#planet 3d

This is a player piano punch sheet or score taken from a rolled up drum. The holes make the keys move. It is a very simple pattern that translates into musical notes.

This is a decoding of “neural spikes” according to scientists at Elon Musk’s brain computer company Neuralink. The spikes are pictured in single boxes, and many spikes when they fire make a pattern. From that pattern you can begin to decode thought. According to Neuralink, “Everything you hear or think is all action potentials, its action spikes and it feels so real, it feels very real.”

This is an early IBM computer punch card, and the basic pattern recognition of the presence or absence of holes in predefined patterns led to everything we know today about computers. Though IBM introduced the first cards in 1928, they had already been in use to ‘program’ cloth in the Jacquard loom in 1804, making gorgeous silks and tapestries.

I think we are basically back in the same position as we were when IBM first made the computer punch card in 1928 in terms of brain research and decoding thought. Neuralink is putting together elements in brain research that 100 years from now, or even 50 years from now will have tremendous repercussions. The company did not invent all of the aspects of technology it is using to put things together, but they did improve upon existing tech and brought it all together. 

THE SURGICAL ROBOT

The Neuralink robot doing an implantation

Neuralink works by extracting electrical signals from neurons. In order to do that it has to implant extremely thin wires in the brain to access and read those signals onto a nanochip. Before it can do that, it has to make a precision surgical robot machine operated by neuroscientists to insert the wires - so it did. Apparently there was some DARPA money thrown into the mix somewhere along the way. The robotic surgeon has to implant tiny wires in between blood vessels and neurons, not on them. This can only be achieved through microscopes and nanometer precision. The implant needle is 24 microns small. Tiny threads are about 1/10 of a human hair, which is about the same size as a neuron. The needle to implant the wire is 24 microns small.  You can open the skull, insert the threads,  put in a tiny chip, and then glue the skull shut. The chip functions as a wireless bluetooth signal.

THE WIRES

This picture shows thin nano threads pasted onto a fingertip. The thinness of this wire is the big issue DARPA was trying to solve with Moldavian wire from Paradromics, that I previously blogged about in 2017, but Musk beat them to the punch. He wrote about his breakthroughs here.

This is the different new design of the threads called a “linear edge”, which are made of layered polymers. They are so are super duper thin they can’t bee seen with a human eye, so they need the special robot that inserts the threads onto the surface of the brain. Another reason is the brain’s surface moves with inhalation and exhalation, and the robot  can account for this natural movement. The wires and chip have to record the output from neurons. They are micro fabricated as precisely as the size of an electron beam. It is important to separate the signal to noise ratio in the chips, as they work with nanometer sizes of light. A new design. 350 nanometers,is smaller than visible light.

This image is the crux of how Neuralink acquires the signal. Here you see the neuron sending out an electrical spike and a thread next to it picking up the impulse. The electrical spike is the bulls eye. The copper colored needle is actually the implanted wire thread. You need to be 60 microns away or less to read the signals, so you really need to be under the skull. This is a graphic representation

This is a photo of the real thing, with lots of wires precision implanted into brain tissue. If you look really closely you will see they skirt around the blood vessels and neuron branches, but don’t touch any of them. Its sort of like an amazing game of darts, but the goal is to miss the bull’s eye of the vessels.

HISTORY OF IMPLANT CHIPS

This is a short history of brain research about making chips to implant in the brain. At this point the Utah array is still the most used implant chip in academia. Neuralinks chips are way, way faster and smaller. Their research builds on a century of neuroscience research and a decade of neuro engineering research. More advanced applications with advanced innovations will follow.

CHIPS AND PODS

These are early iterations of chips and devices made by Neurlink. Looks like Arduino 101, more or less.

The N1 sensor - the beginning of the sophisticated bean sized implant that goes into the skull and contains the chips.

The actual chips are made on super thin nano wafers in a hermetic environment so there is no dust. They are also super duper thin. A 4 x 4 millimeter chip has a thousand electrodes,and  implanting up to  10 chips is feasible. At this point the best FDA approved chip implant for Parkinsons Disease only has 10 electrodes. The Neuralink chips read and write, and are 1000 times more powerful than what is publicly approved. They will get better with newer versions.

Shades of Cyborg Neil Harbisson and his Eyeborg!  The Neuralink implant with four chips receive information from the threads, and sends them to an output area for batteries and firmware. 

This is the size of the output piece, like an earphone.You can upgrade the firmware on the pod on the ear, it is not the actual implant but connects to it. It will be controlled through an iPhone app. Probably Android as well.

SIGNAL PROCESSING

Waves of neural spikes from an array from implanted  threads that are being read out on a computer. monitor The color screen shows the brain at work, and traces of electrodes from single threads. Each trace is a voltage waveform in time. If you focus on one trace, it shows voltage deflections, or spikes per wave. It occurs when a neuron has an action potential, because that is the core information that is recorded.  Then the algorithm is decoded, which means capturing the intended information. You just have to think about something and build up the decoding data from that information or thought, and you can begin to interpret movement, memory, and  many other different types of experiences.

This is a basic diagram of signal processing from the chip - how they get the stuff out of the wires and threads to actually show the spikes. Neuralink said “Everything we care about is contained in the statistics of spikes (inside the brain).” So it goes back to the basic IBM punchcards, or the player piano drum roll - decipher the pattern and you decipher the thought.

Here is their basic logic analogue to digital conversion that is necessary to change action potential spikes to computer code..Calling Alvin Lucier, John Cage, Nam June Paik, David Rosenboom, Richard Teitelbaum - or Duh, this has been going on with the brain, changing analog to digital since 1965 in the music world.

Spike rasters in the brain - The top is the brainwave spikes pointed at by the blue arrows, and the bottom is the beginning of pattern recognition of individual neurons. There should be one pixel per electrode. They are on-chip spike detectors. The methods for detection are thresholding signals or directly characterizing the shape. The Neuralink scientists claim they can identify different neurons from same electrode based on their shape. The engineers had to modify the algorithms and scale them to compress neural data up to 200 times. It takes only 900 nano seconds  to compute the signal, faster than the brain knows that the signal even occurred. They can also stimulate any combination of up to 64 channels

One good use of this in the next few years is creating visual feedback for the blind by targeting the visual cortex to create an image better than a dot matrix image - or computer vision basics meets brain wetware. The scientists want to not only read out, but read into the brain. You can read into the brain by passing a current in the electrode. This causes the cell to fire an action potential, like for cochlear implants, or a way for the eye to restore vision. You can also use this technique in the brain to restore the sense of touch or vision. The visual cortex has maps, a spatial map (orange section of the brain in graphic). If you stimulate a point in that area, a blind person sees a point of light or phosphene. The idea is you can stimulate areas of the brain in the visual cortex to resemble a dot matrix level of the world. There are also parts of the brain that control orientation, color, size and speed of moving objects, and once you figure out what they are, and where to stimulate them, you can generate a more comprehensive image that a blind person can experience. Neuralink wants a device with electrodes that are small enough, but with high density that can do better than a dot matrix image.

The first iteration of their implant will have three different types that can go from a mobile device to a mouse or keyboard on a Bluetooth signal.  Neuralink needs to get FDA approval. Right now they are working on patients with complete paralysis so it is for serious neurological needs, and idea is to make it really safe. 

Who needs Google maps when you can tap into someone’s brain algorithm in their Hippocampus which contains spatial orientation. Here is a rendition of someone who really knows San Francisco, and they can send you pattern signals as you wander through the park and direct you telepathically - voila!

THE PURPOSE

Musk gave a number of reasons for his very public video presentation. The first he shamelessly admitted, was to recruit for new talent to Neuralink. He then framed the motivation of his company as wanting to solve brain ailments, spinal disorders, or catastrophic injuries like a broken neck or spine,. He admitted it won’t happen quickly, and kept mentioning the need for FDA approval. He wants to make his devices as cheap and accessible as a Lasik-like device.

“Hopefully,” he said, “AI is a benign scenario. You can chose to have a neural implant if you want, it is not a mandatory thing... We already have digital super intelligence - our access to a computer and a smart phone. The input speed in the human brain is fast, due to our vision, but the output speed is slow because we have to type information into a computer. We are constrained by human bandwidth and mechanics. “ 

But then he dropped the Neuralink AI bomb -  that he believes that we “ultimately (will) achieve a symbiosis with artificial intelligence” (if you want it at a civilization level scale). Then he added, “two people with Neuralink could have telepathy, a new kind of communication, conceptual telepathy, it has to be consensual”.

So the new “me too” movement will concern having consensual telepathy #metooconsensualtelepathy.

Using just optogenetic light scientists at NorthWestern University changed the social behavior of mice.

For the first time ever, Northwestern engineers and neurobiologists have wirelessly programmed -- and then deprogrammed -- mice to socially interact with one another in real time. The advancement is thanks to a first-of-its-kind ultraminiature, wireless, battery-free and fully implantable device that uses light to activate neurons.  This is the same tech used for wireless payments. Interesting.

This paper represents the first time we've been able to achieve wireless, battery-free implants for optogenetics with full, independent digital control over multiple devices simultaneously in a given environment," said Northwestern bioelectronics pioneer John A. Rogers, who led the technology development. "Brain activity in an isolated animal is interesting, but going beyond research on individuals to studies of complex, socially interacting groups is one of the most important and exciting frontiers in neuroscience. We now have the technology to investigate how bonds form and break between individuals in these groups and to examine how social hierarchies arise from these interactions."

"As they move around, the fibers tugged in different ways," Rogers said. "As expected, these effects changed the animal's patterns of motion. One, therefore, has to wonder: What behavior are you actually studying? Are you studying natural behaviors or behaviors associated with a physical constraint?"

When mice were physically near one another in an enclosed environment, Kozorovitskiy's team wirelessly synchronously activated a set of neurons in a brain region related to higher order executive function, causing them to increase the frequency and duration of social interactions. Desynchronizing the stimulation promptly decreased social interactions in the same pair of mice. In a group setting, researchers could bias an arbitrarily chosen pair to interact more than others.

"We didn't actually think this would work," Kozorovitskiy said. "To our knowledge, this is the first direct evaluation of a major long-standing hypothesis about neural synchrony in social behavior."

You have to pay for the paper, but it is here,

Which begs the question. Can mice become best friends?

Pierre Huyghe, via Serpentine Galleries/Kamitani Lab/Kyoto University and ATR

The artist Pierre Huyghe has made, with the help of Kamitami Lab/Kyoto University and an fMRI, art works of voxel and machine learning representations of reconstructed images of how the brain perceives the world. This is a development Planet 3D has been tracking for years, and its great to see it made it into the big time.

Katami Labs in Kyoto states on their homepage, “Brain signals can be seen as “codes” that encode our mental contents. We study methods for modeling brain functions and representations using information science and technology including machine learning and artificial neural networks. Our approach is based on data-driven predictive models that relate brain data and mind states via analysis of massive neural, behavioral, and multimedia data. Using these models, we aim to understand basic principles of neural information processing, and seek to develop real-life applications such as brain–machine interfaces that exploit decoded brain information.” The newest incarnation of the lab came into existence in 2015, and this is the place Huyghe collaborated with for his works.

An image from the Kamitami lab of a swan. Love the green head.

The artist never shows the original of what the images are all about, but this type of voxel reconstruction has been around since 2011 at Jack Gallant’s Lab at UC Berkeley. 

The actual steps in using an fMRI to code and decode an image according to Katami’s Lab, including the new DGN, or deep generator network

The video that shows the images over time can be accessed here -  The lab says, “To reconstruct visual images, we first decoded (translated) measured brain activity patterns into deep neural network (DNN) features, then fed those decoded features to a reconstruction algorithm. Our reconstruction algorithm starts from a given initial image and iteratively optimizes the pixel values so that the DNN features of the current image become similar to those decoded from brain activity.” The actual paper written in 2017 states, “Here, we present a novel image reconstruction method, in which the pixel values of an image are optimized to make its DNN features similar to those decoded from human brain activity at multiple layers. We found that the generated images resembled the stimulus images (both natural images and artificial shapes) and the subjective visual content during imagery. While our model was solely trained with natural images, our method successfully generalized the reconstruction to artificial shapes, indicating that our model indeed ‘reconstructs’ or ‘generates’ images from brain activity, not simply matches to exemplars.”

Reconstruction of learning datasets for images. The more simple the image (circle) the better the reconstruction

Here are the same images without the DGM (Deep Generator Network) sophisticated smoothing.

Katami’s Github it really fascinating. He gives all the data away, and anyone could theoretically emulate this if they had a nifty fMIRI, which I for one don’t. 

Artificial shape reconstruction including the alphabet

What is amazing is how far they have come in emulating basic shapes and letters. Especially letters. This is what Facebook Building 8 has been going ballistic about. Now its here. So “Is There A Place In Human Consciousness Where Surveillance Cannot Go?”. Hmmmm

Image from [email protected] - note the cute butterfly to make it a warm and fuzzy picture, instead of a despicable Dr. Doom-like image

DARPA wants to write in, or augment your memory code, a development they have been pursuing since 2013 with the Restoring Active Memory Program (RAM). This also includes bolstering the efforts of their collaborators. Robert Hampson, a professor at Wake Forest Baptist Medical Center says that  that without memory, one loses a sense of self. In a compelling experiment he was able to achieve “the first successful implementation in humans of a proof-of-concept system for restoring and improving memory function via facilitation of memory encoding using the patient's own hippocampal spatiotemporal neural codes for memory.”  

Feeding back an exact pattern match of memory from Hampson’s lab

To do this, he focused on epilepsy patients, because many in that population work with therapies using electrodes already implanted in their brains to control seizures. This approach was preceded on the success of numerous animal studies demonstrating how the techniques of pattern matching neurons worked. While undergoing hospital medical procedures, the study asked patients to play video games that involved memory activities. The researchers recorded the activity of specific clusters of neurons in the hippocampus. When they tested patients using a pattern they had extracted from their very own hippocampus, not a randomized pattern, they could improve short term memory in about 35 percent of test subjects. The goal of the study is to help people with stroke, brain injury and memory loss from the diseases of aging. 

The shape of neurons (green) and the synaptic connections of the neuron (white) of a zebra fish, part of ongoing research at University of Southern California on memory. 

The way the study was carried out is the subjects were shown a photo with an image. After a short pause they were asked to choose the photo out of group including other random photos, as per the photo of someone choosing the monarch butterfly above. During this time span their neural patterns were recorded, focusing particularly on when they selected the correct answer.  Then, after 75 minutes the subjects were shown a new set of photos, one of which included the original photos. When the subjects were stimulated with the frequency of their own patterns of correct answers, they were able to identify the correct photos with a 35 percent higher accuracy for this type of short term, episodic memory. These patterns are referred to as “spatiotemporal firing of neural ensembles”. 

Figuring out MIMO Models in rats, picture from Frontiers in Neuroscience between two different rats

This is not entirely new research, as DARPA and their affiliates have already been testing it on individual humans since 2015, and research on rats in 2013 and 2011. Scientists discovered short term memory in humans was like a type of RAM computer memory.  The memory is created from sensory information in the  hippo­campus that also includes the critiical perception of space and time. That consortium of information is held briefly, and if it is accessed again and again it becomes more of a long term memory. The neuron signal moves from a mapped section of the hippocampus called CA3 to a different section, CA1. It is by recreating the CA3 to CA1 signal that an increase in short term memory occurs.  This pattern, once understood, can be recreated with an algorithm.  This algorithm and process is called the MIMO model, or multi-input, multi-output nonlinear model. If you want to figure out how to make one of these MIMO models, this publication from Sweden is very helpful, but not something you can assemble on your own.

Mary Lou Jepsen CEO of OpnWatr does artist James Turrell - or is she really just inside a colossal, colored fMRI?

It began with musician Peter Gabriel thinking about a day in the not so distant future when the world will morph into  a matrix of ‘visible thought’. He named that day “Open Water - the Internet of Visible Thought”.  

Gabriel thought long and hard about this idea, referencing two sources to arrive at his conclusion. The first was the work of Berkeley professor Jack Gallant who is known for mapping visual thoughts, or what he calls the ‘semantic brain’, a topic I blogged about previously on Planet 3D here. His second source involved the work of researcher Mary Lou Jepsen, formerly of Facebook/Oculus, Google X, Intel and MIT Media labs, where she has worked for decades on LCDs and optical displays. Gabriel realized that if Gallant’s work could be enhanced with “optical techniques (and) the advantages of consumer electronics can really assert themselves, the power of AI algorithms will do the rest. This science-fiction future (of reading thoughts) is not only realisable, but because of enormous potential benefits, will inevitably be realised.” 

Mary Lou Jepsen speaking at EmTech, MIT.

He suggested Jepsen  create a company, which she did and graciously named  ‘Openwater’  or ‘OpnWatr’, right after she resigned from  Facebook.  OpnWatr aims to transform cumbersome  brain scan images made in enormous and expensive fMRIs into consumer sized devices. It does this through using diffuse optical tomography combined with new techniques in depth, resolution, size, and cost. 

Jepsen too, was applying the research of Alexander Huth at Jack Gallant’s lab on the semantic brain, calling it ‘telepathic’. The lab’s research had also inspired the use of massive quasi-ballistic photons accelerators that Facebook Building 8′s former director Regina Durbin raved about as a new way to read the semantic brain before she bailed on FBB8 for unknown reasons, something I also covered previously in Planet 3D.

Jepsen wants to up the optical resolution of the semantic brain, making it a billion times stronger in order to track the ‘heartbeat’ of neurons themselves. To do this she had to invent special LCDs or liquid crystal displays and detectors, combining them with special infrared light sources. This process enables the body, and especially blood flow to become translucent. It is also capable of creating a hologram of the brain by inverting the infrared images through the  computational process of phase conjugation. Phase conjugation is a mathematical calculation that makes the brain appear transparent down to resolutions of 1 micron. It looks at voxels of neurons, something neither an MRI nor fMRI can accomplish.

Traditional big, cumbersome fMRI machines. Totally old school.

Jepsen, a survivor of a brain tumor, is adamant that she does not want to shock, stimulate, or drill in anyone else’s brain, the opposite of what a lot of other researchers, especially those in the military are proposing for brain-to-brain communications. Instead she is making an LCD enabled ski-hat, or bandage at consumer electronics price-points. Her goal is to make it “1000 times lower in cost, a million times smaller in size” than an fMri. Her investors have given her a five percent chance of doing it, which is good enough for Mary Lou.

Typical infrared beams go through an optical device with just a few being picked up (orange line). The rest are scattered. This is old school

 Using  this process Jepsen is awed by the possibility of  dumping dreams directly into a repository, and unequivocally states it will be used for this type of telepathy in less than 10 years. She is also acutely aware of the profound ethical and legal implications of attempting this. The issues of who can make you wear a thought revealing headset, such as law enforcement, the military, your parents, your spouse, etc., is an open question. Such a device could also be used both for healthcare, and a personal enhancement tools, and predicts the software development will take longer than the hardware development. The end product will be limited mostly by signal to noise ratio, a problem that will eventually be solved. 

Image from Nature Magazine on phase conjugation speed affecting the quality of optical focusing of infrared light. Moving towards the new school.

Jepsen also looked very closely at the research in the work on optics of Dr. Lihong Wang and his team at Washington University in St. Louis. They were able to harvest not just one beam of infrared light, but all the normally scattered infra red lights to make a more comprehensive image. According to Washington University’s newsletter “The new TRUE technology combines two techniques: focused ultrasonic modulation and optical phase conjugation. Researchers use a type of mirror to record then time-reverse the ultrasound-modulated light emitted from the ultrasonic focus to achieve the best focus.” They accomplished this by using a pretty complex series of imaging and mirrors and a whole lot of mathematical computations.

How neurons actually scatter light, but now with holograms you can see them more clearly

Image from Nature Magazine better focusing of infrared light mimicking its use on types of tissues - very new school.

Using a holgraphic LCD, scattered infrared rays of light are all recaptured through a voxel (in the center) onto a detector, enhancing resolution - new school interpretations.

Putting a flashlight up to your hand in the dark to see what shines through. Very very old school.

Wang’s team used the technique on a mouse ear and got pretty good results, though all they show in the diagram below is graphs and dots. I would really like to see the EAR, of the mouse and what it looks like, not the photo of the poor little mouse’s head held down by all all those metal thingeys. Its actually a sort of  fancy adaptation of putting a flashlight up to the palm of your hand in the dark.

Image from Nature Magazine - optical focusing on the little mouse’s ear. Very new school, despite what the mouse thinks.

Holograms are created from wavelengths of light and be turned into super high resolution images using with LCD chips. They can even focus the light to the size of one neuron which has previously been unobtainable. This technology could eventually replace fMri’s. Right now OpnWatr can focus at a 1000 pixels a second, but has plans to make it much faster. It turns out to be  fashionable, incorporating the analytics of big data and AI software, as well as the ready made supply chain trillion dollar LCD business from China. 

Pretty looking bendable LCD holographic material. The newest of the new.

The main idea is to have this product embedded in a beanie cap or for internal processes, a bandage you can wrap around your stomach. No needles, surgery, pain, invasion or mess. So seeing into your brain is coming to a movie theater near you in the not so distant future. Better get ready.

Tiera and friend get ready to have their brains analyzed

It was only a matter of time starting from when Marion LeBorgne’s original experiment Cloud Brain, held at the Exploratorium in 2015 that dealt with live time brain data analytics went big time, but I really did not expect it to happen in just two years It has now become Intheon - NeuroTechnology, Anytime, Anywhere

It seems Tim Mullen’s well funded company Intheon, formerly known as Qusp (like the artist formerly known as Prince), which riffed off LeBorge’s breakthrough open source experiment hit pay dirt when it team up with Terra Mater Films and Fathom Events to produce MindGamers. MindGamers is a sci-fi movie about a group of students who discover a way to link minds together through a wireless neural network.  

The movie MindGamers and the accompanying MindGamers neurosensors

Tim Mullen at MindGamers premier, footage courtesy VOA

 Fun science experiments where someone else controls your brain forcing you to pull the trigger -  Note: kids do not try this at home (MindGamers)

At the premier, 500 people in Los Angeles and in 500 in New York wore a brainwave EEG headset that let Mullen and his team at Intheon read the electrical activity inside the brains of 1000 participants simultaneously.  After putting on the headset each person was given a small Android tablet to plug it into. A running app then notified the participant it was streaming their brainwaves. Cloud technology measured and recorded measure EEG patterns of focus and how it worked throughout different scenes in the movie.

Interested audience in L.A. with brain sensors watch a movie on mind control

The good old days of just watching movies in 3D without brain monitoring

Audience in N.Y. way more laid back than in L.A., watching their collective brain’s response (photo by Dan Krauss)

What was most interesting is that the government back publication Voice of America (VOA) pushed this article out with exclusive footage, and not a more traditional Hollywood source like Entertainment Tonight. Is this part of the silent but very active arms (brains) race between Facebook, Neuralink, and DARPA/IARPA or even perhaps an unknown something else?

Hot colors (red, orange) mean a big increase in focus. This correlated with a scene in the movie when a woman was going to jump off of a building. It sort of looked like the original Cloudbrain data experiment.

Cloudbrain at the Exploratorium did it first in 2015 . Here  people walk around with EEG brainwave sensors while looking at their group data

This is an Emoshape Affective Toy robotic doll you can construct in 10 minutes that understands your emotions. 

This is a totally wired up person having all their thermal image vital signs registered in Dolby Labs. This does not take 10 minutes.

2018 is going to be the year that biometrics, robotics and consumer technology are going meet and greet all over the place.

Emoshape

Emoshape, hails itself as ‘emotions next frontier’ uses a chip with tons of processing power to work with language to generate EPU (emotional processing units). It uses a patent pending technology that creates emotional states and synthetic emotion in intelligent machines. It claims to distinguish between twelve primary emotions:  anger, fear, sadness, disgust, indifference, regret, surprise, anticipation, trust, confidence, desire and joy. The emotion recognition classifiers says it achieves up to 94 percent accuracy on conversation. It does this with a microcontroller using real-time appraisal computation with reinforcement learning.  An Emotional Profile Graph (EPG) computation functionality allows the AI or robot to experience 64 trillion possible distinct emotional states. Emotional stimuli is stored within the memory bank through emotional patterns or fingerprints.

64 trillion emotional states? Like I don’t have enough to worry about? What does having 64 trillion emotional states actually feel like? It exhausts me just thinking about it. 

12 emotions graphed in realtime - great representations for the synesthesiac crowd.

How the colors are assigned - although I beg to synesthesiacally  differ

It uses “sound waves of our consciousness” (not sure about that one). To drill down further, the Emoshape has two recognition layers that work at the sentence level and uses an form or evolution of the Ekman emotion classification, first developed in 1993. From what I have seen a lot of people are now redefining that classification, but it does give a basic structure to operate within. It then incorporates a  semantic level that refers back to the Emoshape API cloud service  based on Patrick Levy-Rosenthal psychobiotic evolutionary theory. I wondered what that was, so I found an article from Trends In Neuroscience which said, “Psychobiotics were previously defined as live bacteria (probiotics) which, when ingested, confer mental health benefits through interactions with commensal gut bacteria.” Interesting. The Neuroscience article goes on to say “Much psychobiotic research is based on rodent models.”

Original Chatty Cathy Mattel dolls and their wonderful wardrobes

Pull the string - and she says 11 different things!

The buffers of how the emotional stacks are weighted and assembled in Emoshape. 

However, this part looks quite complex and well thought out. And the flow chart below absolutely amps it up.

But still, it all seems semantically based, like a huge taxonomy of words though there is an API for physical interaction, and a psychophysics API, which at this stage seems to mean a relationship between mental phenomenon and physical stimuli - a pretty wide net for sure .

Dolby Labs

Dolby Labs has been around since 1965, but they now place a high premium on people watching movies. This is so they can show their clients that certain colors, sounds, or frequency ranges will elicit a certain response. Neurophysiologist Poppy Crum run the labs. Affective computing is not new news, but the entertainment industry is now all over this one. Netflix and Hulu at this point use eyetrackers.

Poppy Crum at Dolby Labs

But what do these biosensors really talk about? Well, to begin with a subject is outfitted with a 10/20 brain EEG Cap.

Brainwave EEG headset being fitted on TV host Lauren Goode

 Muscle detector and lie detector also being booted up. Thermal images are also captured (see beginning of post)

Raw EEG, GSR and heart rate are all displayed livetime.

Some of the things they discovered that if you are shown a huge screen with flames, your body begins to flush - and that apparently helps shape the stories.

As immersive environments or ‘mixed reality’ grows in influence, it is logical that tracking the emotional responses of users needs to move along as well.

Disney

Disney is now so all over affective emotions and measurements they have teamed with Cal Tech to start using Artificial Intelligence. They work with what they call  “factorized variational autoencoders,” or FVAEs, which after observing an audience member’s face for just 10 minutes,  can predict how a person will react to the rest of the movie. 

It can actually work on any type of time durational data - "We are all awash in data, so it is critical to find techniques that discover patterns automatically," said Markus Gross, vice president at Disney Research. "Our research shows that deep learning techniques, which use neural networks and have revolutionized the field of artificial intelligence, are effective at reducing data while capturing its hidden patterns."

Singing Christmas carols - can you pick out the unhappy one? Bonus points - Disney Research is based in.....Pittsburg - with free research trips to Disney Park

Halloween costume Neurable

Planet 3D has been tracking immersion and biometrics, specifically BCIs for quite a while. Using BCIs as controllers is not new news. Using BCIs for VR is not new either. But now companies are entering the commercial sphere, and that promises to be interesting. Neurable has recently launched, and its certainly ain’t pretty.

First date night Neurable with HTC Vive

Neurable says “Our approach is science-driven and cross disciplinary, incorporating elements of neuroscience, biology, statistics, machine learning and design to create the ultimate user interface. With Neurable, mixed reality can finally achieve its full potential.” Hmmm, what, exactly does that mean?

Well, they use dry electrodes, which at least saves on the cost of conductor paste or saline solution and clean up gear. It’s wireless, and according to MIT Technology Review “The startup expects to offer software tools for game development later this year, and it isn’t planning to build its own hardware; rather, Neurable hopes companies will be making headsets with sensors to support its technology in the next several years.” So they have already opted out of the hardware, which means no messy import snafus with factories and their deadlines  in China.

Their headset seems to work with Event Related Potentials (ERPs) which are difficult to interpret. They are using machine learning to cut down on processing time (I would love to see the hard stats on that!) and they are trying to move away from Alpha, Beta, Theta, Delta, Gamma waves, that don’t tell you much in terms of interpretation, except that they are registering Alpha, Beta, Theta and Gamma waves. . I am not sure what their ERP is about, unless it is the old fall back of the P-300 Mermer signal that Lawrence Farwell and his cohorts in the military cooked up decades ago, when Farwell discovered that when you recognized something (like your mother) your brain spiked a recognition signal in 1/300 of a second (hence the 300 number). Actually, that is pretty much what Neurable did. Their analysis is based on a paper their founder RE Alcaide-Aguirre published in the Journal of Neural Engineering titled “Novel Hold-Release Functionality in a p-300 brain computer interface.” What this means is hold and release, or in layman’s terms start and stop accuracy selection rate was enabled in the p-300 signal - an enormous hurdle for these types of apparatuses.

Mom and Dad can I have one of these for Christmas?

Apparently Neurable also incorporates eye-tracking technology from German firm SMI, so its a one-stop shop for monitoring your vitals. The machine learning aspect is how it intends to work with the patterns of your individual vitals, as opposed to everyone else’s.