<?xml version="1.0" encoding="UTF-8"?><rss xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:atom="http://www.w3.org/2005/Atom" version="2.0" xmlns:cc="http://cyber.law.harvard.edu/rss/creativeCommonsRssModule.html">
    <channel>
        <title><![CDATA[The Substrate - Medium]]></title>
        <description><![CDATA[Storytelling in Neurotechnology - Medium]]></description>
        <link>https://medium.com/the-substrate?source=rss----10fba766a0f5---4</link>
        <image>
            <url>https://cdn-images-1.medium.com/proxy/1*TGH72Nnw24QL3iV9IOm4VA.png</url>
            <title>The Substrate - Medium</title>
            <link>https://medium.com/the-substrate?source=rss----10fba766a0f5---4</link>
        </image>
        <generator>Medium</generator>
        <lastBuildDate>Thu, 19 Nov 2026 10:38:28 GMT</lastBuildDate>
        <atom:link href="https://medium.com/feed/the-substrate" rel="self" type="application/rss+xml"/>
        <webMaster><![CDATA[yourfriends@medium.com]]></webMaster>
        <atom:link href="http://medium.superfeedr.com/" rel="hub"/>
        <item>
            <title><![CDATA[On Fiction & Neuralink]]></title>
            <link>https://medium.com/the-substrate/on-fiction-neuralink-ae5fc8a774fe?source=rss----10fba766a0f5---4</link>
            <guid isPermaLink="false">https://medium.com/p/ae5fc8a774fe</guid>
            <category><![CDATA[neuroscience]]></category>
            <category><![CDATA[neuralink]]></category>
            <category><![CDATA[neurotechnology]]></category>
            <dc:creator><![CDATA[Avery Bedows]]></dc:creator>
            <pubDate>Mon, 31 Aug 2020 14:25:41 GMT</pubDate>
            <atom:updated>2020-08-31T14:17:48.815Z</atom:updated>
            <content:encoded><![CDATA[<p>But first, a hot take: is it just me, or does Neuralink’s robot sort of look like its logo?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/770/1*NY472Mwqh06GhDBzgQHT4w.png" /></figure><p>Anyhow…</p><p>Originally in the realm of fiction, Neuralink is finally materializing into the real. The company began its public journey as a story: a dream dreamt up by Elon in his Fremont dream factory and chronicled by the fabulously talented <a href="https://waitbutwhy.com/">Tim Urban</a>. Now, the company has real technology with real constraints because, unlike in the world of fiction, degrees of freedom in reality are limited.</p><p>Neuralink was announced through loudspeakers with a ~35,000 word <a href="https://waitbutwhy.com/2017/04/neuralink.html">blog post</a> on <a href="https://waitbutwhy.com/">Wait But Why</a> telling a marvelous story about the imperiled and soon-to-be-saved future of humanity. As the article puts it, artificial intelligence threatens the obsolescence of <em>Homo sapiens</em>, but we can secure our significance and safety by integrating biological intelligence and artificial intelligence. This, it explains, was the vision behind Elon Musk’s new endeavor: a neurotechnology company.</p><p>It wasn’t just the linguistic content. Tim Urban’s distinctive style — of which I’m a big fan — incorporates many cartoons, like the now-infamous brain in a wizard hat. In addition to the distant-future hypotheticals, these cartoons also added to the feeling of fiction, or at the very least of being far-fetched.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*XovM5hAHFwv4ly2M.png" /><figcaption>Brain in a wizard hat. Credits: <a href="https://waitbutwhy.com/2017/04/neuralink.html">Wait But Why</a>.</figcaption></figure><p>Fiction is wonderful. It inspires, it draws up strong emotions, and I would argue it has its own Platonic reality — but it’s also inherently hallucinatory.</p><p>Whereas an inherently fictional endeavor like a film will stay fictional, a real endeavor like a company that begins with fiction must obviously become real in order to succeed. If we look at the three “publicity” data points for Neuralink, we have:</p><p><strong>1) </strong><a href="https://waitbutwhy.com/2017/04/neuralink.html"><strong>Wait But Why</strong></a><strong> article.</strong> <em>Lays out a hypothetical vision of the future and verges into science fiction territory; Neuralink is the ultimate solution to the “problem” of biology.</em></p><p><strong>2) </strong><a href="https://www.youtube.com/watch?v=r-vbh3t7WVI"><strong>Summer 2019 event</strong></a><strong>. </strong><em>Shows off ambitious technology that can be implanted in multiple cortical locations simultaneously. Highlights future applications of novel technology in some detail. Keynote discussion on AI symbiosis &amp; existential threat mitigation (including bandwidth argument), and therapeutic applications are secondary to the storytelling.</em></p><p><strong>3) </strong><a href="https://www.youtube.com/watch?v=DVvmgjBL74w"><strong>Summer 2020 event</strong></a><strong>. </strong><em>Demos a scaled back device. Keynote immediately focuses on therapies. Product sloganeering: Fitbit in your skull with tiny wires.</em></p><p>The progression from the original 2017 article to the Summer 2020 event is distinctly an arc of reification: at each point, the degrees of freedom have reduced (which surely reflects a more continuous degrees of freedom reduction during day to day operations). This isn’t bad; it was inevitable, because any real entity that starts as fiction has to become real, or cease to be. Neuralink is maturing.</p><p>Though perhaps an abstract observation, I think it has real significance for the field. When Neuralink was more fictional, it was like a Schrödinger’s pig: it was and it wasn’t; it could be anything and everything. The towering stature of Elon and the implicit perception of Neuralink as infinite (<em>vis a vis</em> its fiction status) has impacted the investment climate in neurotechnology. Why invest in a brain technology when Elon’s already done it?</p><p>Often, when learning about a field from the outside, (royal) we look for core guideposts to align ourselves to. When Neuralink lived as fiction, its presence could expand to an enormous size and become the predominant aligning guidepost for the neuro-naïve (not meant as a knock on people; myself included, most people are naïve about most things, a fact largely attributable to the size of the set of things). As everyone in neurotechnology knows, the field is much larger than Neuralink. <strong>As Neuralink manifests into the real out of the fictitious, people outside of the field will begin to understand that, too.</strong></p><p>This will play to the advantage of other commercial and academic endeavors in neurotechnology. Neuralink has not developed the panacea of neural interfacing, as neurotech Twitter has been eager to kvetch about. They have (according to neural engineers I trust) done difficult integration engineering, and made a cool demo, and have a tough road forward of technical and clinical validation.</p><p>By becoming real, it becomes clear that Neuralink <em>is</em> certain things (effective systems integrators, adept at design and probably user experience, advancing a product forward) and <em>isn’t</em> (yet?) others (a near-term solution to major depression, vision impairment, neurodegenerative disorders, language decoding, and so on).</p><p>As someone formerly on the venture capital side of things and now on the storytelling &amp; strategy side, my advice is this: use this to your advantage. When an investor asks you about Neuralink, tell the story of its adoption of constraint and focus. You, too, have chosen constraint and focus, and it’s different from Neuralink’s. You have space for your story, and they for theirs.</p><p>To some extent, every business begins a fiction. Some of us start with slides, and some of us start with small books. As reality inevitably takes over and the degrees of freedom shrink, what grows are the specificity and differentiability of the businesses—therein lies competition, and therein lies an industry.</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ae5fc8a774fe" width="1" height="1" alt=""><hr><p><a href="https://medium.com/the-substrate/on-fiction-neuralink-ae5fc8a774fe">On Fiction &amp; Neuralink</a> was originally published in <a href="https://medium.com/the-substrate">The Substrate</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Thoughts on CTRL-Labs]]></title>
            <link>https://medium.com/the-substrate/thoughts-on-ctrl-labs-1dfff4a23799?source=rss----10fba766a0f5---4</link>
            <guid isPermaLink="false">https://medium.com/p/1dfff4a23799</guid>
            <category><![CDATA[augmented-reality]]></category>
            <category><![CDATA[neurotechnology]]></category>
            <category><![CDATA[brain]]></category>
            <dc:creator><![CDATA[Avery Bedows]]></dc:creator>
            <pubDate>Tue, 01 Oct 2019 20:29:36 GMT</pubDate>
            <atom:updated>2019-10-01T20:29:36.708Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/984/1*eOu2bE_SoyM4ES8bG-sbfA.png" /></figure><p>While sitting at dinner with an unsuspecting friend, my phone blew up with oddly-late messages from fellow neurotechnology comrades. <a href="https://www.ctrl-labs.com/">CTRL-Labs</a> announced their impending acquisition by Facebook and integration into Facebook Reality Labs (responsible for Oculus et al.), for enough money to buy my guilt-inducingly-expensive New York coffee for a meager 550,000 years. <em>Woah</em> — so very many questions. I figured I’d share (and consider this an open invitation to reach out and throw in your 100 billion cents-worth).</p><p>As a quick note before proceeding, aside from its pragmatic communication implications, I don’t much mind whether or not CTRL-Labs and other EMG technologies hold philosophical credence as true neurotechnologies; what matters is that they’re branded as such. It’s a debate that’s been waged elsewhere, so I’ll leave it be!</p><h3>Q: What does this mean for the neurotechnology field?</h3><h3>Consumer neurotech</h3><ul><li>Consumer neurotechnology has been properly legitimized: there will now be an entry in the technology vernacular for “that neuro company FB bought.” <em>Neurotech</em> will be a true category, and will probably begin to appear on the websites of the more daring venture capital firms. This is non-trivial: when Facebook purchased Oculus in March of 2014, what followed was a huge influx of investments into AR/VR (the successes of which have admittedly been questionable at best). As many VCs are quick to note, VC is a social dance, and billion-dollar surprise exits to a FAANG company is a powerful way to get that game started.</li><li>I’ve spoken to several entrepreneurs in consumer neurotechnology today and, as expected, all of them are excited. This is a win they can point to; a vision they can show might lead to a concrete outcome. It is and will remain motivating for entrepreneurs, their employees, and their present and future investors.</li><li>Even though I don’t personally mind it, this acquisition is going to cause a hassle in nomenclature. To actually understand how CTRL-Labs differs from a central nervous system-BCI requires a bit of basic anatomy/physiology knowledge. For people not in the know (i.e., almost everybody, including most VCs and engineers), this adds an additional layer of complexity/misdirection to how people will think about neuro. Absent detailed information, humans employ underspecified models of semantic domains; that’s what will probably happen in consumer neurotechnology. In the same way, due to my ignorance of quantum computing, any piece of information I learn about quantum computing will apply to the entire field of quantum computing, whereas in reality, it’s just another piece in a complex puzzle. I suspect the same will happen in neuro.</li><li>Control (as opposed to affective computing and the like) is going to be the flagship of neurotechnology, because that’s what <em>CTRL</em>-Labs does. It’s possible that investors, journalists, and technologists will, on average, fall prey to the same underspecified model trap and collapse all “use-cases” into control. (If you want to get a sense for how I create a taxonomy for consumer neurotech, you can skim <a href="https://loupventures.com/consumer-neurotechnology-volume-i/">this</a>).</li></ul><h3>Medical neurotech</h3><p>I’m willing to wager that more of the medical neurotechnology companies I know and will get to know over the coming months are going to emphasize the consumer-crossover angle. “We’re a medical device company now, but we’ll have a patient-facing app which would fit in nicely with the health data focus at [insert your company of choice].” It’s a reasonable notion, and it’s very reasonable for entrepreneurs and investors to be enticed by the idea; who doesn’t like the sound of a $1B pre-revenue acquisition?</p><h3>Venture capital</h3><ul><li>Well, for one, there’s been a large neurotechnology exit now. That’s enticing.</li><li>This is almost certainly going to put neuro onto the radars of generalist VCs and/or specialist VCs who just haven’t specialized in neurotech yet but have plenty of adjacent knowledge in consumer electronics, technically innovative hardware, medical devices, AI of various sorts, etc. This is a good thing, but neurotech founders should think carefully about how to approach the education of non-neuro (read: most) investors. Have primer materials prepared, have current investors on hand willing to help in the educational process, and practice explaining neurotechnology to people in a way that’s impartial and only as informative as your audience wants.</li></ul><h3>Q: What does this acquisition get Facebook?</h3><ul><li>A bunch of talent. A <a href="https://www.ctrl-labs.com/people/"><em>bunch</em></a>.</li><li>I don’t think this acquisition was about neuro; this acquisition was about winning AR/VR. As someone who previously worked on a fairly intricate VR interface, I’d argue the modern missing piece in AR/VR is the ability to have degrees of freedom of control commensurate to what we have over our laptops — i.e., the same DoF of our hands. That’s what Facebook got with this acquisition: the eradication of the barrier between VR and full-depth, intricate applications. To quote a close friend who works in AR/VR, in reference to this acquisition, “It’s an arms race out here” and Facebook just found its new arms in the arms (get it? EMG?).</li></ul><h3>Q: What are the characteristics of EMG use-cases? Pros/cons? Does it obviate the need for head-worn BCIs?</h3><ul><li>In principle, EMG will eventually give your computer access to anything it wants to know about your hands. If we’re constraining ourselves to the physical control mechanisms we’ve evolved to use, then hand kinematics reconstruction is as good as it gets.</li><li>EEG (or other noninvasive modalities), on the other hand, enable interactions we <em>haven’t</em> evolved; whereas as hominids we’ve always used our hands to manipulate things, we have not literally used our brains. The question is whether noninvasive sensing of control signals that don’t cooccur with actual or intended physical motion ultimately leads to as many degrees of freedom as hand manipulation, and what sorts of manipulations are better-suited to abstract control (brain) vs. evolved control (hands/arms).</li><li>EEG is better for motor impairment use-cases because it doesn’t rely on neural signals propagating down the spinal cord.</li><li>We still need central nervous system neurotechnologies to handle cognition and emotion. To appreciate the significance of this, consider the trend (with all its complex positives and negatives) toward understanding the mental states of users in finer-grained detail so that we can adapt technology, and even physical environments, to these states. There are a big class of problems that implicate cognition and emotion, and EMG won’t help us there.</li><li>So, no, CTRL-Labs does not obviate head-worn BCIs!</li></ul><p>These are my initial thoughts. They’ll change and become more detailed over time. Meanwhile, please do reach out and let me know your latest and great perspective — you can find me at “avery at loupventures dot com.”</p><p><a href="https://loupventures.com/research-disclaimer/"><em>Disclaimer</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=1dfff4a23799" width="1" height="1" alt=""><hr><p><a href="https://medium.com/the-substrate/thoughts-on-ctrl-labs-1dfff4a23799">Thoughts on CTRL-Labs</a> was originally published in <a href="https://medium.com/the-substrate">The Substrate</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Neuralink Event: Musings on Messaging]]></title>
            <link>https://medium.com/the-substrate/neuralink-event-musings-on-messaging-61dace55b29f?source=rss----10fba766a0f5---4</link>
            <guid isPermaLink="false">https://medium.com/p/61dace55b29f</guid>
            <category><![CDATA[neurotechnology]]></category>
            <category><![CDATA[neuralink]]></category>
            <category><![CDATA[artificial-intelligence]]></category>
            <dc:creator><![CDATA[Avery Bedows]]></dc:creator>
            <pubDate>Tue, 01 Oct 2019 20:27:17 GMT</pubDate>
            <atom:updated>2019-10-01T20:27:17.544Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/900/1*K-4A7_Y93RCsI7My5tzNSA.jpeg" /></figure><p>As anyone who holds at least one ear to the neurotechnology world undoubtedly knows, Elon took the stage on Tuesday night at the California Academy of Sciences to deliver an Elonian exposé on Neuralink’s progress over the last two years. The company’s primary public debut was two years ago, in April, 2017, when Tim Urban released one of his typical <a href="https://waitbutwhy.com/2017/04/neuralink.html">mega-posts</a> on <a href="http://waitbutwhy.com/">waitbutwhy.com</a>, the subject of which was an ostensibly unending and many-part discussion of Neuralink and Elon’s thesis about “the human/AI symbiosis,” or something to that effect. As most people in neurotechnology also likely know, the vision Tim Urban articulated was — as orator for Elon — that of a nearly indistinguishable fusion between biological brains and artificial ones. If humans don’t fuse with AI, the argument goes, then we’ll be overrun by artificially intelligent agents superior to our biological selves in every way. “Elon Musk Fuses Humans with Machines to Save Humanity” — the headlines basically write themselves…</p><p>Which now brings me to a couple of thoughts I want to share on the subject of Neuralink’s messaging, and what we (the neurotechnology-initiated) can learn from it. Many things have been said about Neuralink’s event and technology by many very smart scientists, and I’ve enjoyed and learned from their commentaries; I don’t have much to add on the technical details that hasn’t already been said better than I could say it. To that end, here are some links in case you’re in the market for rabbit holes:</p><ul><li>The <a href="https://youtu.be/r-vbh3t7WVI?t=5461">event</a> itself</li><li>An enjoyable <a href="https://twitter.com/andrewhires/status/1151319583438983168?s=21">Twitter analysis</a> by Andrew Hires</li><li>A <a href="https://www.youtube.com/watch?v=B2-YiXuXdp8&amp;feature=youtu.be">paper reading group</a> at Numenta</li></ul><p>Now, let’s talk about messaging. After the Odesza-soundtracked teaser video of Neuralink’s chic biotech-meets-tech office and its smiling scientists, the first thing Elon shows us is the following slide. In my opinion, this was one of the most interesting things in the presentation, and the speakers were clearly coached to speak along the lines of the framework laid out.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*31aqf5DEgXUfwL9VPBwQFA.png" /></figure><p>Here’s how these read to me:</p><ol><li>There’s no way we can build an invasive brain interface if we’re not going to use it for medical applications first, and many scientists and engineers are probably skeptical of the transhumanist vision, so if we’re going to hire good people, we need to start on planet earth.</li><li>Now, if we do manage to build an implantation procedure that’s extremely easy and safe to perform (their analogy is Lasik surgery), then we can probably start to hint to people that, hey, why not try this out for fun? Do it for yourself — preserve your brain health, and enhance it while you’re at it. Really, it’s the logical next step after eating blueberries for antioxidants and getting eight hours of sleep every night.</li><li>Alright, we can pull out all the stops. Let’s implant everyone and “connect” us all to the internet so we can remain valuable in an AI-ridden future.</li></ol><p>This slide, and indeed the rest of the presentation, contrasts substantially with the messaging in the Wait But Why article. The speakers were all cautious to distinguish between what they’re working on <em>now</em>, and what is futuristic and “aspirational,” to use the word of the evening.</p><p>Neuralink had two paths to go down: the first would have been to ride the hype machine, a la <a href="https://www.magicleap.com/">Magic Leap</a>, putting out mysterious marketing and keeping internal work secretive. The second option was to open up the science and tone down the marketing. They chose option two, and they pretty much explicitly said they did it for hiring purposes. I can’t be sure, but I suspect they hosted the event <em>reactively</em> to issues with hiring, rather than proactively. If indeed they hosted the event as a reaction, I’d be curious to know what the issues were, and whether or not they were a function of Neuralink’s previous flavor of mystery and secrecy.</p><p>As an additional musing, I think Neuralink did something positive for neurotechnology by bringing their branding down to earth. Understanding neurotechnology requires substantial background information: neuroscience, various engineering disciplines, medicine, regulation, etc. Most people in the general public have few or none of those topics in their toolbox, which means that neurotechnology is easier to understand in terms of sweeping generalizations. Compare this with Musk’s other companies: Tesla is making electric vehicles. SpaceX is sending people to Mars. The Boring Company is building tunnels for transportation. Neuralink is connecting humans to AI. Of the four one-line descriptions, the one that requires the most nuanced interpretation is undoubtedly Neuralink’s. Had Neuralink continued their public-facing transhumanist reductionism of neurotechnology, I would have worried that the general public and non-scientist governmental regulators would have begun to view neurotechnology as a wrecking ball, even more than some already do. Instead, though, Neuralink has primed the public and the media to appreciate how ridiculously complicated it is to develop neural interfaces, and how slow-going it will be. That should bode well in a climate of technology scrutiny.</p><p>As a proud owner of a shirt bearing the infamous <a href="https://images.app.goo.gl/24kAFVTnTS6mj6k4A">wizard hat-wearing brain</a>, it was gratifying to finally see an update from the company that undergirds much of the public interest in neurotechnology.</p><p><a href="https://loupventures.com/research-disclaimer/"><em>Disclaimer</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=61dace55b29f" width="1" height="1" alt=""><hr><p><a href="https://medium.com/the-substrate/neuralink-event-musings-on-messaging-61dace55b29f">Neuralink Event: Musings on Messaging</a> was originally published in <a href="https://medium.com/the-substrate">The Substrate</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[The Story of Rune Labs]]></title>
            <link>https://medium.com/the-substrate/the-story-of-rune-labs-47fddf1c9419?source=rss----10fba766a0f5---4</link>
            <guid isPermaLink="false">https://medium.com/p/47fddf1c9419</guid>
            <category><![CDATA[neurotechnology]]></category>
            <category><![CDATA[health]]></category>
            <category><![CDATA[neuroscience]]></category>
            <dc:creator><![CDATA[Avery Bedows]]></dc:creator>
            <pubDate>Tue, 01 Oct 2019 20:24:14 GMT</pubDate>
            <atom:updated>2019-10-01T20:24:14.536Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/1*f_1H8Fgk8LltTZqgVPtZ3w.png" /></figure><h3>The Story of Rune Labs</h3><p>The nervous system is the substrate of human experience. Logic would therefore predict, and experience would confirm, that some of the most phenomenologically pernicious diseases and disorders are those that implicate the nervous system. This specific insight is the motivation behind our commitment to <a href="https://loupventures.com/category/themes/neurotech/">neurotechnology</a> at Loup, and it’s within this context that I’m extremely excited to share publicly our newest neurotechnology investment: <a href="https://runelabs.io/">Rune Labs</a>!</p><p>Rune, under the direction of CEO Brian Pepin, is developing a software platform for brain data.</p><h3>Laying out some context</h3><p>If there’s one high-level trend I’ve identified in neuroscience and neurotechnology — via literature and via conversation–it’s that circuits are in vogue. The central and peripheral nervous systems are structures of dumbfounding complexity and faced with dumbfounding complexity, the classic human recourse is to abstract, abstract, abstract until we’ve reached a level of granularity that lends itself to comprehensible analysis and scientific investigation. A brain can be investigated at the level of a molecule, such as when studying the structure of a membrane protein. A brain can be investigated at the level of statistical characterization of a large and stochastically-interacting set of molecules, such as when studying ion fluxes and action potentials. A brain can be investigated at the level of microbiology, such as when studying cell migration during neural development. And, lastly and importantly, a brain can be investigated at the level of neural circuits. It’s this last level of abstraction that’s seen strong uptake in recent years, in both research and translational settings.</p><p>The neurocircuitry level of abstraction is prominent in basic science, but it also shows its face in clinical research: in the translational setting, the neurocircuitry view has yielded the field of neuromodulation. Neuromodulation is by no means new, but it’s growing in size and <a href="https://loupventures.com/neuralink-event-musings-on-messaging/">publicity</a>. A non-exhaustive list of modulation technologies that leverage the circuitry approach includes <a href="https://www.sciencedirect.com/science/article/pii/S0149763405000199">vagus nerve stimulation</a>, <a href="https://www.sciencedirect.com/science/article/pii/S0014488609001290">transcranial direct/alternating current stimulation</a>, <a href="https://www.sciencedirect.com/science/article/pii/S0149763417304840">transcranial magnetic stimulation</a>, and <a href="https://jamanetwork.com/journals/jamaneurology/article-abstract/802237">deep brain stimulation</a> (DBS). Here, we’ll focus on DBS. DBS has FDA approval for use in Parkinson’s Disease (PD), and to date, more than <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC6397644/">160,000 patients</a> have been implanted for PD and other indications. “Other indications” refers to the investigative work happening in, for example, treatment-resistant major depressive disorder, treatment-resistant obsessive-compulsive disorder, and addiction disorders (among others).</p><p>Original DBS devices — i.e., most of the devices currently implanted in commercial settings — only have the capability to stimulate the brain. This makes it challenging to customize therapy to individual patients, and impossible to design therapies that use realtime information about circuit activity to determine when/how to stimulate. A new wave of devices is under development that can both record <em>and </em>stimulate; these are referred to as “closed-loop” DBS devices.</p><h3>Confluence of problems</h3><p>Brian founded Rune to solve a confluence of problems. At the most abstract level, brain disease is complex because brains are complex; this means that any data recorded from the brain will be hard to analyze. Furthermore, in the domain of psychiatric disorders wherein the disorders are characterized largely by their qualitative, phenomenological symptomatology, there’s the added challenge of finding ways to measure phenomenological states — which is arguably impossible in a philosophical sense.</p><p>There are also more concrete problems facing the treatment of neurological and psychiatric disorders via neurotechnologies such as DBS.</p><ul><li><strong>Noise</strong>. Any brain measurements, whether electrical or otherwise, will contain some degree of noise. Neurophysiological data is notoriously noisy, meaning certain recording segments can’t be used since they’ll skew analyses.</li><li><strong>Longitudinality</strong>. PD, TRD, etc. all progress and/or fluctuate over long periods of time. PD, for example, is known to have a circadian fluctuation, which is notoriously difficult to capture through discrete clinical visits. Tracking these diseases/disorders must therefore require researchers to capture data over a long time span; this impacts clinical trial design, algorithm/treatment development, and therapy delivery in commercially-implanted patients. In addition to the idea of longitudinality having paradigmatic significance, there are practical concerns: large data sets are hard to work with. Some things that can contribute to the size of electrophysiological data are, for example, the number of recording channels, the sampling rate, and the duration of the recording. Closed-loop DBS, in particular, will create longitudinal electrophysiological recordings that haven’t existed before: with <a href="https://commonfund.nih.gov/sites/default/files/Medtronic%20v3_508.pdf">RC+S</a> being implanted and the Percept trial on the horizon, there is likely to be more invasive human neural data collected in 2020 than every previous year in history combined.</li><li><strong>Patient environment</strong>. The patient’s environment interacts with brain disease, and therefore environmental factors have to be considered and possibly tracked when developing and delivering therapies.</li><li><strong>Patient experience</strong>. If neural data comprises “measurements,” such as the data recorded by a closed-loop DBS system, then in order for a supervised learning algorithm to know how to respond to the measurements, it must decide how those measurements relate to positive patient experience vs. negative patient experience. This amounts to collecting labeled neural data from patients, and infrastructure to do so is generally poor.</li><li><strong>Complex feature spaces. </strong>Given a time series of neural data (or any other kind of neural data), there are many possible ways to transform that data to derive useful “features” for use by a learning or control algorithm. Handling large feature spaces, as well as selecting high-performing features, requires time (and therefore money) as well as computational power.</li></ul><p>Each of these categories can and do fill books. The conclusion, though, is simple: treatment of neurological and psychiatric disorders has miles and miles of road left to build and travel, and many of these miles will be paved by data.</p><p>Time series neural data (electrophysiology and/or fMRI) is so dense and complex that it isn’t human-readable. Furthermore, any model we have that goes beyond more than a few neurons interacting is necessarily a massive simplification and surely throws out many mechanisms. Brian believes these factors motivate the need to use modern deep learning approaches to understand the data and generate new insights. However, to pull this off requires a lot of data that is harmonized at some level by similar labels, sampling characteristics, etc. Data is siloed right now, sometimes even at the level of a single patient clinical visit, preventing harmonization — and that is where Rune comes in: Rune provides unified software interfaces to a large number of patients, researchers, and clinicians that solve the issues outlined above and allow the aggregation of datasets amenable to AI.</p><h3>Rune’s solution</h3><p>Rune is developing a software platform for the labeling, ingestion, cleaning, visualization, and analysis of intracranial brain data. In the technology world, data has scaled, and therefore so has the intricacy of the software engineering necessary to derive value from the data. Brian and his team are bringing the same strategy to neurotechnology.</p><p>In conjunction with a researcher-facing platform, Rune has launched a <a href="https://strive.group/">patient-facing application</a> for Parkinson’s patients to track symptoms; this data is fused with neural recordings on the backend to give physicians and researchers better insight.</p><p>Rune’s software platform moves the needle on the problems we discussed above:</p><ul><li><strong>Noise</strong>. Rune’s software automatically removes chunks of data that are too noisy to use.</li><li><strong>Longitudinality</strong>. The Rune platform is being engineered to handle longitudinal data efficiently; this will expedite R&amp;D of therapies that incorporate longitudinal changes/trends in patient behavior and neural data.</li><li><strong>Patient environment</strong>. Through the patient app, Rune can collect and supports the analysis of data corresponding to environmental variables of interest.</li><li><strong>Patient experience</strong>. The patient-facing application is designed from the ground up to capture patients’ experience with their conditions and to use that data to understand the impact of neurotechnological interventions.</li><li><strong>Complex feature spaces</strong>. Rune has various efficiently-implemented data transformations and visualizations built into the platform, enabling researchers to quickly investigate data and test hypotheses.</li></ul><h3>Where they are and where they’re going</h3><h4>Near-term</h4><p>Today, Rune launched their patient-facing app for PD patients, called <a href="https://strive.group/">Strive PD</a>. The app was designed by people with Parkinson’s, for people with Parkinson’s. It serves as a patient portal to information about their therapy and lets patients record relevant information such as their symptoms and whether or not they took their medication. If patients want, they can share the data with clinicians and caregivers to help find the right mix of therapy. In the context of new therapy development, Rune can adapt its baseline patient app to support new use-cases, including direct wireless connection to implanted devices.</p><p>Rune is also actively partnering with global leaders in the development of DBS therapies. Initially, they’re focusing on accelerating development of improved DBS therapies for Parkinson’s, and as time goes on they’ll expand to other indications.</p><h4>Long-term</h4><p>As with most things in our world these days, the more patients we help, the more patients we can help; in other words, we use the data from present patients to help future patients, and this cycle compounds exponentially as the datasets grow. It seems likely that psychiatric and neurological healthcare continue on their data-centric trajectories, which means there will be a tension between the convenience of non-invasive and/or proxy data — such as data collected from wearables, smartphones, and ambient sensors — and the “closeness to the source” of intracranial brain data. Rune’s long-term vision is to expedite and validate the development and commercialization of new neurological and psychiatric therapies by partnering with other researchers and businesses: as therapies get more complex and personalized, the primary innovation will be in the development of new software and algorithms — right at the nexus of Rune’s expertise.</p><h3>Conclusion</h3><p>If one thing is certain about the direction of brain science, it’s that the quantity and precision of data we have about the brain is only going to increase. Therefore, there is a strong need to leverage that data to develop therapies, and we’re thrilled to be working with Brian and his team at Rune to do just that.</p><p><a href="https://loupventures.com/research-disclaimer/"><em>Disclaimer</em></a></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=47fddf1c9419" width="1" height="1" alt=""><hr><p><a href="https://medium.com/the-substrate/the-story-of-rune-labs-47fddf1c9419">The Story of Rune Labs</a> was originally published in <a href="https://medium.com/the-substrate">The Substrate</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[An Investor’s Take on the Ethics of Neurotech]]></title>
            <link>https://medium.com/the-substrate/an-investors-take-on-the-ethics-of-neurotech-4e79f6159ccc?source=rss----10fba766a0f5---4</link>
            <guid isPermaLink="false">https://medium.com/p/4e79f6159ccc</guid>
            <category><![CDATA[ethics]]></category>
            <category><![CDATA[neurotechnology]]></category>
            <category><![CDATA[venture-capital]]></category>
            <category><![CDATA[neuroscience]]></category>
            <dc:creator><![CDATA[Loup Ventures]]></dc:creator>
            <pubDate>Wed, 05 Dec 2018 17:29:39 GMT</pubDate>
            <atom:updated>2018-12-05T17:35:30.349Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/820/0*WIfXhTv9UK9vSyn9.jpg" /></figure><p><em>By Avery Bedows</em></p><p>A while back, I laid out an <a href="https://medium.com/the-substrate/on-the-human-substrate-and-how-we-can-change-it-85e6bb8ca7cb">argument</a> for why I think neurotechnology is an important and compelling field: it entails investigating, fixing, and enhancing the brain — the substrate of human phenomenological existence.</p><p>Here, I want to discuss the ethics of that field.</p><p>There are futures of neurotechnology I want, and futures I don’t. I want a healthier population with solutions to neurological disorders and other diseases; I want the loss of sensory function to entail a minor surgery, not a lifetime of disability; I want technology that works in the background to keep people feeling good, without making them dependent on it; I want amplified empathy and feelings of social connectedness; I want people to have agency over their identity by having a choice to enhance or change themselves if they’d like, absent social pressure; I want people to feel secure and confident that the technology which understands and influences their brain is trustworthy and positive.</p><p>I don’t want a world where neurotechnology creates a persistent anxiety about invasion of privacy; I don’t want those forsaken science fiction worlds where mind control is a reality and we’re all plugged into machines; I don’t want a world where we collectively resent neurotechnology, where we reflect on it as yet another one of those failures of culture and demonstrations of non-prescience that show up again and again in history.</p><p>I know the following: neurotechnology can mitigate suffering and neurotechnology can expand the horizon of experience.</p><p>I don’t know where technologies will be in 20 years, 50 years, 100 years. I don’t know if science fiction dystopias will materialize or not, but I do know that neurotechnology progress is occurring and will continue to occur. The challenge I therefore see myself and our culture facing is to manage that growth in accordance with prevailing cultural principles; in other words, to be ethical.</p><h4>Introduction</h4><p>Any sufficiently large adjustment to a complex system necessarily entails both positives and negatives — say, for example, introducing Facebook (large adjustment) to human society (complex system). In the technology world, with colorful pitch decks, superlative value propositions, and nine-figure acquisitions, it’s very easy to believe and be led to believe these negatives are trivial — or worse, they don’t exist. It’s easy to believe unwaveringly and with strong emotional conviction that technology is exclusively “good,” despite its obvious nuance. Neurotechnology is either no different, or it’s different in that we need to treat it with even more care.</p><p>In 2002, William Safire offered a definition of <strong>neuroethics</strong> that’s now widely referenced, showing up in nearly every resource on neuroethics I’ve seen. Neuroethics, <a href="http://dana.org/Publications/pressbooks/Details.aspx?id=50134">he says</a>, is “the examination of what is right and wrong, good and bad about the treatment of, perfection of, or unwelcome invasion of and worrisome manipulation of the human brain.”</p><p>Safire challenges an implicit axiom of technocracies, that progress is good. To build up a neuroethics from first principles, we must first take Safire’s cue to challenge this axiom: is progress in neuroscience and neurotechnology good? (Hint: it’s messy.)</p><p>Let’s make this question more specific by first splitting neuroscience/technology progress into categories: progress in basic neuroscience research, progress in the translation of neuroscientific knowledge to medical applications, and progress in the application of neuroscience to augment normal human capabilities.</p><p>The easiest of these three to discuss is the <strong>application of neuroscientific knowledge to medicine</strong>. To argue that it’s bad to mitigate abnormal suffering requires philosophical gymnastics — and the abandonment of multiple cultural values. I’ll go so far as to make the claim that, for our purposes, applying neuroscience to medicine is unequivocally positive.</p><p>This unequivocal positivity of medical neuroscience applications makes it fairly easy to justify progress in certain basic neuroscience research areas: basic research informs and provides the foundation for commercially-driven medicine. Hence, we can reasonably accept that basic neuroscience research is good, at least insofar as that basic research pertains to medicine. There remains the outstanding question of whether, for example, research into the neuroscience of consciousness is fundamentally positive. It seems to be a durable feature of humanity that we ask (yet seldom answer) questions about human cognition, emotion, and consciousness. This is as true now as it was millennia ago. I think, therefore, it’s fair to say that humans have a fairly inherent interest in understanding themselves.</p><p>The pursuit of the neuroscience of cognition and consciousness is just one particular flavor of these age-old inquiries, and it’s not going anywhere. Despite its romantic appeal, neuroscientific knowledge has teeth: the same neuroscience of motivation has both been used to induce negative addictions (like an addiction to social media) and been applied to treat emotional and cognitive disorders. It’s hard to argue that progress to the ends of the former isn’t bad; likewise, it’s hard to argue that progress to the ends of the latter isn’t good. We have to accept this complexity as inevitable and offer principles of ethics to manage it.</p><p>This nuanced balance of negative and positive shows up in our third domain to consider: progress in the use of neuroscience to augment human capacity. The arguments in favor include increasing productivity; expanding emotional, experiential, and sensory horizons; reducing the suffering inherent to modern biological humans; increasing the long-term survivability of the human species; etc. The arguments against: inducing a massive social rift; producing <a href="https://www.amazon.com/Superintelligence-Dangers-Strategies-Nick-Bostrom/dp/1501227742">runaway artificial intelligence</a>; violating “the natural order of things”; etc. Here’s the truth of it: arguments for and against both have merit. In attempting to assess whether augmentative neurotechnology is positive, we’re running a complex calculus of beliefs, truths, and logic — a calculus which has different rules depending on which mathematician you ask.</p><p>For two reasons, I think that augmentative neurotechnology is more-or-less inevitable: first of all, modern humanity has uniformly moved in the direction of augmentation (c.f. arguments that natural language, algebraic symbols, the printing press, modern telecommunications, etc. are all “augmentations,” as it were). Second, people are <a href="https://en.wikipedia.org/wiki/Neuroenhancement">already working on neuro-augmentation</a>, and momentum of that sort is hard to stop. Therefore, as with the case of basic neuroscience research above, the matter at hand is not to determine the “goodness” of augmentative neurotechnology but to encourage augmentative neurotechnology to manifest in ways that minimize its downsides.</p><p>In short, the technocratic axiom that “progress is good” needs to be extended to “progress is good and bad.” My goal, here, is to explore the particulars of this tension and to propose steps for how we can <em>begin</em> to manage it through the vehicle of Loup Ventures.</p><h4>My Definition of Neurotechnology</h4><p>In other pieces, I’ve offered fuller definitions of <a href="https://loupventures.com/bioelectronic-medicine-an-introduction/">medical neurotechnology</a> and <a href="https://loupventures.com/consumer-neurotechnology-volume-i/">non-medical neurotechnology</a>. Here, I’ll partially recapitulate my previous definition of consumer neurotechnology since, as you’ll see below, this will be my ethical area of focus.</p><p>“Neurotechnology,” as used here, refers to devices/products that are explicitly designed to interact with neurons. These can be neurons in the brain, neurons in the spinal cord, neurons in cranial nerves, motor neurons in the body, or any kind of sensory neuron. What this notably excludes are products <em>informed</em> by neuroscience (or its close cousins, cognitive (neuro)science and psychology), but which don’t directly interact with neurons. In addition, neurotechnology includes the collection, storage, processing, and inference performed on neurodata — i.e., data collected from neurons (or other sources of information inherent to the brain).</p><h4>The Benefits of Neurotechnology</h4><p>Concretely, neurotechnology offers promise to solve medical problems implicating the nervous system — and since the nervous system is so critical to our biology, this <a href="https://loupventures.com/bioelectronic-medicine-an-introduction/">encompasses many diseases and injuries</a>. For the science fiction enthusiasts out there, consumer neurotechnology can enable consumers to control their technology in new ways; can enhance their learning; can assist in wrangling control over emotion. I’ve written about this extensively in Volumes <a href="https://loupventures.com/consumer-neurotechnology-volume-i/">I</a>, <a href="http://www.apple.com/">II</a>, <a href="https://loupventures.com/consumer-neurotech-volume-iii/">III</a>,<a href="https://loupventures.com/consumer-neurotech-volume-iv/"> IV</a>, and <a href="https://loupventures.com/consumer-neurotech-volume-v/">V</a> of a series on Consumer Neurotech.</p><p>Abstractly, opportunities in medical neurotechnology derive from the fact that the nervous system is an <a href="http://neuralimplantpodcast.com/dr-manfred-franke-simply-explains-basics-about-neural-activation-therapy">information pathway</a> in the body, and therefore it’s a strong candidate for therapies. Opportunities in non-medical neurotechnology are due to the brain being the ultimate frontier for mastering ourselves. It seems likely that the quest for self-mastery is inherent to the human condition — neurotechnology comprises the tools and mechanisms to discover and act upon our mastery.</p><h4>Therapies and Non-Therapies</h4><p>Without a doubt, the most widely used neurotechnologies right now are therapeutic, and it’s difficult to debate their value (such as <a href="https://www.nevro.com/English/Home/default.aspx">Nevro</a>’s neurostimulation to treat chronic pain, <a href="https://www.mayoclinic.org/tests-procedures/deep-brain-stimulation/about/pac-20384562">deep brain stimulation</a> to manage Parkinson’s symptoms, <a href="https://www.nidcd.nih.gov/health/cochlear-implants">cochlear implants</a> for deafness, and many more). Furthermore, both in the US and elsewhere, not only are medical devices and treatments heavily regulated, but various medical organizations have their own codes of ethics. The FDA has regulatory purview in the US over technologies or drugs that <em>make medical claims </em>(but not necessarily technologies that have biological effects, yet don’t claim to be a therapy), and <a href="https://www.hhs.gov/hipaa/index.html">HIPAA</a> governs the use (and penalizes the abuse) of health data. The American Medical Association has a <a href="https://www.ama-assn.org/sites/default/files/media-browser/principles-of-medical-ethics.pdf">code of ethics</a>, and <a href="https://www.ama-assn.org/delivering-care/ama-code-medical-ethics">additional opinions</a> on more specific matters. Pertaining to neurotechnology directly, the American Association of Neurological Surgeons has their own <a href="https://www.aans.org/-/media/Images/AANS/Header/Govenance/AANS_Code_of_Ethics_11-22-2014.ashx?la=en&amp;hash=124B159D6B41ACF78DFB0110EB55B10E68D5D3DD">code of ethics</a> as well.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*LpSLLEPjFrTQazsR.jpg" /><figcaption>Most current neurotechnologies are used for medical applications.</figcaption></figure><p>Therefore, even though the regulation is far from perfect and there’s a tradeoff between regulation and innovation, I feel confident that existing structures are suited to deal with the ethics of neurotechnology in medicine. Where I’m concerned is non-medical neurotechnologies, for which regulatory applicability is less clear.</p><p>Non-therapies have fewer protections. What’s more, the science fiction thought experiments both futurists and anti-futurists like to run implicate non-therapies more strongly than therapies: here, I’m referring to networked brains, digitally uploaded consciousness, etc. Let me be crystal clear: these don’t exist yet, and it’s possible they never will. But, technology that moves in this <em>general direction </em>has the most ethical pitfalls ahead of it, so that’s where I think it’s most important to help construct guidance.</p><h4>Enhancements</h4><p>Many non-therapies are <strong>enhancements </strong>(although an example of a non-therapy neurotechnology that <em>isn’t</em> an enhancement is, say, a performance test that uses neurotechnology to collect its information).</p><p>Employing what’s likely an imperfect taxonomy, enhancement can be split up into three categories: cognitive, environmental, and social/moral.</p><h4>Cognitive Enhancement</h4><p>“Cognition can be defined as the processes an organism uses to organize information” (<a href="https://nickbostrom.com/cognitive.pdf">Bostrom and Sandberg, 2009</a>). Some examples of cognitive subsystems — i.e., different aspects of cognition — include working and long-term memory, attention, executive processing, language, perception (e.g. auditory, visual), etc. Bostrom and Sandberg (2009) offer a fairly extensive overview of the various types of cognitive enhancement discussed in the literature. Examples of cognitive enhancement include education (evidence for this point: try naming an element from the above list of cognitive subsystems that education doesn’t improve), mental training (such as the <a href="https://en.wikipedia.org/wiki/Method_of_loci">Method of Loci</a>), existing computer software and hardware (e.g. data visualization, data search, and even the simple duo of keyboard and word processor), pharmaceuticals (e.g. Adderall, <a href="https://www.ncbi.nlm.nih.gov/pubmedhealth/PMH0052009/">Modafinil</a>), genetic modifications (<a href="http://medicine.jrank.org/pages/2283/Gene-Therapy-Ethical-Issues-Germ-Line-versus-Somatic-Cell-Gene-Therapy.html">either</a> non-transgenerational somatic gene editing or transgenerational germ-line gene editing), pre- and perinatal enhancement (supplementing a mother’s diet during gestation), neuropriming/neuromodulation (electrochemically modulating the state of the nervous system so as to induce a particular effect, such as faster learning), brain-computer interfaces (currently, these don’t yield much in the way of enhancement; however, in the future they could conceivably be the source of profound enhancement, assuming that future BCIs enable much higher information bandwidth in human-computer communication than modern BCIs allow), and sensory/perceptual prostheses (such as a retinal implant that allows the wearer to “see” electromagnetic radiation in the infrared spectrum).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*zg7iF1yb2j6JtGjg.jpg" /><figcaption>What happens if only a small percentage of the population has radical cognitive enhancement?</figcaption></figure><h4>Enhancement of Environmental Control</h4><p>This is more of a gestalt domain than a fundamental one — I’ll motivate it simply with the idea of using a brain-computer interface to modulate a Nest, and therefore the temperature in your house. This is gestalt insofar as it’s enabled a) by the presence of brain-computer interfaces that can decode the necessary neural signals, and b) the presence of Internet of Things-enabled environments. We can also imagine this in the context of virtual reality, where the environment is controlled with software, and therefore the requirement is reduced to just the presence of a BCI with sufficient neural decoding capabilities.</p><h4>Social/Moral Enhancement</h4><p>Social and/or moral enhancements are already familiar to us. Limited quantities of alcohol lower inhibitions and therefore make it easier to socialize (or so the cultural belief goes). Neurofeedback can be used to modulate people’s <a href="https://www.ncbi.nlm.nih.gov/pubmed/20570245">responses to emotional stimuli</a>. Oral contraceptives <a href="https://www.ncbi.nlm.nih.gov/pubmed/3128427/">increase oxytocin levels</a>. One can imagine behavioral- or neuro-engineering that up-regulates acts of kindness, a sense of empathy, “moral” behaviors (e.g. not lying), etc.</p><p>Interestingly, a study by <a href="https://academic.oup.com/jcr/article-abstract/35/3/495/1856272?redirectedFrom=fulltext">Riis, Simmons, and Goodwin (2008)</a> presents evidence that people are uncomfortable with enhancing traits that are “core” to one’s sense of self (at least insofar as the study’s college-student subjects can generalize to “people” at large) — many of which traits fall within the domains of sociality and morality. As examples, only 9% of polled subjects were willing to enhance “kindness,” and 13% were willing to enhance “empathy,” both of which topics the subjects rated as highly core to their identity. By contrast, 54% of subjects were willing to enhance “foreign language ability,” which subjects rated as not being very core to their identity.</p><h4>Therapies That Become Non-Therapies</h4><p>The considerations around non-medical neurotechnologies include not just the devices themselves, but also the data. In principle, health neurodata (which is governed by HIPAA) can transition into non-health neurodata. Many neurotech applications, like EEG or ECoG data for epilepsy detection, are going to entail the collection of neurodata for use in health applications. It’s natural to want to take that health data and use it to infer health-related things about people, and perhaps investigate large-scale anonymized datasets (e.g., what are the demographic or lifestyle factors that correlate with increased occurrence of a particular type of epileptic seizure?). It’s very easy to transition from here, though, to asking questions like: do people with certain EEG patterns show higher incidence of recidivism following incarceration and/or <a href="https://en.wikipedia.org/wiki/Restorative_justice">restorative justice</a> processes? As we see, health neurodata may be applied to derive medical insights, but it can also be applied in much more controversial and ethically dubious territory.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/0*0q0kYbI83AMWiXhP.jpg" /></figure><p>One question of mine, to which I don’t have an answer right now, is whether HIPAA prevents the eventual use of health neurodata for non-health applications. HIPAA places restrictions on when/why data can be shared with third parties, and these restrictions apply to “protected health information,” which is <a href="https://www.irb.pitt.edu/hipaa">defined</a> as information that</p><ul><li>“Is <strong><em>individually identifiable</em></strong> health information, whether oral or recorded in any form or medium (e.g., narrative notes; X-ray films or CT/MRI scans; EEG / EKG tracings, etc.), that may include demographic information, <strong>and</strong></li><li>Is created or received by a ‘covered entity,’ that is, a health care provider, health plan, or health care clearinghouse, <strong>and</strong></li><li>Relates to the past, present, or future physical or mental health or condition of an individual, to the provision of health care to that individual, and/or to payment for health care services <strong>and</strong></li><li>Identifies the individual directly or contains sufficient data so that the identity of the individual can be readily inferred.”</li></ul><p>I don’t know whether this applies to, for example, EEG data that was collected originally for medical purposes, but then anonymized and used for non-medical ends.</p><p>The danger here isn’t immediate, and it might not even be worth spending time dwelling on; I bring it up for the sake of demonstrating the complexity in the ethics of neurotechnology. More broadly, I think it’s critical to acknowledge the speculative nature of planning for non-medical neurotechnology ethics. By intention, I’m discussing technologies which don’t exist yet, except perhaps as proofs of concept in laboratories, or embedded deep-down in active imaginations. It’s impossible to predict when (and perhaps if) the problems above, and below, will be <em>pressing</em>; for example, when a consumer-ready technology that can non-invasively decode mental speech will be brought to market.</p><p>My philosophy is that it’s better to preempt than to respond. This means that, by necessity, we have to imagine future scenarios. Here, I owe my reader a warning: future scenario imagination is inherently alarmist, but it seems like a reasonable tradeoff to me, since the possibility of “not as much innovation, not as quickly” is better than “too much innovation, too quickly.” I’ll let Facebook’s current situation speak for itself.</p><h4>A Unified Framework</h4><p>With the above in place, I can now lay out the framework I plan to apply to the ethics of investing in neurotechnology.</p><h4>1. The Wall of Discretion</h4><p>There exists a mental wall: behind this wall, we live and hide our inner lives. In front of this wall, we vocalize ourselves. The vocalization is the input to the morality function our culture computes: i.e., we’re morally assessed based on our actions, and not our thoughts and our feelings inside that don’t manifest externally. Neurotechnology threatens this sacred wall by proposing the possibility to characterize the mental contents within, ostensibly with or without our knowledge or consent. I call this wall the <strong>Wall of Discretion</strong>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*VgTJkj9MhsaQOgLS.jpg" /></figure><p>As the Stanford Encyclopedia of Philosophy puts it in their entry on <a href="https://plato.stanford.edu/entries/neuroethics/">neuroethics</a>:</p><blockquote>“In the past, mental privacy could be taken for granted: the first person accessibility of the contents of consciousness ensured that the contents of one’s mind remained hidden to the outside world, until and unless they were voluntarily disclosed…Over the last half-century, technological advances have eroded or impinged upon many traditional realms of worldly privacy. Most of the avenues for expression can be (and increasingly are) monitored by third parties. It is tempting to think that the inner sanctum of the mind remains the last bastion of real privacy.”</blockquote><p>Let’s consider a thought experiment: suppose that Instagram, with its likely ability to classify its own content semantically and to record through the Instagram app how much time you spend looking at each type of content, can build a fairly sound psychological profile of you. How does this make you feel?</p><p>Now, consider a (near?) future in which Facebook builds a device that can <a href="https://techcrunch.com/2017/04/19/facebook-brain-interface/">transcribe</a> the words you’re thinking. How do you feel?</p><p>If you’re like me, and I suspect most people are, the latter scenario is frightening. Maybe even chest-tightening. This is because we’re deeply familiar with putting our external characteristics and behaviors on display and keeping our internal characteristics and behaviors hidden.</p><p>Our psychological profiles are almost <em>meant</em> to be out in the wild; we pride ourselves on our personality traits, and “pride” assumes there’s someone out there who will notice the thing we’re proud of. What’s identity without other people to show that identity to? In contrast, internal language, emotion, intuition, and gestalt percepts are <em>meant</em> to be internal. In other words, we all have a subjective feeling that our thoughts and emotions <em>belong</em> inside.</p><p>To provide further evidence for the distinction between “internal” and “external,” and therefore to further evidence the Wall of Discretion model, consider Myers-Briggs scores. A common question for a new friend is: “What’s your Myers-Briggs?” Horoscopes have a place in pop culture for a reason, too.</p><p>Let’s look at dating apps. They all contain demographic information that defines our preferences and provides other people with a way to determine quickly if they find us appealing. Sometimes it goes beyond demographics, such as with OkCupid, which asks its users an intricate set of questions to try and create better matches — none of which is to mention that some dating apps even have dedicated fields for astrological signs in their profiles.</p><p>It’s evident that we have external selves and that we share those external selves widely. What are the transition points, or the channels, by which we take “stuff” from the inside and bring it outside? One interesting example is making a post on Facebook. Consider the painstaking nature of this process (assuming we’re writing a post with meaningful semantic content, like a political post or a life update or a bit of wisdom you feel compelled to share with the world, as opposed to a photograph from your trip to the Grand Canyon last month). In first person: I sit down at my computer, I open up the tab, I stare at the empty box for a minute or two, I type a few words, delete them, then repeat and gradually craft the post — it’s a meticulous process. I make my edits, I decide which emoji I want to insert based on the reactions I think they’ll evoke, and I choose whether or not to include a photo for emotional impact and/or humor effect. I’ve made quite a few decisions at this point about how to translate what’s behind my Wall of Discretion, both semantically and emotionally, into what’s released externally.</p><p>Another interesting case-study in the Wall of Discretion (WoD) is alcohol. People regularly consume alcohol to make it easier and more natural to move things from within the WoD outside of it. Many of us have a drink or two at social events in order to feel more social (because alcohol lowers inhibitions). There’s much cultural humor (and many anecdotal stories) about people over-imbibing and sharing what’s inside their WoD too freely, then regretting it once they’ve returned to a normal mental state — take, for example, the canonical drunken love profession.</p><p>It’s also reasonable to ask how we decide what types of things are inside the WoD. With intuition and personal experience as my evidence, I’d say the contents within the WoD can be dichotomized: <em>unintentional</em> and <em>intentional</em>. <strong>Unintentional contents</strong> include emotional reactions, implicit bias, physical attraction, etc. Even linguistic thoughts can be unintentional: “This thought popped into my head the other day” implies a lack of agency. The <strong>intentional contents</strong> behind the WoD include deliberate thought processes; beliefs, values, and morals you’ve explicitly chosen to hold; internal decisions you’ve made. Importantly, also within the set of intentional contents are our conscious reactions to unconscious (or unintentional) contents. In other words, it’s not my choice if I feel an emotion towards a sad scene in a movie; but, I can decide how to handle the presence of that emotion, and whether or not to externalize it. Thus, if the contents inside of my Wall of Discretion were exposed, not only would someone be aware of that emotion, but they’d be aware of how I decided to react to it — and that reaction is a measure of my character.</p><p>Recapping, we seem to enjoy and/or value breaking the Wall of Discretion in certain scenarios, primarily social:</p><ul><li>Direct communication</li><li>Mating and signaling social status</li><li>Signifying that we’re comfortable with someone</li><li>Meeting new people &amp; interacting with many people</li></ul><p>From all of this, I see five core principles of the Wall of Discretion.</p><ul><li><strong>Human communication is the intentional transmission of information through the Wall of Discretion. Whenever we communicate, we take a message from inside our conscious realm and express it outwardly; we pass that message through the Wall of Discretion.</strong></li><li><strong>The morality of individuals within our society is judged based on what they’ve transmitted through their WoDs — i.e., their actions, behaviors, and stated beliefs.</strong> This <em>has</em> to be the case, since by definition, what’s inside of the Wall of Discretion is hidden and therefore unobservable by other people.</li><li><strong>What we choose to filter through our WoD varies depending on the recipient: certain information, such as sexual preferences or health status, we’ll share with some people but not others.</strong> This follows directly from the observation that human communication is the transmission of information through the WoD: we communicate different information to different people.</li><li><strong>We’re not used to internal WoD contents being observed. Polygraph tests and torture are examples of WoD-violating practices that we culturally frown upon or outright ban.</strong> The stuff of our inner sanctum — emotion, internal monologue, bias, desires, verbal thoughts that pop into our ahead — is never observed by other people.</li><li><strong>Violating the Wall of Discretion can lead to the observation of unintentional mental states (i.e. the conscious manifestations of unconscious processes) and our intentional reactions to those unintentional states.</strong> Emotion, for example, isn’t something I decide. It occurs (i.e., it derives from neural processes I don’t have conscious access to), and then I observe and react to it consciously. The same set-up applies for implicit bias: it occurs, and then I react to it. Unless I choose to communicate my reactions, they stay behind my Wall of Discretion. Therefore, if my WoD is violated, my intentional reactions to unintentional states can be observed.</li></ul><p>With polygraphs, we’ve finally brought neurotechnology into the WoD picture: since the WoD is a theoretical heuristic for psychological processes that sit on neurobiological substrates, neurotechnology can threaten the internality of contents within the WoD. Here are some considerations:</p><ul><li><strong>Remote vs. local neurotech</strong>: If neurotech evolves such that it can deduce some or all aspects of the WoD internals without actually requiring a device on or in someone’s head, then it will be very hard to prevent nefarious usage. In contrast, if neurotech evolves such that it can transcribe WoD internals only if it’s on the head or very nearby, then covert monitoring is less feasible.</li><li><strong>Willful vs. non-willful neurotech</strong>: if neurotechnology paradigms require the subject to engage in a predefined activity in order to interpret their neurodata (which is often the case with EEG and fMRI), then the subject can simply choose not to comply and barring torture, there’s nothing to be done. On the other hand, if neurotechnology can simply read out internal monologues (very important note: this is <strong>not</strong> possible right now), then things get dicey.</li></ul><p>The worst case is the intersection of remote and non-willful neurotechnology: if such technology comes to exist at some point, it will probably enable profound encroachment upon our Walls of Discretion.</p><p>This discussion of the Wall of Discretion would be incomplete without acknowledging the potential benefits of a neurotech that can cleanly cut through the WoD. Of course, there’s the science fiction dream of direct brain-to-brain communication, which is a means to facilitate profound human connection. There’s also the possibility for, as an example, a psychotherapist to gain detailed insight into a patient’s negative thought processes — down to the words, even — granted the patient’s consent. We can even consider the scenario where an individual has been wrongly convicted, and can willfully submit to a futuristic “truth detection” technology in order to exonerate themselves.</p><p>Concluding my thoughts on the Wall of Discretion: technology that can be used to non-willfully and remotely observe internals behind the WoD should not exist, because it invades privacy above and beyond what we’re comfortable with culturally. That said, it’s feasible for culture to move in a direction where we’re collectively more comfortable with the reality of ad-hoc internal WoD access. I can’t imagine a reality in which I sit down at a coffee shop and intrude on someone else’s inner world — frankly, I see in that possibility a<a href="https://en.wikipedia.org/wiki/Mutual_assured_destruction"> mutually assured destruction</a>. And so, here’s a hypothesis that’s morbidly positive: even in the eventuality of non-willful and remote WoD-breaking technology, I don’t think it will be adopted at scale because if it does, everyone’s worse off.</p><p>One final note: the Wall of Discretion doesn’t capture every issue inherent to neurotechnology. Rather, in my view, it’s an elegant way to conceptualize what is time-asymptotically at stake. For a much fuller treatment of neuroethics, I’d strongly encourage the reader to spend some time studying the Stanford Encyclopedia of Philosophy’s entry on<a href="https://plato.stanford.edu/entries/neuroethics/"> Neuroethics</a> (or reach out to me about a particular area of interest, since I’m not an expert and I’ll enjoy learning with you). In particular, I’ve omitted discussion of <a href="https://www.ncbi.nlm.nih.gov/pmc/articles/PMC3725243/">informed consent</a>, the details of <a href="https://ec.europa.eu/commission/priorities/justice-and-fundamental-rights/data-protection/2018-reform-eu-data-protection-rules_en">data privacy</a>, the moral implications of enhancement, socioeconomic gaps, and <a href="https://journals.sagepub.com/doi/abs/10.1177/0022167815617296?journalCode=jhpa">neuro-essentialism</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/1024/0*XWFMyQHZarWltqs1.jpg" /></figure><h4>2. Thoughts, and Decoding Them</h4><p>The mechanism by which the Wall of Discretion can theoretically be broken is by “decoding thoughts.”</p><p>This discussion begins with the ideas of the conscious and the <a href="https://www.health.harvard.edu/blog/unconscious-or-subconscious-20100801255">unconscious</a> because they underlie the intuitive definition of “thought.” For my purposes here, I’ll just characterize the state of “consciousness” as something that we can report ourselves as possessing; we have some awareness that it’s occurring. “Unconsciousness,” on the other hand, is characterized by the fact that we <em>can’t</em> report it, and we <em>aren’t</em> aware when processes of the unconsciousness are occurring, nor we can we describe those processes.</p><p>Unconscious processes strongly influence conscious processes, so the model of considering them separately is inherently flawed. It’s useful, though, because we don’t actively <em>perceive </em>unconscious thought factoring into conscious thought (by definition), and therefore much of the intuitive emotion behind “thought” really pertains to conscious processes, which are in turn influenced by unconscious processes.</p><p>What, then, does it mean to decode “thought”? Put another way, how might we break the Wall of Discretion using neural signals, as opposed to through intermediary modalities such as language (writing and speaking), observing behavioral data, dance, the visual arts, etc.? To make this a well-posed problem, we have to identify concepts or states that can be deemed as entities and then identified through the observation of neural substrates. Using the English descriptions of these concepts (instead of describing them by a hypothetical set of neural interactions, such as “population A of neurons exhibits a firing rate normally distributed around mean firing rate x…”):</p><ul><li>Emotion (e.g., location along the<a href="http://www.apple.com/"> Pleasure-Arousal-Dominance</a> axes)</li><li>Language</li><li>External visual information (information we observe through our eyes)</li><li>Internal visual information (imagined sight)</li><li>Spatial knowledge and sensation</li><li>Procedural knowledge (e.g., how to play a Beethoven’s Moonlight Sonata on the piano)</li><li>Impending decision (e.g., the selection between coffee beverages)</li><li><a href="http://kirwaninstitute.osu.edu/research/understanding-implicit-bias/">Implicit bias</a></li><li>Sexual and romantic preferences</li><li>Auditory information</li><li>Motor intention</li></ul><p>Listed above are some of the types of contents we keep within, or transfer out of, the Wall of Discretion. It’s these that non-medical neurotechnology promises to expand and threatens to uncover.</p><h4>3. The Wall of Discretion Should Remain Intact</h4><p>The Wall of Discretion should not be broken unless someone wants it to be, in which case they can only exercise that will over their own WoD. This is my view for two reasons (which need more research to turn into well-evidenced arguments, rather than strings of hypotheses).</p><h4><strong>Morality preserves life</strong></h4><ul><li>Human life ought to exist (axiom);</li><li>Culture enables prolific human life (hypothesis);</li><li>Morals keep cultures together by facilitating group cohesion (hypothesis);</li><li>Morals depend on filtered human emotion and thought; i.e., social pressure acts to moderate and counterbalance internal urges (critical hypothesis [note that this hypothesis depends on our current biological configuration in which there are competing signals, such as those from the neocortex and the limbic system]);</li><li>Therefore, if the Wall of Discretion is broken unwillingly and at scale, then under current biology, it’s unclear what happens to moral systems, which in turn poses a threat to culture and therefore to the proliferation of human life.</li></ul><h4><strong>We should minimize suffering</strong></h4><ul><li>We shouldn’t create suffering (a moral axiom);</li><li>The invasion of privacy has a distinctly negative feeling, and can, therefore, be considered as a form of suffering (personal observation and subjective claim);</li><li>Breaking the Wall of Discretion involuntarily is interpreted emotionally as having one’s privacy invaded (personal observation);</li><li>Therefore, to minimize suffering, we shouldn’t unwillingly break the Wall of Discretion, because to do so is to cause suffering.</li></ul><p>So, going forward, I assume that the Wall of Discretion should stay intact (and that willful breaking of the WoD is better than local non-willful breaking of the WoD, which in turn is better than remote non-willful breaking of the WoD).</p><h4>Ethics and Investors</h4><p>If there’s a practical way to quantify the regard we have at Loup Ventures for the ethics of our investments, it would be by measuring the number of hours of phone conversations we’ve dedicated to it (note my use of “hours,” not “minutes”). At Loup, we want to be investing in technologies that are, in aggregate, a positive force. Doug Clinton, one of our Managing Partners, has put in significant work to articulate a broader code of ethics, and it boils down to the following statement: <strong>Invest in high-integrity founders making products people love and that our Limited Partners will be proud to back.</strong></p><p>Why do we care about this? It’s unclear yet if there’s strict business value, but there’s definitively a social obligation. We have an obligation to society to be conscientious of the technologies we back, and of our resulting impact. I hold this obligation to be true of anyone — investor, engineer, poet, painter, or otherwise.</p><p>As investors, we have certain leverage points that others don’t, meaning that our efforts in neuroethics can go a long way.</p><ol><li>We’re connected to many stakeholders: startups, regulators, consultants, other investors, employees, entrepreneurs, etc.</li><li>By virtue of the enabling relationship between venture capitalists and entrepreneurs, the venture community can set the standards of conduct for entrepreneurs.</li><li>As investors, we have downstream impact on public sentiment through the successes (and failures) of our portfolio companies, and even through direct press.</li><li>We write checks that help entrepreneurs realize their products.</li></ol><h4>Taking Action</h4><p>With the Wall of Discretion as the underlying principle and a quick argument for why VCs can and have to be effective ethics gatekeepers, I can now begin to think about how Loup can work to uphold ethical neurotechnology development.</p><p>Broadly, we want to create a pipeline for ourselves that ultimately gives a green light or a red light to a particular neurotech investment opportunity. An ethically sound neurotechnology company doesn’t mean we’ll invest, but an unethical company means we won’t.</p><p>We’ve been engaging in an internal process to generate this pipeline. The procedure has included reading, talking to experts (ethicists, other investors, companies, scientists/engineers, etc.), and synthesizing the results of this research. Once we have a pipeline in place, we’ll publish the principles behind it and maybe the pipeline itself (in support of transparency and in advocacy of ethical neurotech). We’ll articulate this pipeline to our LPs, to our existing neurotech portfolio companies, and to prospective neurotech portfolio companies.</p><p>Ethics are best upheld by social forces, and ultimately collective opinion decides ethics as opposed to individuals (or individual organizations). That said, we chose to start this process as an internal push rather than external push since the feedback cycle on principles of ethics is much more efficient internally. Even though it was developed internally at Loup and not with formal external input, the forthcoming end-result of this internal process — a set of guidelines for our investments in neurotechnology — is something that can benefit the entire community; at the least, it can drive conversation and inspire additional collaborative work.</p><h4>Sketch of a Pipeline</h4><p>We haven’t finished this yet, but we want to share our progress so far. The initial model is to use a three-filter pipeline:</p><p>“Filter 1” contains a set of criteria for whether or not it’s worth introducing a given neurotechnology to society.</p><p>“Filter 2” checks a prospective investment’s work against a blacklist (no-go), graylist (exercise caution, but not immediately no-go), and whitelist (all clear) of neurotechnologies and their applications (this idea was inspired by conversations with <a href="https://www.boundless.ai/">Boundless Mind</a>, a behavioral engineering company that has put substantial thought into the <a href="https://www.boundless.ai/blog/how-to-score-leads-for-values-alignment/">ethical applications</a> of their technology). This is a list we’ll develop internally, and if other funds adopt a similar pipeline, they’ll have to apply their own philosophies and judgments to generate their unique versions of these lists. If we encounter a technology that isn’t already listed, we’ll have a process for making a <em>de novo</em> decision, and of course, use that decision to expand the lists appropriately.</p><p>“Filter 3” comprises a checklist of items that our portfolio companies will have to complete either before our investment or once it’s been made.</p><p>All three of these filters, and the structure of the pipeline itself (i.e., the utilization of three linearly-arranged filters), needs to be adaptive. To make it adaptive, we will a) define certain technological trigger points that would cause us to re-evaluate our pipeline or particular components of it, and b) we’ll implement an annual review and reconsideration of the pipeline.</p><h4><strong>Filter 1</strong></h4><p>As a first-order filter for making ethical investments, we can select a set of criterion for whether a neurotechnology is worth introducing to society. In my view, the burden of proof should be on the technology: we assume a technology shouldn’t exist until convinced otherwise. Make no mistake, this is a strong philosophical stance: augmentative neurotechnologies shouldn’t exist unless they strictly improve certain “endpoints” (e.g., human performance, human well-being) for a definable population. Note that this coincides well with what makes a business successful.</p><p>To determine that a neurotechnology ought to exist, we need to see evidence that</p><ol><li>There’s positive value added by the technology;</li><li>The technology doesn’t have reasonable feasibility of unwillingly breaking the WoD (particularly in a remote, non-willing manner); and</li><li>The technology doesn’t facilitate negative mental impacts via means other than WoD breaking (e.g., reducing the intrinsic enjoyment of an experience through over-quantification).</li></ol><h4><strong>Filter 2</strong></h4><p>Filter 2 is comprised of a blacklist, a graylist, and a whitelist. We’re not ready to put these lists out to the public yet since they’re not complete, but to give an example of something that’s blacklisted: we won’t invest in neurotechnology that’s primarily intended to be weaponized. As another example, we won’t invest in neurotechnology that’s being directly applied to decode someone’s internal contents against their will (e.g., a hypothetical lie detection test that works without the subject’s consent).</p><h4><strong>Filter 3</strong></h4><p>The third filter is less stringent; it’s more about the continued maintenance of ethical practices within the prospective portfolio company. For example, we could require that the prospective CEO has a conversation with a trained neuroethicist to discuss ethical implications of their technology. We may further require a simple but informative neuroethics training for new hires at the prospective portfolio company. Essentially, Filter 3 covers operational elements that aim to build a culture of ethics into our portfolio companies. If the management team of a company expresses a commitment to completing this checklist and has passed the first two filters, then we would give them the final ethics green-light.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/927/0*bU1uS9-8NRhhhedi.png" /></figure><h4><strong>Implementing trigger points</strong></h4><p>Above, I mentioned that one of the characteristics of this pipeline is its built-in adaptability. That largely comes from a “trigger-point” model, where we define then continuously update a set of triggers that would cause us to think carefully about a given neurotechnology and understand the nuances of its development and applications in order to make more precise ethical determinations in that domain.</p><p>Some trigger points could include:</p><ul><li><em>The ability to reliably modulate emotion on-demand.</em> The better case of this is that this can only occur with the consent of a user; the worse case is if this can be done without consent because that’s an unwilling manipulation of contents behind the Wall of Discretion.</li><li><em>The ability to fully decode covert speech with a non-invasive device.</em> A fine case of this is if it can only occur with the consent of the user since this constitutes a willful breaking of the WoD. The worse case is if this is feasible without consent, in which case we have unwilling WoD breaking.</li><li><em>The proliferation of consumer EEG devices where data is centrally collected and privacy is put in the hands of business organizations, and furthermore, the neurodata can be demonstrated to yield insights into real-time characterization and can predict user actions above and beyond behavioral data</em>. The fine scenario is if users are made aware in clear, understandable terms the capabilities of a system that performs these actions, and are kept up to date on new capabilities. Additionally, users should have the ability to opt out of prediction, characterization, and the related data storage at <strong>any time</strong>, and should be able to access reports on if/how their behavior has been manipulated using insights from the neurodata. These conditions ensure that this scenario is a willful breaking of the WoD. (A note: one way to verify the “fineness” of this scenario is to look at the distribution in a population of comfort with a given application or software capability, on a scale of very uncomfortable to very uncomfortable. If the comfort/discomfort distribution is similar to existing WoD-breaking technology, like targeted advertising, then we can be confident that this hypothetical neurotechnology is alright — assuming, of course, that we’re willing to consider statistics-level sentiment as a signal on which to base our ethical determination.) A worse case is if users are unaware of the capability of the hypothetical software that uses EEG data to make decisions since this covertly breaks the WoD.</li><li><em>The hyper-quantification of subjective experience</em>. The fine case here is invisible-but-consented quantification of subjective experience that can be used in the background of a software application, physical environment, or other another service (e.g., the gym) to facilitate positive emotion. In this case, it’s a willful breaking of the WoD. The worse scenario is consented <em>and</em> visible quantification of subjective experience, since it might<a href="https://academic.oup.com/jcr/article-abstract/42/6/967/2358309"> reduce the intrinsic enjoyment</a> of experiences. I tend to favor technominimalism under the anecdotally-informed hypothesis that human experience is richest in the absence of overt technological interaction (“natural = good” isn’t a sound justification, nor do I buy it, although I need to flesh out a better-formed argument for technominimalism). In general, the goal is to keep “neuroinsights” and neurodata out of people’s fields of view.</li><li><em>Cognitive enhancement that is a) extremely expensive, b) creates a meaningful increase in performance, and c) is unlikely to reduce in price anytime in the near future</em>. This is a trigger-point because it invokes the question of a socioeconomic split. This is most strongly a possibility in the case of an invasive, enhancing neurosurgery that costs tens or hundreds of thousands of dollars out of pocket, since it wouldn’t be covered by insurers.</li><li><em>Weaponization of any form of neurotechnology</em>. A discussion of the pros and cons of weapons is beyond the scope of this piece, but our heuristic is to not fund technologies that can and/or will knowingly be used to induce suffering.</li><li><em>Introduction of any neurotechnology that strictly </em><strong><em>decreases </em></strong><em>the probability of well-being or the economic success of an individual (e.g., using lie detection as a hiring tool)</em>. Note the importance here of considering the individual over the corporation.</li></ul><h4><strong>Conclusion</strong></h4><p>What I’ve undertaken here is an attempt to do the following:</p><ol><li>Explain why neuroethics is worthwhile to think about in general;</li><li>Explain why neurotechnology investors have particular leverage in supporting ethical neurotechnology;</li><li>Define an underlying principle/idea that can be used to guide the discussion of the ethics of neurotechnologies; and</li><li>Sketch out a pipeline for performing ethical vetting on neurotech investments.</li></ol><p>My thinking on the ethics of neurotech, and how I work with my colleagues at Loup to actionably implement this abstract thinking, is fluid and responsive. I encourage you to reach out with any feedback, ideas, and/or criticisms.</p><p><em>Disclaimer: We actively write about the themes in which we invest or may invest: virtual reality, augmented reality, artificial intelligence, and robotics. From time to time, we may write about companies that are in our portfolio. As managers of the portfolio, we may earn carried interest, management fees or other compensation from such portfolio. Content on this site including opinions on specific themes in technology, market estimates, and estimates and commentary regarding publicly traded or private companies is not intended for use in making any investment decisions and provided solely for informational purposes. We hold no obligation to update any of our projections and the content on this site should not be relied upon. We express no warranties about any estimates or opinions we make.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=4e79f6159ccc" width="1" height="1" alt=""><hr><p><a href="https://medium.com/the-substrate/an-investors-take-on-the-ethics-of-neurotech-4e79f6159ccc">An Investor’s Take on the Ethics of Neurotech</a> was originally published in <a href="https://medium.com/the-substrate">The Substrate</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Consumer Neurotech: Volume V]]></title>
            <link>https://medium.com/the-substrate/consumer-neurotech-volume-v-2d6fa504c051?source=rss----10fba766a0f5---4</link>
            <guid isPermaLink="false">https://medium.com/p/2d6fa504c051</guid>
            <category><![CDATA[brain-computer-interface]]></category>
            <category><![CDATA[brain]]></category>
            <category><![CDATA[neuroscience]]></category>
            <category><![CDATA[neurotech]]></category>
            <category><![CDATA[venture-capital]]></category>
            <dc:creator><![CDATA[Loup Ventures]]></dc:creator>
            <pubDate>Wed, 24 Oct 2018 17:50:52 GMT</pubDate>
            <atom:updated>2018-12-05T17:32:53.352Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/820/0*KRVitAYS-dt7FdYU.png" /></figure><p><em>By Avery Bedows</em></p><h4><strong>Other Neurotech: The Odd Ones Out</strong></h4><p>In <a href="https://loupventures.com/consumer-neurotechnology-volume-i/">Volume I</a> of this series on Consumer Neurotech, we established a framework for thinking about Consumer Neurotechnologies and their value propositions. In <a href="https://loupventures.com/consumer-neurotech-volume-ii/">Volume II</a>, we discussed neurotechnologies that have the value proposition of <strong>Control</strong> (or, using our taxonomy from Volume I, neurotechnologies whose “Effect-type” is Control). In <a href="https://loupventures.com/consumer-neurotech-volume-iii/">Volume III</a>, we analyzed neurotech with an effect-type of <strong>Feedback</strong>. In <a href="https://loupventures.com/consumer-neurotech-volume-iv/">Volume IV</a>, we looked at neurotechnologies that directly modulate the nervous system. In this Volume V, we’ll look at some wildcard consumer neurotech that doesn’t fit in any of the preceding categories.</p><h3><strong>Neurotech for Temperature Neuromodulation</strong></h3><h4>Fiction</h4><p><em>You once fervently argued with a Chicago bartender regarding the devilishly polarizing topic of Chicago winters — what’s worse, brown-gray snow-turned-slush, or the biting cold? The bartender sided errantly with the slush; you offered mathematical proof for the cold. Today is a case-in-point: how are you supposed to prepare for the board meeting in 30 minutes if you’re shivering? It’s 8am and the heating still hasn’t kicked in at the office; you think about setting a reminder, but reconsider because your shivering probably precludes the dexterity required to type semi-correct words into the iPhone’s Reminders app. You opt for Siri instead, and the act of speaking draws your attention to the blue cold on your cheeks. Next to your monitor sits a watch-like device with a large rectangular panel. You put the band on, and the LED indicator signals that it’s turned on, and pulses of heat extend across your wrist. Five minutes later, with the assistance of your cup of coffee, your vitriol towards the Chicago cold goes away. You find it refreshingly easy to enter into a quick flow, and do your meeting prep with eyes closed, fingers playing the keyboard, letters as music. Walking into the boardroom, as social air traffic control helps you plan your descent to the mahogany table, you muse: there’s something elegant about the contrast between warmth indoors and the frigidity outside.</em></p><p>• • •</p><p>Every sense in the body is underpinned by the nervous system, and this opens up opportunities to control/modulate the senses. One such sense is temperature.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/0*TElrCbZaIlRIsTAe.jpg" /></figure><p>Image: <a href="https://www.digitaltrends.com/cool-tech/embr-wave-review/">https://www.digitaltrends.com/cool-tech/embr-wave-review/</a></p><h4><strong>Embr Wave</strong></h4><ul><li>Medium of innovation: <strong>Hardware</strong></li><li>Device-type: <strong>Stimulate</strong></li><li>Form-factor: <strong>Armband</strong></li><li>Effect-type: <strong>Physiological</strong></li></ul><p>Neurotechnology can leverage temperature-sensitive neurons in the skin to modulate subjective temperature perception, which is the goal of the <a href="https://embrlabs.com/">Embr Wave</a>. The Wave looks like a watch with the watch-face on the inside of the wrist. Instead of a watch face, the Wave has a heating and cooling element that’s used to coax a user’s nervous system into perceiving that the user is warmer or colder than before the Wave was turned on. This phenomenon is similar to running your wrists under warm water when you feel chilly.</p><p>The device costs $300, and the August 2018 manufacturing batch contained 1500 units (as specified on their website at the time of our research).</p><h3><strong>Takeaways</strong></h3><ol><li>Like many other consumer neurotechnology products, the Embr Wave has a 30-day money-back guarantee, reflecting the variability in its effectiveness for different people.</li><li>The Embr Wave falls in the category of consumer neurotechnology that’s “wear-as-needed.” When a user is feeling too warm or too cold, they put the device on and start a 3–5 minute heating or cooling cycle. The device is no longer needed after that. Realistically, the Wave will live most of its life in bags, on desks, and in drawers — it’s hard to imagine users wearing around a clunky-looking inverse watch that remains dead weight most of the time.</li></ol><h3><strong>Use-Cases and Market Size</strong></h3><p>We didn’t identify any useful heuristics for sizing this market, so we’re omitting a market size here.</p><h3><strong>Sensory Enhancement</strong></h3><p><em>Sometimes it’s fun to reenact moments from movies where the protagonist stands still in the middle of a crowd. You like standing in the middle of a sidewalk during lunchtime, sitting on the ground in high-foot-traffic city parks during warm summer months, pushing against the crowd entering a baseball stadium, ambling among the miniature mountains of a densely-populated white-sand beach. In the cinematographic versions of this, you can pause playback to observe the facial expressions of the extras, listen to them walk and speak, understand their emotion. In real life though, on your beaches and walkways, the experience is transient. Your senses don’t have much time to experience it all, and you find this frustrating because you feel drawn to observe and understand the details in the world that surrounds you. You recently saw a new piece of technology online that caught your attention more than most do: it’s a wristband that turns the sounds of the world into vibrations. As a collector of sensory richness, you had a hunch that this would be a product for you. Today, you’re at the beach again. You’ve been standing in one spot for long enough that the white grains have conformed to your feet, filtering down between your toes. Your eyes are closed because today is about a different sense. Normally, you’d call it sound, but today it’s something…different. It’s richer and extended. Today, you hear in the waves and in the harmonizing cinematic laughter from a mother and her daughter detail you’ve never encountered before because this is a new sensory experience.</em></p><p>• • •</p><p>In pop culture and folk neuroscience, humans have five senses. Human curiosity, amplified through science fiction and fantasy, has for years posed the question: <a href="https://www.ted.com/talks/david_eagleman_can_we_create_new_senses_for_humans">can that set of five senses be expanded</a>?</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/266/0*4RBTJuOks7Ezs9Nw.png" /></figure><p>Image: <a href="https://neosensory.com/buzz/">https://neosensory.com/buzz/</a></p><h4><strong>NeoSensory</strong></h4><ul><li>Medium of innovation: <strong>Integrated</strong></li><li>Device-type: <strong>Stimulate</strong></li><li>Form-factor: <strong>Armband</strong></li><li>Effect-type: <strong>Sensory</strong></li></ul><p>A company called <a href="https://neosensory.com/">NeoSensory</a> has tackled this question for the sense of hearing. NeoSensory lists four products on their website. <em>Clarify</em> is a wrist-worn device that associates vibrations with sounds to help disambiguate words for people with a hearing impairment. <em>Buzz</em> is similar to <em>Clarify</em> but focuses on enhancing hearing for users without a hearing impairment. <em>Vest</em> can convert a variety of information into haptic sensory perception in order to grow intuition for that information (e.g., stock market data by converting the stock market data into patterns of vibration). <em>Exoskin </em>is a $400 haptics vest for use with virtual reality; NeoSensory partnered with <a href="https://highfidelity.com/">High Fidelity</a> to produce this product, which was also featured in Season 2 of the TV show, <em>Westworld</em> (see NeoSensory’s <a href="https://www.facebook.com/NeoSensory/">Facebook page</a> for more details). No information on the pricing of <em>Clarify</em>, <em>Buzz</em>, or <em>Vest</em> is available on their website.</p><h3><strong>Takeaways</strong></h3><ol><li>NeoSensory highlights the fine distinction between therapeutic and augmentative uses of neurotechnology. <em>Clarify</em> and <em>Buzz </em>are similar products: both are wrist-worn haptics devices that use haptic vibrations to accompany audible sound. One focuses on helping people who have a hearing impairment, and the other focuses on enhancing people without a hearing impairment. This begs the oft-invoked question: what problem are you solving? <em>Clarify</em> addresses the obvious problem of hearing impairment. <em>Buzz</em>, on the other hand, requires logical gymnastics to explain the problem. “Question: what problem are you solving? Answer: the problem is that we aren’t enhanced yet.” For example, imagine a futuristic memory prosthesis that gives its users external storage of human memories. This device solves the problem that “people are bad at remembering things.” But, the critical consideration is whether or not people <em>notice </em>they’re bad at remembering. If people notice, then the memory prosthesis has a visceral draw; if people don’t notice, then it doesn’t. Likewise, with regards to <em>Buzz</em>, do people ever wish their hearing was better?</li><li>However, there’s an argument to be made that this is the wrong framing. Consider the iPhone: people weren’t clamoring for mobile internet, yet the iPhone introduced it as an entirely new category because the experience was profoundly different and enabling. It’s possible that <em>this</em> question, “What new experience are you enabling, and how profound is it?” is more salient.</li><li>We suspect that enhancing neurotechnologies will only succeed when either a) they simultaneously solve an <em>a priori</em> evident problem, or b) when potential purchasers can access compelling demonstrations before buying. It’s worth noting that this scenario dichotomy only applies to a culture where neurotechnological enhancement is <em>not </em>the norm; if neurotechnological enhancement is the norm, then social pressure will act as a forcing function. Entertaining nigh-science fiction notions of neurotechnology-mediated intelligence enhancement, as in the case of the memory prosthesis: does this eventuality mean that human augmentation will be self-propagating since even people who don’t initially want to use it will realize they’re at a disadvantage unless they do? Compare this to smartphones: because such a large percentage of the population uses them, if someone doesn’t own a smartphone, they’re at a disadvantage in the professional realm, and arguably the personal realm as well.</li><li>The <em>Vest</em>, as advertised, is bulky; even if it were able to give stock traders an edge, it would face massive friction for adoption. This is another case where it would be interesting to figure out what works best to get potential users to make a purchase: quantitative evidence versus emotional appeals.</li></ol><h3><strong>Use-Cases and Market Size</strong></h3><h4><strong>Deafness</strong></h4><p>There are 28.8M American adults who <a href="https://www.nidcd.nih.gov/health/statistics/quick-statistics-hearing">can benefit</a> from hearing aids, but only 20% (5.76M) of those people <a href="https://report.nih.gov/nihfactsheets/viewfactsheet.aspx?csid=95">actually use them</a>. Partially, this is because hearing aids range in price from $1,500 to $5,000 dollars; additionally, hearing aids carry social and aesthetic disincentives. Intuitively, there’s a meaningful market for a non-hearing aid product to assist those with hearing impairments.</p><p>To estimate this market size, we have two components: the percentage of people who already own hearing aids who will use <em>Clarify</em> in addition to their existing hearing aid, and the percentage of people who don’t already own hearing aids but who will purchase <em>Clarify</em> since it’s a cheaper and less socially intrusive solution. There’s no available pricing information on <em>Clarify</em>, so we assume it’s on par with other technologies like the Embr Wave, and we set the price at $300 for our estimates. We assume that penetration rates will be higher for people who aren’t already using hearing aids than for those who are, since presumably <em>Clarify</em> will have less marginal value for the latter group. Our estimate is that, <strong>in 2025, <em>Clarify</em> sales will be $103,700,000</strong>.</p><h4><strong>General auditory enhancement</strong></h4><p>As far as we know of, this is the first consumer sensory enhancement. There’s no way to predict reception to this type of sensory enhancement product without first testing its reception experimentally. For consumer neurotechnologies like this, where it’s legitimately new and without precedent, it will be important to find ways to test public sentiment before launching or investing in a product. Ideally, these tests would incorporate social dynamics and attempt to understand how the product would be received when introduced to different groups. For example, what would happen if this were blanket-marketed to San Francisco-based software engineers, as opposed to Google-employed software engineers, as opposed to New York-based software engineers? Would the characteristics of these groups (geography and trade as the shared feature, versus employer as the shared feature) impact how sentiment towards sensory enhancement changes over time (the sentiment-time curve)? Any product needs product-market fit, but it’s harder when there’s no precedent, and possibly strong pushback (e.g., “Auditory enhancement ruins the authenticity of the human experience.”).</p><h3><strong>Neuroanalytics</strong></h3><p><a href="https://sparkneuro.com/">SPARK Neuro</a> is an advertising optimization company: using EEG, galvanic skin response (GSR), facial expressions, and eye-tracking, they derive two simple signals about user engagement with advertising content. SPARK generates an attention signal, ranging from 0 to 10, and an emotion signal, ranging from -10 to 10. Customers can submit video clips or other creative material for testing, and specify their desired demographic. SPARK, in turn, will send the company a monthly engagement report about their creative content. This is made possible because SPARK is constantly running subjects.</p><h3><strong>Takeaways</strong></h3><ol><li>The use of neuroanalytics to create digital content is like “indirect neurofeedback.” Instead of directly giving neuro/biofeedback to a user and using that to directly achieve a desired mental state, there’s an intermediary — the advertisement — which gets adjusted (not in realtime) to maximize a desired mental state or behavior in the user.</li><li>In spirit, this is no different from what advertising, speech-writing, book-writing, etc. have always tried to do: maximize engagement.</li><li>The question, therefore, is whether or not there’s something fundamentally different about using neurotechnology to enhance the engagement-optimization process. In the eyes of policy and ethics, should this be treated differently than focus groups and surveys?</li><li>Let’s suppose that SPARK and its customers want to understand engagement to an advertisement for a very specific demographic. Further, let’s suppose that neural responses are relatively similar between demographics (which is plausible given <a href="https://www.nature.com/articles/s41467-017-02722-7">this research</a> demonstrating that similar neural responses predict friendships). Additionally, in order to get around the thorny issue of whether or not it’s ethical to use knowledge of neural signals to influence someone’s behavior, let’s suppose a law is enacted that requires user consent before neural signals can be used, for example, in a mobile application to influence the user’s behavior. In this particular case, the neural signals of SPARK’s demographically-matched test subjects become proxy signals for people who haven’t yet given consent. How should the neurotechnology community handle this scenario?</li><li>In a <a href="https://spark.adobe.com/page/kEF8uXzmAUkhL/">video demonstration</a> from SPARK, their engagement metrics are overlaid on an advertisement and the results are surprisingly intuitive. In a sense, any kind of neuroanalytics will yield a more quantitative version of what humans already try to do qualitatively with our social brains. The difference between behavioral analytics and neuroanalytics is that neuroanalytics uses direct physiological measures, which in principle are better representations of the underlying system of study: the brain.</li></ol><h3><strong>Use-Cases and Market Size</strong></h3><p>The global advertising market is expected to be $558B in 2018, and given that advertising is a statistics game (“What percentage of viewers will buy this product after viewing the ad? Will that percentage, multiplied by the total viewership, multiplied by the profit from the product sale, outweigh the advertising spend?”), the higher the purchase conversion rate can go, the better. Beyond this, it’s difficult to come up with a sound heuristic to size the neuroanalytics market, so we omit a market size here.</p><h3><strong>Conclusion</strong></h3><p>In this final Volume of our Consumer Neurotech series, we took a look at some consumer neurotech that falls outside of our previous classification system. While this post concludes our deep-dive on consumer neurotech, stay tuned for a follow-up post where we discuss some high-level conclusions from our research.</p><p><em>Disclaimer: We actively write about the themes in which we invest or may invest: virtual reality, augmented reality, artificial intelligence, and robotics. From time to time, we may write about companies that are in our portfolio. As managers of the portfolio, we may earn carried interest, management fees or other compensation from such portfolio. Content on this site including opinions on specific themes in technology, market estimates, and estimates and commentary regarding publicly traded or private companies is not intended for use in making any investment decisions and provided solely for informational purposes. We hold no obligation to update any of our projections and the content on this site should not be relied upon. We express no warranties about any estimates or opinions we make.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=2d6fa504c051" width="1" height="1" alt=""><hr><p><a href="https://medium.com/the-substrate/consumer-neurotech-volume-v-2d6fa504c051">Consumer Neurotech: Volume V</a> was originally published in <a href="https://medium.com/the-substrate">The Substrate</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Consumer Neurotech: Volume IV]]></title>
            <link>https://medium.com/the-substrate/consumer-neurotech-volume-iv-6c1d8dae96f0?source=rss----10fba766a0f5---4</link>
            <guid isPermaLink="false">https://medium.com/p/6c1d8dae96f0</guid>
            <category><![CDATA[neuroscience]]></category>
            <category><![CDATA[brain-computer-interface]]></category>
            <category><![CDATA[venture-capital]]></category>
            <category><![CDATA[neurotechnology]]></category>
            <dc:creator><![CDATA[Loup Ventures]]></dc:creator>
            <pubDate>Thu, 18 Oct 2018 19:37:19 GMT</pubDate>
            <atom:updated>2018-12-05T17:33:24.114Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/820/0*w5eChvA216yOU2Sz.png" /></figure><p><em>By Avery Bedows</em></p><h4>Neurotech for Brain Modulation</h4><p>In <a href="https://loupventures.com/consumer-neurotechnology-volume-i/">Volume I</a> of this series on Consumer Neurotech, we established a framework for thinking about Consumer Neurotechnologies and their value propositions. In <a href="https://loupventures.com/consumer-neurotech-volume-ii/">Volume II</a>, we discussed neurotechnologies that have the value proposition of <strong>Control</strong> (or, using our taxonomy from Volume I, neurotechnologies whose “Effect-type” is Control). In <a href="https://loupventures.com/consumer-neurotech-volume-iii/">Volume III</a>, we analyzed neurotech with an effect-type of <strong>Feedback</strong>. Here, in Volume IV, we’ll explore neurotechnologies that directly modulate the nervous system.</p><h3>Neurostimulation for Learning</h3><h4>Fiction</h4><p><em>Eight squats left. Seven. Now, six. You silence the verbal monologue ringing in your head, attempting to focus all of your awareness on visualizing the flatness in your back, helping to construct the perfect angle with your upper legs as you lower yourself. You shift your focus to your feet and micro-adjust so your weight is distributed in the sweet spot, a little in front of your heel. Four squats left. You let the high-tempo electronic music take hold and carry you the rest of the way. Last one, and you put the bar back on the rack. Now you peel off your headphones — you run your fingertip over the electrodes in the band while you catch your breath. Training for your powerlifting competition is a game of masterful repetition. It can be hard to see the improvement day over day, and part of your will to continue comes down to belief in the power of addition: small improvements sum over time to noticeable changes, to heavier plates on the bar. Today, though, you notice that your squats felt meaningfully easier. The competition is in two months, and you can feel yourself on the path to your new personal record.</em></p><p>• • •</p><p>“Learning,” no matter which of its many conceptions we consider, ultimately comes down to cellular and molecular-level changes in the nervous system. As such, it’s reasonable to ask what would happen if we could make these changes occur more efficiently, or if we could make them more likely to occur? Below, we review one such piece of technology that takes the approach of modifying the brain at a molecular level in order to make it easier to learn motor tasks.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/270/0*oFbsPnfJJX26ZL99.jpg" /><figcaption>Image: <a href="https://www.haloneuro.com/get-halo-sport">https://www.haloneuro.com/get-halo-sport</a></figcaption></figure><h4>Halo Sport</h4><ul><li>Medium of innovation: <strong>Hardware</strong></li><li>Device-type: <strong>Stimulate</strong></li><li>Form-factor: <strong>Headphones</strong></li><li>Effect-type: <strong>Neuromodulation</strong></li></ul><p>The <a href="http://haloneuro.com/">Halo Neuroscience</a> headset is a $600 neurostimulation device that uses transcranial direct current stimulation (tDCS). Halo has developed a set of headphones that feature electrodes to perform the tDCS; the headphones’ band typically sits over the motor cortex, and therefore the tDCS impacts the part of the wearer’s cortex that controls movement. According to Halo, the tDCS “primes” neurons in the motor cortex so they become more plastic — when an athlete wears the headphones before and during a workout, they can learn movements more quickly. Halo focuses their product on elite athletes who consistently train, and they’ve conducted decent studies comparing physical improvement of athletes who do and don’t use Halo’s neurostimulation. A large part of Halo’s strategy appears to be sponsoring athletes and sharing success <a href="https://www.youtube.com/watch?v=uW-ltuS6rXg">stories</a>. Although Halo’s primary marketing is around athletics, in <a href="https://www.youtube.com/watch?v=fVUvgUSX9hU">this video</a> a pianist describes the apparent benefits of using the Halo headphones while learning a new piano piece.</p><h3><strong>Takeaways</strong></h3><ol><li>Halo operates on a bold principle: “We run electrical current through your brain, and as a result, you learn more quickly” (our phrasing, not Halo’s). For companies like Halo that claim to enhance their users, there is a need to prove the efficacy of the product, because the knee-jerk reaction to such claims is intense skepticism. This proof can take either or both of two forms: anecdotal/qualitative and quantitative. Halo has undertaken both, through their sponsorship of influencer athletes (anecdotal/qualitative) and through conducting scientific studies of their product to provide data (quantitative).</li><li>To the above point, an open question is the extent to which willingness to buy a consumer neurotechnology product will be based on testimonial, in contrast with scientific data. This will certainly depend on the particular product and market; as such, it’s advisable for consumer neurotechnology companies to quantitatively assess the preferences of their target audience through a survey or other methodology.</li><li>The gold standard for evidence of product efficacy is evidence provided by a third-party. For a product like the Halo, where there isn’t much at stake in the decision to use or not us the product, this isn’t a huge deal. However, setting a precedent is important because there are technologies for which third-party verification is critical, as in the case of neuroimaging applied to <a href="http://www.noliemri.com/">lie detection</a>, where non-commercial academic review has found <a href="https://www.sciencedirect.com/science/article/pii/S136466130700335X">significant problems</a> with the first-party commercial claims.</li><li>Although with this device, the risk faced by the user is mostly limited to a financial risk (which they mitigate with a 30-day satisfaction guarantee — although, is 30 days long enough to experience the benefits of the headphones?), an ethical question still arises: what responsibility does Halo or any other consumer neurotechnology company have to establish and communicate to potential users the variability in its effectiveness? With a 30-day money-back guarantee, this question isn’t all that important. It becomes more significant, though, looking forward to the distant future when invasive neurotechnology could be used for purposes of enhancement. This also bears resemblance to the developing and marketing drugs that might be toxic for some, but highly effective for others.</li><li>Related to the point above, this inter-individual variability is going to be a marketing problem. If neurobiological and engineering constraints make it such that a product will only work for 50% of its users, reviews for the product will be highly polarized: imagine a barrage of 1-star reviews from users who see no results with Halo, and a set of 5-star reviews from athletes who’ve seen their performance improve markedly. How should Halo manage this marketing conundrum?</li></ol><h4>Use-Cases and Market Size</h4><h4>Athletics</h4><p>Halo’s primary use-case for motor learning enhancement is athletics. To estimate the US market size, we split up our analysis between college athletes and professional athletes; it’s possible that non-collegiate and non-professional athletes would purchase a Halo, but given the $600 price-point, the inherent product skepticism, and the fact that amateur athletes usually don’t track their performance as closely as collegiate and professional athletes — we assume this market will be small, and therefore omit it from our analysis.</p><p>Broadly, we model this as <strong>[Number of Athletes] x [Market Penetration] x [Average Selling Price]</strong>. We estimate the number of <a href="https://www.bls.gov/ooh/entertainment-and-sports/athletes-and-sports-competitors.htm">professional</a> and <a href="https://www.ncaa.org/sites/default/files/Recruiting%20Fact%20Sheet%20WEB.pdf">collegiate</a> athletes and use the price of the Halo headset as our ASP estimate. We make assumptions for the market penetration growth curves, although our models are easily adjustable to accommodate different values for these assumptions.</p><p>With conservative market penetration assumptions, we estimate that the annual market for enhanced motor-learning will be only <strong>$500,000 for professional athletes in 2025</strong> and <strong>$12,090,000 for collegiate athletes in 2025</strong>. Both of these numbers are smaller than the grand-slams sought in venture capital-funded businesses. The reason these markets are so small is that the total addressable market is limited — simply, there aren’t very many athletes. In order to increase revenue given the small TAM, Halo could introduce an annual subscription.</p><h4>Music</h4><p>Even though Halo has official marketing material around the musician use-case, we don’t think this market will take off. Musicians are probably less inclined to purchase performance-enhancing gear than athletes are; the field of athletics is already receptive to optimization (e.g. using fluid dynamics to design swimsuits), and a core principle of athletics is metrics-based competition. In contrast, music is aesthetic and appraised as art, sans quantification. We think this serves as a useful example of an important point: <strong>consumer neurotechnology products that emphasize enhancement should target markets that have an “optimization mentality.”</strong> Without this mentality, the value proposition doesn’t impress, and might even yield disdain.</p><h4>Surgery</h4><p>Transcranial direct current stimulation has been <a href="https://www.ncbi.nlm.nih.gov/pubmed/29951631">investigated</a> for its utility in training surgeons. The reasoning goes: <em>Surgeons must use fine motor control during surgery, so therefore if we make it easier for the motor cortices of surgical students to adapt to new movements, then perhaps we can train surgeons more effectively.</em></p><p>As of 2008, there were <a href="http://www.acshpri.org/documents/ACSHPRI_Surgical_Workforce_in_US_apr2010.pdf">136,000</a> active surgeons. The surgical student market is even smaller than the professional athlete market, so we assume that any company (Halo or otherwise) who wanted to enter this market with a tDCS product would need a subscription model to be profitable. Given that there are <a href="https://en.wikipedia.org/wiki/List_of_medical_schools_in_the_United_States">141 medical schools</a> in the U.S. and generating a back-of-the-napkin estimate for the number of surgical students per medical school (based on other statistics), in addition to estimating the ASP to be the price of the Halo headphones and estimating an annualized subscription cost of $300/head/year, our estimates show the <strong>2025 market to be only</strong> <strong>$520,000.</strong></p><p>Each of the markets we’ve covered have been, by themselves, fairly small. The Halo device and its particular constraints and business models represents only one take on a broader concept: upregulating human learning. Halo tackles the learning of movements; one can imagine future neurostimulation technology that makes it easier to learn to recognize visual patterns (helpful in e.g. pathology or radiology) by stimulating the visual cortex. We could further envision the upregulation of even higher-order learning — in a sense, we already have “technology” that does this: caffeine. Even though our projections have Halo’s markets as potentially small, the sheer number of domains where human learning is relevant (and the size of the academic and professional education markets) makes for a large opportunity.</p><h3>Neurostimulation for Affective State</h3><h4>Fiction</h4><p><em>Every Wednesday, it seems you rediscover the meaning of “Hump Day.” That mid-week hump hits right around 3pm — emerging from the food coma, you dwell on the thought of still having two days left before your severely-needed R&amp;R. And that’s not even thinking about the essay you promised you’d edit for that friend who decided to apply to grad school after a decade in the workforce…and how is it, again, that you’ll finish the 20 more slides on this presentation before going home? Coffee is an option, but you’ve already had two cups today and you feel yourself approaching addiction. You sit for a few minutes and think, willing the haze to clear slightly so you can arrive at a plan of action. Suddenly, it dawns on you. You reach into the left drawer on your desk and fumble around with your fingertips for a few seconds, searching for the pod-like shape among the sea of pencils and dangerous papers eager to doll out paper cuts indiscriminately. You grab the device, pull it out, and nestle its curved adhesive pad onto the matching curvature midway down your neck. After opening the companion app, and setting the Energy intensity to level 7, the tingling underneath the adhesive pad starts to kick in. Sixty seconds later, you peel the stimulator off, return it to its home quickly, and crisply close the drawer. As you pull your chair up closer to your desk with a game-face on, you think you might be feeling slight butterflies about getting to give this presentation to the CEO tomorrow.</em></p><p>• • •</p><p>One obvious direction for neurotechnology to explore is that of modulating emotion and physiological state. We refer to these collectively as “affect modulation,” although we acknowledge this terminology doesn’t capture the difference between emotion and its physiological manifestation; rather, we operate under the assumption that physiology may be used to deduce and modulate emotional state. In the above scenario, we describe what it might be like to one day use a consumer-available neurostimulation device that generates a sense of energy.</p><p>Here, we’ll discuss two examples of devices that aim to achieve affect modulation through neurotechnology (we’re exclusively considering technologies that do this by electrically stimulating the nervous system; importantly, we <em>aren’t</em> considering ingestible substances like caffeine and other drugs).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/0*qqD8xmKFhqaHBPPw.jpg" /><figcaption>Image: <a href="https://experiencenervana.com/blog/product/nervana-generator-and-headphones-unit-complete-set/">https://experiencenervana.com/blog/product/nervana-generator-and-headphones-unit-complete-set/</a></figcaption></figure><h4>Nervana</h4><ul><li>Medium of innovation: <strong>Hardware</strong></li><li>Device-type: <strong>Stimulate</strong></li><li>Form-factor: <strong>Headphones</strong></li><li>Effect-type: <strong>Physiological</strong></li></ul><p>The first product is called <a href="https://experiencenervana.com/">Nervana</a>, a non-invasive vagus nerve stimulation device that claims induce a sensation of relaxation and calm. The device, which costs $300, synchronizes vagus nerve stimulation with audio of the user’s choice: the user plugs their phone, MP3 player, etc. into the Nervana Generator using an audio jack, and the Generator outputs to Nervana’s headphones, which contain electrodes in the earpiece to provide electrical stimulation to the user. We want to note that Nervana provides <strong>no scientific data</strong> about the effectiveness of their product, so we have no idea if it works or not; we’re including it since it’s an interesting concept.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/0*VMAMyzL7uFr4xwcu.png" /><figcaption>Image: <a href="https://www.slashgear.com/thync-wearable-claims-it-can-change-the-way-you-think-03386499/">https://www.slashgear.com/thync-wearable-claims-it-can-change-the-way-you-think-03386499/</a></figcaption></figure><h4>Thync (no longer sold)</h4><ul><li>Medium of innovation: <strong>Hardware</strong></li><li>Device-type: <strong>Stimulate</strong></li><li>Form-factor: <strong>General Headset</strong></li><li>Effect-type: <strong>Physiological</strong></li></ul><p>The second device for modulating affective state was manufactured by Thync. <a href="https://www.thync.com/">Thync</a> is a non-invasive neurostimulation company. Their original product, the Thync Calm and Energy Wearable, was intended to evoke sensations of energy or relaxation. Their second product, the Thync Relax Pro, was specifically focused on inducing sensations of relaxation. According to their website, Thync has since pivoted into treatment of psoriasis, an autoimmune skin condition, for which they’re conducting clinical trials and will pursue FDA clearance as a medical treatment. Our discussion here will be limited to the non-medical uses of the original two Thync devices. It’s unclear from public information why Thync pivoted, and therefore it’s difficult to take lessons from it.</p><p>The Thync device, at least for the relaxation mode, works by stimulating afferent fibers of the <a href="https://en.wikipedia.org/wiki/Facial_nerve">facial</a> and <a href="https://en.wikipedia.org/wiki/Trigeminal_nerve">trigeminal</a> nerves, as well as cervical <a href="https://en.wikipedia.org/wiki/Spinal_nerve">spinal nerves</a> (C2/C3). The device used an adhesive pad to stick a stimulator to one side of the user’s forehead (facial and trigeminal nerves) and an additional electrode attached to the back of the user’s neck (cervical spinal nerves). According to research <a href="https://www.nature.com/articles/srep13865">published</a> by Thync in the journal Nature, their neurostimulation technique has downstream effects on the physiological mediators of stress. For those savvy with neuroanatomy and neurophysiology, this diagram from the paper delineates a possible pathway for the stimulator to achieve its effects. Note that “TEN” stands for “transdermal electrical neurosignaling,” which is the fancy name Thync uses to distinguish its technology from other neurostimulation techniques.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/290/0*hDx8ZUdw4C5vyDYb.jpg" /><figcaption>Image: <a href="https://www.nature.com/articles/srep13865">https://www.nature.com/articles/srep13865</a></figcaption></figure><p>The original device cost $200, although the adhesive strips cost more (we aren’t sure what the specific price was). The Relax Pro cost $150, with a $30 monthly subscription to continuously send users the adhesive pads. Note that neither of these devices are available anymore — the reviews we’ve found are from early 2017. In terms of efficacy, reviews like <a href="https://www.technologyreview.com/s/535641/the-electric-mood-control-acid-test/">this one</a> from MIT Technology Review note that the original Thync devices worked for some people, but not for others.</p><h3>Takeaways</h3><ol><li>As with Halo, Thync demonstrates the fact that consumer neurotechnology will have to deal with severe value proposition variability. The products will work for some people, but not for others. This comes down to differences in the nervous systems between individuals, so it’s going to be tough to mitigate.</li><li>The value of this product is directly measured in emotional and subjective experience, in contrast to normal consumer electronics like cellphones, or even consumer neurotechnology like the Halo headphones whose value will probably fuse the quantitative (<em>My maximum squat went up by 10%</em>) with the qualitative (<em>My workout felt great when I wore the headphones</em>). The closest analogue is narcotics. On one hand, if affect-modulation technologies work, then they’re likely less risky than drugs since they don’t require inputting substances to the body. On the other hand, to the best of our knowledge, it’s unknown whether relaxation effects will attenuate over time. The attenuation could cause addiction behavior (more stimulus required for the same reward); but, if there isn’t attenuation, and the reward is consistent…would this prevent addictive behavior? These questions are worthwhile basic science questions.</li><li>Should affect modulation companies have an obligation to investigate whether their products can become addictive? There’s probably no legal imperative, but there might be a PR imperative as the expectation of technology companies rightfully moves toward building products that are conscious of their emotional impact.</li></ol><h4>Use-Cases and Market Size</h4><p>To build our model for the affect modulation market, we split the market into three subcategories: Positivity, Relaxation, and Energy. This could easily be extended to include other categories like Focus. We assume that devices will cost $200 (based on the original Thync), and most would employ subscription models for discardable physical components like adhesive pads and companion software — we estimate this as $15 / mo. Our model makes assumptions for market penetration split up by each of the three categories, and factors in churn.</p><p>Importantly, market penetration is relative to the “receptive subset” of the total addressable market. For example, if the total addressable market is every individual in the United States, only a subset of these individuals will be receptive to the idea of using neurostimulation to feel more positive or feel more relaxed. We guess these numbers right now, but we (or anyone else) could easily conduct a survey to replace our guesses with empirical point-estimates. Additionally, we make the limiting assumption that the receptive subsets of the TAM will be constant from 2018–2025 (the extent of our projections); this assumption is invalid because the receptive subset will probably grow over time as affect-modulation products become more popular. This growth rate could also be point-estimated empirically, although it would be more challenging. One additional assumption that could lead to an overestimate is that we consider Positivity, Relaxation, and Energy to require separate devices; it’s possible that, as with the original Thync, more than one of these effects will be achievable by a single device.</p><p>With the above considerations, noting conservative estimates for market penetration, we estimate that the <strong>U.S. market for affect-modulation technologies in 2025 will be $499,000,000</strong>.</p><h3>Conclusion</h3><p>In Part IV of our Consumer Neurotech series, we discussed consumer neurotech that directly modulates the human nervous system. In Part V, we’ll address some technologies that fall in a wildcard category.</p><p><em>Disclaimer: We actively write about the themes in which we invest or may invest: virtual reality, augmented reality, artificial intelligence, and robotics. From time to time, we may write about companies that are in our portfolio. As managers of the portfolio, we may earn carried interest, management fees or other compensation from such portfolio. Content on this site including opinions on specific themes in technology, market estimates, and estimates and commentary regarding publicly traded or private companies is not intended for use in making any investment decisions and provided solely for informational purposes. We hold no obligation to update any of our projections and the content on this site should not be relied upon. We express no warranties about any estimates or opinions we make.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=6c1d8dae96f0" width="1" height="1" alt=""><hr><p><a href="https://medium.com/the-substrate/consumer-neurotech-volume-iv-6c1d8dae96f0">Consumer Neurotech: Volume IV</a> was originally published in <a href="https://medium.com/the-substrate">The Substrate</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Consumer Neurotech: Volume III]]></title>
            <link>https://medium.com/the-substrate/consumer-neurotech-volume-iii-21e669351f64?source=rss----10fba766a0f5---4</link>
            <guid isPermaLink="false">https://medium.com/p/21e669351f64</guid>
            <dc:creator><![CDATA[Loup Ventures]]></dc:creator>
            <pubDate>Thu, 18 Oct 2018 19:31:49 GMT</pubDate>
            <atom:updated>2018-12-05T17:34:57.308Z</atom:updated>
            <content:encoded><![CDATA[<h4>by Avery Bedows</h4><p>In <a href="https://loupventures.com/consumer-neurotechnology-volume-i/">Volume I</a> of this series on Consumer Neurotech, we established a framework for thinking about Consumer Neurotechnologies and their value propositions. In <a href="https://loupventures.com/consumer-neurotech-volume-ii/">Volume II</a>, we discussed neurotechnologies that have the value proposition of <strong>Control</strong> (or, using our taxonomy from Volume I, neurotechnologies whose “Effect-type” is Control). In this third Volume, we’ll learn about and analyze neurotech that leverages a phenomenon known as “neurofeedback.” Neurofeedback is a straightforward idea: <em>record</em> some aspect of neural activity, and feed that activity back to the brain through one of the <em>senses</em>. By doing so, the neurofeedback can help a user achieve a desired brain state.</p><h3><strong>Neurofeedback for Attention</strong></h3><h4>Fiction</h4><p><em>This morning, walking into the espresso bar en route to work, you ran into a friend you hadn’t seen in months. When you’d seen her last, she told you she thought her boyfriend might propose soon. This morning, you updated her on your project at work, your son’s first words, and other little things, but you could tell she was waiting. Finally, you had yielded the microphone and she promptly broke the news by holding up her hand. When you saw a glistening engagement ring, you felt a rare, precious vicarious excitement. You were chatting away excitedly but realized if you stayed any longer you’d be late, so you hurried over to the office riding the high on life. You’ve finally just now squeaked into your swivel chair, and your priorities are evident: text friends about the engagement first, second, third…get work done later. The adult in you tries to wrestle control back, and you start listening to music through your headphones. When you first put them on, they’re constantly dinging, telling you that you’re unfocused. The dings work like chimes in wind, catching your attention and slowly zoning you into trance-like productivity. 15 minutes later, you’re in flow, and there’s a small smile from this morning still cemented onto your face.</em></p><p>• • •</p><p>One well-known application of EEG is for assessing focus and attention: certain patterns in EEG signals indicate greater or lesser degrees of attention. Given the volume of research, patents, and practical utility of attention enhancement, we’re likely to see an abundance of products that leverage these signals. Here, we’ll discuss two such products on the market.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/0*xq8ZI590qLttf7uL.jpg" /><figcaption>Image: <a href="https://www.kickstarter.com/projects/mindset/headphones">https://www.kickstarter.com/projects/mindset/headphones</a></figcaption></figure><h4><strong>Mindset</strong></h4><ul><li>Medium of innovation: <strong>Integrated</strong></li><li>Device-type: <strong>Record</strong></li><li>Form-factor: <strong>Headphones</strong></li><li>Effect-type: <strong>Feedback</strong></li></ul><p><a href="https://www.thinkmindset.com/">Mindset</a> is a pair of headphones that integrates EEG sensors into the headband. The goal of the $280 headphones is, as per their website, “[With] our powerful EEG technology, you will train your subconscious to recognize and tune out the things that distract you.” According to Mindset’s <a href="https://www.thinkmindset.com/science/">explanation</a> of the science behind their product, they use the principle of neurofeedback to provide an auditory cue through the headphones whenever a user loses focus. Through this feedback, over time, the user’s brain will become trained to stay more focused. Mindset is launching in the coming months; preorders are active on their website.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/0*qhKdmYqz4mnydPC6.jpg" /><figcaption>Image: <a href="https://www.kickstarter.com/projects/1571399980/neuroplus-improve-focus-with-brain-sensing-headset">https://www.kickstarter.com/projects/1571399980/neuroplus-improve-focus-with-brain-sensing-headset</a></figcaption></figure><h4>NeuroPlus</h4><ul><li>Medium of innovation: <strong>Integrated</strong></li><li>Device-type: <strong>Record</strong></li><li>Form-factor: <strong>General Headset</strong></li><li>Effect-type: <strong>Feedback</strong></li></ul><p><a href="https://www.neuro.plus/">NeuroPlus</a> is a $100 EEG device that’s paired with attention-enhancing games. Like Mindset, NeuroPlus <a href="https://www.neuro.plus/children.html">leverages neurofeedback</a>: “[Users] play brain-controlled video games that give real-time feedback on how well they’re paying attention. The more they focus, the better they’ll do in a variety of fun, interactive training games. Over time, this practice leads to improved focus in everyday life.” Users are granted access to the games for $30 / mo. for an individual, $50 / mo. for a group of 2–4 people, and each additional person beyond the fourth adds $10 / mo. to the subscription price. A large part of NeuroPlus’s target market appears to be children with ADHD, and they include a 60-day money-back guarantee so that users can have a fair shot at determining whether the device and software work for them.</p><h3><strong>Takeaways</strong></h3><ol><li>Mindset and NeuroPlus both integrate attention training into pre-existing use-cases: for Mindset, it’s music-listening; for NeuroPlus, it’s gaming. This is a theme in consumer neurotech: it’s useful to find pre-existing contexts in which to use the technologies. To show how this makes sense, suppose we’re comparing Mindset with Thync (a device we’ll discuss in Volume IV that attaches to the user’s temple and stimulates nerves — the goal is to help a user relax or feel energized, i.e. to modulate their affect). Mindset benefits because users have already developed behaviors for using headphones and opening up an application (e.g. Spotify) when they want to wear their headphones. In contrast, Thync requires the new behavior of placing adhesive strips on the device, then putting the device itself onto their forehead with no purpose other than to modulate affect; this is a high-friction requirement. We suspect that not only will it lower friction to deliver neurotech value through pre-existing use-cases, but it will also make prospective users more comfortable with the idea of technology that records or stimulates neural signals — comfort is born of familiarity.</li><li>The fact that neurotechnology <em>can</em> be packaged into existing use-cases will be a boon to its uptake. Contrasting with mixed and virtual reality, neurotechnology doesn’t need a “killer use-case” because it will piggyback on use-cases that are <em>already</em> killer, such as listening to music.</li><li>Mindset takes advantage of the fact that they can place EEG sensors into hardware that people are already used to using (<a href="https://www.haloneuro.com/">Halo Neuroscience</a> does something similar, although Halo’s headphones are less ergonomic and some athletes don’t wear massive over-ear headphones while working out, perhaps to avoid pooled sweat). NeuroPlus, in contrast, requires a dedicated but low-profile EEG headband.</li><li>It’s interesting that the actual usage of neurotechnology generally recedes into the background: <strong>neurotechnology isn’t necessarily the primary experience a user is aware of when interacting with a product</strong>. This is true of the CTRL-kit, the Muse (see below), the Halo headphones, the Mindset headphones, the NeuroPlus games, and more. Consider that even the semantics of “augment” or “enhance” make this evident: we have to augment <em>something</em>, enhance <em>something</em>. These verbs need their nouns.</li></ol><h4><strong>Use-Cases and Ma</strong>rket Size</h4><p>We’ve split the market size here into two categories: attention training for people with ADHD (like NeuroPlus), and attention training for inducing flow state (like Mindset).</p><h4>ADHD</h4><p>For a given year, we can estimate this market size by combining a total addressable market scaled for market penetration with the price of the hardware and annualized cost of a subscription. <a href="https://www.addrc.org/adhd-numbers-facts-statistics-and-you/">6.7M children</a> in the United States have been diagnosed with ADHD, so we use this as the TAM in our calculations. We use NeuroPlus’s price point of $100 for the hardware, and a 12 x $30 / mo. = $360 annual subscription cost. We estimate that <strong>in 2025, the American market for ADHD-targeted attention training will be $113,000,000</strong>. In 2003, there was <a href="https://www.healthaffairs.org/doi/pdf/10.1377/hlthaff.26.2.450">over $2B</a> of U.S. spending on ADHD medication, so the ceiling on this market is high.</p><h4>Flow</h4><p>Given the ubiquity of using headphones to help focus, we assume for this calculation that headphones will be the most common form-factor for focus enhancement. As such, our starting point is that the smart headphone market is <a href="https://www.statista.com/statistics/694067/smart-headphone-market-size-by-product-in-the-us/">expected</a> to be $0.995B in 2018 and $2.825B by 2022. In 2016, Bose had 8% of <a href="https://www.statista.com/chart/7993/headphone-market-share/">wireless headphone sales</a>. We’ll assume that this market share stays constant. A pair of Bose QC35 headphones cost $350, so using the 2018 figure from above, we can estimate that Bose sold roughly 227,430 headphones in 2018. In 2022, that would mean 645,700 headsets sold assuming constant market share. With linear growth, this would grow to 960,700 Bose headphone sales in 2025. We assume the size of the Bose market represents the upper bound for the total number of headphones that will incorporate neurofeedback features by 2025 (including and in addition to Mindset). We use an exponential curve to model the growth of neurofeedback headphone sales from 2018 (estimated at 5,000) to 2025 (estimated at the upper bound of 960,700). Using this as our model, we estimate that <strong>in 2025 there will be $269,000,000 in neurofeedback headphones sold </strong>for the purpose of more easily entering flow and staying focused during productivity sessions.</p><h3><strong>Neurofeedback for Meditation</strong></h3><h4>Fiction</h4><p><em>Every morning, you wake up and tell yourself with stern-voiced and earnest conviction that when you come home this evening, you’ll write another page of the short story whose idea was seeded months ago by a chance interaction on the street. Every night, you come back from work carrying a boulder’s weight of tension in your shoulders, and the creative in you sits cowering in his corner, terrified of the stress and refusing to be coaxed out. Tonight is no different: you walk in the door, take your backpack off, sling it around and let it slam down on the floor. Every morning, your glance at your laptop is loving and longing, wishing for the chance to sit down and have a conversation with a blank screen, mediated by keys. Every night, with this one as no exception, your eyes seem to have the same magnetic polarity as your computer: by a mysterious cognitive law of physics, you can’t bring yourself to look at it. The only escape from this internal battle is to sit down on your couch. You close your eyes, wishing that energy and motivation would return to bring you back to your early-morning fervor. Didn’t you hear on a podcast that meditation can help clear the head after a long day? Wait — you still haven’t opened that meditation headband your mom curiously gifted you for your birthday three months ago. You dig it out from under accumulated magazines and daily material acquisitions and follow the instructions for set-up. A little while later, after fumbling around through the on-boarding in your tired state, you thank yourself for purchasing such a well-crafted comforter as you lay back with the headset on and the app open, ready to give it a try. Thirty minutes later you emerge and feel strangely refreshed. Tonight, you decide, you’ll sleep — but tomorrow, you have a new approach to tackle that tension and maybe, just maybe, you’ll wind up in an invigorating tango with a word processor, bringing that street-encounter to literary life.</em></p><p>• • •</p><p>In a similar vein to its application to attention, EEG-based neurofeedback paradigms can be applied to meditation as well.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/0*2-Gk8wNEEgDBMd8D.jpeg" /><figcaption>Image: <a href="https://medium.com/pixeldreams/my-40-day-journey-into-meditation-with-muse-the-brain-sensing-headband-12e7b060c6fa">https://medium.com/pixeldreams/my-40-day-journey-into-meditation-with-muse-the-brain-sensing-headband-12e7b060c6fa</a></figcaption></figure><h4>Muse</h4><ul><li>Medium of innovation: <strong>Integrated</strong></li><li>Device-type: <strong>Record</strong></li><li>Form-factor: <strong>General Headset</strong></li><li>Effect-type: <strong>Feedback</strong></li></ul><p>The major player in this space right now is <a href="http://www.choosemuse.com/">Muse</a>, built by InterAxon. The Muse is a $200 headband with EEG sensors and comes with a companion app to help users meditate more effectively. The app delivers sound to a user through their headphones and uses the real-time EEG signal to determine whether or not a user is in a meditative state. The app will change the sounds based on the EEG signal. Muse offers the Muse PRO software to professional therapists and meditation guides for ~$35 / mo. With the subscription comes access to Muse Connect, a SaaS platform for using Muse to assist with a professional practice. This subscription also includes a reduced price for practitioners’ clients who purchase the headset as part of a referral program, in addition to access to case studies, best practices, and more. Muse also has a developer program where developers can gain access to raw EEG data from the headset in order to use it for their own applications or for research.</p><h3><strong>Takeaways</strong></h3><ol><li>One of our team-members tried the Muse a couple years ago, and they were neutral-to-negative on the experience of the neurofeedback. Two important caveats are that a) the Muse and its software have likely improved since then, and b) our team-member had no guidance (and therefore lower motivation to keep trying). This latter point may, hypothetically, explain why Muse is marketing the device heavily through meditation guides and therapists: they might be finding that individual users aren’t using the product properly/consistently.</li><li>This insight is applicable to consumer neurofeedback as a whole: neurofeedback inherently takes time to work, and this means that users will have to use the products for a while <em>sans value add</em> (note the similarity to the exercise, dieting, and broader meditation industries). That’s a big ask to make of a user, and Muse’s way around this is to use another human — the professional meditation guide or therapist — as a forcing function. NeuroPlus takes a similar approach with the parents of children with ADHD. This contrasts with Mindset, whose product inherently adds instantaneous value by playing music, regardless of whether the neurofeedback actually works. While Mindset will only succeed in the long-run if the neurofeedback succeeds, the inherent value should make it easier for users to try out Mindset for a sufficiently long period to see improvements.</li><li>Control devices like CTRL-kit or various Commoditized EEG applications have immediate value.</li><li>The meditation and self-care markets are rapidly growing (more on this below), so this could be a promising opportunity area in the sense that people are actually taking the time to use meditation apps like <a href="https://www.headspace.com/">Headspace</a> and <a href="https://www.calm.com/">Calm</a>, which require repeated usage to work. What are the success factors for these apps? How much effort do they require of the user? Why are users coming back? How long does it take for users to perceive the value-add? Perhaps the answers to these questions will be informative for neurofeedback products.</li><li>Neurofeedback products contrast with neuromodulation products like Nervana and Thync (to be discussed in Volume IV), which are “on/off” in that the user just has to turn them on in order to receive value from the product, whereas neurofeedback products take time. It’s important to consider the psychological impact of on/off technologies. A 2017 <a href="https://www.ncbi.nlm.nih.gov/pubmed/28223484">paper</a> by Weible et al. reported on an experiment where meditation-like brain states were switched on and off in mice using a neural interface. When these brain states were on, the mice showed behavior consistent with reduced anxiety. Extrapolating into the far future, neurotechnology could conceivably turn many things we currently have to work for into “on/off” experiences; we think it’s important to begin investigating the psychological and cultural impacts this shift might cause.</li></ol><h4><strong>Use-Cases and Market Size</strong></h4><p>We tailored this projection to Muse’s current business model of selling both direct-to-consumer and through professional meditation guides and therapists. There are <a href="https://www.webwire.com/ViewPressRel.asp?aId=214152">2450 meditation studios/centers</a> in the U.S., and we estimate three professionals per studio. The number of behavioral therapy practitioners is <a href="https://www.ibisworld.com/industry-trends/specialized-market-research-reports/life-sciences/health-practitioners/behavioral-therapists.html">~100,000</a>. Summed, these give us an estimate for the number of professionals who might purchase subscriptions to Muse PRO.</p><p>There are conflicting reports on the size of the meditation market: according to one source, <a href="https://nccih.nih.gov/research/statistics/NHIS/2012/mind-body/meditation">18M adults</a> in the US (8%) meditated in 2012. According to another, only <a href="https://www.webwire.com/ViewPressRel.asp?aId=214152">9.3M adults</a> (3.7% based on 2017 population) in the US meditated during the year-long period covering parts of 2016–2017. We’ll take the conservative approach and use 9.3M as our estimate for the number of meditators in the US right now (if the number of children and teenagers who meditate is non-trivial, then this is certainly an underestimate).</p><p><a href="https://www.webwire.com/ViewPressRel.asp?aId=214152">In 2017</a>, there were 1000+ meditation apps available, which conjointly with websites and online courses generated $100M+ in revenue. The total meditation market was $1.21B in 2017, and is expected to have 11.4% average annual growth to $2.08B by 2022. There seems to be big growth in meditation app uptake. Calm had MAU that <a href="https://blog.apptopia.com/self-care-apps-growing">grew 81%</a> YoY in 2017–2018; self-care app installs are up 36% YoY, and consumer spending in the top 10 self-care apps was up by 40% YoY.</p><p>To calculate how many digital meditation app users there are, we divide the market value for apps by the total market value: $100M / $1.21B = ~8.3%. This is probably an overestimate, since many meditators don’t pay for meditation. Therefore, if we’re going to use this as a proxy for figuring out the number of meditators, we need to scale the denominator accordingly. We estimate that the number of paying meditators is only 25% the number of total meditators, so we divide 8.3% by 4 to estimate that 2% of total meditation users use digital meditation apps. Finally, multiplying the 9.3M meditators of 2016–2017 by 2%, we arrive at 186,000 digital meditators in 2017 (since our math is loose, we’re comfortable removing the ambiguity of “2016–2017” from our estimate).</p><p>As described above, digital meditation is growing very quickly; hence, we’ll assume a 20% YoY growth sustained through 2025. All told, we calculate there will be ~666,500 digital meditators in 2025, which seems a sane estimate (only 7% of the 9.3M adult meditators in 2016–2017).</p><p>Finally, we scale this by market penetration and account for Muse’s business model to yield our final estimate of <strong>a $30,280,000 market in 2025</strong>.</p><h3><strong>Neurofeedback for Sleep</strong></h3><h4>Fiction</h4><p><em>You read about empathy often these days, and normally you think you’re fairly decent at it. You frequently imagine yourself as your waiter when you dine out with your colleagues, and act in a way you think will make the interaction meaningful and invigorating. You read competing views on political topics and feel torn by testimonials at odds with each other. Your empathy has a hard limit, though: those so-called “morning people” are full of it. Mornings are Atlassian at best, Sisyphusian at worst. You’re a different person for those 45 post-wakeup minutes than for the rest of the day. Your eyelids experience gravity more strongly than anything else, and your soul feels destined for the underworld. Mornings are, unequivocally, the worst. Two kind and clever colleagues of yours became tired of your morning mood and pitched in to purchase a headband that supposedly helps improve your sleep and your wake-up experience. That was two weeks ago. No, maybe three? Hold on a second…you’ve only been awake for five minute and you’ve been thinking about all of this — that never used to happen! Coherent thoughts at this hour? Unheard-of. You roll out of bed lithely (unprecedented) and open up your phone, checking your sleep stats. The graph is moving slightly, but definitively, upwards. Well, you think, I suppose I have an extra 30 minutes to kill, now. You glance over at the waist-high stack of books in the room’s corner. Might as well…</em></p><p>• • •</p><p>Sleep problems are ubiquitous, and as with attention and meditation, the underlying troubled system is the brain. Therefore, sleep is a prime candidate for neurofeedback using EEG and other biosignals.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/0*pZaZaxVW5S4tReNu.png" /><figcaption>Image: <a href="http://techapple.net/2017/08/dreem-sleep-solution-quality-asleep/">http://techapple.net/2017/08/dreem-sleep-solution-quality-asleep/</a></figcaption></figure><h4><strong>Dreem</strong></h4><ul><li>Medium of innovation: <strong>Integrated</strong></li><li>Device-type: <strong>Record</strong></li><li>Form-factor: <strong>General Headset</strong></li><li>Effect-type: <strong>Feedback, User Characterization</strong></li></ul><p><a href="https://dreem.com/">Dreem</a>, one of the leading companies using neurotech to improve sleep, estimates that across Europe and the U.S., 1/3 of the adult population has <a href="https://s3-us-west-1.amazonaws.com/rythm/rythm-dreem-whitepaper.pdf">poor sleep quality</a>. Dreem is a $500 headband that contains EEG sensors, an accelerometer, and a pulse oximeter to measure blood oxygenation. With these biosignals, Dreem uses algorithms and a companion app to add four types of value to users who wear the headband while they sleep: they fall asleep faster, get deeper sleep, wake up more refreshed, and can analyze their sleep over time and therefore improve it. Users wear the headband at night and can check their sleep stats in the morning on their phone.</p><h3><strong>Takeaways</strong></h3><ol><li>Among Dreem’s four discrete value propositions, three require no effort from the user: 1) to fall asleep faster, the headset provides audio cues. 2) Similarly, audio is used to facilitate deep sleep. 3) Dreem induces a gentle wake-up by playing the user’s alarm at different times depending on their sleep stage. Achieving the 4th value proposition of long-term sleep improvement, however, requires the user to consistently check in with the sleep data collected by the headset, and compare this to their lifestyle; ultimately, if a user wants to benefit from this information, they have to enact changes in lifestyle themselves. This duality of value proposition — effortless and effortful — probably works to Dreem’s advantage since even for users who don’t put in the effort, the device can still be useful.</li><li>7/60 reviews on <a href="https://www.trustpilot.com/review/www.dreem.com">TrustPilot</a> complained about the comfort of the Dreem device. Discomfort will reduce the addressable market for Dreem and any other head-worn consumer neurotech.</li><li>The Dreem companion app features a Sleep Score that characterizes how well a user slept. Referring to this score, one <a href="https://dreem.com/en/sleep-coach">testimonial</a> from their website reads, “It’s very addictive! It’s very interesting to see the scores of the night.” Quantitative self-improvement can become addictive — what are the ethics around using biosignals as the quantitative value that causes addictive usage behavior, as opposed to external signals like a score in a video game?</li><li>Dreem offers a high-quality <a href="https://s3-us-west-1.amazonaws.com/rythm/rythm-dreem-whitepaper.pdf">white paper</a> about the science behind their product. The paper contains information about how the biosignals are processed, where and how Dreem uses algorithms, and the efficacy results. In our view, this ought to be the gold standard for articulating the science behind a neurotechnology product. It has ethical value, and it’s also a smart business decision: customers will be more comfortable using the product, investors will be more comfortable investing, and influencers will be more comfortable influencing.</li></ol><h4><strong>Use-Cases and Market Size</strong></h4><p>The total addressable market for sleep products is large: <a href="http://maxime.elbaz.free.fr/nos_actions/Insomniie-%25E9pid%25E9mio-internationale.pdf">in the United States</a>, 131M people have sleep problems. This TAM should be scaled back by a factor representing the proportion of people who can sleep comfortably with a head-worn device. To do this, we use the fact that 7/60 reviews on TrustPilot mentioned discomfort. We round this up to 10/60 to make the clean estimate that 1/6 people will find the headset too uncomfortable to sleep with. This would shrink the TAM to 109.2M people.</p><p>We estimated market penetration ad-hoc based on our sense of what’s reasonable. We suppose that 10,000 Dreem units are sold in 2018, and 1,000,000 units are sold in 2025. We also assume the ASP of this product will decrease over time from $500 to $200. <strong>Our estimate for Dreem’s wearable sleep enhancement market in 2025 is $200M</strong>. Assuming that other companies come along with unique hardware and software innovations, Dreem’s market will be only a subset of the total size of the neurotechnology-enhanced sleep market.</p><h3>Conclusion</h3><p>In Volume III of our five-part series on Consumer Neurotechnology, we took a look at neurotech that leverages neurofeedback to drive value to users. In Volume IV, we’ll explore consumer technologies that directly modulate the human nervous system. Stay tuned!</p><p><em>Disclaimer: We actively write about the themes in which we invest or may invest: virtual reality, augmented reality, artificial intelligence, and robotics. From time to time, we may write about companies that are in our portfolio. As managers of the portfolio, we may earn carried interest, management fees or other compensation from such portfolio. Content on this site including opinions on specific themes in technology, market estimates, and estimates and commentary regarding publicly traded or private companies is not intended for use in making any investment decisions and provided solely for informational purposes. We hold no obligation to update any of our projections and the content on this site should not be relied upon. We express no warranties about any estimates or opinions we make.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=21e669351f64" width="1" height="1" alt=""><hr><p><a href="https://medium.com/the-substrate/consumer-neurotech-volume-iii-21e669351f64">Consumer Neurotech: Volume III</a> was originally published in <a href="https://medium.com/the-substrate">The Substrate</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Professor Sapolsky,]]></title>
            <link>https://medium.com/the-substrate/professor-sapolsky-ab239ec82191?source=rss----10fba766a0f5---4</link>
            <guid isPermaLink="false">https://medium.com/p/ab239ec82191</guid>
            <dc:creator><![CDATA[Avery Bedows]]></dc:creator>
            <pubDate>Sun, 14 Oct 2018 16:54:42 GMT</pubDate>
            <atom:updated>2018-10-14T16:54:41.548Z</atom:updated>
            <content:encoded><![CDATA[<p>Professor Sapolsky,</p><p>Thanks for the great piece. I’m always a fan of your writing.</p><p>How would you respond to the following spin: with modern technology, we’re developing a *different* sort of recognition and familiarity, rather than the *wrong* sort. Certainly, I don’t think it’s feasible to replace seeing someone’s face as the primary means of identifying them, nor spending quality in-person time as the primary means of establishing the warm familiarity that most of us crave deeply. But, I can speak from experience that it’s possible to recognize someone cognitively, and to develop an emotional familiarity with them, simply by reading their texts. For example, with one of my past significant others, I could intuit within one or two text messages what kind of mood she was in — evidence to the fact that, insofar as social interactions are concerned, digital/semantic signals can be substituted for natural biological ones. I’ve also built close friendships primarily through a digital medium, and those friendships enjoy the same in-person comfort as friendships developed face-to-face — direct evidence that digital signals can be substituted for natural ones when forming recognition &amp; familiarity.</p><p>It’s important to ask 1) how does this neo-recognition and neo-familiarity stack up to the natural version (e.g. in terms of recognition accuracy or the mutuality of subjective feelings of familiarity), and 2) what are the personal and social consequences of substituting digital signals for biological ones?</p><p>Ultimately, I think this comes down to a matter of morals: should we view natural signals as “better”, or should we take the view that developing new signals is good and in the spirit of human advancement (which spirit is, itself, a moral). I don’t know where I fall on this.</p><p>Avery</p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=ab239ec82191" width="1" height="1" alt=""><hr><p><a href="https://medium.com/the-substrate/professor-sapolsky-ab239ec82191">Professor Sapolsky,</a> was originally published in <a href="https://medium.com/the-substrate">The Substrate</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
        <item>
            <title><![CDATA[Consumer Neurotech: Volume II]]></title>
            <link>https://medium.com/the-substrate/consumer-neurotech-volume-ii-adac8e560151?source=rss----10fba766a0f5---4</link>
            <guid isPermaLink="false">https://medium.com/p/adac8e560151</guid>
            <category><![CDATA[brain-computer-interface]]></category>
            <category><![CDATA[virtual-reality]]></category>
            <category><![CDATA[neuroscience]]></category>
            <category><![CDATA[venture-capital]]></category>
            <category><![CDATA[neurotechnology]]></category>
            <dc:creator><![CDATA[Loup Ventures]]></dc:creator>
            <pubDate>Thu, 04 Oct 2018 16:26:01 GMT</pubDate>
            <atom:updated>2018-12-05T17:35:58.815Z</atom:updated>
            <content:encoded><![CDATA[<figure><img alt="" src="https://cdn-images-1.medium.com/max/820/0*giSD39Ad1wciUzdk.png" /></figure><p><em>By Avery Bedows</em></p><h4>Neurotech for Control</h4><p>In the <a href="https://loupventures.com/consumer-neurotechnology-volume-i/">first part</a> of this series on Consumer Neurotech, we established a framework for thinking about Consumer Neurotechnologies and their value propositions. With the initial groundwork laid, we can finally dig into the technologies themselves. Here, in Volume II, we’ll discuss neurotechnologies that have the value proposition of <strong>Control</strong> (or, using our taxonomy from Volume I, neurotechnologies whose “Effect-type” is Control). Within the scope of this piece, this encompasses two categories: <strong>Commoditized EEG </strong>and <strong>EMG Human-Computer Interfaces</strong>.</p><p>It’s always useful to convey <em>intuition</em> about technology before discussing it. In this spirit, we present short fictional scenarios that describe a hypothetical neurotechnology <em>in situ</em>. The narratives intentionally take literary and science fictional license, while staying well within the realm of what we believe will come to market within the next 5–10 years.</p><h3><strong>Commoditized EEG</strong></h3><h4>Fiction</h4><p><em>Your phone buzzes; frenzied, you grab your notebook off the table and throw it into your shoulder bag. You have one foot out the door when you realize why your bag feels so light: it’s missing the laptop which, in the aftermath of last night’s Wikipedia binge, must be wedged between couch seats. Glancing at your phone, you see that the autonomous Uber fleet car will wait another 76 seconds, and this number stays seared into your mind’s eye as you start racing back into the living room and make a dive into the couch cushions, expertly fishing out your computer. You’ve got it in your bag; 25 seconds left, now. You run outside, open the backseat door of the fleet car, and as you exhale and situate yourself into the seat with two whole seconds to spare, you catch the sun outside glancing off a shiny, small piece of metal in the headrest. You close your eyes, buckle, and rest your head back. The electrode embedded into the headrest establishes effortless contact with your scalp, and starts recording neural signals. Behind the scenes, an algorithm notices you have patterns of brain activity that correlate with stress, and responds by darkening the windows. Your last thought before slipping into a rejuvenating nap is that yes, good job, car, the Chopin piece is a great choice, how did you know?…</em></p><p>• • •</p><p><em>As a general policy, you don’t check personal notifications at work. Back when you were working your first job out of school, there was a particularly grueling performance review, the theme of which was, “Your productivity needs to improve, or we’re going to let you go.” You left that meeting and knew exactly where all the wasted time went: you’d been handing your time out left and right to the rectangular notifications floating into and out of the top-right corner of your MacBook Pro. Plans for the weekend; coordinating the date for tomorrow night; cat GIF. You were terrified, and subsequently enacted your policy. It suffices to say your productivity improved dramatically thereafter. Today, nearer to a decade later than you’d like to acknowledge, you find yourself still on the defensive with notifications, even with these new-fangled augmented reality glasses. Last year, your employer introduced the glasses for you to use as you sit at your desk and tinker with the electronics for the toy company’s smart stuffed animal. They’re pretty useful, actually: reference documents float wherever you want, making the building process much smoother. There are also, of course, the notifications, which you religiously ignore. It’s 2pm, and a little “ding” flows through the headset and into your ears. It’s a notification. You’re well-trained to ignore these by now, but you catch the name, and see that it’s your son. Strange, you think. Doesn’t he have class right now? With one hand holding a solder and another holding the breadboard in place, you focus on the notification for the quick fraction of the second the system needs, and now the box slides over. Floating above the middle of your desk, it unfolds into an open message: “Hi, I have a fever, can you come pick me up from school please?”</em></p><p>• • •</p><p><a href="https://en.wikipedia.org/wiki/Electroencephalography">Electroencephalography (EEG)</a> has been used widely since it was first described by Hans Berger in 1929. EEG plays many roles in medical diagnostics, but increasingly it’s used as a control component in brain-computer interfaces. The promise of using brain signals to control technology is the stuff of science fiction, and many are slowly working to make this a reality. Within a decade or so, we think EEG technology will become fairly commoditized; in this section, we discuss products working towards that goal.</p><p>In the first fiction scenario above, an EEG sensor is built into an autonomous car so the vehicle can adjust parameters like lighting and music to support the emotional state of the rider. In the second scenario, an array of EEG sensors built into a mixed reality headset enables the user to interact with the interface hands-free. What we described above are manifestations of commoditized EEG that we think will be on the market in a 5–10 year timeframe. A number of consumer-focused standalone EEG hardware devices are already available today (there are also many research- and clinical-grade headsets, but these are woefully unergonomic and cost tens of thousands of dollars).</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/261/0*Dn0VcsXUwbqkEjAN.jpg" /><figcaption>Image: <a href="http://neurable.com/news/game-you-can-control-your-mind">http://neurable.com/news/game-you-can-control-your-mind</a></figcaption></figure><h4>Neurable BCI Development Kit</h4><ul><li>Medium of innovation: <strong>Integrated</strong></li><li>Device-type: <strong>Record</strong></li><li>Form-factor: <strong>Glasses</strong></li><li>Effect-type: <strong>Control, Developer</strong></li></ul><p>The <a href="http://neurable.com/developers">Neurable BCI Development Kit</a> includes both hardware and software: the hardware is a headband with dry EEG electrodes that can attach to the HTC Vive VR headset, and the software is an SDK that enables developers to leverage the EEG signals to control virtual reality applications. One of their demonstration videos features a simple <a href="https://www.youtube.com/watch?v=QXN6_GEmY_A)">virtual reality game</a> in which users can pick up objects just by thinking, then throw those objects at targets, again just with thought. Neurable has a <a href="https://unity3d.com/">Unity</a> plugin that enables developers to quickly start using brain control for their virtual reality applications.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/0*UN-VwFT2f6CdwSPL.png" /><figcaption>Image: <a href="https://www.emotiv.com/epoc/">https://www.emotiv.com/epoc/</a></figcaption></figure><h4><strong>Emotiv Epoc+</strong></h4><ul><li>Medium of innovation:<strong> Integrated</strong></li><li>Device-type: <strong>Record</strong></li><li>Form-factor: <strong>General Headset</strong></li><li>Effect-type: <strong>Control, Feedback, User Characterization, Developer</strong></li></ul><p>The <a href="https://www.emotiv.com/epoc/">Emotiv Epoc+</a> is an $800, 14-channel EEG device. It can be used with an ~$80 / mo. / seat subscription to the <a href="https://www.emotiv.com/emotivpro/">EmotivPRO</a> software suite, which gives researchers and developers convenient ways to observe and use the neural signals recorded by the headset. Users can pay an additional $80 (one-time fee) for the <a href="https://www.emotiv.com/emotiv-brainviz/">BrainViz</a> software to visualize EEG data. Emotiv also offers a software package called <a href="https://www.emotiv.com/emotiv-bci/">EmotivBCI</a>, a free tool that lets developers use the EEG headset for brain-computer interfaces. Developers can train the headset to distinguish up to four commands, such as the hypothetical command to “open this notification.” EmotivBCI also features signals about the user’s mental state, which could in principle be used in a manner similar to the autonomous car scenario above.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/800/0*r-trtTbPkDZMEC-8.png" /><figcaption>Image: <a href="https://openbci.com/">https://openbci.com</a></figcaption></figure><h4><strong>OpenBCI</strong></h4><ul><li>Medium of innovation:<strong> Integrated</strong></li><li>Device-type: <strong>Record</strong></li><li>Form-factor: <strong>General Headset</strong></li><li>Effect-type: <strong>Control, Feedback, User Characterization, Developer</strong></li></ul><p>OpenBCI creates a series of partially DIY, open-source EEG electronics boards. They offer a <a href="https://shop.openbci.com/collections/frontpage/products/pre-order-ganglion-board?variant=13461804483">4-channel version</a> for $200, an <a href="https://shop.openbci.com/collections/frontpage/products/cyton-biosensing-board-8-channel?variant=38958638542">8-channel version</a> for $500, and a <a href="https://shop.openbci.com/collections/frontpage/products/cyton-daisy-biosensing-boards-16-channel?variant=38959256526">16-channel version</a> for $950. Customers can also purchase a 3D printed <a href="https://shop.openbci.com/collections/frontpage/products/ultracortex-mark-iv">headset</a> with electrodes, and this can cost between $350 and $800 depending on the degree of assembly the customer wants to undertake on their own.</p><h4><strong>Takeaways</strong></h4><ol><li>For standalone EEGs, there’s an opportunity around software for processing signals and making them accessible to developers. Emotiv’s EmotivPRO software capitalizes on this opportunity, and we’re aware of others pursuing this path as well. The goal is to create a simple interface for non-neuroscientist/neuroengineer developers that lets them incorporate aspects of Control and User Characterization into their products. EEG analysis is complicated, and many patents exist around just the manipulation of EEG data; abstracting the signal processing makes it possible for non-specialists to leverage the benefits of EEG.</li><li>It’s reasonable to consider what an EEG marketplace might look like. Hypothetically, we can see two types of products: those that leverage the EEG for Control or User Characterization in <strong>another user-facing application</strong>, and those that <strong>give developers new ways to use the EEG signal</strong> itself. The latter seems like it could lend itself to a marketplace akin to the <a href="https://marketplace.atlassian.com/">Atlassian Marketplace</a>, e.g. where Neurable developers could purchase additional apps that perform complex and specialized processing on the raw EEG data; Neurable would take a cut of these purchases/subscriptions. In other words, we think there’s an opportunity for a “data processing marketplace” that offers different ways to act on the data created by the hardware. In contrast, it doesn’t make sense to build a standalone marketplace for user-facing EEG-enabled software, since the software will run on another device or operating system (e.g., iOS), and be distributed through that platform natively (e.g., iOS App Store).</li><li>We don’t see standalone scalp EEGs ever succeeding at scale like smartphones, because EEG listens to users but can’t tell them anything. In contrast, smartphones “listen” through their touch screen and “speak” by displaying visuals on the screen; augmented reality glasses “listen” to gesture, voice, and gaze input, and “speak” through their displays as well. Imagine asking a user to constantly wear an augmented reality headset with its lenses removed: it would never happen because that device couldn’t “speak” to the user.</li><li>We <em>do</em>, however, think that most future XR headsets will incorporate EEG, since those devices already need to be worn on the head. Neurable is focusing their efforts here: as Neurable’s CEO <a href="https://www.youtube.com/watch?v=QXN6_GEmY_A">remarks</a>, virtual/augmented/mixed reality will never find it’s “killer application” until it finds its “killer interface.” EEG in the form-factor of XR headsets is likely a large part of this killer interface — anyone who has spent time using or developing XR applications understands the difficult constraints around using hands or voice for input. Thus, in evaluating the size of the Commoditized EEG market, it makes sense to consider that consumer EEG will likely proliferate as a function of XR, and not of its own accord.</li></ol><h4><strong>Use-Cases and Market Size</strong></h4><p>To estimate the size of the Commoditized EEG market, we estimate the markets that will <em>involve</em> Commoditized EEG as part of the respective technologies. By our logic above, we believe this to be mostly limited to the XR market, where XR headsets will incorporate EEG sensors, just like smartphones incorporate microphones. This market has two components: the revenues generated by the introduction of EEG into hardware, and the revenues generated by EEG-enabled software.</p><p>To model the hardware component, we use our internal AR/VR models and estimate the percentage of EEG-enabled devices for a given year.</p><p>To model the software component, we assume that every application available on an XR headset which leverages EEG will either use OS-level integrations (such as an EEG version of a mouse click) or will require custom analysis of EEG signals (a custom analysis seems reasonable because there are many patents pertaining to EEG signal processing methodologies for very specific applications). Therefore, if we take the number of XR devices with EEG capabilities, multiplied by the number of applications the average user uses per year, multiplied by the fraction of applications requiring unique signal processing (a guesstimate), then we get a gross estimate of how many different seats will be required for signal processing packages. If we assign an average price per seat to this, then we can roughly estimate a market size.</p><p>We must therefore estimate three additional values: the number of apps a user uses per year, on average; the number of apps requiring custom EEG signal processing; and, the average spend per user per year per application (glossing over the distinction between consumer and enterprise). We estimate the number of applications based on a report from <a href="https://www.appannie.com/en/insights/market-data/global-consumer-app-usage-data/">App Annie</a>; we guess the percentage of custom-EEG applications will be 20% of all applications and choose this small value because EEG signal processing is a specialized skill, subjecting EEG-enabled applications to talent constraints; finally, we ad-hoc guess the annual spend per user per application to be $5 (this represents an average of the few expensive applications and the many cheap/free applications). We’re more bullish on AR/MR than on VR; our estimate is that <strong>in 2025, EEG-enabled Augmented/Mixed Reality technology will be involved in $71.6B in revenue.</strong></p><h3><strong>EMG Human-Computer Interfaces</strong></h3><h4>Fiction</h4><p><em>Step, step, step — you pace forward, across the room from the projector. About-face, then begin the journey back. Four years ago when you founded your company, you dreamed of many things. You dreamed of the day you’d open a door to your own office in your own building. You dreamed of the day you’d have enough of a staff tackling bugs and engineering challenges that your primary cognitive output would be strategic planning, the domain where you feel your talent can shine unencumbered. You turn again and keep pacing. Omnigraffle is open on the screen, and between steps you pause for a moment to fidget with the armband on your dominant forearm. You reflect on the fact that, somehow, these dreams are now the reality — here you are, deep inside a flow-chart, planning out the next five years of your business, walking around your own office in your own building. Beginning a gratitude journal was one of the best decisions you ever made; with a twitch of your hand, you pull open your journal on the screen, planting hands on hips and feet towards the screen. With no keyboard necessary, you subtly twitch your forearm again, and your words of appreciation show up on the screen. After thirty rapid seconds of typing, you’re glowing inside, and — oh! — another potential R&amp;D direction comes to mind. You’re walking again, now with arms crossed, and with just a pinky, you bring Omnigraffle back to the front, ready to add to the chart. Twitch, and there the idea is, up on the screen.</em></p><p>• • •</p><p>Electromyogram armbands can translate muscle movements and/or twitches into digital signals which can then be interpreted to understand the motion (or intended motion) of a user’s hand or arm. Described above is the effortless use of a near-futuristic EMG armband for Control. If a consumer product can record high-fidelity EMG signals, these signals can be used to reconstruct arm and/or hand movement. This opens up new ways to control technology and can be achieved in an ergonomic form-factor since EMG hardware can be miniaturized into an armband.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/172/0*sZ3eg22z07J93QLQ.png" /></figure><p>Image: <a href="https://www.myo.com/">https://www.myo.com</a></p><h4><strong>Myo Gesture Control Armband</strong></h4><ul><li>Medium of innovation: <strong>Integrated</strong></li><li>Device-type: <strong>Record</strong></li><li>Form-factor: <strong>Armband</strong></li><li>Effect-type: <strong>Control, Developer</strong></li></ul><p>To motivate the EMG opportunity, we’ll discuss two well-funded startups that are developing EMG Control interfaces. The first is <a href="https://www.thalmic.com/">Thalmic Labs</a>, which builds the <a href="https://www.myo.com/">Myo Gesture Control Armband</a>. These armbands are already shipping and cost $130 per unit. Thalmic Labs also has a marketplace for Myo-compatible applications; some apps are developed in-house and some by third-party companies, such as a $50 music software product called <a href="http://precisionmusic.technology/shop/leviathan">Leviathan</a>. Leviathan lets users interface with other music software using the Myo armband. By way of example, Leviathan lets DJs adjust the volume of a track by moving their arms. Thalmic Labs has raised <a href="https://www.crunchbase.com/organization/get-myo%23section-overview">$135.6M</a>.</p><figure><img alt="" src="https://cdn-images-1.medium.com/max/300/0*l8uM5FuyMALjJdyC.png" /></figure><p>Image: <a href="https://www.geekwire.com/2018/vulcan-capital-among-investors-ctrl-labs-startup-unique-brain-computer-interface/">https://www.geekwire.com/2018/vulcan-capital-among-investors-ctrl-labs-startup-unique-brain-computer-interface/</a></p><h4><strong>CTRL-kit</strong></h4><ul><li>Medium of innovation: <strong>Integrated</strong></li><li>Device-type: <strong>Record</strong></li><li>Form-factor: <strong>Armband</strong></li><li>Effect-type: <strong>Control, Developer</strong></li></ul><p>The second company pursuing an EMG armband is<a href="https://www.ctrl-labs.com/"> CTRL-labs</a>. CTRL has <a href="https://www.crunchbase.com/organization/ctrl-labs%23section-locked-charts">raised $39M</a> to develop an armband that can record muscle twitches. The ultimate goal is to replace typing and other input interfaces by capturing a user’s “intention” to move without the user having to actually contract the muscle and execute the movement. CTRL has a <a href="https://www.ctrl-labs.com/2018-oreilly-ai-keynote-video-by-ceo-thomas-reardon/">demonstration</a> where they show their ability to reconstruct a user’s intended hand movement when another person is holding the user’s fist closed. CTRL is still in the development phase: the development kit is supposed to ship in late 2018. There’s no indication about the price of the developer kit, but given that the highest-end keyboards and mice cost a few hundred dollars, it’s hard to imagine spending more than that on a device to control a computer.</p><h4><strong>Takeaways and Use-Cases</strong></h4><ol><li>To list some use-cases that are already possible, or may soon be: a surgeon can <a href="https://developerblog.myo.com/using-myo-during-surgery-with-adora-assistant/">control equipment</a> in an operating room without becoming distracted from the current task. A DJ can adjust music volume and lighting <a href="https://youtu.be/xWLBAv56xDw">using her arms</a>. Complex machinery like factor components and automobiles can be intuitively controlled. When paired with AR/VR, this technology can enable engineers and artists to use their hands as their primary tool for creating and modifying 3D models. A programmer can hop between files by twitching her pinky. Anyone can type rapidly with their hands on hips or arms folded. Extrapolating, the big insight of EMG armband technology is that controlling computers and machines will require less physical activity, particularly for tasks that require extensive movement right now, such as typing and 3D model manipulation. EEG, in contrast, will probably be better for quick, frequent commands like navigating through menus and selecting items on-screen.</li><li>Sizing this market is unrealistic because it would involve exhaustively enumerating the use-cases — if this technology eventually works as advertised, the market will be gigantic: perhaps as large as the set of use-cases that currently require a mouse and/or keyboard. Mice and keyboards aren’t large market opportunities, but EMG is subject to different constraints because a) the components of an armband will probably be more expensive, and b) EMG armbands won’t be sold with every computer, so the business models will operate differently (imagine requiring a user to wear a different armband for every computer or tablet they use regularly).</li><li>EMG armbands have two characteristics that make us think they can succeed at scale: a) the armband form-factor is already socially acceptable (watches, smartwatches, fitness trackers, bracelets, etc.); and b) EMG armbands can provide many degrees of freedom to control a computer. The CTRL-kit, for example, seems to be able to capture as many degrees of freedom as the human hand has.</li><li>Thalmic Labs is making a platform play by attempting to own the marketplace for apps that use its interface. We aren’t sure whether this is sustainable. If we assume the ultimate goal is to make this control interface ubiquitous, then many apps in different ecosystems like iOS or a mixed reality platform will utilize the control input. Large-scale proliferation of EMG armbands seems incompatible with a marketplace business model, since so many of the integrating applications are distributed through incumbent marketplaces.</li></ol><h3>Conclusion</h3><p>In Volume II of our five-part series on Consumer Neurotechnology, we discussed neurotechnologies that enable their users to control computers in new ways. In Volume III, we’ll look at technologies that directly modulate the human nervous system. Stay tuned!</p><p><em>Disclaimer: We actively write about the themes in which we invest or may invest: virtual reality, augmented reality, artificial intelligence, and robotics. From time to time, we may write about companies that are in our portfolio. As managers of the portfolio, we may earn carried interest, management fees or other compensation from such portfolio. Content on this site including opinions on specific themes in technology, market estimates, and estimates and commentary regarding publicly traded or private companies is not intended for use in making any investment decisions and provided solely for informational purposes. We hold no obligation to update any of our projections and the content on this site should not be relied upon. We express no warranties about any estimates or opinions we make.</em></p><img src="https://medium.com/_/stat?event=post.clientViewed&referrerSource=full_rss&postId=adac8e560151" width="1" height="1" alt=""><hr><p><a href="https://medium.com/the-substrate/consumer-neurotech-volume-ii-adac8e560151">Consumer Neurotech: Volume II</a> was originally published in <a href="https://medium.com/the-substrate">The Substrate</a> on Medium, where people are continuing the conversation by highlighting and responding to this story.</p>]]></content:encoded>
        </item>
    </channel>
</rss>