Models of Simon

Kumaraswamy Vela Velupillai

• Publisher: Taylor & Francis
• Year: 2017
• Print ISBN: 9780415311588
• eBook ISBN: 9781134385454
• Edition: 1
• Topic: Business
• Subtopic: Business General
• Index: Business

Herbert Simon (1916-2001) is mostly celebrated for the theory of bounded rationality and satisficing. This book of essays on Models of Simon tackles these topics that the he broached in a professional career spanning more than 60 years. Expository material on the fundamental concepts he introduced are re-interpreted in terms of the theory of computability.

This volume frames the behavioural issues of concern for economists, such as: hierarchy, causality, near-diagonal linear dynamical systems, discovery, the contrasts between the notion of heuristics, and the Church-Turing Thesis of Computability Theory. There is, consistently, an emphasis on the historical origins of the concepts Simon worked with, in emphasising Human Problem Solving and Decision Making – by rational individuals and institutions (like Organizations). The main feature of the results in the book are its emphasis on the procedural aspects of human problem solving, decision making and the remarkable way Simon harnessed many tools of mathematical logic, mathematics, cognitive sciences, economics and econometrics.

This long-awaited volume is an important read for those who study economic theory and philosophy, microeconomics and political economy, as well as those interested in the great Herbert Simon's work.

1
Introduction

My economist friends have long since given up on me, consigning me to psychology or some other distant wasteland. If I cannot accept the true faith of expected utility maximization, it is not the fault of my excellent education in economics…..

….

My traumatic exposure in 1935 to the budgeting process un the Milwaukee recreation department had immunized me against the idea that human beings maximize expected utility, and had made me an incorrigible satisficer…. [T]he theory of scientific discovery to which my study … has led me … is not a theory of global rationality, but a theory of human limited computation in the face of complexity.

– Simon (1989a, pp. 394–395; italics added)

It is, therefore, important to recall that a few years earlier Simon (1986, p. S.223; italics added) had clarified this notion of complexity and its implication for deductive prediction:

In this kind of complexity, there is no single sovereign principle for deductive prediction.

My own precept for Simonian Human Problem Solvers was thus (Rustem & Velupillai, 1990, p. 432; italics added):

The search for simplicity in the growth of complexity is the exercise of reason

Simon was the quintessential Reasonable Man, whose task was to search systematically for solutions to problems, even in the absence of any ‘single sovereign principle of deductive prediction’. Searching systematically, in my interpretation, was an act of procedural rationality. Hence, human problem solvers were procedurally implementing systematic search procedures. The space over which a solution was systematically searched for was complex; simplifying it was an act of the reasonable human problem solver. It is here that the reasonable human problem solver was characterized by boundedly rational systematic search, which led to (often but not always) a satisficing solution.

This ‘incorrigible satisficer’, the self-confessed ‘closet-engineer since the beginning of [his] career (Simon, 1991, p. 109), the monomaniacal (Simon, 1957, p. vii) rational artificer, the compelling storyteller traversing Mazes without Minotaurs (ibid., chapter 15 & Simon, 1991, chapter 11), the renaissance man par excellence, master applied mathematician, founding father of cognitive behavioural science, with a supreme mastery of the methodology of scientific method and processes of discovery and much else, was said to have ‘had it put together at least 40 years ago [i.e., in 1939!]’ (Newell, 1989, p. 400). ‘The central idea’, in putting it all together, being ‘bounded rationality’ (ibid., p. 400).

It would be a very foolish person, or one who is outrageously audacious, who would contradict Allen Newell’s characterization of the central theme of Herbert Simon’s lifelong research program. I hope I am not ‘foolish’, nor do I think of myself as ‘outrageously audacious’. Yet, the central theme that informs my interpretation of Models of Simon is not bounded rationality; it is Human Problem Solving. I emphasize the Human aspect of Problem Solving to draw parallels with Alan Turing’s Computability (Machine) approach to Solvable and Unsolvable Problems (Turing, 1954) – a parallelism, bordering on similarity, emphasized in Simon’s Machine as Mind in The Legacy of Alan Turing (Simon, 1996a), explored in Velupillai (2013, pp. 339–341). The contrast should be with the (partially) intuitionistic, constructive, Brouwer–Heyting–Kolmogorov (BHK) approach to proof procedures (Brouwer, 1923, 1925; Heyting, 1930 a, 1930b; Kolmogorov, 1932).

In the rest of this Introduction I try to explain – even, at times, justify – my (peculiar) stance, which goes ‘against the stream’ of current and standard interpretations of Simon’s construction of his vision (which was in constant flux on many fronts, but not all). In addition, this Introduction tries, also, to summarize the contents of each of the chapters and the various appendices. It is doubtful that the summaries are a surrogate for the actual contents of the different chapters and appendices, where their contexts are also made reasonably clear.

Without a uniform point of view that I can adopt, it would be impossible to tell a coherent story of Simon’s vision. For such a uniform approach it is necessary that I interpret Simon’s theories, the methods with which he formed, developed, implemented and experimented with them from some unified stance. If not it will be a patchwork quilt, which has its own charm, even usefulness, but it is not one that I think Simon would approve. I am, of course, not trying to develop a unified stance to satisfy (sic!) Simon’s (posthumous) approval, even if only in spirit. However, it – the unified stance – impressed upon me, as I read, reread, worked with, worked over and tried to understand the many-faceted world that Simon strode with a mastery of diverse tools, concepts, methods and theories.

Thus, in may seem at odds with any unified stance when one views his work on the Hawkins–Simon conditions for stability, on causality, identification and aggregation, on evolution due to semi-(or near-)decomposability, to a variety of aspects of organization and administrative theories, to empirical microeconomics, even to game theory and macrodynamics, to discovery, creativity, axiomatics and the philosophy of science, to artificial intelligence and the weird and wonderful world of cognitive – or classical – behavioural social sciences. Without exception, Simon’s contributions to social, human, natural and pure sciences, both in their theoretical and empirical aspects, are absolutely original. If a second-rate (perhaps, actually, third-rate) intellectual like myself tries to interpret and summarize these outstandingly original visions and contributions it can only lead to third-(or fourth-)rate results and less than worthy stories of the work of a – I choose this word deliberately – magician, one who wrought, out of the material available to all and sundry, a world of possibilities that enriched experience, both theoretically and from a policy point of view.

But if I am able to find a convincing and unified theoretical stance, to tell the story of Simon’s many-faceted visions, then it might – at least – mitigate the narrative’s potential low status and, who knows, may even contribute to a development of one (or more) of the many frontiers he broached, and created.

Thus, my search – I nearly qualified it by using ‘sic’! – for a unified (theoretical) stance, at least one that is consistent with his explicitly expressed visions of the many-faceted world he created and strode, like the colossus he was.

My first attempt at formulating a unified theoretical vision was to begin and end chronologically, to begin at the beginning with bounded rationality, (henceforth, BR) refined around the end of the first third of his professional life with the addition of satisficing (hereafter referred to as SAT) embellished with the pioneering of cognitive behavioural science (CBE, where the ‘C’ could refer, interchangeably, to ‘Cognitive’ or ‘Classical’) and artificial intelligence; then, in that properly fertile professional time of ‘half-life’, came the monumental Human Problem Solving, summarizing, essentially, the novel approach to problem solving as a dynamic process, underpinned by a model of computation that was peopled by solvers who were procedurally rational.

This first attempt led to my ‘unified theoretical stance’ in terms of a model of computation – in particular, the Turing Machine but also using the Church–Turing Thesis as I then understood it, interpreting heuristics as algorithms in the sense of computability theory – and buttressed by computational complexity theory. This enabled me to embed BR and SAT as elements defining procedural rationality and use the idea of the duality between computational processes and dynamical systems.

One important observation must be made at this point. I do not think, or ascribe to, Simon’s world, peopled by procedurally rational agents, indulging in problem solving in diverse domains, was stochastic or probabilistic in any ad hoc sense; any probabilistic underpinning came from an algorithmic information stance, which was also the foundation on which randomness was based, and this was embraced wholeheartedly by Simon, all the way from the time of the Dartmouth conference, where artificial intelligence – AI, henceforth, in the Turing tradition was enunciated – to the end of his life (but especially so in Simon (1989a, 1996a)).

This was the first attempt helped me organize my thoughts and interpretations of Simon’s contributions fairly adequately, but was, I felt, inadequate from many theoretical and empirical points of view. It was, however, a rereading of Section 6 of Models of Discovery (henceforth, MoD) that led to my current unified theoretical stance. To this must be added the influence of a serious (‘nth’) reading of Martin Davis’s classic Computability and Unsolvability (in its Dover version as Davis, 1982).

My current unified theoretical stance, to interpret the many-faceted world of Simon is based entirely on computability theory because I am able to interpret mathematical logic from a recursion theoretic point of view (Davis, ibid., chapter 8) and understand the interplay between the underpinnings of predicate logic in axioms and inference rules and computation rules, for the way Human Problem Solving leads to search for proofs of the discovery of structures with some notion of order, by creative, procedurally rational agents (or organizations or processes of evolution), in a dynamically evolving environment. It is as if Simon was trying to tame an evolving pattern of a jigsaw puzzle, pro tempore, which is why chess played in important role in his research (and time eluded his obvious path towards GO, or Weiqi,¹ where the surprise element in an evolutionary process plays an important part in the thought processes implemented by Human Problem Solvers playing this game, when pitted against Machine-based strategies).²

But above all, it is the inspiration of the Ramsey Theorem, and Ramsey Theory, in general, together with a particular uncomputability between the former and Busy Beaver functions that were instrumental in disciplining my unified stance of Simon’s world of Human Problem Solving. The duality I discovered, between R(k, l)³ in Ramsey Theory and S (m, n)⁴ in the theory of Busy Beavers, together with the recursive graph theorem of Brattka (2008), was instrumental in understanding the essentially constructive nature of human problem solving in chess, Cryptarithmetic and (Recreational) Games that form the backbone of Newell and Simon (1972).

Essentially, five precepts form the backdrop against which I formulate the unified vision to tell this story of Models of Simon (stated in the chronological sequence, whi...

Table of contents

• Cover
• Title
• Copyright
• Dedication
• Contents
• List of Figures
• Preface
• Appendix to the Preface
• Acknowledgements
• 1 Introduction
• 2 Herbert Simon: a life
• 3 Classical – or Cognitive – Behavioural Economics (CBE) versus Modern Behavioural Economics (MBE)
• 4 The zeitgeist: cybernetics, servomechanisms, information, communication and computation
• 5 Heuristics versus algorithms
• 6 Computable and computational complexity theoretic bases for cognitive behavioural economics
• 7 Notes on hierarchic, near-decomposable, causal, evolutionary dynamics: The Architecture of Complexity
• 8 Gandy’s Mechanisms Modelling the Rationality of Organizations by March & Simon
• 9 Towards a Classical Behavioural Finance Theory
• 10 Intuitionistic foundations for economic theory: a Brouwer–Simon research programme
• 11 A crooked path along The Gravel Walks
• Appendix 1: Labyrinths and mazes without Minotaurs – Borges’s library, Simon’s mazes and Turing’s librarian
• Appendix 2: Herbert Simon’s pioneering methodological work in confirming public predictions
• Appendix 3: Professor Herbert Simon: an obituary
• Appendix 4: Letter from Herbert Simon
• Author Index
• Subject Index