There’s a great journal called Behavioral and Brain Sciences (BBS), published by Cambridge University Press. It allows people to present their theories and evidence, and get discussion on it from other researchers, from many different disciplines and perspectives. There are also many journals relating to Behavioural Science, Brain Science, Cognitive Science, Cognitive Linguistics, Cognitive Neuroscience and Cognitive Psychology that are surprisingly readable even to the lay person. But the advantage of BBS is the diversity of opinions and the richness of the discussion which helps clarify things more than a single author could possibly do.
Of course, popular magazines like Scientific American and New Scientist are targeted more to budding scientists, and you don’t have to have a PhD to subscribe to those — and they are written by science journalists who are good explaining things to non-scientists, and might just happen to have the goal of encouraging people to become scientists.
I guess with a father who is a neuroscience professor and a mother who is a clinical psychologist, both whom have shelves full of relevant journals, it is not surprising that I have developed an interest in this area. And I’ve never looked back since the day my father first pointed me in the direction of BBS.
Then my mother likes Science Fiction, and has a good library of Science Fiction books, and she particularly likes the kind of Science Fiction that is based on real science, not just shooting round the galaxy breaking the laws of physics. But she particularly likes the kind of Science Fiction that makes you think about thinking, about how the brain works, or could work, or could work better, or how telepathy could work.
Yes, ironic isn’t it!
Since my aim here is to help school students like me understand various things about Brain and Quantum Science, I’m going to try to keep a balance between citing the popular press (see Further Reading) and the research literature (see References). Because it is always important to take things back to the original source, as otherwise the Chinese whisper effect can distort things (R. A. Sanderson, personal communication; see also entries on Chinese Whispers in Wikipedia and Britannica, as well as the Merriam-Webster and Collins dictionaries).
Definitions
•
Cognitive Science: The collective science that help us understand how the human brain and mind work, with a much broader emphasis on interdisciplinary science including the Behavioural and Biological Sciences, Linguistics, Cognitive Psychology and Cognitive Neuroscience, Psycholinguistics and Philosophy of Science (
https://plato.stanford.edu/entries/cognitive-science/).
•
Brain Computer Interface (BCI) or Brain Machine Interface (BMI): A subfield of Human Computer Interface focussed on using Cognitive Neuroscience technologies to allow control of a computer or other device in a hands-free way. These may be invasive (involving electrodes inside the head) or non-invasive (involving electrodes on the head). Controversially, the term BCI is often used for what would more accurately be called a Brain Muscle Interface because the electrodes are placed in places where it is impossible to pickup brain signal (EEG) and the BCI device is really operated by inducing the much stronger muscle signals (EMG) — e.g. through jaw clenching, eye or eyebrow movement, etc. Eye-tracking and eye-gaze interfaces can also be hard to distinguish from true BCI interfaces. (
https://www.sciencedirect.com/topics/neuroscience/brain-computer-interface).
The term EEG can thus be misleading, and it is helpful to clarify some related terms. Some of these, like the Stentrode™ technology developed at Melbourne University, and being commercialized by Synchron, represent radical advances that blur the traditional invasive vs non-invasive distinction, and their website is helpful in tracing the history of the BCI technologies (
https://synchron.com/history):
• Electroencephalogram/Electroencephalography (EEG): EEG has its origins in experiments by Richard Caton in 1875. The letterism ‘EEG’ refers both to the signal and the method of obtaining a signal non-invasively by placing electrodes on the head. Many different kinds of electrodes can be used, and low density EEG may involve a dozen electrodes or less, while high density EEG may involve a couple of hundred electrodes.
• Electromyogram/Electromyography (EMG): EMG measures electrical signals originating in muscles, rather than the brain, e.g. from jaw muscles.
• Electrocardiogram (ECG/EKG): ECG measures electrical signals originating in the muscles of the heart, and is thus a specific form of EMG.
• Electrooculogram (EOG): EOG refers to unwanted ‘artefact’ when EEG equipment measures electrical signals (EMG) originating in the muscles that control the eyes, with very strong signals being generated from eyeblinks or eye movements (saccade).
• Electrocorticogram/Electrocorticography (ECog): ECog has its origins just over a century after the first EEG experiments, with Philip Kennedy implanting the first ECog electrodes in 1988 (in Monkeys). A more general term for ECog is intracranial electroencephalography (iEEG). ECog generally involves placing dense grids of electrodes directly on the surface of the cortex, and thus involves cutting into the scalp and skull to place them. This invasive techniques is thus not appropriate for use on volunteer or student subjects and is limited to subjects with neurological disorders or brain injury which may be addressed with such invasive surgical intervention.
• Stereotactic Electroencephalography (sEEG): sEEG is actually a form of iEEG that involves drilling holes into the head to place individual needle or wire-like electrodes in specific places, and is being pioneered by the University of Pittsburgh and the associated Children’s Hospital Medical Centre (with patent application WO2021174061A1 filed in 2021).
• Stentrode™: A revolutionary approach pioneered by Melbourne University and Synchron (with patent application WO2017070252A1 filed in 2017) is to insert an expanding stent through a blood vessel. This device avoids both drilling and wires using similar wireless power technology to what is used to charge a phone, allow wireless entry or contactless credit card transactions.
Then there are all sorts of other kinds of brain imaging technologies, some also used scan the whole body, or parts other than the brain. All of the things we’ve discussed so far have been variations on the use of electrodes: sometimes inside the head; sometimes on the surface of the head (or the hair); sometimes on the skin (of the face or elsewhere); often above some kind of muscle…
These other imaging techniques are more like X-rays – but in some cases, safer.
• Computer Assisted Tomography (CT/CAT) — basically 3D X-rays where the X-ray source is rotated or spiralled around the patient.
• Electron Beam Tomography (EBT) — basically the same as CAT but a beam of negatively-charged electrodes (like in an old Cathode Ray Tube television) can be manipulated electronically, avoiding the physical rotation and allowing for much faster scanning.
• Positron Emission Tomography (PET) — basically the opposite of EBT in two senses, positrons are the positively charged siblings of electrons, and rather than being sent into the body by the scanner, they are emitted by radioisotopes (radiotracers) injected into the patient.
• functional Magnetic Resonance Imaging (fMRI) — this is a scan, but is more like EEG in the sense that it is non-invasive and doesn’t require dosing the patient with radiation, or injecting radioactive substances. Blood haemogobin contains iron, and different levels of oxygenation causes it to be more magnetic (deoxygenated → paramagnetic) or less magnetic (oxygenated → diamagnetic). Since active neurons need oxygenated blood, blood races to feed them when they are active, so that for a few seconds afterwards there is more oxygenated blood.
• Magnetoencephalography (MEG) — involves using SQUIDs (not the fishy kind but superconducting quantum interference devices). Again it is relatively non-invasive, and because the magnetic fields are orthogonal to electric fields it provides different information from EEG as well as higher spatial resolution than EEG and much higher temporal resolution than fMRI. But MEG equipment is relatively rare (there are only around a hundred around the world as they are very expensive).
• functional Near-Infrared Spectroscopy (fNIRS) — like fMRI this is focussed on the changes in the flow and oxygenation of haemoglobin, but using near infrared light rather than magnetic effect — the skin is almost transparent to IR light in the 700-900nm range). It has the advantage of being portable, so is more of a competitor for EEG in that sense, and can be placed using the same 10-20 electrode arrangement as EEG (which is based around angular intervals of 10° and 20° for low density EEG, and interspersed electrode/sensor locations for higher densities). It is portable with similar temporal resolution to EEG but higher latency, and thus uncertainty (blood flows much more slowly than electrical currents), but may suffer less from artefacts due to vibration/movement whereas EEG electrodes can lose contact with the skin.
Hardware
In addition to the medical and research grade EEG/BCI equipment which costs hundreds of thousands of dollars, there is also consumer grade or hobbyist BCI equipment costing as little as a couple of hundred dollars. This is largely oriented to the gaming market, and in fact some of these BCI companies have actively discouraged use for serious purposes by use of proprietary software libraries, nasty countermeasures, and exorbitant ongoing license fees, for access to the captured signal. Examples are noted as we discuss significant market entrants.
Furthermore, there are other kinds of HCI that are worth including in the discussion, including in particular eye-tracking or gaze-tracker hardware. Eye-gaze tracking has been used since the 1800s to understand how we read (Javal 1907; Huey, 1899/1908).
Eye-tracking hardware is now also available as much cheaper consumer grade products, but again manufacturers restrict usage for non-authorized applications with limited licenses for the game devices, and expensive licenses for the required software libraries needed for even amateur research and development purposes. Again, such restrictions on usage of the consumer level device are noted in the discussion of the specific product concerned.
Gaming EEG/BCI hardware may also include motion sensors to provide an additional form of input for games.
Electrodes
A superficial difference between EEG electrode setups is whether they use a cap (feeding through holes), or a band or headset. A more significant difference distinguishing different kinds of EEG hardware is whether the electrodes are wet or dry. In particular, traditional medical and research electrodes have used a messy conductive gel or expensive (silver-containing) paste, while many modern systems, professional or hobbyist, tend to be dry (no gel or paste or saline needed).
Electrodes can also be flat, ring-like (with the gel squirted into the hole), or comb-like (with multiple feet to get through your hair). Traditionally, electrodes had to make actual contact with the scalp, or at least low-impedance contact with the scalp through a conductive gel or paste or liquid. Some experimental electrodes make use of capacitive effects and don’t make actual contact with the scalp (here electrons on one side of a gap repel those on the other side, which means alternating current can effectively get through, and the higher the frequency the lower the impedance, because electrons are pushed less distance before they pushed in the opposite direction).
Products
We will now move to discuss some of the hobbyist/gamming brands and some important models of EEG/BCI headsets.
First though, it is important to be careful. Many products claiming to collect EEG don’t seem to actually be able to pick up EEG, but instead rely on EMG — that is they cheat and work off muscle signal rather than actual brain signal. While it is easier to make contact direct on skin rather than through hair, electrodes that are located on the skin at the sides of the head, and in particular on the forehead, cannot pick up clean EEG as the signal is swamped by the much stronger muscle activity.
If there are multiple electrodes on the top of the head as well, then they will pick up satisfactory EEG, and it is then also possible to use signal processing techniques to clean up those peripheral electrodes because of the cleaner and less muscle-dominated mixed-source signal we have at the central electrodes.
We do not discuss the NeuroSky or Mindflex systems, as they only have peripheral electrodes, and it is dubious as to how much brain signal they are really able to extract.
The consumer product that did make a breakthrough in getting real EEG cheaply was the Emotiv (
https://www.emotiv.com/) EPOC (a spinout of technology from Melbourne University). This uses conductive saline solution (salt water) to moisten flat electrode pads for very good EEG signal until the pads dry out. Its fourteen electrodes are well distributed around the head. Unfortunately, this is one of the companies/products where access to raw data and their software libraries involves very expensive subscription-style licenses, although there are some third parties that provide alternatives.
Emotiv also brought out a cheaper 5-electrode model that uses a really clever hydrophilic polymer electrode that pulls sweat from the user and doesn’t dry out, maintaining good connections continuously. Unfortunately only one electrode is in a central position, and the others are in positions that are strongly contaminated by muscle. To get a voltage with real EEG you need to be able to measure it across two electrodes, one of which is typically chosen as a reference (or sometimes the average of several is used as a reference). But this reference needs to be in a clean signal position, and at least some of the other electrodes need to be in clean signal positions to allow signal processing and noise cancelling algorithms to work.
The other very important recent development is OpenBCI (
https://openbci.com/). OpenBCI’s open-source hardware and software is based around the Texas Instruments ADS1299 chip specifically designed for EEG (although the ADS1298 family is also usable for EEG and EMG and is arguably more flexible).
Their classic design uses a rather ugly 3D-printed headset with screw-in electrodes, but these days they also offer caps and headbands, and third party electrodes. The system is used by researchers as much as hobbyists, and saves developers of new EEG sensors the hassle of designing their own circuits and printed circuit boards around the TI ADS129* family of chips.
Their headline product is the 8-channel Cyton board, with or without an 8-channel piggybank board, which together provide sixteen channels that are wirelessly, sampled at 125Hz. This is a rather slow sampling rate due to its use of BLE (Bluetooth Low Energy) for the wireless, and means it is only suitable for use below around 50-60Hz (mains frequency will interfere within that range).
Typical usage of EEG hardware for BCI is implementing grid-like keyboards based on Steady State Visual Evoked Potentials (typically operated in a range from 12 to 24 Hz, or more broadly from 4 to 48Hz, but then is more risk of confusion with alpha waves (below 12 Hz) or contamination by EMG (above 24Hz) or mains frequencies (between 48 and 62Hz).
The rows and columns of the grid are flashed in sequence, and when the row and column VEP or SSVEP is detected we have the coordinates of square you are looking at. However, this can also be done with an eye-tracker.
Another SSVEP eye-tracking approach is to have strong LEDs (light emitting diodes) flashing at different frequencies at the corners, and the mix of those (relatively prime) frequencies allows determination of the fixation point. (Cottrell et al., 2012).
Microsoft introduced support for eye-trackers in Windows 10, and teamed up with leading eye-tracker manufacturer, Tobii, to provide a consumer grade product (the model 4C). However, for this product there are also license restrictions and technical impediments for using this for research or your own HCI project (although a legacy Matlab toolbox exists that is sufficient to allow developing a BCI-like typing interface).
Wetware
I’ve mentioned things like alpha waves and SSVEPs, and to understand what is going on when we use the above non-invasive hardware that sits outside our heads, it is important to understand something of the wetware that sits inside our heads.
The brain and central nervous system (CNS) are made up of neurons, some of which are configured as receptors (for the senses) and some as effectors (for the muscles). Individual neurons branch and connect to other neurons chemically across gaps called synapses, which may be excitatory or inhibitory. The neurons themselves open and close channels that allow ions to pass across their cell membranes, or hinder them.
Neurons
Once the activation of a neuron builds to a sufficient threshold level, it discharges suddenly producing an electrical spike and transmitting its chemical neurotransmitters across the synaptic gaps to other neurons. It is these spikes that EEG is detecting.
The neurons that are involved in a given perceptual or motor ‘event’ or ‘task’ are all involved in stimulating each other, with different kinds of neurons operating in different ways and with different speeds. The back and forward transmissions produce complex patterns, including specific frequencies that relate to the distances between the neurons involved, as well as the number of neurons involved.
Spikes
EEG spikes are labelled with N for negative or P for positive, depending on their ‘direction’ or ‘polarity’, along with a number that gives the approximate time in milliseconds after a triggering sensory event or intention. One important one is the P300, which is often used in BCI.
Spikes can also be measured backwards from a motor event, such as pressing a key, and the intention to press a key also tends to occur around 300ms before the motor event happens, and can be detected in the premotor cortex.
For sensory events, we have SEPs (Sensory Event Potentials), with Auditory and Visual events being particularly important in BCI (i.e. AEP and VEP). When we have a repeated event, this tends to produce spikes at a matching frequency. For a flashing light this is called a steady state VEP (SSVEP) and SSVEPs are recorded in the V1 visual cortex at the back of the head (occipital lobe). SSVEP is also a key signal for BCI.
If you shut your eyes and there is no visual input, your brain tends to fall into a relaxed state with only low frequency, long distance, interactions between neurons. This 8 to 12Hz frequency is called an alpha wave, and an open/closing eyes test is one of the standard ways of seeing if our EEG equipment is working properly — with the resulting low frequency signals being detected most strongly in the occipital lobe around the primary visual cortex (V1).
EEG is typically measured in broad bands (although their precise definition varies according to source). These include notably alpha (7-14Hz), beta (14-28Hz) and gamma (28-56Hz). The beta band is most strongly associated with active processing and is commonly used in BCI. The Flinders University researchers showed that the gamma band tended to be very strongly contaminated by muscle, invalidating much of their previous work (Whitham et al., 2007-11; Pope et al. 2009).
To explore what was really muscle (EMG) and what was really neural (EEG), the Flinders researchers created a unique dataset in which they paralysed subjects (with two anaesthetists on hand to ensure they breathed) and were able to do a direct contrast between muscle-free and muscle-contaminated EEG with the same subjects and tasks. The dangers involved to subjects were such that ethics was granted on the condition that it couldn’t be advertised and that only volunteers with the medical understanding to appreciate the risks could be used as subjects — thus the subjects were generally drawn from the research group itself.
Using this dataset they were then able to characterize which electrodes and frequencies were worst contaminated, and to develop signal processing pipelines that removed all the contamination of central electrodes, and (given enough central electrodes) most of the contamination of peripheral electrodes (Fitzgibbon et al. 2007-2016; Janani et al. 2017-18).
Cortex
The cortex is kind of the skin of the brain, the outer layers. And this is pretty well all that EEG can pick up directly, although with advanced signal processing and a dense array of electrodes it is possible to probe deeper (without actually digging or drilling).
Working back to front, we have the occipital lobe and visual cortex at the back, and in front of that visual association areas and somatosensory association areas. We actually have far more than five senses, and the somatosensory system relates to our internal sensing of our own body, while association areas make connections between different parts of the body (which allows us to learn about and understand the whole sensorimotor landscape of an event).
At the sides, we have the auditory cortex towards the back and the speech cortex nearer the front, with language processing in association areas adjacent to them.
Generally, speech processing and logical thought take place in the left half of the brain for naturally right-handed people, while the right brain is more associated with artistic and intuitive thought and creativity. Some early BCI experiments on subliminal audio messages (that affect you but are not consciously seen), suggest that the syntactic processing takes places in the left hemisphere while the semantic processing takes place in the right — and the syntactic processing was bypassed for the subliminal audio (Powers et al., 1998).
The two hemispheres of the brain are connected by the corpus callosum, and when this is severed there is a mismatch where things seen with one eye (sent to opposite hemisphere) may be understood practically/functionally but can’t be named correctly, or vice-versa (Sachs, 1985). The corpus callosum is slightly different between the sexes (bigger in females), and also gets smaller in old age, and some functional differences between the sexes have been characterized (Davatzikos and Resnick, 1998).
Across the top, from ear-to-ear, we have a band of sensory cortex (about receptors/perception) then a band of motor cortex (about muscles/movement) then a band of pre-motor cortex (where intentions to move start). These are a kind of distorted human projection (homunculus) where the more important bits have more real estate (like hands and face).
Of course, the identified mouth area is adjacent to the areas for language, including grammar (syntax), and for producing speech; while the ears area is adjacent to the areas for understanding language and meaning (semantics) and for understanding speech.
In front of the motor areas, we have the prefrontal cortex — just behind the forehead. This is associated with attention, decision making, planning and working memory.
Software
The management, running and testing of BCI/EEG systems is quite complex, and there are many software packages in Python or Matlab for performing the signal processing and managing stimulus presentation and synchronization.
However, the most accessible and easy to use software is probably the BCI2000 software which directly implements a BCI SSVEP speller keyboard and has an active user community to help get things working:
https://www.bci2000.org/bbs/.
It is not appropriate here to discuss particular software in detail, or provide set-up instructions, and generally the libraries and tutorial information provided by the hardware manufacturer and/or BCI2000 (plus the help of the user community) will be sufficient to get going.
So in this section, we discuss briefly the general principles of how the EEG is collected and processed, and what algorithms and neuroprocessing phenomena are used to implement BCI.
EEG signals are very weak, as they must pass through skull and scalp and hair, so high amplification is needed, — as well as great care to maintain good contact.
Moreover, each electrode receives signal from multiple sources, neural and muscular, even signals from the eyes muscles and from as far away as the heart. The basic approach to handling this is called Blind Signal Separation or Blind Source Separation (BSS), the aim being to separate out the sources mixed together in the electrodes. A basic principle is that you need as many electrodes as you have sources you want to separate out.
A secondary goal here is to localize the sources so that you can figure out where they all came from. This is the so-called inverse problem, since given location information about both the electrodes and the sources, and the speed and attenuation of the propagation between them, it is possible to calculate what signal we would expect — this is the much simpler forward problem, and if you know the location information you are no longer ‘blind’, so no longer doing BSS.
The simplest approach is called Principal Component Analysis, and minimizes the variances (or sum of squared errors) in the reconstruction of the signals from the extracted components. It is usually the first step in any analysis, and also allows estimation of the number of sources — and thus compression of the data to a smaller number of putative sources.
A higher order family of approaches called Independent Component Analysis seeks to minimize the higher order (powers of) error as well. But this is mathematically more difficult (although there are libraries of many such algorithms, and their many variants).
Other techniques also try to take into account the delays.
Similar principles apply to soundwaves and are used by noise cancelling microphones and microphone arrays, and indeed can allow you to home in on an arbitrary point in 3D space if you have at least four non-colinear microphones.
Similarly for radiowaves, modern cellular and WiFi systems use multiple antennae at different distances, they seek to localize the receivers and beam multiple low-power delay-synchronized signals towards them, and receive them back the same way.
This means high power can be sent to the receiver rather than blasting it in all directions; and when receiving, the different sensors can combine signals together with the appropriate delays and increase noise resilience.
This approach is called beamforming, as it focuses the signal in both space and time, while the unwanted signals tend to cancel out because they are not in phase.
A closely related approach is the Surface Laplacian, which used first and second order differences amongst a cluster of electrodes on the curved surface of the scalp to focus in on an internal point where signals of interest are expected to occur.
For some BCI typewriters, this might be over the part of the motor and premotor cortex associated with particular fingers (where the corresponding locations are only millimetres apart in the homunculus). For other very common BCI typewriters, they may be focussed over the occipital cortex in order to pick up Visual Evoked Potentials (VEP or SSVEP).
Using appropriate pipelines of such signal processing techniques it is in principal possible to home in on brain signal and eliminate almost all of the muscle signal if there are enough central electrodes, in which case the muscle signal can largely filtered out of even the peripheral signals that are directly over muscle (although it is important to use difference signals between pairs of nearby signals, because with a common reference, too many bits of resolution are lost in coping with the huge EMG amplitudes, while close electrodes have similar signals and can focus their 16 or 24-bits of resolution on the differences.
Note that 16-bit resolution is accurate to ± one part in 32767, while 24-bit resolution is accurate to about ± one part in eight million, providing the analogue part of the circuit is up to scratch. This is a very important consideration when you consider that the EMG signals at peripheral electrodes can be many thousands of times greater than the EEG signals in amplitude.
Various such pipelines have been explored, and even optimized using information from the Flinders paralysis experiment (Fitzgibbon et al., 2007-16; Janani et al., 2017-18).
With appropriate application of these signal processing pipelines, it is possible to successfully analyse gamma-band EEG, and it is possible to get good results from consumer/hobby grade EEG equipment so long as there are electrodes in the required locations and the software and interface allows for sufficient temporal resolution: for hobby grade systems with fixed frames, the electrodes are very often in poor positions, and if there are not multiple electrodes in central positions, little actual EEG can be recovered; for wireless systems using Bluetooth/BLE, performance will tend to be poor due to the sampling frequency and thus low temporal resolution. (Grummett, 2014-2015).
The Emotiv EPOC or EPOC+ remains a good compromise system, and is the only fixed frame system recommended. The 3D printed frame of the OpenBCI system allows some flexibility, but best results will be achieved by using an electrode cap or flexible band with good electrodes.
Generally 802.11 WiFi is preferred to 802.15 Bluetooth standards as it provides much greater bandwidth.
Of course, wired systems can also be used successfully, but dragging cables can often pull the electrodes out of position and make them lose contact with the scalp.
Further Reading
- Behavioural and Brain Sciences, Target Article by EM Pothos and JR Busemeyerand. Can quantum probability provide a new direction for cognitive modeling? Volume 36 , Issue 3 , June 2013 , pp. 255 – 274. https://doi.org/10.1017/S0140525X12001525
- Communications of the ACM, Keith Kirkpatrick, November 2023, pp. 13–15. Accessibility and Inclusion through Technology. https://doi.org/10.1145/3623642
- New Scientist, Editorial, 9 September 2023, p10. Implant lets you type by thinking. https://www.newscientist.com/issue/3455/
- New Scientist, Timothy Revell on 26 May 2023. Elon Musk's brain implant firm Neuralink gets approval for human trial. https://www.newscientist.com/article/2375886
- New Scientist, Matthew Sparkes on 12 May 2021. AI lets man with paralysis type by just thinking about handwriting. https://www.newscientist.com/article/2277272
- Scientific American, Dario Gil on December 26, 2020. The Ethical Challenges of Connecting Our Brains to Computers. https://www.scientificamerican.com/article/the-ethical-challenges-of-connecting-our-brains-to-computers/
- Scientific American’s George Musser on The Strangeness of Physics and Telepathy. https://bigthink.com/hard-science/george-musser-on-the-strangeness-of-physics-and-telepathy/
- Scientific American, Nature’s Miryam Naddaf on August 24, 2023. Brain-Reading Devices Allow Paralyzed People to Talk Using Their Thoughts. https://www.scientificamerican.com/article/brain-reading-devices-allow-paralyzed-people-to-talk-using-their-thoughts/
References
- Carabali CA, Willoughby JO, Fitzgibbon SP, Grummett T, Lewis T, DeLosAngeles D, Pope KJ (2015). "EEG source analysis of data from paralysed subjects", 11th International Symposium on Medical Information Processing and Analysis
- Caton R (1875). "Electrical currents of the brain". British Medical Journal. 2 (765): 278.
- Cottrell J, Fitzgibbon SP, Lewis TW, Powers DMW (2012). "Investigating a Gaze-Tracking Brain Computer Interface Concept Using Steady State Visually Evoked Potentials", 2012 Spring Congress on Engineering and Technology, pp.1-4
- Davatzikos C, Resnick SM (1998). "Sex differences in anatomic measures of interhemispheric connectivity: Correlations with cognition in women but not men." Cerebral Cortex 8 (7): pp.635–40. https://doi.org/10.1093%2Fcercor%2F8.7.635
- Fitzgibbon SP, DeLosAngeles D, Lewis TW, Powers DMW, Whitham EM, Willoughby JO, Pope KJ (2015). "Surface Laplacian of scalp electrical signals and independent component analysis resolve EMG contamination of electroencephalogram", International Journal of Psychophysiology 97 (3), pp.277-284
- Fitzgibbon SP, Lewis TW, Powers DMW, Whitham EM, Willoughby JO, Pope KJ (2013). "Surface Laplacian of Central Scalp Electrical Signals is Insensitive to Muscle Contamination", IEEE Transactions on Biomedical Engineering 60 (1), pp.4-9
- Fitzgibbon SP, Powers DMW, Pope KJ, Clark CR (2007). "Removal of EEG noise and artifact using blind source separation", Journal of Clinical Neurophysiology 24 (3), pp.232–239
- Fitzgibbon SP, DeLosAngeles D, Lewis TW, Powers DMW, Grummett TS, Whitham EM, Ward LM, Willoughby JO, Pope KJ (2016). "Automatic determination of EMG-contaminated components and validation of independent component analysis using EEG during pharmacologic paralysis", Clinical Neurophysiology 127 (3), pp. 1781-1793
- Fuss IG, Navarro DJ (2013). "Open Parallel Cooperative and Competitive Decision Processes: A Potential Provenance for Quantum Probability Decision Models." Topics in Cognitive Science 5 (4), pp.818–843. https://doi.org/10.1111/tops.12045
- Freedman, DZ, van Nieuwenhuizen (1985). "The Hidden Dimensions of Spacetime", Scientific American, 252 (3) pp.74-83. https://www.jstor.org/stable/pdf/24967594
- Grummett TS, Leibbrandt RE, Lewis TW, DeLosAngeles D, Powers DMW, Willoughby JO, Pope KJ, Fitzgibbon SP (2015). "Measurement of neural signals from inexpensive, wireless and dry EEG systems", Physiological Measurement 36 (7), p.1469
- Grummett TS, Fitzgibbon SP, Lewis TW, DeLosAngeles D, Whitham EM, Pope KJ, Willoughby JO (2014). "Constitutive spectral EEG peaks in the gamma range: suppressed by sleep, reduced by mental activity and resistant to sensory stimulation", Frontiers in Human Neuroscience 8, p.927
- Huey EB (1908/1968). The Psychology and Pedagogy of Reading, MIT Press, Cambridge MA.
- Janani AS, Grummett TS, Lewis TW, Fitzgibbon SP, Whitham EM, DelosAngeles D, Bakhshayesh H, Willoughby JO, Pope KJ (2017). "Evaluation of a minimum-norm based beamforming technique, sLORETA, for reducing tonic muscle contamination of EEG at sensor level", Journal of Neuroscience Methods 288, pp.17-28
- Janani AS, Grummett TS, Lewis TW, Fitzgibbon SP, Whitham EM, DelosAngeles D, Bakhshayesh H, Willoughby JO, Pope KJ (2018). "Improved artefact removal from EEG using Canonical Correlation Analysis and spectral slope", Journal of Neuroscience Methods 298, pp.1-15
- Javal LÉ (1907). Physiologie de la lecture et de l’écriture. Annales d’Oculistique, Paris, pp.137-187.
- Kaku M (2006). Parallel Worlds: A Journey Through Creation, Higher Dimensions, and the Future of the Cosmos. Anchor.
- Kennedy PR, Bakay RA (1998). Restoration of neural output from a paralyzed patient by a direct brain connection. Neuroreport 9, pp.1707-1711
- Kuhn TS (1962). The Structure of Scientific Revolutions. The University of Chicago Press.
- Kuhn TS (1970). Logic of Discovery or Psychology of Research? In Lakatos, Irme; Musgrave, Alan (eds.). Criticism and the Growth of Knowledge. Cambridge University Press. pp. 1–24.
- Kunjan S, Lewis TW, Grummett TS, Powers DMW, Pope KJ, Fitzgibbon SP, Willoughby JO (2016). "Cross subject mental work load classification from electroencephalographic signals with automatic artifact rejection and muscle pruning", Brain Informatics and Health BIH 2016, Omaha
- Oxley TJ, Opie NL, John SE, Rind GS, Ronayne SM, Wheeler TL, Judy JW, McDonald AJ, Dornom A, Lovell TJH, Steward C, Garrett DJ, Moffat BA, Lui EH, Yassi N, Campbell BCV, Wong YT, Fox KE, Nurse ES, Bennett IE, Bauquier SH, Liyanage Kishan A, van der Nagel NR, Perucca P, Ahnood A, Gill KP, Yan B, Churilov L, French CR, Desmond PM, Horne MK, Kiers L, Prawer S, Davis SM, Burkitt AN, Mitchell PJ, Grayden DB, May CN, O'Brien J (2016). "Minimally invasive endovascular stent-electrode array for high-fidelity, chronic recordings of cortical neural activity", Nature Biotechnology 34, pp.320-327.
- Pope KJ, Fitzgibbon SP, Lewis TW, Whitham EM, Willoughby JO (2009). "Relation of gamma oscillations in scalp recordings to muscular activity", Brain Topography 22 (1), pp.13-17
- Popper K, (1934/1959). The Logic of Scientific Discovery (2 ed.). Martino Publishing.
- Powers DMW, Dixon SE, Clark CR, Weber DL (1996). "Cocktails and brainwaves-experiments with complex and subliminal auditory stimuli." Australian New Zealand Conference on Intelligent Information Systems (ANZIIS 96), pp.68–61.
- Powers DMW, Fitzgibbon SP, Clark CR (2007). Brain Computer Interface, Joint HCSNet-HxI Workshop on Human Issues in Interaction and Interactive Interfaces
- Powers DMW, Clark CR, Fitzgibbon SP, Pope K (2007). Removal of EEG Noise and Artifact Using Blind Source Separation, Journal of Clinical Neurophysiology 24 (3), pp.232-243.
- Roll, W. G., & Williams, B. J. (2010). Quantum theory, neurobiology, and parapsychology. In S. Krippner & H. L. Friedman (Eds.), Mysterious minds: The neurobiology of psychics, mediums, and other extraordinary people. Praeger/ABC-CLIO. Pp.1–33
- Sacks O (1985). The Man Who Mistook His Wife for a Hat and Other Clinical Tales. Pan Books.
- Whitham EM, Pope K, Fitzgibbon S, Lewis T, Clark C, Loveless S, Willoughby J (2007). Scalp EEG during paralysis: augmented gamma activity can be detected during cognitive processing, Scientific Meeting of the International Brain Research Organisation, Melbourne
- Whitham EM, Pope KJ, Fitzgibbon SP, Lewis T, Clark CR, Loveless S, Broberg M, Wallace A, DeLosAngeles D, Lillie P, Hardy A, Fronsko R, Pulbrook A, Willoughby JO (2007). Scalp electrical recording during paralysis: Quantitative evidence that EEG frequencies above 20Hz are contaminated by EMG, Clinical Neurophysiology 118 (8), 1877-1888
- Whitham EM, Fitzgibbon SP, Lewis TW, Pope KJ, DeLosAngeles D, Clark CR, Lillie P, Hardy A, Gandevia SC, Willoughby JO (2011). Visual Experiences during Paralysis, Frontiers in Human Neuroscience 5
- Whitham EM, Lewis TW, Pope KJ, Fitzgibbon SP, Clark CR, Loveless S, DeLosAngeles D, Wallace AK, Broberg M, Willoughby JO (2008). Thinking activates EMG in scalp electrical recordings, Clinical Neurophysiology 119 (5), 1166-1175
- Whitham E, Fitzgibbon S, DeLosAngeles D, Lewis T, Pope K, Clark C, Lillie P, Hardy A, Gandevia S, Willoughby J (2009). Experiences during paralysis: vision and humour, Australian Neuroscience Society Inc. 29th Annual Meeting
- Yazdani N, Khazab F, Fitzgibbon SP, Luerssen MH, Powers DMW, Clark CR (2010). Towards a brain-controlled Wheelchair Prototype, 24th BCS International Conference on Human-Computer Interaction
