As befits an institute founded on a spirit of interdisciplinarity, interactions between the biological and computational sciences have always been strong at ÅÝܽ¶ÌÊÓƵ, particularly between the brain sciences and artificial intelligence (AI). Indeed the earliest research and teaching in AI, dating back to the mid 1960s, was to be found in the School of Biological Sciences (BIOLS), long before it took off on other parts of the campus. Most of that pioneering work came out of the Laboratory (later Subject Group) of Experimental Psychology whose founding head, Stuart Sutherland, was a strong advocate of AI and computational modelling of brain processes.
Sutherland was encouraged by the founding dean of BIOLS, John Maynard Smith, who had originally trained as an engineer and had been influenced as a young researcher by the great cybernetics thinkers of the time, including Alan Turing and Ross Ashby (Maynard Smith, 2008). Cybernetics was an interdisciplinary forerunner of computer science and AI which had, in the 1950s, had a significant impact on neuroscience, psychology and theoretical biology, among other fields. It is worth noting that both Sutherland’s and Maynard Smith’s doctoral supervisors, J.Z. Young and J.B.S. Haldane respectively, were prominent supporters of research in cybernetics.
Sutherland and Maynard Smith were able to attract brilliant young scientists with a keen interest in using computational ideas and methods in biology, including Max Clowes (visual processing) and Brian Goodwin (developmental biology), and later senior AI and computational neuroscience pioneers such as Christopher Longuet-Higgins and Albert Uttley (who had been a member of the Ratio Club, an important post-war cybernetic dining club). Sussex rapidly gained an international reputation for such research.
In 1974, partly propelled by what Maynard Smith termed the Clowes-Sutherland exclusion principle, Clowes, philosopher Aaron Sloman, psychologist Alistair Chalmers and philosophical psychologist Margret Boden founded a radical new interdisciplinary teaching programme that eventually became the influential School of Cognitive and Computing Sciences (COGS). (See Margret Boden’s essay in this volume for further details). Meanwhile, the strand of work involving computational modelling that Sutherland had initiated in Experimental Psychology continues to this day (now within the School of Psychology), with important contributions over the years from the likes of George Mather, Chris Darwin, Zoltan Dienes and Roland Baddeley.
In the late 1980s and early 1990s, COGS attracted a number of researchers with strong biological interests, particularly in neuroscience and evolutionary theory. Their work often involved computational modelling of biological phenomenon, as well as the development of biologically inspired methods and theories in AI and cognitive science. This group included Harry Barrow, Andy Clark, Dave Cliff, Inman Harvey, Phil Husbands and Mike Wheeler, with strong support from Boden, Sloman and other senior members of COGS. In 1991 they formed the Evolutionary and Adaptive Systems group (EASY) which quickly rose to international prominence. This development was greatly aided by reinvigorated links with BIOLS, with members of the neuroscience, evolution, and EASY groups regularly attending each other’s seminars. Collaborations between the two schools began to grow.
An important stimulus in the growth of the EASY group was the considerable strength, in BIOLS, of invertebrate neuroscience. There was a shared conviction among the neuroscientists, computational modellers and AI researchers that the study of relatively ‘simple’ systems was the best route to understanding general mechanism underlying the generation of behaviour in animals. Such understandings could potentially be used in new approaches to AI.
Neuroscience research at Sussex had flourished from the time of the institution’s establishment in 1963, but it received a very significant boost in 1990 when the university was awarded the status of an Interdisciplinary Research Centre (IRC) in Simpler Systems Neuroscience by the UK government’s Science and Engineering Research Council (SERC). Support for the idea of providing long-term substantial funding for major centres of excellence in areas of research with strategic importance had its origins in government policy of the mid 1980s. In response to this policy each of the UK’s research funding agencies, the Research Councils, invited IRC applications from universities in a limited number prescribed areas of scientific and social scientific research judged to be of national importance.
It was clear from the terms of reference of IRCs that successful proposals would have to demonstrate not only research excellence but also a strong interdisciplinary ethos and a commitment from the industrial sector to engage with the academics in realising the aims and objectives of the IRC. Needless to say, all of the UK’s top research universities prepared and submitted IRC applications – the competition for these high-status centres, especially in view of the unprecedented scale of their long-term funding, was going to be fierce.
One of the areas of research selected by the SERC for IRC funding in 1988/9 was Simpler Systems Neuroscience. By this time the SERC had already recognised this as a strategically important area. Indeed this recognition was reflected in the establishment of the Invertebrate Neuroscience Initiative (INI), which earmarked funds for research on invertebrate nervous systems in the mid 1980s. Largely through the efforts of Paul Benjamin, the ÅÝܽ¶ÌÊÓƵ had received a grant from the INI and it was Paul who first realised that Sussex might be well-placed to submit a much more ambitious IRC grant. Thus in 1988, under Paul’s leadership, a competitive bid for an IRC in Simpler Systems Neuroscience at Sussex began to take shape.
Sussex of course was not the only UK University capable to mounting a competitive bid for an IRC in this area. The University of Cambridge for example was particularly strong, as was UCL. Sussex neuroscientists were in no doubt of the need to make the strongest case that they possibly could. A decision was therefore made to ask Michael O’Shea, who had just returned to the UK from the University of Geneva, to help strengthen the Sussex bid by agreeing to become the Director the IRC and bring his research team to Sussex should the application be successful. Having succeeded to get through earlier elimination rounds, the Sussex University application faced its last obstacle – a head to head contest between Sussex and Cambridge, the two finalists. In the spring of 1990 an international panel of experts were tasked with deciding which should be awarded the IRC. Having visited both sites they made their decision: Sussex would have its IRC in Simpler Systems Neuroscience. Apart from Benjamin and O’Shea, original members of the IRC were Richard Andrew, Jonathan Bacon, Tom Collett, Julian Burke and Mike Land.
One of the reasons SERC decided to invest significantly in invertebrate neuroscience was that the simpler brains of invertebrate organisms provided model systems that yielded insights into the cellular and molecular mechanisms underlying mental functions of far more complex brains. In addition to this basic science rationale, there were potential practical benefits of such research that were of significant interest to industry. For example, the agrochemical industry was particularly interested in discovering new targets for novel pesticides based on our investigations of the neurochemistry of insect brains. Companies in the IT sector were interested for an entirely different reason, the same one that motivated the growing collaborations between the IRC and the EASY group: potential new insights for AI. The IRC’s work was revealing the fine details of neural networks which controlled the behaviour of some of the most successful inhabitants (insects and molluscs) of our planet. These networks evolved over eons to maximise survival in a dangerously unpredictable and unforgiving world. Simple neural networks studied in the IRC could inspire the design of artificial nervous systems for mobile robots, enabling them to behave autonomously and adaptively in challenging ‘real-world’ environments.
In 1995 an opportunity to extend the IRC-EASY collaborations arose through the auspices of the University’s Research Development Fund (RDF). Gordon Conway, the Vice Chancellor at the time, decided to use the RDF to provide significant funding to set up a new interdisciplinary research centre. Invitations to bid for the money were extended across campus. O’Shea and Husbands coordinated a successful proposal to establish a Centre for Computational Neuroscience and Robotics (CCNR).
The CCNR was created in 1996 as a joint venture between the IRC and COGS. Its remit was to explore and exploit the interface between neuroscience and computer science and engineering. The RDF award, which underpinned the CCNR’s funding for the first three years, helped to pump-prime new avenues of research, thus facilitating a major new initiative at Sussex. Importantly it meant that Sussex was well-placed to capitalise on new funding opportunities as the Research Councils (especially the EPSRC) began to recognise the strategic benefits of neuroscience-inspired computer science and computational approaches to biology. Also from the outset the CCNR attracted significant support from industry, particularly from BT.
Since its birth, the CCNR has attracted several million pounds of research funding, and many dozens of PhD students and post-doctoral researchers have passed through its doors, most going on to very successful careers in academia or industry. From the start it has had dedicated lab and office space ensuring that the mingling of biologists and AI researchers was not just confined to shared seminars. It quickly developed a leading international reputation and has continued to attract outstanding researchers from all over the world. The CCNR received helpful support from various deans of schools over the years, but without doubt the most important early ingredient in its success was the close proximity, and shared personal of the IRC with its interdisciplinary ethos and healthy disregard for discipline boundaries.
The initial aim of the centre was to encourage a two-way flow of ideas and methods between neuroscience and AI in order to provide new ways to help understand natural nervous systems and to pioneer fresh approaches to the development of robot control. Although research in these areas is still central, the remit of the CCNR has broadened to encompass a range of related cross-disciplinary activities. Major strands of work include: modelling and analysing biological neural systems, developing artificial neural systems, insect and robot navigation, bio-inspired evolutionary and adaptive approaches to robotics, the theory of natural and artificial evolution, modelling complex adaptive systems, development and cognition, and the application of bio-inspired technology to industrial problems. Major past achievements include Adrian Thompson’s seminal work on using bio-inspired techniques in electronics design. Since its earliest days the centre has also been involved in the use of adaptive technology and biological methods in the arts and has had many successful collaborations with artists and musicians. Since 2000 it has run an artist-in-residence programme with internationally well known artists such as Paul Brown, Jon McCormack and Stelarc being involved.
Today the CCNR continues to flourish and three of its earliest PhD graduates, Anil Seth, Andy Philippides and Paul Graham, are now established Sussex academics (the first two in the School of Informatics and the third in the School of Life Sciences). Indeed Seth, who returned to Sussex after several years in the USA, now co-directs the Sussex Centre for Consciousness Sciences, established with colleagues in the Brighton and Sussex Medical School and the School of Psychology, helping to push interactions between the biological and computational sciences in important new directions.
References
Maynard Smith, J. (2008) An interview with John Maynard Smith. In P. Husbands, O. Holland, M. Wheeler, (Eds), The Mechanical Mind in History, MIT Press, 373-382.