One can do worse in predicting a future technology than. following the creative leaps of prophetic science fiction

Transcription

1 1 P age Cognition, Computers and Cooperation Bruce J. West Army Research Office Research Triangle Park, NC Background One can do worse in predicting a future technology than following the creative leaps of prophetic science fiction writers, such as H.G. Wells description of Captain Nemo s nuclear powered submarine, the Nautilus. Here we start with another incomparable author, George Orwell, and his thought police. But before we begin let me issue a note of caution. Most attempts at predicting the future of science in general and technology in particular, along with their impact on the Army of the future, are an embarrassment to those that made them. This is due to the many ways such predictions can fail and do fail. Setting aside the fact that scientific discoveries are, by and large, completely unpredictable, for the purposes of this workshop let us focus on technology forecasting, while being fully conscious of the many pitfalls to which such predictions are prone. However, before donning the mantle of the Mad Scientist, let me briefly mention what I think are the main

2 2 P age ways in which forecasters fail in making these kinds of predictions and how I hope to avoid such failure. The first mode of futurology failure, in order of frequency of occurrence, is assigning too much significance to a technology, merely because it is attractive and captures the imagination. For example, backpack rockets for the individual soldier have been predicted to be a game changer since the turn of the last century, but have had little actual success in overcoming the problems of stability and affordability. The second most frequent mode by which forecasts fail is omission. In this mode one does not properly assess the importance of a nascent technology. For example, almost no one anticipated the impact of the personal computer on society, much less its impact on the Army. Consequently, when this disruptive technology burst onto the technology landscape it pushed everything else to one side. The Army is still working to integrate the networking effects of personal computers into its strategic and tactical thinking. The third and final mode of failure is guessing the development of the wrong branch of alternative technologies. One example of which is the marketplace choosing VHS over the beta

3 3 P age recorder. It certainly was not technical superiority that determined the adoption of the VHS, but some other aspect of the market that is not yet understood. But the point became moot when VHS, in its turn, was replaced by the DVD. In spite of this potential minefield, I am proposing to you today a technology that will revolutionize the Army over the next quarter century. My selection is at the nexus of the mindcomputer interface (MCI) and Network Science (NS). The understanding of MCI has been pursued by the Department of Defense for the past half century, whereas the research program of NS is only a decade old. In themselves these two areas do not constitute a technology, but they encompass an area of scientific investigation that entails a vast number of new technologies. The mental control of machines, using a computer interface, is even more attractive than the rocket backpack. However, the synthesis of NS and MCI can avoid the first mode of failure through the research successes that have materialized. The potential impacts of NS and MCI on the soldier and the battlefield have certainly not been overlooked, and with every new research success another potential application is added to an ever expanding list of applications. Finally the marriage of

4 4 P age NS with MCI is unique, without competitors, and in this sense its technological offspring will have singular effects on the Army of the future. How the Brain Works Scientists have been working on models of the workings of the human brain for at least a century. Recent research finds that the workings of the brain require coordination/cooperation among a scattered mosaic of functionally specified brain regions that underlie a warrior s ability to attend, perceive, think, decide, remember and act. This research shows that coordination/cooperation within the brain is a self-organized dynamical process that emerges spontaneously as a consequence of the nonlinear interaction among participating neuronal elements. Neuroergonomics is a new discipline that seeks to understand how the human brain functions in relation to sensory, cognitive and motor processes during its performance in real-world tasks in order to develop and apply fundamental neuroscience to human factors and network interfaces. The spontaneous activity of the brain is characterized by partially coordinated/cooperative states in which component

5 5 P age activities (individual neural regions) are neither completely independent of each other (segregated) nor linked in a fixed mutual relationship (integrated). Changes in the dynamical balance between the coupling among neural ensembles and the expression of each ensemble s intrinsic properties may be seen to create meaningful information. Thus, at least some cognitive processes appear to arise as the result of transitions to fully coordinated/cooperative states that take the form of synchrony in different frequency bands. A psychologist would say that an individual displays tendencies, whereas the mathematics would represent such tendencies by the coexistence of convergence to and divergence from non-stationary transient processes. This non-stationarity is a consequence of there being no stationary states for the brain, with the convergence and divergence corresponding to integrating and segregating of information. Such non-ergodic encryption of information will be at the heart of decoding EEG time series and its understanding will be the basis for the science supporting BCI technology. The mathematics of networks i, in describing the dynamics of the brain, will replace the notion of a stationary state by a

6 6 P age more functionally useful idea of a ghost or transient state. The self-organizing behavior of coordination/cooperation describes the tendency towards and away from these ghost states. This is truly a process to which the concept of equilibrium is alien, and although the brain may rest, it never stops. One candidate formalism for describing this behavior is criticality that has recently gained traction in neuroscience and over the next few decades will be fully developed in a neuromorphics context to provide an understanding of how the processing techniques of the brain are altered under stress. Thought-driven Systems The true integration of man and machine not only addresses the physical nature of the two, but also the psychological acceptance of that symbiosis. BCI requires sensors to detect the electrical activity of a sender s brain and these signals are encoded by a computer algorithm for subsequent transmission to another computer, where the message is decoded back into a form that can be interpreted by the brain of a receiver. The transmitted message might be an instruction to a robot to stop

7 7 P age or go or an abstract communication to a soldier such as danger on your right. Recently the movement of a monkey s arm was used to generate signals at Duke University, which were transmitted to a remote location. The remote site was an MIT lab, where these neurological signals were interpreted by a computer. The decoded message was used to move a robot arm in precisely the same way the monkey s arm had moved. This scientific understanding of neurological signals, manifest in the computer programs, is providing the new technology that will enable the future warrior to control complex machines with mental commands. Once science understands how the variations in brain waves are interpreted in

8 8 P age terms of muscle movement, decoding EEG time series, it will be able to recreate such movement with mechanical devices. A brain-actuated robot or UAV would enable a human operator to have an at-a-distance presence in a remote hostile environment and would revolutionize reconnaissance. The medical applications of such technology to prosthesis devices are limitless, everything from the traditional moving of limbs, to the feeling within those limbs becomes possible. Even such devastation as produced by spinal cord injuries will be overcome through thought-controlled exoskeletons. Today research on the decoding of EEG signals has been able to identify areas of the brain activated by imagined speech and such signals are processed to determine to which speaker in a cocktail party a subject is attending. EEG signals have also been directed to remotely control wheelchairs, cars and quadcopters and toy applications of this technology is already available on the internet. Tomorrow a single warfighter in an outpost will, with this thought-control technology, guide and control swarms of unmanned vehicles to lead an attack, and scout out the terrain in advance of the main force. All done through the silent communication of

9 9 P age thoughts that are inherently immune to external environmental sound and light. This would reduce the sensor-to-shooter time from the minimum of 20 minutes experienced in Afghanistan, to an essentially real time response. Such an individual, deep within enemy territory, will contain a biological marker to enable his comrades to distinguish him from the enemy and avoid the fate of the Civil War General Stonewall Jackson, who fell to friendly fire. Such genetically engineered proteins as the biological marker will also determine the properties of the camouflage material from which a uniform and lightweight armor are made. Just as in the case of the neurological signals for the movement of the monkey s arm transmitted over the Internet, and the remote control of vehicles by the human brain, in the next quarter century the thoughts of a warrior will be converted into the operation of sensor, information and weapon networks, all of which will be interconnected. The information contained in the action potential generated by the human brain will be directly converted via data processing into the operation of fire control networks, communication networks and the motion of vehicles.

10 10 P age Summary In networks-of-networks, every scale is linked to every other scale, either directly, or indirectly through phase transitions, and a formalism of complex networks capable of predicting that influence of change across multiple scales is now being developed. For example, scientists have formulated a principle that indicates that the maximum information exchanged between networks are matched in complexity. This principle highlights how situational awareness can degrade over time and how brain-to-brain coupling, such as has been achieved in rats ii, is optimal for information transfer. The dynamics of the brain, like that of other, out of equilibrium, complex networks, would be described by a Network Science in which the notion of a stationary state is replaced by a transient or ghost state. The adaptive behavior of coordination/cooperation describes the tendency towards and away from these ghost states, leaving only transients to describe the dynamics of such things as decision making. Imagine the security of soldiers being able to communicate with one another through a kind of synthetic telepathy, while on the battlefield. The messages would not be subject to background

11 11 P age noise being directly channeled into one s brain using a thought helmet. i B.J. West, M. Turalska and P. Grigolini, Networks of Echoes, Springer, NY (2014) ii M. Pais-Vieira, M. Lebedev, C. Kunicki, J. Wang, and M.A.L. Nicolellis, Sci. Rept. 3, 1319 (2013).