Martin Taylor

Maurice Martin Taylor died in the afternoon of Friday 17 March 2023, after several months of terminal illness. His 87th birthday was nine months prior, on 14 June 2022.

By temperament and by education Martin was well prepared for a creative career as a scientist. He was privileged in having exemplary forebears who employed their considerable intellectual and aesthetic capacities in socially beneficial ways, and in having mentors, family and others, who guided the development of his own very productive career. He has written that they set a high standard of example and expectation. Appreciation that others may be smarter or more talented than oneself can be salutary. Their example seems to have fostered a degree of modesty as well as aspiration. His interests in communication and interaction were a central motif of his career, and came to fruition in his work with Kent McClelland and others on collective control.

His experience with computing began early. In 1952, while he was still in high school, he watched one of the first large-scale computers being delivered in sections through a second-story window of the Physics Building at the University of Toronto (acquired thanks to the penury of a Tory government in England). A few years later, he was employed as an undergraduate student assistant in its operation. He received his BASc in Engineering Physics there in 1956, with a focus on communications. At Johns Hopkins he earned an MSE in operations research. This discipline constructs mathematical and computational models of systems, the better to understand their capacities and limitations in allocating and controlling limited resources, providing guidance for decision-makers and managers in organizations. The pertinence to communication and collective control is easy to see in hindsight.

His 1960 PhD at Johns Hopkins was in experimental psychology, and like Bill Powers he entered a career applying both physics and psychology to problems of medical research. He worked on perception and control in what came to be called ‘human factors’, with applications in design of systems and of training in their use. For about 35 years, he says (in PPC ), “I split my time about 50-50 between computer hardware and software development and research into human perception in vision, hearing, and touch, before the two threads came together around 1980 in the form of multi-modal human-computer interaction research.”

Even before his arrival at Johns Hopkins Martin had begun developing Layered Protocol Theory (LPT), a framework for modeling communication and collective control. Around 1990, he realized that LPT is a ‘special case’ of PCT. A lesser mind might have failed to see the relationship, or might have tried to fit PCT into the familiar framework in which he had invested so many years of work and advocacy. A lesser man might have begrudged the outward appearance of subordinating his life’s work to PCT, and we’d have heard no more of him. It is a mark of Martin’s commitment to science and of his character as a human being that for the last 33 years of his life he turned whole-heartedly to working out the ramifications of this deceptively simple science, Perceptual Control Theory.

He applied tools of mathematics and principles of physics and biophysics to better understand the capacities and limitations of control systems, and especially how to model communication and collective control among autonomous hierarchical control systems within a public environment. To understand why this was not always welcomed by all participants in CSGnet conversations, some context is helpful. In the years just prior to his arrival, prominent individuals (Carver & Scheier, William Glasser, and others) had seemed ready to understand and advance PCT, but instead appropriated the name and terminology within their existing systems of books and workshops, while mistaking and obscuring the fundamental principles of control. Secondly, in 1991, discussions moved from a private email exchange to a public listserv called CSGnet hosted at Indiana University by Gary Cziko. Suddenly there were greater numbers of newcomers bringing their preconceptions and their personality needs. In this circumstance, some construed Martin’s challenging questions and proposals as more of the same. His steadfast pursuit of core issues, as he saw them, may be clearly seen in the CSGnet archives.

In 2011, Martin initiated a separate discussion forum called ECACSnet “for two main reasons, firstly because of the very limited half-life of discussion threads on CSGnet, which meant that the wheel had to be reinvented many times, and secondly because discussion of the complexity of real-world perceptual control seemed to be crowded out on CSGnet in favour of one-loop and two-level control systems acting in a homogeneous environment” (Martin Taylor 2014. There was more discussion that year about setting up an alternative forum platform, but it did not result in any change. Five years later in 2019, just before the COVID‑19 pandemic, IAPCT migrated the CSGnet conversations to Discourse. Martin was a significant supporter of this vital transition and of IAPCT in general.

Martin’s crowning achievement, The Powers of Perceptual Control (PPC), was the focus of the last decade of his life, or more. The writing first took form as his chapter for The interdisciplinary handbook of Perceptual Control Theory: Living control systems IV. He arrived at the title early. Although in Warren’s 2013 outline of projected contributions to LCS IV his initial chapter title was “Language and Culture as Malleable Artifacts”, soon a few months later he told Rupert Young that he didn’t have time to work with him on parallel programming to model a polyflop because he was devoting most of his time to writing PPC. In 2019-2020 he and Warren cut his chapter of the Handbook in half because it had grown too big. He began preparing the excised portion as a journal article, but the more he delved into the properties and consequences of control systems the more it grew. In this 4-volume book he explores the universe of hypotheses and conjectures that flow logically from PCT and from an analytical consideration of the properties, capacities, and limitations of control systems.

Like Bill Powers, Martin was a subscriber to and an assiduous reader of important journals. In January of 2021, an article in Science rattled his thinking and writing and threw almost every volume and chapter of PPC into reorganization, no doubt causing much crumpling of paper. In the Introduction to PPC, he says:

Rattling may be the most important concept newly introduced to PCT in this work …. It is a statistical measure in the same family as ‘variance’ and ‘uncertainty’. Like them, it can be measured only as a measure of a collection of individual measurements, which may be distributed over time, over a set of measures of the same variable over individuals or over a set of individuals. Rattling relates not only to structural organisation within individual life forms, but to structural relationships among societies of living beings, which could be groves of trees just as well as animate beings such as fish or humans, or entire ecologies.

Rattling is closely related dynamically to changes in organisational structure, often minor, often smooth, but sometimes dramatic and sudden. Introduced by Chvykov et al. (2021), it is a measure of the uncertainty of the rate of change of a collective measure such as variance. Rattling applies to complete organisational structures rather than to individual entities in the organisation that interact with each other.

Also in 2021, mathematical results reported in Nature (Andrejevic et al. 2021) provided a metaphor for the events that induce rattling and sometimes reorganization in a control hierarchy, and conversely for the reorganized control processes that result from this. “Although one must never take a metaphor as exactly representing the analogue — as Korzybski (1933) said: ‘The map is not the territory’—nevertheless, a good metaphor is often helpful in understanding the analogue.” This metaphor “helps one to understand the refinement and consciously experienced discrimination of perceptual categories as a living entity matures.”

Do perceptions constitute a map of the ‘territory’ of Reality? The philosophical implications of PCT have led many a discussion down a rabbit-hole. Most PCT research and modeling has been limited to individual control systems in a simplified environment, where the deeper ontological and epistemological issues can be ignored. This is analogous to Galileo’s early experiments rolling balls down inclined planes, which disclosed fundamentals of physics (mass, force, inertia, gravity). Only after these were understood could anyone derive ballistic trajectories and planetary orbits, or discover unknown planets from perturbations in the orbits of known planets. As we move from single control systems in laboratory conditions to interactions of interacting systems, we must account for their shared environment, what exactly it means that they share it, and how their respective perceptions are related. In hindsight, we see that even in simplified laboratory cases, the investigator may be unwittingly participating with the subject in a circumstance of collective control. The relationship between the input quantity q i (a quantitative measurement), a neural firing rate p (sometimes measured by a neuroscientist), and a subjectively experienced perception (sometimes conscious) is a more prominent issue when we assert that two individuals are controlling the same variable. Martin had time and incentive to consider these conceptual difficulties deeply, as they were in the background of his work from the beginning, and his framing of the issues has always been incisive and cogent.

Like many of us, Martin hoped for social benefits from widespread understanding and adoption of PCT. He felt that the greatest interest and persuasion will result from demonstrating the power of PCT in matters of greatest public interest, social phenomena such as human relationships, economics, and politics, language and culture, conflict and comity. He knew that for such demonstrations PCT must build upon the foundation that has been laid and extend beyond the psychology of individual control systems to collective control. In the canonical diagram of an individual control loop, a simple box labeled ‘Environmental feedback function’ represents cause-and-effect sequences in the environment which link output to perceptual input. This suffices for a laboratory setup with tightly constrained influences. That ‘Environmental feedback function’ box needs more detail when it is shared. In social situations, two or more control loops pass through the same shared environment, and their respective environmental feedback paths may intersect.

Martin collaborated with Kent McClelland in the theory, investigation, and modeling of collective control. For these projections of purpose through the environment, Kent proposed the vivid imagery of ‘gossamer threads’ passing through perceived objects, and often intersecting in them. To be more precise than appealing imagery requires some new terminology and methods, just as in any other science. Something in the environment which can be used as means of controlling diverse perceptions they called an ‘atenex’ (ATomic ENvironmental NEXus); when a physical property of an atenex actually is used as a cause-effect link within the environmental feedback path for controlling some perception, it is called an ‘atenfel’ (ATomic ENvironmental FEedback Link). Sometimes combinations of these are required, as for example a screw and screwdriver can be used for fastening, but separately neither suffices for that purpose. The corresponding terms are molenex and molenfel, on the analogy of atoms and molecules.

These concepts are allied with the notion of an ‘affordance’, without Gibson’s naive Realism. They support the important distinction between the social institutionalization of collectively controlled variables and their ad hoc employment as means of controlling other variables. In PPC, Martin shows how they provide means to clarify the philosophical difficulties of Reality and consciousness, with perceptual control the interface between them. In PPC (I.2.4), he says:

Most objects we think of as physical objects are atenexes, in that they offer many different properties that could be the environmental part of atenfels for controlling different perceptions. Later, we will come close to claiming that all we can know of their possible existence in Real Reality is the total bundle of atenfels that they might embody. In other words, we will claim that the atenex is the reality, the object just a possible means of producing the atenfels. In Volume II and in Volume IV, we will use a metaphor based on crumpling paper to clarify a complex relationship between consciousness, category and object perception, and the relation between language and the analogue hierarchy.

Many of us have been impressed by Martin’s intellectual brilliance. Kent said “he was one of the smartest people I’ve had the privilege of knowing.” Phil Farrell said “He introduced me to PCT and became my PCT mentor way back in the 1990s. At Defence and Civil Institute of Environmental Medicine (DCIEM, now called DRDC Toronto Research Centre) we always had great badminton games and great conceptual formulation sessions about Layered Protocol Theory and General Protocol Grammar. What a beautiful mind. He will be missed.”

But Martin had no personal need to make this impression on others. His was a gentle presence with an incisive voice at our conferences. He argued forcibly for ideas and principles, but not as means to assert dominance or superiority. Warren Mansell said

I met Martin in Manchester for our IAPCT annual conference. He was very kind and thoughtful and wonderfully intelligent. He was well aware of his physical frailty even then and I have always been stunned by his resilience and continued intellect and creativity throughout his last years. I personally believe that his work on PCT is the most extensively considered and meticulously connected to wider theories, processes and applications of any writer. At some point I am sure his writing will grow in their readership and following.

Eva de Hullu likened him to Nestor, the old warrior and wise counselor to the Greeks at Troy:

When I went to the 2019 IAPCT conference in Manchester, it was a new world for me. Here was a group of people with intense and genuine interest in a subject that I recently came to care about. I had a large variety of encounters. I was tested by some, greeted grandfatherly, recognized as a fellow PCT human and embraced by others. Martin stood out as someone special, as our Nestor. Walking through Manchester I remember some of us watching over him: he looked fragile then but was perfectly able to get along and walk with the group. Talking to him, I got a glimpse of the immense architectural structure [of PPC] in his mind, built with PCT building blocks. He explained to me, with kind patience, how protocols worked. He invented new words for constructions that were needed in his PCT architecture, like “atenfels”. I know that Martin really wanted people to read his book, to get to understand what he was seeing. That was a difficult part for me. The book is immense and when I started reading a part to answer one question, I would be confronted with many more questions. I could rarely find the patience or time needed to go that far. Reading his work is a confrontation with how much there is to know and understand, and how little I actually know yet. I feel comforted by the idea that his book is there in the future for me to find a few answers and many more questions and to enjoy the curiosity and the joy of learning that he has kept up until his last breath. I feel very grateful for having had the chance to meet him.

Curiosity and joy of learning, yes, with a mix of confidence and humility that could seem paradoxical, these are characteristics that Martin and Bill had in common. The confidence stems from recognition of the rock-solid lasting importance of PCT and of its revolutionary consequences yet to come; the humility from understanding PCT well enough to recognize how much we do not yet understand, and how much of its power and scope remains to be realized. Both men expected that it might take centuries to develop PCT to its full potential. Both were acutely hopeful that a wider public understanding of control will ameliorate social and political arrangements, reduce wasteful and destructive conflicts, and release the Anthropocene epoch from its possibly suicidal beginnings into a bright future of humanity on this planet and perhaps elsewhere.

This obituary and further reactions are posted on Discourse: and on Twitter.