Aaron Sloman's first degree was in mathematics and physics, at CapeTown in 1956. A Rhodes Scholarship brought him to Oxford in 1957, where, after meeting philosophers with incorrect beliefs about mathematics, he switched and in 1962 completed a DPhil in philosophy of mathematics, defending Kant's view of mathematics. He taught philosophy at Hull then, from 1964, at Sussex University. In 1969 he was introduced to AI by Max Clowes, and decided that designing working AI systems was the best way to make progress in epistemology and philosophy of mind, language, mathematics, science etc. In 1971 he criticised the 1969 logicist AI manifesto by McCarthy and Hayes, and was invited to spend 1972-3 among AI researchers in Edinburgh, one of the few AI centres at the time, where he learnt about computer science, AI, cognitive science, started programming in earnest, and met leading researchers in AI, including international visitors.
Back at Sussex, he collaborated with Max Clowes, Margaret Boden and others, developing teaching and research in AI and computational cognitive science, and contributed to the Poplog AI development system, used for teaching and research in academe and industry. In 1978, he published The Computer Revolution in Philosophy: Philosophy science and models of mind (revised edition freely available online) He also helped to found the School of Cognitive and Computing Sciences (COGS). In 1991 he moved to Computer Science in Birmingham, helping to build up teaching and research in AI, including modelling of motivation and emotional architectures, and robotics. Retired in 2002, but continued full time research.
After encountering AI in 1969 Sloman suspected Kant's philosophy of mathematics could be defended using a design for a "baby" robot that could grow up to be a mathematician, like Archimedes, or Euclid, using mechanisms consistent with, but more precisely defined, than Kant's proposals. Nearly half a century later it's clear that nobody knows how to do that. An invitation to contribute to the Elsevier Turing Centenary Volume (https://goo.gl/9eN8Ks), including commenting on Turing's amazing Morphogenesis paper, led him to ask why Turing had written about chemical morphogenesis, and what he would have done had he lived several more decades.
The conjectured answer was a proposed multi-strand Meta-Morphogenesis (M-M) project on which Sloman has been working (unfunded) ever since, trying to understand how a physical universe can make it possible for biological evolution to produce the minds of ancient mathematicians (long before modern logic, algebra and cartesian representations of geometry), and other species with high levels of spatial intelligence, e.g. nest-building birds, squirrels, elephants and many others, abilities still completely lacking in current AI and unexplained by psychology and neuroscience (though often mis-described).
"Meta-" indicates that evolution's products repeatedly extend the powers of evolution, mainly through production of new construction-kits, all based on, but different from, the fundamental physical construction kit. Turing suggested in 1938 that mathematical discovery needs both intuition and ingenuity and suggested that Turing-machines can provide ingenuity (if suitably programmed) but not intuition. Sloman's research project will explore ways of modifying the Turing-Machine specification to fill the gap: a multi-membrane-Super-Turing machine, perhaps using sub-neural chemistry.
Achievements and awards
Sloman's 1962 thesis remains relevant, originally unpublished, but now freely available online, along with papers in philosophy of science, meta-ethics, semantics, philosophy of language, philosophy of information, philosophy of mathematics, philosophy of mind, vision (e.g. including functions of biological vision), affective phenomena - emotions, moods, etc. (topics seriously misconstrued by AI modellers and others), consciousness, evolution, epigenomics (jointly with Jackie Chappell) and programming.
His 1978 book (CRP), a progress report, made predictions about uses of computers that have recently come true, and discussed philosophical and psychological problems to be solved through (not yet achieved) developments in AI. The Epilogue disposed of the "AI Singularity" problem (still not on the horizon, despite misinformed dire predictions!). The Poplog development system, to which he contributed ideas, code, documentation and management, eventually became a successful commercial product, winning a UK "Smart" award for $5Million sales: now a free, open source system, including the Birmingham SimAgent toolkit.
CRP and work on Poplog led to invitations and collaborations, including a two year professorial fellowship from GEC. He was elected a fellow of AAAI, then a fellow of AISB, and a fellow of ECAI (Now EURAI). In 2008 Sussex University awarded him an honorary DSc. Since starting the Turing-inspired Meta-Morphogenesis project in 2012, working alone, and unfunded, he has received international conference/workshop invitations, has given tutorials (e.g. Edinburgh ESSENCE school, IJCAI 2016, the PACS Symposium Seoul 2016, Diagrams 2018) and published progress reports on the project. There remain huge problems. Deep learning won't help. Perhaps chemistry will??