Can anyone who has successfully made a design for training at the infralow range provide assistance?

What is it that you have done in the design in regards to filtering, settings, etc?
I currently have silver chloride electrodes, a number of pocketneurobics amps (Qwiz and Uwiz), Bioera, and a Cyton amp.

Comments

  • I have used Cyton with Bioera to create multiple Infra-low Neurofeedback designs following the Othmers method and with very successful results, I am sure it works because I was getting the same results they are reporting in the symptom profiles in the protocol guide, I have used it with multiple people (friends and family) and the results are very satisfying

    Regrading the filters and settings I am using a standard low pass filter and setting the Max frequency to 1.5 x Training Frequency, to make sure that the training frequency is not attenuated, but you can skip that if you like, I did it because of a something Siegfried Othmer wrote or said in a discussion somewhere,

    The frequencies I am using are the same ones in the latest Cygnet from 0.0001 mhz to 10 mhz, working with frequencies that low requires a specific way to process the signal to get good feedback, basically the signal is very very slow so averaging and normal signal processing won't work, the way they do it is by following the actual signal as it speeds up and slows down going from 0 and increasing in amplitude throughout the session

    This part is very important to understand, the infra-low signal in this frequency range does not complete a full cycle or even half a cycle in even a 1 hour session, it takes days for this very slow signal to complete a cycle so we are only interested in what the signal does as the amplitude changes speeding up and slowing down and mirroring that in the feedback

    Bioera provides multiple elements that can help with processing this signal, the most important one is Range Mapper as it allows mapping the infra-low signal values to values you can use in providing feedback

    For example, let's say you are providing feedback using a video that expands and shrinks following the signal, you may want to make the input range say 0 -> 0.0000001 and output range 0 -> 400, and pass that to an Html Player element that reads the values and modifies the video element size accordingly

    Say it takes a minute for the signal to go from 0 to your max input range, what you do after the signal reaches the max input range is update the range to be say 0.0000001 -> 0.0000002 and use the same output range, that's actually what they are doing currently in their latest version of the software, it's like setting a flash light on a mountain climber as he climbs, you are setting the light on a point above them say 10 meters up and when they reach it you move the light another 10 meters up and so on, our interest is to follow the signal as it progresses from 0.0000001 -> 0.0000002 and mapping that in our shrinking video, then once it reaches 0.0000002 the input range updates to 0.0000002 -> 0.0000004, and the video window resets and starts shrinking again following the signal, with the same analogy as our climber and the flash light

  • This is the technical part, but this part is not the important part, the really important part is knowing what locations to train for which symptoms, and how to choose the optimal frequency, how to know if you should go lower or higher, what's the starting frequency based on the symptom profile, do you train right, left or both, these questions are very important and the success of the training largely depends on understanding the training model the Othmers use, and this requires taking the courses and watching the annual summit and reading the protocol guide multiple times and having a lot of practice

    A word of caution, Infra-low neurofeedback is a very powerful tool and it can achieve miraculous results if used correctly, but it can also mess up your brain pretty bad leaving you with horrible side effects if you did it wrong, so before diving in to train yourself and loved ones please take a lot of time to read and watch the Othmers materials many times and go slowly and carefully till you are sure you know what you're doing

    Let me know if you have any other questions and I will try to answer them

  • What filter type and order do you use? What voltage values do you get and what is your optimal reward frequency? Also, do you use inhibits?

  • I am using a low pass filter with default settings and the voltage values when the design starts begin at 0 and gradually goes up depending on the frequency you use, it can go to 3 - 4 millivolts in 1 - 2 minute and then reverses direction to 0 if you are using a high frequency like 10 milli hertz, but it you're using a lower frequency like 0.001 mhz it will progress very slowly from 0 reaching 0.000001 in 2 minutes then 0.000002 in 5 minutes and so on, by the end of an hour session it may only be up to 10 microvolt, that's what I mean by it could take days for a signal in that frequency range to make a full cycle (please note the actual values will differ but should follow the same ideas)

    The optimal reward frequency is different for each person and their symptoms and can only be determined after training for a few sessions while monitoring the after session effects

    Regarding Inhibits, I have implemented them also based on the Othmers explanation

  • I’m asking how you’re implementing the range. Let’s say your optimal reward frequency is 1.3 mHz, what would range be? Would it be 0 to 2.6 mHz with 1.3 mHz being the center frequency?

  • No I am using the formula 1.5 x Training frequency, I think the center frequency only applies in band pass filters we are using a low pass filter so the cutoff frequency is 1.5 x 1.3 mhz

  • What type of filter and what order are you using?

  • Like butterworth, chebyshev, vessel, etc...

    Also the filter order

  • I am using the defaults in Bioera Filter element

  • How do you turn on DC coupling in Bioera for the Cyton? Any help would be appreciated.

  • edited October 2019

    The signal is DC coupled by default, that's why the signal amplitude reaches thousands of microvolts, thats the DC offset, if you want to filter it you need a high pass filter at 0.3 or 0.5 hz, then you will get a signal in the tens of microvolts range

  • Interesting stuff for anyone wishing to develop software for neurofeedback training.

  • Thanks for all the help. I made a rudimentary ILF design where I am just looking at the signal using auto-range. I don't mind not having any kind of fancy feedback mechanism like watching a video.I have another question; how do you make the inhibits for ILF? My understanding is that their inhibits are actually different from the inhibits they were doing with the old method. Could you describe how you make them or tell me what elements you use? If you could post a screenshot that would be great too, but I'll appreciate any help. Thanks again in advance!

  • edited October 2019

    Regarding the inhibits, the way they work is by having multiple inhibit bands each band covering a small range of frequencies, (1 - 3.5), (3.5 - 6), (6 - 8), (8 - 10), (10 - 12), (12 - 15), (15 - 18), (18 - 22), (22 - 28), (28 - 35), (35 - 45), You do that by a band pass filter for each range.

    From my understanding what the Othmers do is inhibit any sudden increase in any of the bands above the average ranges for that client, so let's say over the past 30 seconds 90% of the time the values for the first band is below "x", then if the value suddenly becomes above "x" then that triggers an inhibit

    These inhibits are then combined using logical operation (OR) so if the signal at any band is above the inhibit ratio (success rate) an inhibit is triggered.

    The reason it's called success rate is that it indicates how often you succeed by not triggering an inhibit, if the ratio is 50% then half the time you will get an inhibit, if it's 85% then you will get an inhibit 15% of the time ...etc

    This is accomplished using a "Time ratio" element, you pass the RMS of the filter output over 0.1 second, and set the time ratio (10 means 90% success) and feed the output as the threshold to a threshold element and set the input value to the threshold element to the 0.1s RMS, so whenever the 0.1s RMS of the band filter output is above the threshold calculated by the "Time ratio" element the "Threshold" element gives an ON signal, that ON means an inhibit is triggered, you then use a "Logical Mixer M" element to combine the output of all the Thresholds for all the bands and create a single Inhibit ON/OFF

  • tuialinharestuialinhares Belo Horizonte, Brazil

    That's a very interesting thread.
    I've been trying to use infralow protocols in BioEra and got to a solution that looks like yours.
    My concern on using the lowpass filter for getting the infralow signal is that the delay of the filter response is very high. For example, 150 seconds for a 1 mHz filter. Wouldn't such a big lag be so much for the brain to understand that it's a feedback of the brainwave activity?

  • What happens in practice is quite the opposite, whenever they went lower in frequency the effects became stronger to the point where they can calm a person down in minutes and also the reverse if the frequency is not well optimized

    What the ILF method is doing is to take the direct output of the filter and display that to the brain so the brain can monitor the changes in the signal (filter output) as it speeds up and slows down and somehow this process leads to amazing effects

    Currently the Cygnet software can go down to 0.0001 mhz, and many people are optimizing at the bottom frequency or very close to it, so the filter lag issue is not affecting the results at all, maybe because no signal processing is done they just display the real-time signal as it changes right now

  • In response to a question I was asked via private messaging on signal smoothing, I think it may help anyone trying to build their own ILF NF

    "In the othmers method they mention that they do Neurofeedback with the actual signal without any modification so no smoothing, the real signal is fed back to the user,

    This is only possible because it's slow enough to view, so each cycle takes minutes so the user can view the actual signal without processing and this gives a better effect in training

    The only processing that is done is related to how the signal is shown, so what I do is show a video that changes it's size based on the signal, but the video can only grow or shrink 300 - 400 pixels, while the signal can go up to thousands, so the video shrinks with the signal, 1 pixel every 1 microvolt, When the signal reaches 300 I reset the video size to original and start over, when the signal reaches 600 I do the same ...etc

    That way I can track the actual signal with only a minor flicker in the video size every 300 micro volts, then it continues tracking the actual signal

    In slower frequencies I introduce a scale factor, as the signal may take several seconds to go up 1 microvolt so that way the video will be resizing too slowly, so the scaling factor makes every 0.1 microvolts 1 pixel, to only make the resize noticeable while not modifying how the signal slows down and speeds up"

  • Thanks @compujohnny, that's really great to know! Do you have a specific source/reference for where you got the details, in particular the filtration methods the Othmers are describing where you watch the frequency go up and down over a longer period? Are you using this or, are you using a standard IIR filter?

    Thanks

Sign In or Register to comment.