Fullscreen HTML5Med Res (??MB)HTML4

INTERF TO THE by DR DAVID MADDISON S While interfacing to the human brain might seem the stuff of science fiction, there is much work being done in this area, as well as work on animals and insects. You can even do it yourself and it can have many practical aspects. cience fiction is full of scenarios in which a person’s own brain is interfaced directly to a computer or a machine (or another person) and is used to interact with, or control it. Examples include the people in The Matrix trilogy, the Daleks in Dr Who and the Borg in Star Trek. And who can forget that 1983 sci-fi film Brainstorm, the whole theme of which was the development, use (and misuse) of a Brain-Computer Interface (BCI). BrainStorm can be viewed on YouTube at http://youtu.be/cOGAEAJ4xJE In this article we will primarily focus on methods of interfacing the human brain with computers and machines, so called brain-computer interfaces or BCIs. Australia is a world leader in the Cochlear implant but these devices do not interface directly to the brain. Rather, they connect to existing nerve fibres and are in the related category of neuroprosthetics. A brain-computer interface can be defined as a system for reading information from the brain to enable control of a machine or the transmission of an item of communication or thought. It is also a system of feeding information into the brain to enable the brain to interpret a sensation from some external sensory device. In other words, information is transmitted to and from the brain to a machine without the engagement of the usual senses, the peripheral nervous system or limbs. Reading the brain To interface the brain to a computer, information has to be first read from the brain. There are several means by which information can be acquired from a brain for the purpose of brain-computer interfacing. Electroencephalography (EEG) has the advantage that it is relatively cheap and simple to do and can provide useful information in a clinical setting. It is also non-invasive and so is amenable to a wide variety of brain computer interfacing techniques, provided useful information can be obtained. There is a distinct advantage that changes in brain activity can be read very rapidly compared to other slower methods that rely on a change of blood flow, such as functional magnetic resonance imaging (fMRI), for example. EEG also has a number of disadvantages. A scalp-reAn EEG headset, as used in a clinical setting. Worldwide, the location of EEG electrodes is standardised according to the so-called 10-20 system (see right) whereby electrodes are positioned according to anatomical landmarks. Results from different researchers will therefore correspond to the same electrode locations (there are also higher resolution electrode placement schemes such as the 10-5 system and others). In clinical applications typically 19 electrodes are used plus an earth and system voltage reference. The voltages measured are of the order of microvolts and are amplified by 1,000 to 100,000 times. 12  Silicon Chip siliconchip.com.au FACING BRAIN ... yes, it is really happening! corded EEG represents a coarse measure of brain activity due to the poor electrical conduction and thickness of the skull and the subsequent dispersion of electrical signals. It only measures the collective excitation of large numbers of neurons behaving in a synchronised manner that also happen to be oriented in the correct direction to provide an electrical signal that conducts toward the scalp. Individual neurons or small groups of neurons cannot be read directly. The EEG output consists of rhythmic signals in various frequency ranges and also transient activity. Typically (but not always) these rhythmic signals are classified in terms of a number of frequency bands. These are usually Delta (<4Hz), Theta (4-7Hz), Alpha (8-15Hz), Beta (16-31Hz), Gamma (32+Hz) and Mu (8-12Hz). All these bands are associated with a certain biological significance and activities in the brain. Electrocorticography Electrocorticography (ECoG) is a form of EEG in which the electrodes are placed on the surface of the brain (cerebral cortex). It has the advantage that much higher spacial resolutions can be obtained and the sampling of much smaller groups of neurons. Different types of electrodes can also be used. Of course, it has the distinct disadvantage that it is intrusive and requires the skull to be opened. For the purposes of brain computer interfacing it would only be done (at this point in time) for life-critical applications such enabling a quadriplegic to operate a robotic arm or wheelchair. Electrical activity in the brain The brain consists of specialised cells called neurons and glial cells. The neurons are the cells responsible for information processing while the glial cells mostly have support roles. Neurons are electrically active and can communicate with other cells in the brain by a branched conducting fibre called an axon that extends from the body of the cell and which can communicate with many other nearby or far away neurons. Neuron to neuron communication constitutes the essence of how the brain works. The architecture of this connectivity between neurons Research conducted at the Brain Institute at the University of Utah showing three types of electrocortical arrays in simultaneous use. The numbered electrodes are part of an ECoG array sitting on the surface of a human brain, the green wires terminate in a micro-ECoG grid and the black square with the gold-coloured electrodes is a “Utah Electrode Array” (UEA) which has an even finer resolution than the micro-ECoG grid. In this work the electrodes are used to discover and remove areas of the brain responsible for epileptic seizures but data read from such electrodes can also be used, for example, to convert speech-related brain signals into words, control machinery such as a robot arm, a wheelchair or even an aircraft or to work in any other application requiring a direct brain-computer interface. Note that while it is obviously an invasive procedure to have such electrodes implanted beneath the skull, these particular electrodes sit on the surface of the brain and do not penetrate it where damage may be done in sensitive areas. siliconchip.com.au January 2015  13 Synaptic transmission of information showing neuron body (soma) and attached dendrites and axons. Information enters a neuron via a dendrite and leaves via an axon. Neurotransmitter molecules pass across the synaptic gap. Each electrical impulse will cause a connected neuron to be either excited or inhibited. The collective excitement or inhibition of very large numbers of neurons is what can be detected by an EEG signal. is known as a neural network. The way axons transmit electrical signals is by means of electrochemical pulses involving sodium and potassium ions being transported in different directions through the neural cell membrane. These electrochemical pulses are known as action potentials and typically last less than one millisecond and propagate at speeds of 1 to 100 metres per second. Some neurons are inactive most of the time while others may be constantly active and fire at a rate of 5 to 50 times per second. A neuron’s axon is connected to other neurons via junctions called synapses which make contact with another part of the neuron’s body called the dendrite. There is a very high level of connectedness; each axon may have many thousand synaptic connections to neurons or possibly other cell types. According to the latest estimates the human brain has an average of 86 billion neurons and 100 trillion synapses. The axons are the “wires” that connect most of the functional elements of the brain with each other. Once an electrical signal or action potential arrives at a synapse, specialised chemicals known as neurotransmitters are released and these bind with the target neuron or other cell. Many different neurotransmitters exist (around 100 have been identified so far) and can exert many different simple 14  Silicon Chip or complex influences on the target (or post synaptic) neuron but fundamentally will cause the post synaptic neuron to be either inhibited or excited. As each neuron is connected to large numbers of other neurons the total numbers of inhibitory or excitatory signals received will determine whether that neuron will either not fire or fire and not pass or pass information to the next neuron in the network, and so on. Many of these synaptic junctions are dynamically reconfigurable by changing the nature of the signals that travel through them and are thought to be involved in learning and memory. Since the connections are not “set in stone”, some reconfiguration of the brain is possible and this is the basis of neuroplasticity, the ability of the brain to reconfigure itself to compensate for damage. This plasticity has only been seriously recognised in recent years and also suggests that electrode placement for the purpose of brain-computer interfacing is not extremely critical. It suggests that the brain will eventually be able to learn how to control an interface no matter where in the brain it is located (within reason) by a sufficient amount of learning. Electrical signals in the brain or action potentials are the way neurons communicate with each other. Action potentials are subject to some basic but important rules. Firstly, there is a minimum threshold voltage below which no signal will be propagated along an axon so electrical “noise” will not cause signals to propagate. Secondly, it is “all or nothing”; each action potential has the same strength, independent of the strength of a stimulus. Thirdly, there is a refractory period after the action potential in which no further action potentials can be generated. This helps ensure that the action potential propagates in only one direction and not back to its point of origin. Most people are familiar with the terms “grey matter” and “white matter”. If one takes a cross-section of a human brain, it will be seen that the outer layers are dark in colour (grey) while the inner parts are light in colour (white). The difference arises from the fact that axons are lighter in colour due to their insulating myelin sheaths while neurons are darker in colour. These colour differences show that the outer parts of the brain contain mostly neurons and the inner parts of the brain contain mainly axons or the “wiring” of the brain. Non-invasive brain interface While EEG and other methods can be used to read information from the brain, the information has to be meaningful and somehow express the subject’s intent if they are to do something useful like control a machine. Like any new task, practice is necessary so that the appropriate synaptic connections can be strengthened in order to learn the desired behaviour. The following methods describe ways BCI devices can be controlled without intrusive implanted electrodes. An EEG signal can be influenced by imagined movements and biofeedback methods whereby an individual learns with many training sessions to influence an EEG signal in a way that can be detected and used to drive a machine. Silent vocalisation of words can also be sensed and used to drive the interface. The Steady State Visual Evoked Potential (SSVEP) is siliconchip.com.au a control system whereby a subject looks at one or more flashing screens or symbols. The signal from the flash is relatively easy to detect in an EEG signal and the intent of the subject can be inferred from the frequency of the flashing area they are looking at. It may be annoying for people to use, however. The P300 wave, or more specifically now known as two waves, the P3a and P3b, are another way information can be read from the brain. These occur after a low probability event is observed and recognised among a series of “standard” events. These waves are useful to monitor for brain-computer interfacing because they are relatively consistent across most people and using them can be learned with minimal training. One example of using this brainwave for communication in the disabled is the use of a P300 matrix speller. A test subject is presented with a 6x6 matrix of letters and numbers and individual rows and columns are illuminated in a pseudo-random manner. The subject selects a letter by concentrating on the character they want and their P300 wave is detected at that time. Using this method with a scalp EEG results in letter selection rates of 1.4 to 4.5 characters per minute. This was able to be increased to 17 characters per minute by Peter Brunner and others in 2011 with an implanted 96-electrode array. Hybrid systems have also been developed combining the SSVEP mentioned above and the P300. See YouTube video http://youtu.be/08GNE6OdNcs “Emotiv BCI2000 Video.mp4”. Writing information to the brain Mentioned above were several methods that could be used to read information from the brain. It is also possible to “write” information to the brain. This can be done via implanted electrode arrays, transcranial magnetic stimulation (TMS) where a powerful magnetic field is pulsed through the skull or focused ultrasound (FUS) where a focused ultrasound beam is transmitted through the skull. All these methods excite groups of neurons within their field of influence and cause them to fire. The earliest experiments with interfacing animal brains to machines happened in 1969. The experiment was by E.E. Fetz at the University of Washington School of Medicine in Seattle and involved training a monkey to move a biofeedback meter needle by activating neurons in its motor cortex, the region of the brain responsible for the execution of movement. The activity of these neurons was read from an implanted tungsten micro-electrode. Following work by Fetz in interfacing a monkey brain to a machine, in the 1980s Apostolos Georgopoulos at Johns Hopkins University found a mathematical relationship between the electrical signals from motor cortex neurons and the direction the animal wished to move. This lead to the development of computer models that relate movement to neural signals and are the basis of models that now translate complex neural signals into commands to operate machines such as robot arms. Monkey controls robot arm Professor Miguel Nicolelis from Duke University in North Carolina was the first to interface a monkey brain to a robot arm which it could move. By 2000 the group had managed to reproduce a monkey’s siliconchip.com.au A monkey using a brain-controlled robotic arm to grab food to feed itself. The monkeys were able to effortlessly control the robot arm as though it were a natural part of themselves. arm motion in a robot arm by monitoring neural signals from the monkey. The monkey had no direct control over the arm, it just reproduced its movements. Subsequently, monkeys first trained to reach and grab objects on a computer screen using a joystick. This joystick also controlled a robot arm which the monkeys could not see. They were learning the simply task of moving things in two dimensions on the computer screen before being shown the actual robot arm which could move in three dimensions which the monkeys learned to control. In this work an electrode array monitored an area on the motor cortex of around 50 to 200 neurons. Other groups have done similar work and a group lead by Andrew Schwartz at the University of Pittsburgh in 2008 interfaced a monkey to a robot arm with an electrode array which recorded signals from 15-30 neurons and which enabled the monkey to feed itself. A video of a monkey operating a robot arm can be seen at http://youtu.be/gnWSah4RD2E “Monkey controls robotic arm with brain computer interface”. Visual imagery from the brain Although the stuff of science fiction, scientists are mak- Open-source brain computer interface There is a successfully funded Kickstarter project called OpenBCI to develop an open source platform to enable anyone with an interest to monitor their own or another person’s brainwaves via a wearable EEG monitor with a view to developing products controlled by the brain. Each board supports eight electrodes but these can be daisychained together to increase the electrode count. Apart from the electronics and software there is also a 3D printable headset to mount the electronics package. See http://openbci. com/ January 2015  15 can be seen at http://youtu.be/nsjDnYxJ0bo “Movie reconstruction from human brain activity”. Reading the subject matter of dreams Image (top row) presented to a cat and reconstruction (bottom) of that image as read from the brain using electrodes implanted in a region of the brain that processes visual information. ing good progress in reading visual imagery from inside the brain. Examples include reading images seen by the eye directly from the brain and also determining some content of dreams. In one of the first demonstrations of reading visual imagery from a brain a cat had electrodes implanted in its brain and it was made to watch various scenes. The data from the electrodes was processed with some basic mathematical filtering and the original image was reconstructed. It certainly seems from the reconstructed images, however, that the animal imposed its own cat-like interpretation on the features on the human face. This work was done in 1999 at the University of California, Berkeley with a research team lead by Professor Yang Dan. Naturally, this brain reading was invasive by virtue of the fact that electrodes needed to be implanted on the brain. Apart from cats, visual imagery has also been read from human brains. This work was done in 2011 by scientists at the University of California, Berkeley lead by Professor Jack Gallant. In this case non-invasive function magnetic resonance imaging (fMRI) techniques and computational modelling were used to read and interpret brain activity. Subjects watched video clips and the moving images were read from their brains. To extract this video information from the brains of experimental subjects they had to lay still inside a fMRI machine while watching two different sets of trailers from Hollywood movies. The fMRI machine was used to measure the blood flow through the visual cortex of the brain which is the part responsible for vision. The fMRI data was then broken down into three dimensional versions of pixels known as “voxels”. One of the researchers said “We built a model for each voxel that describes how shape and motion information in the movie is mapped into brain activity”. As the video was being played to the subject the change in each voxel, corresponding to changes in brain activity in that region, was correlated with the video image being presented at the time. A problem of using fMRI for this type of work is that the blood flow which fMRI measures changes much more slowly than the electrical neural signals. This problem was overcome by the development of a two stage model that separately describes the neural signals and blood flow. However, the scientists who did this work were careful to point out at the time that the technology to read people’s thoughts is many decades away. A video of the experiment 16  Silicon Chip Japanese researchers Yukiyasu Kamitani and colleagues at the Advanced Telecommunications Research Institute International in Kyoto, Japan have been working on reading the subject matter of people’s dreams. In work published in 2013 they showed that they could tell what a person was dreaming about. The research involved asking volunteers to have a mid-afternoon nap in a fMRI machine and when they had reached the earliest stages of sleep (stage 1 or 2) they were woken and asked to give a verbal report of what they were dreaming about. This was repeated at least 200 times for each subject. Next, these verbal dream reports were analysed by researchers who reduced them to key words and concepts. Researchers next went online to build a vast visual database of images that mostly closely corresponded to the subject matter of the verbal reports provided by the dreamers. Researchers then did further fMRI scans on the dreamers while they were awake and asked them to watch the images that had been collected that corresponded to the subject matter reported from their 200 plus dream sessions. This enabled brain activity patterns to be read from that individual that corresponded to the visual imagery they were watching. These activity patterns were used to train a decoder computer to correlate patterns of brain activity with certain types of visual imagery. After the decoder was trained it was possible to enter measured brain activity and it could then correlate that with the visual imagery now known to produce this pattern and thus the subject matter of the dream could be predicted. The predictive capacity of the system was quite coarse. For example, it could tell if someone was dreaming of driving in a car but not what type of car. Also, the decoder has to be trained individually for each person. It cannot be used to read dream subject matter without individualised training. See YouTube video http://youtu.be/inaH_i_TjV4 “Dream decoding from human brain”. Transmitting thoughts from one person to another In early 2014, a team lead by Alvaro Pascual-Leone, Director of the at the Berenson-Allen Center for Noninvasive Brain Stimulation at Beth Israel Deaconess Medical Center (BIDMC) and Professor of Neurology at Harvard Medical School in Boston succeeded in reading a thought from one person and transmitting it to another person 8,000km away via the Internet. Together with researchers in France and Spain, the thoughts of a person in India were transmitted to a person in France. The words transmitted were the greetings “hola” and “ciao”. In reality it was not words that were transmitted but a binary code. The sender evoked imagery of using either their hands or feet. The brainwaves of the sender in India were read by an EEG and it was determined if they were imagining using either their hands or feet. Hands corresponded to a “0” and feet to a “1”. The chosen number was transmitted over the Internet to France and the receiver’s brain was stimulated via the process of transcranial magnetic stimulation (TMS). The TMS stimulation was interpreted as a flash of light (phosphene) for a 1 and siliconchip.com.au no flash for a 0 and thus the simple message was decoded. Connecting two rat brains together The brains of two rats were electronically linked such that what one rat did was duplicated by another rat at a distant site. A team lead by Miguel Nicolelis of Duke University in North Carolina and collaborators in Brazil published this work in early 2013. One rat called the “encoder” learned various tasks and signals from a cortical micro-electrode array implanted in it were monitored. The electrical signals from the encoder rat’s brain were then transmitted to the same area of a “decoder” rat’s brain. The encoder’s electrode arrays consisted of 32 electrodes connected to the rat’s primary motor cortex of the brain which is responsible for movement. The decoder rats had 4 to 6 micro-stimulation electrodes implanted in the same area. When the decoder rat received signals from the encoder rat’s brain it interpreted the action meant by those signals and performed the same task (pressing the same lever) as the encoder rat. Even when the decoder rat was untrained and unfamiliar with the task the decoder rats would press the correct lever around two thirds of the time which while not perfect is still a remarkable result. The encoder rat was located in Brazil while the decoder rat was located in the USA. A video of the experiment can be seen at http://youtu.be/w_qbkYDlhDY “Brain-to-brain interface transmits brain activity directly from one rat to another” Human-to-animal control Transmitting a thought from one person to another is impressive but so too is transmitting a command from a person to animal. Seung-Schik Yoo of Harvard Medical School in Boston lead the team. A person was connected to an EEG machine and used the technique of steady state visual evoked potential (SSVEP) to trigger a signal for a rat to move its tail. The rat’s brain was stimulated in the area that controls tail movement by the technique of focused ultrasound (FUS) and the rat moved its tail. The experiment can be seen at https://www.youtube. com/watch?v=VaJjHgyHnEc “Human moves rat’s tail with thoughts alone”. See also http://youtu.be/TpFdM_e76Fw “LEGO goes with the brain: A robot remotely controlled with steady-state visual evoked potentials”, Still images taken from video showing the presented image (top) and the corresonding image read from a human brain using functional magnetic resonance imaging (fMRI). (From http://spectrum.ieee.org/geek-life/tools-toys/this-isyour-brain-on-fmri) (sic) in which a robot is controlled by a person using SSVEP techniques. Human-to-human control Researchers at the University of Washington have enabled one person to control motion in another person. The first person thought of an action to move their hand to press a button but did not actually move their hand. The electrical activity in the brain associated with this intention was recorded with an EEG headset and transmitted via the Internet. The brain of a receiving subject was stimulated via the process of transcranial magnetic stimulation (TMS) which induced an electrical signal in the brain of the subject over an area responsible for hand movement causing them to physically move their hand to press a button. This may sound scary in some senses but it is important to note that this work is currently at a very basic level and there is no indication that mass mind control or robot-like zombie people will be walking our streets any time soon. See http://youtu.be/rNRDc714W5I “Direct Brain-to-Brain Communication in Humans: A Pilot Study”. Human vision & movement An obvious application for interfacing the brain is to provide vision for blind people. Retinal implants (“bionic eyes”) are one such approach but if this is not suitable the vision areas of the brain can be stimulated directly. Data from a camera is processed and sent to an electrode array implanted on the visual cortex of the brain. Where Scheme by which a thought was transmitted from one person to another over the Internet. From http://abcnews. go.com/Technology/scientists-transmitthoughts-brain/story?id=25319813 siliconchip.com.au January 2015  17 this has been done the subjects have gained some limited level of functionality to enable them to do basic tasks and even driving a car slowly in a car park was demonstrated in one instance. BCIs have been used to help disabled people control computer cursors for communication, wheelchairs and robotic arms to help them with household tasks. See YouTube videos http://youtu.be/mJQ0HqThU4c “Two-Dimensional Cursor Control Using EEG”, http://youtu.be/qQ7AJnVKc_g “Mind Typing and PC Control with Brain-Computer Interface (BCI)”, http://youtu.be/gvR0kHm9fwo “BCI driving a wheelchair” and http://youtu.be/76lIQtE8oDY “One Giant Bite: Woman with Quadriplegia Feeds Herself Chocolate Using Mind-Controlled Robot Arm”. Neurogaming Neurogaming is a new computer gaming modality where characters and games are controlled by BCI technology as well as other sensors such as heart rate monitors, eyetrackers and sensors to detect muscle movement. Such technology can also be used for virtual reality training for different professionals and has also been suggested for the treatment of various disorders such as PTSD, ADHD and other behavioural and cognitive disorders. Future uses Anything that requires human input for control is open to the possibility of direct control via a brain-computer interface. For precise and high levels of control it may be necessary to have implanted electrode arrays since at the moment scalp EEG readings are fairly coarse in nature although if training with EEG headsets started at a young age, better results might be achievable. The military also have some interest in controlling fighter jets and other machines with the mind (whether the pilot is in the cockpit or a remote operator). Firefox (1982) was a science fiction movie which features an aircraft with a mind-controlled weapons system but the English-speaking pilot tasked to retrieve the plane could not get it to work until he realised he had to think in Russian, not English. Brain-controlled toys A number of toys have been produced or are under development which are controlled by the brain. One such toy is a radio-controlled helicopter called the Puzzlebox Orbit which us controlled via a NeuroSky EEG headset (see below). Instructions for a do-it yourself conversion of a cheap radio controlled helicopter to mind control using consumer EEG headsets is described at http://www.instructables.com/ id/Brain-Controlled-RC-Helicopter/ Note that on that web page on the right hand column you will see links to other brain control DIY projects. Consumer EEG headsets Interfacing the brain is not just restricted to laboratories. There are a large number of consumer grade EEG headsets available for the purpose of brain computer interfacing. They are all capable of measuring a number of mental states and some can measure facial muscle movement and eye movement as well. A full description of these devices is not possible here but you may wish to research them yourselves. These devices have between 1 and 14 electrodes. Some of these headsets are also appropriate for professional use and research. The devices include: Emotiv EPOC, Emotiv Insight, HiBrain, iFocusBand, Mindball, Mindflex, MindSet, MindWave, Muse, MyndPlay BrainB, Neural Impulse Actuator (discontinued and detected muscle movement only), NeuroSky, OpenBCI (this is an open hardware project, see box), Star Wars Force Trainer (discontinued), Xwave headset (discontinued) and Xwave Sonic (discontinued). Of particular interest is that Emotiv Systems is a Sydneybased company with international offices, founded by former Young Australian of the Year, Tan Le. For an overview of some features of one of the Emotiv headset models see the YouTube video at http://youtu.be/bposG6XHXvU “Emotiv’s New Neuro-Headset”. A lot of open-source software has been developed to support the output of some of these and other EEG devices. An example is OpenViBE, which is a general purpose and highly capable software platform for real-time acquisition, processing and classification of brain waves for all aspects of brain-computer interfaces including biofeedback, robotinterfacing, diagnosis, biofeedback and game control. OpenViBE can be used by anyone even if they are not familiar with programming. Several open-source Matlab toolboxes have also been developed for interpreting data Scheme for brain to brain interface with human subjects. A sender imagines hand movement to press a fire button but does not actually move his hand. The intent to press the button is detected via EEG signals and the signal is transmitted via the Internet. The person receiving the signal is stimulated to press a button as their brain is stimulated via transcranial magnetic stimulation (TMS). (From www.washington.edu/ news/2013/08/27/researcher-controls-colleaguesmotions-in-1st-human-brain-to-brain-interface/). 18  Silicon Chip siliconchip.com.au from various EEG devices. With any EEG device, receiving unwanted electrical noise from muscles can be a problem with these devices so a special effort has to be made to avoid unwanted movement, especially of the facial area, when using these devices. SILICON CHIP readers may be interested in experimenting with some of these devices and software tools. Many of these devices can be connected to smart phones for purposes such as meditation, biofeedback or playing games (neurogaming) or other possible purposes such as assisting the disabled to communicate, for research, software usability testing and so-called neuromarketing where a person’s reaction to advertising material is monitored. BCI2000 In addition to the open source software mentioned above to analyse EEG signals, BCI2000 (www.schalklab.org/research/bci2000) is an open-source suite of software for all aspects of brain-computer interface research and can be used for data acquisition, stimulus of neurons and brain monitoring applications. It is free for non-profit and educational use and supports numerous types of instrumentation and runs on Windows, OS X and Linux. It has been under development since 2000 by the BrainComputer Interface R&D Program at the Wadsworth Center of the New York State Department of Health in Albany, New York with substantial contributions from various other groups. BCI2000 is designed to easily interface with various equipment and software in real time via a network-based interface so that, for example, a robot arm running its own software could be made to be easily controlled by neural signals processed by BCI2000. In addition, Matlab scripts can be executed within BCI2000. BCI2000 has an additional benefit that all data is stored in a standardised format along with the system configuration and event markers so that it can easily be shared with other researchers. To see an example of BCI2000 in use see http://youtu.be/ suKTlrzaU9g “Playing the Game ‘Pong’ with EEG”. Here a 32-channel EEG is acquired and analysed from each of two subjects to extract control signals which move the electronic game paddles. Ethical issues As with any new technology certain ethical issues need to be considered, especially with intrusive brain interfaces such as cortical electrode arrays. While few would question the need for such intrusive interfaces in life-critical applications such as controlling a wheelchair or robot arm, one might question the appropriateness of such an interface for a non-critical application such as connecting to the Internet. On the other hand many would argue that a person is entitled to do as they will with their own body as long as that person pays for it. Other issues relate to the reversibility or otherwise of intrusive BCI interface procedures. Most implants, no matter what type, leave some sort of permanent impact on the body and may not be removable without doing damage. What issues arise if better models of interface are developed and old ones need to be removed? siliconchip.com.au Cyborg Roaches! We make no judgement on the ethics of doing this but some people have built their own remote controlled living cockroaches with parts from a kit as featured in this video. http://youtu.be/V2zNOP6RqRk “Amazing! Real Creating a Cyborg Cockroach (Bugs Robot)”. It is not known whether this would work with typical Australian cockroaches. It is not a joke! Note that the developer does consider ethical issues and addresses them on their web page at https://backyardbrains.com/products/roboroach Alternative therapies also need to be considered. For example, with advances in stem cell research it is conceivable that in the near term future spinal cords could be repaired and the necessity to have an electrode implant for brain control of a wheelchair might become unnecessary (but people already in receipt of such implants might be able to re-purpose them). Conclusion Brain-computer interfacing has an exciting future and it is likely that the first major uses will be to assist disabled people to communicate and move. Neurogaming, like much computer gaming is likely to have many spin-offs such as virtual reality and treatment of various disorders. Later developments might include control of cars, aircraft and many other machines as well. Some people may consider the technology “inhuman” and may choose to preserve what they see as their humanity. Controlling animals with BCI may bring many benefits such as in search and rescue but may also raise ethical challenges. Neuroplasticity ensures that most people should be able to learn to use BCI and most likely do useful things with non-intrusive BCI such as EEG headsets. Other ethical challenges are raised due to appropriateness of the technology for certain uses and cost. In the medium to long term future the rights of people not to have their mind read (should that prove to be possible) need to be seriously considered. BCI is potentially very useful for the disabled but biological cures using stem cells for conditions such as a severed spinal cord may be better and not far off. The nightmare scenarios from science fiction seem a long SC way off, if they happen at all. YouTube videos of interest: Visual Image Reconstruction from Human Brain: http://youtu.be/daY7uO0eftA A Remote Controlled Rat: http://youtu.be/G-jTkqHSWlg Cyborg insects: http://youtu.be/dSCLBG9KeX4 Computer records animal vision in Laboratory – UC Berkeley: http://youtu.be/piyY-UtyDZw January 2015  19