Voluntary Self-Regulation

Question 1

What is real-time fMRI neurofeedback?

Answer 1

• an emerging technique allowing voluntary control over specific brain regions/activity
  
  • can be used to study causal brain-fuction relationships via assessing how this learned self-regulation of brain activity affected perception or behaviour
  • e.g. evidence that self-regulation leads to behaviour that also normally is specific to that part of the cortex
  • studies also found therapeutic effects for chronic ppain patients, Parkinson’s disease and tinnitus

Question 2

Can neurofeedback be useful for cognitive enhancement or clinical purposes?

Answer 2

• it needs to show that the effects persist in situations in which neurofeedback (fMRI scanner) is not available anymore
• and that effects can last much longer than during training period
• there is evidence that:
  
  • once learned, self regulation can be perfromed in transfer runs without feedback information (but only immediately after training period)
  • generally evidence is aprase
  • evidence that neurofeedback leads to plastic brain changes 1 day after training (another study found that effects lasted 2 months)
  • OCD patients doing neurofeedback of orbitofrontal cortex showed persistent changes in resting state connectivity and reduction in anxiety few days after the training
• EEG neurofeedback (exist much longer than fMRI): usually no follow-up material collected
  
  • some studies foound effects of EEG neurofeedback to last 12 months, suggesting that effects remain stable

Question 3

What did this study (Robineau et al., 2011) attempt to do?

Answer 3

• becausee EEG findings might not be trasnferrable to fMRI:
  
  • rescanned 5 participants 6 and 14 months after succesful learning of self-regulation of differential visual cortex aactivity (learned to control interhemispheric balance)
    
    • at 6 months follow-up: same neurofeedback information as during training - top up session)
    • 14-month follow-up: no feedback information (transfer)
• Hypotheses:
  
  1. that self-regulation is an acquired skill maintainable over longer periods of time
  2. participants would still be able to self-regulate visual cortex activity in follow-up sessions with similar performance as in initial training

Question 4

Materials and methods

Answer 4

• participants were trained to control differential feeback between target region of interest in early visual cortex (ROI_target) and its homolog in opposite hemisphere
  
  • ROI’s were delineated in seperate functional localiser scans
  • participants had to maintainfixation on a central point while a flickering checkerboard wedge (100%contrast, 8HZ contrast reversal, etc.)
  • checkerboard was presented for three blocks 30s alternating in left and right visual field, interleaved with baseline blocks
  • activity difference was fed back to participants in form of a visual thermometer
  • neurofeedback runs were interleaved with transfer runs in which participants trained without feedback

Question 5

MRI Data Acquisition

Answer 5

• 3T MRI scanner
• Functional images obtained with single-shot gradent-echo with a 12-channel phased array coil
  
  • first three EPI volumes were automatically discarded to avoid T1 saturation effects
  • T1 wighted structural image was acquired at beginning of each session
• visual stimuli were presented on a projection screen at rear of scanner bore

Question 6

Participants

Answer 6

• 5 females between 25 and 40 years
• normal or corrected to normal vision
• approved by ethics comittee + informed consent
• before follow up sessions, participants received reminders on what they had learned and how the task worked
  
  • sessoion also included reminders about neurofeedback thermometer and cognitive strategy (i.e. covert shifts in attention and imagery of moving stimuli)
• instructed to fixate on central fixatioon point, breath steadily and remain as still as possible

Question 7

Follow-up scanning protocol

• first and second session
• training and transfer runs
• thermometer signal

Answer 7

• sessions started with a T1-weighted structural image of whole brain
  
  • used for coregistration of current head position (using Turbo-Brain-Voyager)
  • coregistration was used to position the ROIs with those in the initial neurofeedback session (localise)
  • same ROIs that were trained were also targeted in follow-up sessions
• first follow up: 3-7 months after initial training
  
  • participants did three neurofeedback training runs in which they received neurofeedback (activity difference between visual ROIs, left and right hemisphere) in form of a visual thermometer
• second follow-up”: 9-14months after initial training
  
  • participants perfromed three transfer runs (no neurofeedback was provided) everything else was identical
• training and transfer runs were composed of four 20s baseline blocks and 30s up-regulation blocks
  
  • baseline blocks: fixation cross is black, instruction to count back mentally from 100 with steps of -3 to maintain basline stability
  • up-regulation blocks: fixation cross is white. instruction to regulate brainactivity and increase feedback signal
• thermometer feedback signal:
  
  • consisted of thin horizontal line, moved up or down depending on differential feedback line, updated every 2s
  • computed as difference between percentages of signal changes of the two ROIs
  • to avoid brisk fluctuations, temporal filtering of previous three time points were made through means

Question 8

Statistical Analyses to test the hypotheses:

Answer 8

1. H1 (that the learning effect found in initial study were still present in the follow-up group)
2. H2 (their performance was similar as in the initial training):
   
   • to test these hypotheses an ANOVA was conducted using R²
     
     • based on performance difference between sessions (difference contrast)
     • factor: contrasts between sessions (three levels)
     • and subject (to account for intersubject variability)
   • because the H2 was that performance would be maintained in follow up sessions performance differences were considered
     
     • normally absolute values of regulation performance would be used in neurofeedback studies
   • data of all learners would be included (also the three that only perfromed initial training without follow-ups)
     
     • often longitudinal studies exclude those participants that have missing values due to doing t-tests
       
       • that would not optimally use the data (assuming homogeneity)
     • in this way they got a more stable estimate of variance and increase in degrees of freedom in reference student t-distribution
   • one one-sample t-test was performed for the subgroup of participants who perfromed the follow up test, to show that beta-values would be sig. positive in last follow-up
   • analyses were perfromed seperately for training and transfer runs

Question 9

Results:

Answer 9

• subgroup of those in follow up sessions succesfully learned to control the differential feedback signal
  
  • beta estimates increased sig. between session 1, 2 and 3
  • t-test (only subgroup) also showed significantly positive values in last session
  • these findings suggest a reliable modulation of brain activity during regulation compare to baseline blocks
  • performance did not differ during follow-up from the last session of initial training

Question 10

What significance and additional value does this studies finding by Robineau et al. provide?

Answer 10

• the findings suggest that voluntary control over brain activity can be maintained for a much longer time than previously though or known (14months) and that it is transferrable to situations without neurofeedback
  
  • this has special implications for its utility in clinical use as it has been found to positively effect chronic pain, Parkinson’s disease, tinnitus, depression, OCD, spider phobia and addiction
    
    • especially that transfer effects can last so long without neurofeedback which is seldomly available when needed
      
      • this studiy included transfer runs in which neurofeedback was not given, this may have increased likelihood of long lasting transfer effects
  • previous studies showed a lastingness of between 1 day and and 2 months after training
    
    • participants were trained over several days, which may have contributed to these long-lasting effects as they had slept in between and sleep is very important for encoding
    • confirms EEG studies that found a lastingness of 12 months
  • as these self-regulation skills can be applied voluntarily by participants that is in contrast to concommitant training effects that effected resing state changes

Question 11

What were limitations of this study by Robineau et al.?

Answer 11

• small sample sizes (four in first follow-up and five in second follow up), results should be sonsidered a preliminary demonstration
• no control group (without neurofeedback attemting to self-regulate)
  
  • hence cannot exclude possibility that just training (without neurofeedback) led to results
    
    • nonetheless, in initial study non-learners (I assume they did not get neurofeedback) practiced self regulation but did not show improvements
    • other studies in which participamts received sham-feedback or no feedback, no learning occurred (at least not for: anterior cingulate cortex (ACC), inferior frontal gyrus, and visual cortex)
  • also, visual detection task from initial training were not repeated in follow-up task

Question 12

What would be the conclusion for the present study?

Answer 12

• in a sample of healthy participants learned self-regulation of diferential visual cortex activity can be maintained over long periods of time and transferred to situations without neurofeedback
  
  • probably also apply for other neurofeedback target regions as the visual cortex has been shown to be less responsive to neurofeedback learning (also shown by large numbers that failed to learn neurofeedback self-regulation)
  • compared to normal ROI based feedback the one of this study was probably harder to control (as uncorrelated Gaussian noise is additive; reduced signal to noise ratio (SNR)
  • implies that neurofeedbakc training leads to plastic changes and not just short-lived changes
  • underlying mechanisms still unknown, but findings are promising for novel experimental therapy