A. Newell and H. A. Simon were two of the most influential scientists in the emerging field of artificial intelligence (AI) in the late 1950s through to the early 1990s. This paper reviews their crucial contribution to this field, namely to symbolic AI. This contribution was constituted mostly by their quest for the implementation of general intelligence and (commonsense) knowledge in artificial thinking or reasoning artifacts, a project they shared with many other scientists but that in their case was theoretically (...) based on the idiosyncratic notions of symbol systems and the representational abilities they give rise to, in particular with respect to knowledge. While focusing on the period 1956-1982, this review cites both earlier and later literature and it attempts to make visible their potential relevance to today's greatest unifying AI challenge, to wit, the design of wholly autonomous artificial agents (a.k.a. robots) that are not only rational and ethical, but also self-conscious. (shrink)

Is attention both necessary and sufficient for consciousness? Call this central question of this treatise, “Q.” We commonly have the experience of consciously paying attention to something, but is it possible to be conscious of something you are not attending to, or to attend to something of which you are not conscious? Where might we find examples of these? This treatise is a quest to find an answer to Q in two parts. Part I reviews the foundations upon which the (...) discourse on Q is built. Different inputs to Q produce different answers. After consideration of the many ways “attention” and “consciousness” have been defined, I settle upon phenomenal consciousness and Executive Attention (defined as a suite of strategies for structuring cognition for further processing implemented by the executive of working memory) as the most interesting inputs to Q, and the ones on which Part II focuses. Attention without consciousness seems relatively easy to establish empirically, but consciousness without attention is much harder. The putative candidates all seem to have major problems, but I build a strong abductive case for the hitherto ignored case of foveal phenomenal overflow. We consciously see far more detail in our foveal fields than we can Executively Attend, although there is a serious obstacle to our ever confirming that empirically—identifying conscious content relies on Executive Attentional report. Triangulating the capacity limitations of attention, consciousness, and working memory strengthens this case for consciousness without attention, and suggests that cognition must work something like my “Witches’ Hat Model,” on which content can become conscious outside of Executive Attention or working memory. I conclude with some reflections on the implications of my arguments for the discourse on Q, and for other discourses such as the ontologies of attention and consciousness, theories of consciousness, some other cognitive concepts, and ethical considerations in humans, animals, and machines. A conclusive answer to Q continues to elude us. It may perhaps be an ultimately insoluble conundrum. But it is the very essence of humanity to seek an answer, and in so doing, to improve our understanding of our own nature: “The proper study of mankind is man.”1. (shrink)

Working memory is a foundational construct of cognitive psychology, where it is thought to be a capacity that enables us to keep information in mind and to use that information to support goal directed behavior. Philosophers have recently employed working memory to explain central cognitive processes, from consciousness to reasoning. In this paper, I show that working memory cannot meet even a minimal account of natural kindhood, as the functions of maintenance and manipulation of information that tie working memory models (...) and theories together do not have a coherent or univocal realizer in the brain. As such, working memory cannot explain central cognition. Rather, I argue that working memory merely redescribes its target phenomenon, and in doing so it obfuscates relevant distinctions amongst the many ways that brains like ours retain and transform information in the service of cognition. While this project ultimately erodes the explanatory role that working memory has played in our understanding of cognition, it simultaneously prompts us to evaluate the function of natural kinds within cognitive science, and signals the need for a productive pessimism to frame our future study of cognitive categories. (shrink)

A leading theoretical framework for naturalistic explanation of mind holds that we explain the mind by positing progressively "stupider" capacities ("homunculi") until the mind is "discharged" by means of capacities that are not intelligent at all. The so-called homuncular fallacy involves violating this procedure by positing the same capacities at subpersonal levels. I argue that the homuncular fallacy is not a fallacy, and that modern-day homunculi are idle posits. I propose an alternative view of what naturalism requires that reflects how (...) the cognitive sciences are actually integrating mind and matter. (shrink)

Short abstract (98 words). Reasoning is generally seen as a means to improve knowledge and make better decisions. However, much evidence shows that reasoning often leads to epistemic distortions and poor decisions. This suggests that the function of reasoning should be rethought. Our hypothesis is that the function of reasoning is argumentative. It is to devise and evaluate arguments intended to persuade. Reasoning so conceived is adaptive given humans’ exceptional dependence on communication and vulnerability to misinformation. A wide range of (...) evidence in the psychology of reasoning and decision making can be reinterpreted and better explained in the light of this hypothesis. (shrink)

To explain why, in scientific problem solving, a diagram can be “worth ten thousand words,” Jill Larkin and Herbert Simon (1987) relied on a computer model: two representations can be “informationally” equivalent but differ “computationally,” just as the same data can be encoded in a computer in multiple ways, more or less suited to different kinds of processing. The roots of this proposal lay in cognitive psychology, more precisely in the “imagery debate” of the 1970s on whether there are image-like (...) mental representations. Simon (1972, 1978) hoped to solve this debate by thoroughly reducing the differences between forms of mental representations (e.g., between images and sentences) to differences in computational efficiency; to carry out this reduction, he borrowed from computer science the concepts of data type and of data structure. I argue that, in the end, his account amounted to nothing more than characterizing representations by the fast operations on them. This analysis then allows me to assess what Simon’s approach actually achieves when transported from psychology to the study of scientific representations, as in Larkin and Simon (1987): it allows comparing, not representations in and of themselves, but rather the computational roles they play in particular problem-solving processes—that is, representations together with a particular way of using them. (shrink)

Logic has been a—disputed—ingredient in the emergence and development of the now very large field known as knowledge representation and reasoning. In this book (in progress), I select some central topics in this highly fruitful, albeit controversial, association (e.g., non-monotonic reasoning, implicit belief, logical omniscience, closed world assumption), identifying their sources and analyzing/explaining their elaboration in highly influential published work.

My dissertation explores the ways in which Rudolf Carnap sought to make philosophy scientific by further developing recent interpretive efforts to explain Carnap’s mature philosophical work as a form of engineering. It does this by looking in detail at his philosophical practice in his most sustained mature project, his work on pure and applied inductive logic. I, first, specify the sort of engineering Carnap is engaged in as involving an engineering design problem and then draw out the complications of design (...) problems from current work in history of engineering and technology studies. I then model Carnap’s practice based on those lessons and uncover ways in which Carnap’s technical work in inductive logic takes some of these lessons on board. This shows ways in which Carnap’s philosophical project subtly changes right through his late work on induction, providing an important corrective to interpretations that ignore the work on inductive logic. Specifically, I show that paying attention to the historical details of Carnap’s attempt to apply his work in inductive logic to decision theory and theoretical statistics in the 1950s and 1960s helps us understand how Carnap develops and rearticulates the philosophical point of the practical/theoretical distinction in his late work, offering thus a new interpretation of Carnap’s technical work within the broader context of philosophy of science and analytical philosophy in general. (shrink)

Mercier & Sperber (M&S) argue for their argumentative theory in terms of communicative abilities. Insights can be gained by extending the discussion beyond human reasoning to rodent and robot navigation. The selection of arguments and conclusions that are mutually reinforcing can be cast as a form of abductive reasoning that I argue underlies the construction of cognitive maps in navigation tasks.

Cognitive science is an interdisciplinary research endeavor focusing on human cognitive phenomena such as memory, language use, and reasoning. It emerged in the second half of the 20th century and is charting new directions at the beginning of the 21st century. This chapter begins by identifying the disciplines that contribute to cognitive science and reviewing the history of the interdisciplinary engagements that characterize it. The second section examines the role that mechanistic explanation plays in cognitive science, while the third focuses (...) on the importance of mental representations in specifically cognitive explanations. The fourth section considers the interdisciplinary nature of cognitive science and explores how multiple disciplines can contribute to explanations that exceed what any single discipline might accomplish. The conclusion sketches some recent developments in cognitive science and their implications for philosophers. (shrink)

We conducted a computer-based psychological experiment in which a random mix of 40 tautologies and 40 non-tautologies were presented to the participants, who were asked to determine which ones of the formulas were tautologies. The participants were eight university students in computer science who had received tuition in propositional logic. The formulas appeared one by one, a time-limit of 45 s applied to each formula and no aids were allowed. For each formula we recorded the proportion of the participants who (...) classified the formula correctly before timeout and the mean response time among those participants. We propose a new proof formalism for modeling propositional reasoning with bounded cognitive resources. It models declarative memory, visual memory, working memory, and procedural memory according to the memory model of Atkinson and Shiffrin and reasoning processes according to the model of Newell and Simon. We also define two particular proof systems, T and NT, for showing propositional formulas to be tautologies and non-tautologies, respectively. The accuracy was found to be higher for non-tautologies than for tautologies ( p T was.89 and for non-tautologies the correlation between latency and minimum proof length in NT was.87. (shrink)

The Lisp programming language is often described as the first functional programming language and also as an important early AI language. In the history of functional programming, however, it occupies a rather anomalous position, as the circumstances of its development do not fit well with the widely accepted view that functional languages have been developed through a theoretically-inspired project of deriving practical programming languages from the lambda calculus. This paper examines the origins of Lisp in the early AI programming work (...) of the mid-to-late 1950s, and in particular in the work of Allen Newell, Cliff Shaw and Herbert Simon. Their 1956 program, the Logic Theory Machine, introduced new ideas about data and program structures that were articulated in response to perceived limitations in existing programming technique. Later writers, notably John Backus, have described these features as constituting a “programming language style” distinct from the traditional style that preceded it. The paper examines the origins of the earlier style in practices of manual computation, analyses the key technical differences between it and the style first manifested in the Logic Theory Machine, and concludes that programming practice and experience play a large and underappreciated role in the development of programming styles and languages. (shrink)

In this article, I revisit Pierre Bourdieu's concept of habitus and contrast it with Herbert Simon's notion of bounded rationality. Through a discussion of the literature of economic sociology on status and Fligstein's political-cultural approach, I argue that this concept can be a source of fresh insights into empirical problems. I find that the greater the change in the social environment, the more salient the benefits of using habitus as a tool to analyze agents' behavior.

The relationship between philosophy and research on artificial intelligence (AI) has been difficult since its beginning, with mutual misunderstanding and sometimes even hostility. By contrast, we show how an approach informed by both philosophy and AI can be productive. After reviewing some popular frameworks for computation and learning, we apply the AI methodology of “build it and see” to tackle the philosophical and psychological problem of characterizing perception as distinct from sensation. Our model comprises a network of very simple, but (...) interacting agents which have binary experiences of the “yes/no”-type and communicate their experiences with each other. When does such a network refer to a single agent instead of a distributed network of entities? We apply machine learning techniques to address the following related questions: i) how can the model explain stability of compound entities, and ii) how could the model implement a single task such as perceptual inference? We thereby find consistency with previous work on “interface” strategies from perception research. While this reflects some necessary conditions for the ascription of agency, we suggest that it is not sufficient. Here, AI research, if it is intended to contribute to conceptual understanding, would benefit from issues previously raised by philosophy. We thus conclude the article with a discussion of action-selection, the role of embodiment, and consciousness to make this more explicit. We conjecture that a combination of AI research and philosophy allows general principles of mind and being to emerge from a “quasi-empirical” investigation. (shrink)

The Greeks had a ready answer for what happens when the mind suddenly finds the answer to a question for which it had been searching: insight was regarded as a gift of the Muses, its origins were divine. It served to highlight the Greeks'' belief that there are some things which are not meant to be scientifically explained. The essence of insight is that it comes from some supernatural source: unpredicted and unfettered. In other words, the origins of insight are (...) unconscious, and hence, unexplainable. Wittgenstein felt that, as long as there continues to be a noun expression like to have a moment of insight which functions in the same way as the expression to have a hunger pang, thereby inducing us to treat moment of insight as the name of an experience, then people will keep stumbling over the same puzzling difficulties and find themselves staring at something which no explanation seems capable of clearing up. To the founders of AI, this argument reeked of obscurantism. The moment of insight, they felt, is indeed a mystery, but it is one that begs to be explained in causal terms. Indeed, the problem of insight served as one of the leading problems in the evolution of AI. Hence anyone interested in the foundations of AI is compelled to examine the manner in which the early pioneers of the field sought to explain the eureka experience. In this paper I will look at some of the key conceptual developments which paved the way for Newell and Simon''s theory of GPS: the fundamental changes in the notion of the unconscious — the emergence of the cognitive unconscious — which took place in the nineteenth- and early twentieth-century. In so doing, I hope to clarify what Wittgenstein may have had in mind in his strictures against mechanist attempts to analyse the nature of insight. (shrink)

Cognitive science is, of course, not really a new discipline, but a recognition of a fundamental set of common concerns shared by the disciplines of psychology, computer science, linguistics, economics, epistemology, and the social sciences generally. All of these disciplines are concerned with information processing systems, and all of them are concerned with systems that are adaptive—that are what they are from being ground between the nether millstone of their physiology or hardware, as the case may be, and the upper (...) millstone of a complex environment in which they exist. Systems that are adaptive may equally well be described as “artificial,” for as environments change, they can be expected to change too, as though they were deliberately designed to fit those environments (as indeed they sometimes are).The task of empirical science is to discover and verify invariants in the phenomena under study. The artificiality of information processing systems creates a subtle problem in defining empirical invariants in such systems. For observed regularities are very likely invariant only within a limited range of variation in their environments, and any accurate statement of the laws of such systems must contain reference to their relativity to environmental features. It is a common experience in experimental psychology, for example, to discover that we are studying sociology—the effects of the past histories of our subjects—when we think we are studying physiology—the effects of properties of the human nervous system. Similarly, business cycle economists are only now becoming aware of the extent to which the parameters of the system they are studying are dependent on the experiences of a population with economic events over the previous generation.In artificial sciences, the descriptive and the normative are never far apart. Thus, in economics, the “principle of rationality” is sometimes asserted as a descriptive invariant, sometimes as advice to decision makers. Similarly, in psychology, the processes of adaptation (learning) have always been a central topic, at one time a topic that dominated the whole field of research. Linguistics, too, has suffered its confusions between descriptive and normative attitudes towards its subject. But we must avoid the error, in studying information processing systems, of thinking that the adaptive processes themselves must be invariant; and we must be prepared to face the complexities of regression in the possibility that they themselves may be subject to improvement and adaptation.It might have been necessary a decade ago to argue for the commonality of the information processes that are employed by such disparate systems as computers and human nervous systems. The evidence for that commonality is now overwhelming, and the remaining questions about the boundaries of cognitive science have more to do with whether there also exist nontrivial commonalities with information processing in genetic systems than with whether men and machines both think. (shrink)

Mercier & Sperber (M&S) claim that the phenomenon of belief bias provides fundamental support for their argumentative theory and its basis in intuitive judgement. We propose that chronometric evidence necessitates a more nuanced account of belief bias that is not readily captured by argumentative theory.

Computational modeling has long been one of the traditional pillars of cognitive science. Unfortunately, the computer models of cognition being developed today have not kept up with the enormous changes that have taken place in computer technology and, especially, in human-computer interfaces. For all intents and purposes, modeling is still done today as it was 25, or even 35, years ago. Everyone still programs in his or her own favorite programming language, source code is rarely made available, accessibility of models (...) to non-programming researchers is essentially non-existent, and even for other modelers, the profusion of source code in a multitude of programming languages, written without programming guidelines, makes it almost impossible to access, check, explore, re-use, or continue to develop. It is high time to change this situation, especially since the tools are now readily available to do so. We propose that the modeling community adopt three simple guidelines that would ensure that computational models would be accessible to the broad range of researchers in cognitive science. We further emphasize the pivotal role that journal editors must play in making computational models accessible to readers of their journals. (shrink)

The popular debates about the future organization of work through artificial intelligence technologies focus on the replacement of human beings by novel technologies. In this essay, we oppose this statement by closely following what has been developed as AI technologies and analyzing how they work, specifically focusing on research that may impact work organizations. We develop this argument by showing that the recent research and developments in AI technologies focus on developing accurate and precise performance models, which in turn shapes (...) organizational patterns of work. We propose that the increased interest in the relationship between human cognition and performance will shortly bring human cognition to the focus on AI systems in workplaces. More specifically, we claim that the cognitive load measurement will shape human performance in manufacturing systems shortly. (shrink)