcentral gyrus, parietal lobe

• Somatosensory cortex - motor preparation affected?

The ERPs were converted to source densities and entered into eLORETA, to localize the sources of the differences in ERP amplitude. The source was localised to an area best matched by Brodmann area 5, in the Postcentral Gyrus, parietal lobe, or the somatosensory association cortex. The time course of this difference, at the bottom shows the slightly right-oriented localisation of these differences.

Research reveals that BA5 influences the primary motor cortex’s excitability, affecting hand motor function. However, BA5 has not been identified among those areas involved in sustained attention fMRI studies, such as presupplementary motor area (pre-SMA) and the dorsal premotor cortex. The difference in activity in this area could be related to differences in latency rather than amplitude. As the P3 peak is partly response-locked, the differences in activity in the somatosensory association cortex may reflect differences in the speed of movement preparation or execution. Further analysis of the data could elaborate this hypothesis by studying the lateralized readiness potential, and possible differences in amplitude and starting points.

Resting state data

• We also measured resting state EEG (eyes open and closed)
• Divided into regions of interest (ROIs)
• L & R frontal, central, occipital
• Significant increases in alpha & decreases in delta relative frequency after sleep restriction
• Most prominent in right central ROI (C4, T4, CP4, TP8)

Resting state changes - right central ROI, eyes closed

Relative alpha frequencies

Relative delta frequencies

Individual variation in effects of sleep restriction

Relative alpha change

Relative delta change

Subjective sleepiness change

Conclusions

• Mild sleep restriction caused changes in resting state frequencies in right central areas
• Small increase in reaction times in a vigilance task (PVT)
• Increase in ERPs (P3) at central sites during the PVT
• Source localisation using eLORETA suggests the post-central gyrus (Brodmann Area 5) as the origin of this difference
• This area has not previously been associated with sustained attention - motor preparation/execution differences?
• Individual variations in effects of sleep restriction - behavioural and neural effects correlate

Questions?