Philosophy and Economic Theory, New School University, New York NY
Journal of Evolution and Technology – Vol. 26 Issue 2 – October 2016 – pgs 60-81
Brain-computer interfaces (BCIs) are any manner of technology that might link the human brain to communications networks such as the Internet. In more detail, a brain-computer interface, brain-machine interface (BMI), neural prosthesis, etc., is typically a computational system implanted in the brain that allows a person to control a computer or other electronic device using electrical signals from the brain (Peters 2014). Individuals use BCIs to generate alphanumerical characters on a computer screen in the following way. The BCI equipment registers the electrical output of the brain when the eyes are focused on a particular location or quadrant of a computer screen – on the “q” in a stretched-out string of letters, for example – and outputs the letter onto the computer monitor (Mayo Clinic 2009).
The primary aim of BCIs at present is repairing human cognitive and sensorimotor function. One of the most widely adopted uses is cochlear implants, where a small computer chip is substituted for damaged control organs in the inner ear. The chip transforms sound waves into electrical signals that are interpretable by the brain (Friehs 2004). Vision restoration is another application: here, implantable systems transmit visual information to the brain. Two-way BCIs are another form of the technology under development, using both output and input channels for communication between the brain and the external world. There would be the usual BCI communication output from the brain in the form of translating neuronal activity into electronic commands to move robot arms, wheelchairs, and computer screen cursors. In addition, feedback from this activity could be input back into the two-way system via electrical brain stimulation that delivers signals into the brain (Bo 2015).
BCIs comprise an active area of research and could start to integrate advances from adjacent fields such as neuroscience, nanomaterials, electronics miniaturization, and machine learning. For example, one neuro-imaging research project is starting to make guesses as to what participants see during brain scans, purporting to be able to distinguish between a cat and a person (Smith 2013). Merging this kind of functionality with BCIs might produce new applications. Other experimental BCI projects have been proposed. One is Neocortical Brain-Cloud Interfaces: autonomous nanorobots that could connect to axons and neuronal synaptic clefts, or embed themselves into the peripheral calvaria and pericranium of the skull (Boehm 2016). Another project, Brainets, envisions linking multiple organic computing units (brains) to silicon computing networks (Pais-Vieira 2015). A third project is Neural Dust, in which thousands of 10-100 micron-sized free-floating sensor nodes would reside in the brain and provide a computing processing network (Seo 2013).
So far BCIs have been conceived primarily as a solution for medical pathologies. However, it is possible to see BCIs more expansively as a platform for cognitive enhancement and human-machine collaboration. The BCI functionality of typing on a keyboard with your mind suggests the possibility of having an always-on brain-Internet connection. Consider what the world might be like if each individual had a live 24/7 brain connection to the Internet. Just as cell phones connected individual people to communications networks, BCIs might similarly connect individual brains to communications networks. I propose a variety of BCI applications and concepts throughout the rest of this paper, all of which are speculative and not in development to my knowledge.
In one sense, ubiquitous BCIs are expected. It is contemplated that communications technology, already mobilized to the body via the cell phone, could be “brought on board” even more pervasively. BCIs are merely a next-generation improvement to the current situation of people constantly staring at their phones. In another sense, though, BCIs are not only a “better horse” technology: they are also a “car” in that it is impossible to foresee the full range of future applications that might be enabled from the present moment. BCIs pose a variety of practical, ethical, and philosophical issues. Life itself and the definition of what it is to be human could be quite different in a world where BCIs are widespread. Some of the immediate practical concerns of BCIs could include invasiveness, utility, reversibility, support, maintenance, upgradability (hardware and software), anti-hacking and anti-virus protection, cost, and accessibility. Beyond practical concerns, there are ethical issues regarding privacy and security. For example, neural data privacy rights are an area where standards need to be defined (Swan 2014a).
There could be at least three classes of BCI applications introduced in graduated phases of risk and complexity: biological cure and enhancement; information and entertainment; and self-actualization (realization of individual cognitive and artistic potential). Each of these merits separate discussion.
2.1 Health and enhancement BCI applications
One first class of BCI applications could relate to cure and enhancement. These applications can be framed as providing an “Apple HealthKit or Google Fit for the brain.” The idea is to employ BCIs as a constant health monitor, pathology resolver, and neural optimizer. One of the great promises of BCI technology is that applications such as daily health checks might run automatically in the background to improve our lives. Periodic health checks could be orchestrated seamlessly by ambient quantified-self smart infrastructure (essentially the next generation of unobtrusive sensors worn on the body such as smartwatches). Personal biometric data could be transmitted to longitudinal health profiles in electronic medical records. This could facilitate the development of advanced preventive medicine. Preventive medicine is maintaining a state of health by detecting and resolving potential conditions in the 80 per cent of their life cycle before they become clinically diagnosable (Swan 2012b). Neural data streamed from BCIs to secure health data banks could finally start to allow amassing of “big health data,” datasets large enough to study the longitudinal norms of brain patterns and cognitive well-being. A variety of health management and neural performance enhancement applications could ensue.
Personal biometric data collected by cellular telephone applications are an example of how personal data from BCIs might be treated. Norms for collecting and storing personal biometric data are starting to be codified. Ostensibly, neural data is just a special case of personal biometric data, with additional sensitivities. Apple HealthKit, for example, automatically captures 200 health metrics per day via the iPhone and seamlessly uploads them to the Internet cloud for subsequent on-demand analysis (Swan 2015d). Google Fit on the Android platform performs a similar function (Welch 2014). However, despite the potential benefits of automated health data collection, appropriate social and legal contracts with technology providers are not yet completely in place. Individuals may not fully grasp how their personal data is being collected, stored, and used (including being sold to third parties). This is important since personal medical information is a valuable asset. Health data may be worth ten times more to hackers than financial data such as credit card numbers and personal identity (Humer 2014). Even though cell phone users “must explicitly grant each application the permission to read and write data to the HealthKit store” (iOS Developer Library 2015), health-tracking data may be collected without full user awareness (other than by having agreed to the initial phone activation agreement). When installing applications, it can be easy for users to accede quickly to requested permissions without fully understanding what they entail in terms of granting access to personal biometric data.
2.2 Information and entertainment BCI applications
A second class of BCI applications is related to information and entertainment. One application could be brain-based information requesting. Information query could be both pushed and pulled: automatically pushed to users per pre-specified settings, or pulled (requested) by users on demand. Data notifications could be presented in the mind’s visual space. This would be the analog to information cards or short data messages being posted to a phone or smartwatch. Here, the information could be presented in the brain, for example as an unobtrusive notification in the lower visual field. BCIs could have Google Now type functionality, making contextual guesses about information that might be relevant in the moment (such as transportation delay information).
BCIs could be the interface for immersive experience, conceptually similar to internalizing virtual reality headsets inside the body. The idea would be to have an onboard Oculus Rift, Meta 2 (Jabczenski 2016), or MindMaze (Lunden 2016). This could allow “HUD-sharing (heads-up display),” as is possible in video games now, and beyond: deeper levels of experience sharing and the “transparent shadowing” (Boehm 2016) of others for purposes ranging from learning to entertainment. A variety of contexts for experience sharing have been suggested, for example apprenticeship, scientific discovery, sports matches, music concerts, political rallies, and sex (Kurzweil 2006). In one example, Greg Bear’s science fiction novel Slant (1998) explores an updated version of Brave New World feelies (movies with sense and touch, not just audio and video). Individual experience feeds could become marketable not just for entertainment purposes, but also for personal and societal record keeping.
Consider that in the future you might grant different levels of access to your personal experience feed. This could include selecting the payment models based on the situation, for example, fee-based or open-source contributions. Live events could provide interesting situations of sharing personal experience into the group memory feed. Computing algorithms could aggregate arbitrarily many contributor threads into a single summary. The crowdfile (e.g., a group experience file) for an event could be a new means of recording human history. After the fact, the event crowdfile could be accessed just as Wikipedia, Twitter, and YouTube are now. During the event, remote participants could join the live crowdfile, similar to live-streams now. To the extent that individual experience files or their “diffs” (salient differences from the aggregate file) could be stored expediently, various accounts of history could be kept simultaneously. Finally, multiple accounts of events could be available from different standpoints (similar to instant replay from different cameras). Any assessment of public opinion such as political polling could undergo a substantial shift as many more individual and collective reactions might be known in detail, and also in real-time. With BCIs, to the extent shared by the owner, experience files could become like any other digital content: a creative object for others to take up, reformulate, repurpose, reinterpret, “mash up,” and share back out to the public venue. Just as the Selfie (a self-taken photograph) was the killer app of photos, and moment-showing was the killer app of video blogs (Dedman 2007), some form of “Brain Selfies,” “Brainies,” “Experiencies,” or “myWorld-ies” might be the killer app of BCIs.
2.3 Self-actualization BCI applications
A third class of BCI applications is related to self-actualization. This refers to a full realization of one’s potential for self-development. Per Abraham Maslow’s theory (1943), self-actualization represents the growth of an individual toward the fulfillment of the highest level of needs, those related to meaning. Carl Rogers (1961) further posited a human drive or tendency for self-actualization. Here, this is understood as becoming one’s potentialities, expressing and activating all of the capacities of the human organism. This could include the expression of one’s creativity, a quest for spiritual enlightenment, the pursuit of knowledge, and the desire to give to society: anything an individual self-determines as meaningful. Actualization is not merely experiential but generative; it is developing oneself actively and bringing this into concrete expression in reality. There are fascinating possibilities for how BCIs might help with intellectual, creative, and artistic self-actualization. Beyond health tracking and entertainment, one of the strongest aspects of what might be at stake with BCI technologies is the possibility of realizing more of our human potential, and this could be a strong motivation for adoption (Swan 2016a).
The potential future applications of BCIs discussed so far relate primarily to individuals; however, BCI technologies might be mobilized similarly into other classes of applications to support group activities. We are inherently social creatures and lead interactive lives with others in the context of a social fabric, and new technologies could continue to facilitate these interactions. One of the most potent applications for BCI group applications could be the speculative notion of the cloudmind or crowdmind. Most broadly, a cloudmind would be, as the term suggests, a cloud-based mind, a mind in the Internet cloud. This would be some sort of processing or thinking capability (hence “a mind”) that is virtual, located in Internet databanks without having a specific body or other physical corporeality. A crowdmind might comprise large numbers of minds operating together.
There could be different kinds of cloudminds. One might be a basic machine mind: algorithms quietly crunching in the background, maybe as the result of the next generations of big data analysis programs. Other types of cloudminds might be different forms of human-machine minds (e.g., a person plus a cloud-based thinking assistant or companion such as Siri or Her (Jonze 2013)). There could be different forms of multiple minds pooled together (mindpools), combinations of human minds, human-machine minds, or machine minds. The use of the word “mind” in the expression cloudmind could be misleading since the familiar example of mind is the human mind, and machine intelligence is not in possession of the full range of capacities of the human mind such as general purpose problem solving, volitionary action (free will), and consciousness. However, “mind” is meant generally here to denote an entity that has some sort of capacity for processing and “thinking,” perhaps initially in the narrow sense of finding solutions to specified problems, but possibly expanding as processing tasks become more broadly “thinking” oriented. The general definition of a cloudmind is a cloud-based thinker with some sort of analytic processing power.
3.1 Prototypical cloudminds
The notion of a cloudmind is perhaps not so much a new idea as a new label that connotes a greater range of functioning. Prototypical cloudminds already exist in the sense of automated cloud-based systems that coordinate the processing activity of multiple agents. One such prototype is Mechanical Turk, an algorithmic system for organizing individuals to perform online tasks that require human intelligence. In this category of crowdsourced labor marketplaces, there are many other examples such as Topcoder, Elance, and Upwork (formerly Odesk). A second cloudmind prototype is the notion of humans as a community computing network. The idea is that humans, in their everyday use of data, perform a curation, creation, and transfer function with the data. Humans actively transform, mold, steward, and produce data in new forms by interacting with it. Data is active and living, dynamically engaged by humans as a community computer, each person a node operating on data and re-contributing the results back into the network (Swan 2012a). A third kind of cloudmind prototype is “big data,” the extremely large data sets that are analyzed computationally to reveal patterns (such as Amazon and Netflix recommendation engines). This “algorithmic reality” is an increasingly predominant feature of the modern world (Swan 2016b). Big data takes on entity-level status in the notion of the cloudmind, where big data is envisioned as a whole, quietly crunching in the background. The dual nature of technology (having both “good” and “evil” uses) can be seen in big data. On one hand, big data might be seen as contributing to our lives in helpful ways including by reducing the cognitive load required to deal with administrivia. On the other hand, a worry is that big data may not be just guessing our preferences but starting to manufacture them for us (Lanier 2014).
3.2 Cloudmind starter application: Sell unused brain processing cycles to the cloud
In the future, cloudminds involving human brain power might be facilitated by BCIs or other ways of linking human cognitive processing to the Internet. The key feature is the live 24/7 connection, not just generally to the Internet, but specifically to other brains and machine thinkers. One way that individuals might start to explore and adopt BCI cloudmind applications is in a “starter application” idea of selling permissioned braincycles to the cloud. This is a parallel concept to selling self-generated electricity from solar panels back into the power grid. This initial and basic cloudmind application might involve the sharing of unused brain processing cycles. The structure could be timesharing cognitive processing power during sleep cycles or other down time, conceptually similar to participating in community computing projects such as SETI@home or protein Folding@home. The idea would be to securely and unobtrusively share one’s own unused resources, downtime braincycles. There could be diverse compensation models for this, including remuneration and donation.
3.3 Cloudmind health app: Virtual patient modeling
More advanced cloudmind applications could correspond to the three classes of individual BCI applications discussed above: health tracking, information and entertainment, and actualization. With pathology resolution and enhancement applications, the daily health check could include longitudinal neural data-logging to Electronic Medical Records (EMRs), which could be integrated into virtual patient modeling systems. The personal health simulation could include different possible scenarios of how patient wellness could evolve from the simulated impact of various drugs or lifestyle choices. The system might model any variety of responses to personal health questions, such as recommending a nootropics stack to maximize cognitive enhancement given a particular individual’s genomic profile. Virtual patient simulations could be part of any future EMR (Bangs 2005; Uehling 2004), and instantiated as a cloudmind application with a cloudmind’s full range of intelligent processing capabilities. Virtual patient EMR files could be shared more widely (by permission) with family groups and health databank repositories for remunerated research studies and clinical trials. There could be a new concept of “virtual clinical trials” to accompany any physical-world clinical trial. Simulated patient responses could be a supplemental mode of information, particularly to model safety and efficacy. Collecting data initially for medical purposes is already practical and cost effective enough to justify the effort. There is the additional benefit of creating valuable digital health assets that might be mobilized later for many other purposes, for example, to invite participation in user-permissioned cloudmind projects.
3.4 Cloudmind information app: Crowdminding an IoT collaboration archipelago
One of the most obvious information-related BCI cloudmind applications could be thought-commanded Google searches and information look-up. Consider how many steps can be required now to obtain simple data elements such as a weather forecast or a movie time. This can involve having to turn on a phone and go through a series of screens, with variable response times as the phone negotiates network connectivity.
Beyond information query, another cloudmind application could be commanding Internet-of-Things (IoT) connected objects in the environment. There are numerous examples of individuals feeling as if they are one with objects and equipment, for example a submarine commander or airline pilot experiencing the ship or aircraft as an extension of their own body (Takayama 2015). A modern example is remote workers piloting telepresence robots in the main office, where again the robot feels like an extension of the individual’s physical body (Dreyfus 2015). IoT cloudminds could provide an expanded version of this: using one’s mind to control physical objects in a local or remote environment via BCI. A security guard could command a whole building, for example. An IoT home security system could be operated remotely via BCI cloudmind. The science fictional idea of linking humans together as one, in one recent example using Naam’s drug Nexus (Naam 2012), could be extended to include linking humans and objects. There could be a joint agent that is a human plus IoT objects, functioning together as one cloudmind entity. On one hand, my being a cloudmind with my IoT objects is merely an extension and formalization of the human-machine fusing phenomenon that already occurs in intensive machine operation (“better horse” technology). On the other hand, the functionality of cloudminding myself into a collaboration archipelago of intelligent action-taking capability with IoT-enabled smart objects is a revolutionary new kind of concept (“car” technology).
3.5 Cloudmind actualization app: Digital self
One implication of a simulated digital patient self as a standard part of health records is the possibility of having a digital self more generally. There could be a more fully embodied digital self, a version (or versions) of me that exists electronically. Already there are many versions of digital me’s as digital selves existing online for many purposes. There are digital profiles for different websites: avatars, digital personae, and “fake me” accounts. Any form of my digital profile could be said to reasonably comprise some version of me, including those that explore dimensions of me otherwise not manifested in the physical world. There are digital self projects, such as CyBeRev and Lifenaut, which explicitly aid in the creation of digital selves (Zolfagharifard 2015). Even now, it is possible that algorithms could assemble digital selves of people from existing online footprints such as photos, social media, academic and blog writing, email communication, file storage, and other aspects of digital presence.
Digital selves might be mobilized for many online operations, including eventual participation in cloudminds. The lowest-risk starter applications for digital selves could be related to backup, archival, and storage – a digital self as a biographical record and memory-logging tool. My digital self could become more active as a digital assistant self, a virtual agent version of me deputized to conduct a certain specified slate of online activities. These activities could include purchase transactions, information search and assembly (for example an automated literature search), and more complicated automatable operations such as drafting email, blog entries, and forum posts based on previous content. Digital selves could be an interesting way to extend and monetize one’s own self as a computing resource, and provide a possible solution for the transition to the automation economy (one’s digital self engages in remunerable online work).
In the scope of their activity, digital selves could participate in computing projects that are increasingly complicated and remunerative, and might eventually lead to cloudminds. Joining a cloudmind project through a limited digital self could be a comfortable and gradual adoption path to cloudminds that builds trust and familiarity. Participating in a cloudmind with a digital copy, including one with an expiration date, could be less risky than participating with one’s “real” physical self. Over time, the digital self could incorporate more richness and fidelity from the underlying person, in order to be more active as an agent with volition and decision-making, not just passive storage. Eventually, BCI neural-tracking data could be integrated to produce an even more fidelitous digital self that includes the neural patterns of how an individual actually experiences and reacts to the world. The longer-term conceptualization of the digital self could be an entity that records, stores, simulates, and runs a full “me” node: a digital agent, and eventually a clan of digital agents, operating just like me.
3.6 Cloudmind actualization app: Subjectivation
Regarding personal development and actualization applications, this could be a central motivation for joining a cloudmind, participating with either a traditional “meatspace-mind” or (with less risk) a digital self mind. Of the extensive slate of BCI cloudmind applications, including health tracking, life logging, archiving, sharing, information requesting, fun and entertainment, and IoT archipelago control, one of the real killer apps might be the personal actualization potential that BCI cloudminds could deliver. Cloudminds could be employed in different levels of elective engagement that is productive, generative, and creative: pooled productive activity toward a goal. There could be a wide range of reasons for joining a cloudmind including compensation, fun, new experience, productive use of one’s mind, contributing to problem-solving efforts, and self-actualization (growth and development opportunities). Cloudmind collaborations could consist of meaningful and remunerable work, and personal mental engagement and development that is fun, creative, and collaborative. One of the deepest incentives for exploring improved connection and cognition through BCI cloudmind collaborations could be the possibility that they facilitate our individual growth and development as humans, our subjectivation or self-forming (Robinson 2015). BCI cloudminds could allow us to actualize our potential to become “more” of who we are and might be, more quickly and effectively, thereby hastening and accelerating our capacities to be more intelligent, capable, and creative participants in life (Swan 2016a).
3.7 Crowdminds to remedy possibility space myopia
From a practical perspective, one hope or assumption might be that problem solving could be made more effective with technology tools such as BCI cloudminds. Problem solving is a central activity we engage in as humans, and any means of improving our capability to do this might be useful and valuable. Minds (irrespective of type) collaborating together might solve problems more expediently than the “classical” (e.g. current) human methods of individual breakthroughs, competition, and team striving. Cloudmind infrastructure might support improved human problem solving, and enable new kinds of human-human brainstorming, human-machine collaboration, and progress in the development of machine minds. It may be that we have tackled only a certain circumscribed class of problems so far, one limited to human understanding and articulation, whereas the universe of problems and problem-solving techniques could be much larger. We should understand the possibility space (the full universe of possibilities) for many phenomena to be much larger than the part that is human-viewable or human-conceivable. A simple example is the electromagnetic spectrum, where only a small portion is viewable, but where our tools have vastly expanded our reach. This “possibility space myopia” has been documented in many domains, for example in computing algorithms (Wolfram 2002), intelligence (Yudkowsky 2008), perambulatable (i.e. able to walk) body plans (Lee 2013; Marks 2011), mathematics and logic (Husserl 2001), and the size of the universe (Shiga 2008). Collaborative methods between humans and machines, such as BCI cloudminds, might extend our reach into a larger possibility space of problems and their resolution.