Deep learning is a kind of machine learning which happens in a certain type of artificial neural networks called deep networks. Artificial deep networks, which exhibit many similarities with biological ones, have consistently shown human-like performance in many intelligent tasks. This poses the question whether this performance is caused by such similarities. After reviewing the structure and learning processes of artificial and biological neural networks, we outline two important reasons for the success of deep learning, namely the extraction of successively (...) higher level features and the multiple layer structure, which are closely related to each other. Then some indications about the framing of this heated debate are given. After that, an assessment of the value of artificial deep networks as models of the human brain is given from the similarity perspective of model representation. Finally, a new version of computational functionalism is proposed which addresses the specificity of deep neural computation better than classic, program based computational functionalism. (shrink)

An introduction to functionalism in the philosophy of psychology/mind, and review of the current state of debate pro and con. Forthcoming in the Routledge Companion to the Philosophy of Psychology (John Symons and Paco Calvo, eds.).

Defending or attacking either functionalism or computationalism requires clarity on what they amount to and what evidence counts for or against them. My goal here is not to evaluate their plausibility. My goal is to formulate them and their relationship clearly enough that we can determine which type of evidence is relevant to them. I aim to dispel some sources of confusion that surround functionalism and computationalism, recruit recent philosophical work on mechanisms and computation to shed light on (...) them, and clarify how functionalism and computationalism may or may not legitimately come together. (shrink)

I argue that Stich's Syntactic Theory of Mind (STM) and a naturalistic narrow content functionalism run on a Language of Though story have the same exact structure. I elaborate on the argument that narrow content functionalism is either irremediably holistic in a rather destructive sense, or else doesn't have the resources for individuating contents interpersonally. So I show that, contrary to his own advertisement, Stich's STM has exactly the same problems (like holism, vagueness, observer-relativity, etc.) that he claims (...) plague content-based psychologies. So STM can't be any better than the Representational Theory of Mind (RTM) in its prospects for forming the foundations of a scientifically respectable psychology, whether or not RTM has the problems that Stich claims it does. (shrink)

A distinction between the use of computational models in cognitive science and a philosophically inspired reductivist thesis is developed. PF is found questionable for phenomenal states, and, by analogy, dubious for the nonphenomenal introspectible mental states of common sense. PF is also shown to be threatened for the sub-cognitive theoretical states of cognitive science by the work of the so-called New Connectionists. CMM is shown to be less vulnerable to these criticisms.

The viewpoint that consciousness, including feeling, could be fully expressed by a computational device is known as strong artificial intelligence or strong AI. Here I offer a defense of strong AI based on machine-state functionalism at the quantum level, or quantum-state functionalism. I consider arguments against strong AI, then summarize some counterarguments I find compelling, including Torkel Franzén’s work which challenges Roger Penrose’s claim, based on Gödel incompleteness, that mathematicians have nonalgorithmic levels of “certainty.” Some consequences of (...) strong AI are then considered. A resolution is offered of some problems including John Searle’s Chinese Room problem and the problem of consciousness propagation under isomorphism. (shrink)

This paper describes the efforts of those who work with informational machines and with informational analyses to provide a basis for understanding consciousness and for speculating on what it would take to make a conscious machine. Some of the origins of these considerations are covered and the contributions of several researchers are reviewed. A distinction is drawn between functional and phenomenological approaches showing how the former lead to algorithmic methods based on conventional programming, while the latter lead to neural network (...) analyses. Attention is drawn to the many open questions that this approach generates and some speculation on what future work might bring is included. (shrink)

Some philosophers have conflated functionalism and computationalism. I reconstruct how this came about and uncover two assumptions that made the conflation possible. They are the assumptions that (i) psychological functional analyses are computational descriptions and (ii) everything may be described as performing computations. I argue that, if we want to improve our understanding of both the metaphysics of mental states and the functional relations between them, we should reject these assumptions.

The Hypothesis of Extended Cognition (HEC) holds that that not all human cognition is realized inside the head. The related but distinct Hypothesis of Extended Mentality (HEM) holds that not all human mental items are realized inside the head. Clark & Chalmers distinguish between these hypotheses in their original treatment of cognitive extension, yet these two claims are often confused. I distinguish between functionalist theories on which functional roles are individuated according to computational criteria, and those on which functional (...) roles are individuated according to rational criteria. I then present an argument for a modest version of HEC from computational functionalism, based on Clark & Chalmers’ original argument. In doing so I articulate a successor to their parity principle, and review studies by Wayne Gray et al. that provide plausible evidence for actual cognitive extension. I then respond to a new criticism of HEC by Mark Sprevak using the modest account I have developed, arguing that Sprevak conflates HEC and HEM. (shrink)

Kripke’s argument against functionalism extended to physical computers poses a deep philosophical problem for understanding the standard view of what computers are. The problem puts into jeopardy the definition in the standard view that computers are physical machines for performing physical computations. Indeed, it is entirely possible that, unless this philosophical problem is resolved, we will never have a good understanding of computers and may never know just what they are.

It is here argued that functionalist constraints on psychology do not preclude the applicability of classic forms of reduction and, therefore, do not support claims to a principled, or de jure, autonomy of psychology. In Part I, after isolating one minimal restriction any functionalist theory must impose on its categories, it is shown that any functionalism imposing an additional constraint of de facto autonomy must also be committed to a pure functionalist--that is, a computationalist--model for psychology. Using an extended (...) parallel to the reduction of Mendelian to molecular genetics, it is shown in Parts II and III that, contrary to the claims of Hilary Putnam and Jerry Fodor, there is no inconsistency between computational models and classical reductionism: neither plurality of physical realization nor plurality of function are inconsistent with reductionism as defended by Ernest Nagel. Employing the results of Part I, the conclusions of Parts II and III are generalized in Part IV to cover any version of functionalism whatsoever; thus, functionalism and reductionism are shown to be consistent. It is urged in conclusion that although a de facto form of autonomy is defensible, there are sound methodological grounds for unconditionally rejecting any principled version of the autonomy of psychology. (shrink)

The computational view of mind rests on certain intuitions regarding the fundamental similarity between computation and cognition. We examine some of these intuitions and suggest that they derive from the fact that computers and human organisms are both physical systems whose behavior is correctly described as being governed by rules acting on symbolic representations. Some of the implications of this view are discussed. It is suggested that a fundamental hypothesis of this approach is that there is a natural domain (...) of human functioning that can be addressed exclusively in terms of a formal symbolic or algorithmic vocabulary or level of analysis. Much of the paper elaborates various conditions that need to be met if a literal view of mental activity as computation is to serve as the basis for explanatory theories. The coherence of such a view depends on there being a principled distinction between functions whose explanation requires that we posit internal representations and those that we can appropriately describe as merely instantiating causal physical or biological laws. In this paper the distinction is empirically grounded in a methodological criterion called the " cognitive impenetrability condition." Functions are said to be cognitively impenetrable if they cannot be influenced by such purely cognitive factors as goals, beliefs, inferences, tacit knowledge, and so on. Such a criterion makes it possible to empirically separate the fixed capacities of mind from the particular representations and algorithms used on specific occasions. In order for computational theories to avoid being ad hoc, they must deal effectively with the "degrees of freedom" problem by constraining the extent to which they can be arbitrarily adjusted post hoc to fit some particular set of observations. This in turn requires that the fixed architectural function and the algorithms be independently validated. It is argued that the architectural assumptions implicit in many contemporary models run afoul of the cognitive impenetrability condition, since the required fixed functions are demonstrably sensitive to tacit knowledge and goals. The paper concludes with some tactical suggestions for the development of computational cognitive theories. (shrink)

Some philosophers have conflated functionalism and computationalism. I reconstruct how this came about and uncover two assumptions that made the conflation possible. They are the assumptions that (i) psychological functional analyses are computational descriptions and (ii) everything may be described as performing computations. I argue that, if we want to improve our understanding of both the metaphysics of mental states and the functional relations between them, we should reject these assumptions. # 2004 Elsevier Ltd. All rights reserved.

In the article I discuss functionalist interpretations of Husserlian phenomenology. The first one was coined in the discussion between Hubert Dreyfus and Ronald McIntyre. They argue that Husserl’s phenomenology shares similarities with computational functionalism, and the key similarity is between the concept of noema and the concept of mental representation. I show the weaknesses of that reading and argue that there is another available functionalist reading of Husserlian phenomenology. I propose to shift perspective and approach the relation between (...) phenomenology and functionalism from a methodological perspective, specifically taking into account the functionalist explanatory strategy called functional analysis. I discuss the notion of function in Husserl’s works and Husserl’s idea of functional phenomenology. The key argument I develop is that in functional phenomenology we can find an explanatory strategy which is analogous to the strategy of functional decomposition used in functional analysis. I conclude that the proposed functionalist reading of phenomenology opens a new approach to the integration of phenomenology with cognitive sciences. (shrink)

“Triviality arguments” against functionalism in the philosophy of mind hold that the claim that some complex physical system exhibits a given functional organization is either trivial or has much less content than is usually supposed. I survey several earlier arguments of this kind, and present a new one that overcomes some limitations in the earlier arguments. Resisting triviality arguments is possible, but requires functionalists to revise popular views about the “autonomy” of functional description.

In his recent article "The Computational Model of the Mind and Philosophical Functionalism," Richard Double argues that there are some fairly forceful a priori arguments showing that Philosophical Functionalism cannot provide adequate explanations for phenomenal states, the nonphenomenal conscious states of common sense, and the theoretical states of cognitive psychology and linquistics. In this paper it is argued that none of Double's arguments are successful.

More than a decade ago, philosopher John Searle started a long-running controversy with his paper “Minds, Brains, and Programs” (Searle, 1980a), an attack on the ambitious claims of artificial intelligence (AI). With his now famous _Chinese Room_ argument, Searle claimed to show that despite the best efforts of AI researchers, a computer could never recreate such vital properties of human mentality as intentionality, subjectivity, and understanding. The AI research program is based on the underlying assumption that all important aspects of (...) human cognition may in principle be captured in a computational model. This assumption stems from the belief that beyond a certain level, implementational details are irrelevant to cognition. According to this belief, neurons, and biological wetware in general, have no preferred status as the substrate for a mind. As it happens, the best examples of minds we have at present have arisen from a carbon-based substrate, but this is due to constraints of evolution and possibly historical accidents, rather than to an absolute metaphysical necessity. As a result of this belief, many cognitive scientists have chosen to focus not on the biological substrate of the mind, but instead on the abstract causal structure_ _that the mind embodies (at an appropriate level of abstraction). The view that it is abstract causal structure that is essential to mentality has been an implicit assumption of the AI research program since Turing (1950), but was first articulated explicitly, in various forms, by Putnam (1960), Armstrong (1970) and Lewis (1970), and has become known as _functionalism_. From here, it is a very short step to _computationalism_, the view that computational structure is what is important in capturing the essence of mentality. This step follows from a belief that any abstract causal structure can be captured computationally: a belief made plausible by the Church–Turing Thesis, which articulates the power. (shrink)

In this article, I argue that if tacit knowledge of grammar is analyzable in functional‐computational terms, then it cannot ground linguistic meaning, structure, or sound. If to know or cognize a grammar is to be in a certain computational state playing a certain functional role, there can be no unique grammar cognized. Satisfying the functional conditions for cognizing a grammar G entails satisfying those for cognizing many grammars disagreeing with G about expressions' semantic, phonetic, and syntactic values. This (...) threatens the Chomskyan view that expressions have such values for speakers because they cognize grammars assigning them those values. For if this is true, semantics, syntax, and phonology must be indeterminate, thanks to the indeterminacy of grammar‐cognizing (qua functional‐computational state). So, the fact that a speaker cognizes a grammar cannot explain the determinate character of their language. (shrink)

An effective method is a computational method that might, in principle, be executed by a human. In this paper, I argue that there are methods for computing that are not effective methods. The examples I consider are taken primarily from quantum computing, but these are only meant to be illustrative of a much wider class. Quantum inference and quantum parallelism involve steps that might be implemented in multiple physical systems, but cannot be implemented, or at least not at will, (...) by an idealised human. Recognising that not all computational methods are effective methods is important for at least two reasons. First, it is needed to correctly state the results of Turing and other founders of computation theory. Turing is sometimes said to have offered a replacement for the informal notion of an effective method with the formal notion of a Turing machine. I argue that such a view only holds under limited circumstances. Second, not distinguishing between computational methods and effective methods can lead to mistakes when quantifying over the class of all possible computational methods. Such quantification is common in philosophy of mind in the context of thought experiments that explore the limits of computational functionalism. I argue that these ‘homuncular’ thought experiments should not be treated as valid. (shrink)

With mind-brain identity theories no longer dominant in philosophy of mind in the late 1950s, scientific materialists turned to functionalism, the view that the identity of any mental state depends on its function in the cognitive system of which it is a part. The philosopher Hilary Putnam was one of the primary architects of functionalism and was the first to propose computational functionalism, which views the human mind as a computer or an information processor. But, in (...) the early 1970s, Putnam began to have doubts about functionalism, and in his masterwork _Representation and Reality_, he advanced four powerful arguments against his own doctrine of computational functionalism. In Gödel, Putnam, and Functionalism, Jeff Buechner systematically examines Putnam's arguments against functionalism and contends that they are unsuccessful. Putnam's first argument uses Gödel's incompleteness theorem to refute the view that there is a computational description of human reasoning and rationality; his second, the "triviality argument," demonstrates that any computational description can be attributed to any physical system; his third, the multirealization argument, shows that there are infinitely many computational realizations of an arbitrary intentional state; his fourth argument buttresses this assertion by showing that there cannot be local computational reductions because there is no computable partitioning of the infinity of computational realizations of an arbitrary intentional state into a single package or small set of packages. Buechner analyzes these arguments and the important inferential connections among them -- for example, the use of both the Gödel and triviality arguments in the argument against local computational reductions -- and argues that none of Putnam's four arguments succeeds in refuting functionalism. _ Gödel, Putnam, and Functionalism_ will inspire renewed discussion of Putnam's influential book and will confirm Representation and Reality as a major work by a major philosopher. Jeff Buechner is Director of the Bioethics Institute and Lecturer in Philosophy at Rutgers University-Newark. (shrink)

This paper proposes a model of the Artificial Autonomous Moral Agent (AAMA), discusses a standard of moral cognition for AAMA, and compares it with other models of artificial normative agency. It is argued here that artificial morality is possible within the framework of a “moral dispositional functionalism.” This AAMA is able to “read” the behavior of human actors, available as collected data, and to categorize their moral behavior based on moral patterns herein. The present model is based on several (...) analogies among artificial cognition, human cognition, and moral action. It is premised on the idea that moral agents should not be based on rule-following procedures, but on learning patterns from data. This idea is rarely implemented in AAMA models, albeit it has been suggested in the machine ethics literature (W. Wallach, C. Allen, J. Gips and especially M. Guarini). As an agent-based model, this AAMA constitutes an alternative to the mainstream action-centric models proposed by K. Abney, M. Anderson and S. Anderson, R. Arkin, T. Powers, W. Wallach, i.a. Moral learning and moral development of dispositional traits play here the fundamental role in cognition. By using a combination of neural networks and evolutionary computation, called “soft computing” (H. Adeli, N. Siddique, S. Mitra, L. Zadeh), the present model reaches a certain level of autonomy and complexity, which illustrates well “moral particularism” and a form of virtue ethics for machines, grounded in active learning. An example based on the “lifeboat metaphor” (G. Hardin) and the extension of this model to the NEAT architecture (K. Stanley, R. Miikkulainen, i.a.) are briefly assessed. (shrink)

The most cursory examination of the history of artificial intelligence highlights numerous egregious claims of its researchers, especially in relation to a populist form of ‘strong’ computationalism which holds that any suitably programmed computer instantiates genuine conscious mental states purely in virtue of carrying out a specific series of computations. The argument presented herein is a simple development of that originally presented in Putnam’s (Representation & Reality, Bradford Books, Cambridge in 1988) monograph, “Representation & Reality”, which if correct, has important (...) implications for turing machine functionalism and the prospect of ‘conscious’ machines. In the paper, instead of seeking to develop Putnam’s claim that, “everything implements every finite state automata”, I will try to establish the weaker result that, “everything implements the specific machine Q on a particular input set (x)”. Then, equating Q (x) to any putative AI program, I will show that conceding the ‘strong AI’ thesis for Q (crediting it with mental states and consciousness) opens the door to a vicious form of panpsychism whereby all open systems, (e.g. grass, rocks etc.), must instantiate conscious experience and hence that disembodied minds lurk everywhere. (shrink)

The main claim of this paper is that notions of implementation based on an isomorphic correspondence between physical and computational states are not tenable. Rather, ``implementation'' has to be based on the notion of ``bisimulation'' in order to be able to block unwanted implementation results and incorporate intuitions from computational practice. A formal definition of implementation is suggested, which satisfies theoretical and practical requirements and may also be used to make the functionalist notion of ``physical realization'' precise. The (...) upshot of this new definition of implementation is that implementation cannot distinguish isomorphic bisimilar from non-isomporphic bisimilar systems anymore, thus driving a wedge between the notions of causal and computational complexity. While computationalism does not seem to be affected by this result, the consequences for functionalism are not clear and need further investigations. (shrink)

Three recent, influential critiques (Stich 1978; Fodor 1981c; Block 1980) have argued that various tasks on the agenda for computational psychology put conflicting pressures on its theoretical constructs. Unless something is done, the inevitable result will be confusion or outright incoherence. Stich, Fodor, and Block present different versions of this worry and each proposes a different remedy. Stich wants the central notion of belief to be jettisoned if it cannot be shown to be sound. Fodor tries to reduce confusion (...) in computational psychology by dismissing some putative tasks as impossible. Block argues that the widespread faith in functionalism is just not warranted. I argue that all these critiques are misguided because they depend on holding cognitive psychology to taxonomic standards that other sciences routinely rise above. (shrink)

Saul Kripke has proposed an argument to show that there is a serious problem with many computational accounts of physical systems and with functionalist theories in the philosophy of mind. The problem with computational accounts is roughly that they provide no noncircular way to maintain that any particular function with an infinite domain is realized by any physical system, and functionalism has the similar problem because of the character of the functional systems that are supposed to be (...) realized by organisms. This paper shows that the standard account of what it is for a physical system to compute a function can avoid Kripke's criticisms without being reduced to circularity; a very minor and natural elaboration of the standard account suffices to save both functionalist theories and computational accounts generally. (shrink)

The main claim of this paper is that notions of implementation based on an isomorphic correspondence between physical and computational states are not tenable. Rather, ``implementation'' has to be based on the notion of ``bisimulation'' in order to be able to block unwanted implementation results and incorporate intuitions from computational practice. A formal definition of implementation is suggested, which satisfies theoretical and practical requirements and may also be used to make the functionalist notion of ``physical realization'' precise. The (...) upshot of this new definition of implementation is that implementation cannot distinguish isomorphic bisimilar from non-isomporphic bisimilar systems anymore, thus driving a wedge between the notions of causal and computational complexity. While computationalism does not seem to be affected by this result, the consequences for functionalism are not clear and need further investigations. (shrink)

Most philosophers appear to have ignored the distinction between the broad concept of Virtual Machine Functionalism (VMF) described in Sloman&Chrisley (2003) and the better known version of functionalism referred to there as Atomic State Functionalism (ASF), which is often given as an explanation of what Functionalism is, e.g. in Block (1995). -/- One of the main differences is that ASF encourages talk of supervenience of states and properties, whereas VMF requires supervenience of machines that are arbitrarily (...) complex networks of causally interacting (virtual, but real) processes, possibly operating on different time-scales, examples of which include many different procesess usually running concurrently on a modern computer performing various tasks concerned with handling interfaces to physical devices, managing the file system, dealing with security, providing tools, entertainments, and games, and possibly processing research data. Another example of VMF would be the kind of functionalism involved in a large collection of possibly changing socio-economic structures and processes interacting in a complex community, and yet another is illustrated by the kind of virtual machinery involved in the many levels of visual processing of information about spatial structures, processes, and relationships (including percepts of moving shadows, reflections, highlights, optical-flow patterns and changing affordances) as you walk through a crowded car-park on a sunny day: generating a whole zoo of interacting qualia. (Forget solitary red patches, or experiences thereof.) -/- Perhaps VMF should be re-labelled "Virtual MachinERY Functionalism" because the word 'machinery' more readily suggests something complex with interacting parts. VMF is concerned with virtual machines that are made up of interacting concurrently active (but not necessarily synchronised) chunks of virtual machinery which not only interact with one another and with their physical substrates (which may be partly shared, and also frequently modified by garbage collection, metabolism, or whatever) but can also concurrently interact with and refer to various things in the immediate and remote environment (via sensory/motor channels, and possible future technologies also). I.e. virtual machinery can include mechanisms that create and manipulate semantic content, not only syntactic structures or bit patterns as digital virtual machines do. -/- Please note: Click on the title above or the link below to read the paper. I prefer to keep all my papers freely accessible on my web site so that I can correct mistakes and add improvements. -/- http://www.cs.bham.ac.uk/research/projects/cogaff/misc/vm-functionalism.html -/- This is now part of the Meta-Morphogenesis project: http://www.cs.bham.ac.uk/research/projects/cogaff/misc/meta-morphogenesis.html. (shrink)

The paper criticizes standard functionalist arguments for multiple realization. It focuses on arguments in which psychological states are conceived as computational, which is precisely where the multiple realization doctrine has seemed the strongest. It is argued that a type-type identity thesis between computational states and physical states is no less plausible than a multiple realization thesis. The paper also presents, more tentatively, positive arguments for a picture of local reduction.

The most cursory examination of the history of artificial intelligence highlights numerous egregious claims of its researchers, especially in relation to a populist form of ‘strong’ computationalism which holds that any suitably programmed computer instantiates genuine conscious mental states purely in virtue of carrying out a specific series of computations. The argument presented herein is a simple development of that originally presented in Putnam’s (Representation & Reality, Bradford Books, Cambridge in 1988 ) monograph, “Representation & Reality”, which if correct, has (...) important implications for turing machine functionalism and the prospect of ‘conscious’ machines. In the paper, instead of seeking to develop Putnam’s claim that, “everything implements every finite state automata”, I will try to establish the weaker result that, “everything implements the specific machine Q on a particular input set ( x )”. Then, equating Q ( x ) to any putative AI program, I will show that conceding the ‘strong AI’ thesis for Q (crediting it with mental states and consciousness) opens the door to a vicious form of panpsychism whereby all open systems, (e.g. grass, rocks etc.), must instantiate conscious experience and hence that disembodied minds lurk everywhere. (shrink)

The most cursory examination of the history of artificial intelligence highlights numerous egregious claims of its researchers, especially in relation to a populist form of ‘strong’ computationalism which holds that any suitably programmed computer instantiates genuine conscious mental states purely in virtue of carrying out a specific series of computations. The argument presented herein is a simple development of that originally presented in Putnam’s monograph, “Representation & Reality”, which if correct, has important implications for turing machine functionalism and the (...) prospect of ‘conscious’ machines. In the paper, instead of seeking to develop Putnam’s claim that, “everything implements every finite state automata”, I will try to establish the weaker result that, “everything implements the specific machine Q on a particular input set ”. Then, equating Q to any putative AI program, I will show that conceding the ‘strong AI’ thesis for Q opens the door to a vicious form of panpsychism whereby all open systems,, must instantiate conscious experience and hence that disembodied minds lurk everywhere. (shrink)

Though there is yet no consensus on the right way to understand ‘physicalism’, most philosophers agree that, regardless of whatever else is required, physicalism cannot be true if there exists fundamental mentality. I will follow Jessica Wilson (Philosophical Studies 131:61–99, 2006) in calling this the 'No Fundamental Mentality' (NFM) constraint on physicalism. Unfortunately for those who wish to constrain physicalism in this way, NFM admits of a counterexample: an artificially intelligent quantum computer which employs quantum properties as part of its (...) cognitive operations. If one of these quantum properties serves a proper functional role in the artificial intelligence, then that property counts as a mental under the physicalism-friendly theory of mentality called “realizer functionalism”, which says that a lower-order property is mental if it satisfies an appropriate higher-order functional description. Further, if this quantum property is both fundamental and mental, then NFM must rule that it is not physical. Yet the existence of such an artificially intelligent quantum computer, which possesses mental properties solely in virtue of the functional roles those properties play, is surely consistent with the truth of physicalism. This ought to motivate NFM proponents to reformulate their view. (shrink)

The mainstream view in cognitive science is that computation lies at the basis of and explains cognition. Our analysis reveals that there is no compelling evidence or argument for thinking that brains compute. It makes the case for inverting the explanatory order proposed by the computational basis of cognition thesis. We give reasons to reverse the polarity of standard thinking on this topic, and ask how it is possible that computation, natural and artificial, might be based on cognition and (...) not the other way around. (shrink)

Multiple realizability (MR) is traditionally conceived of as the feature of computational systems, and has been used to argue for irreducibility of higher-level theories. I will show that there are several ways a computational system may be seen to display MR. These ways correspond to (at least) five ways one can conceive of the function of the physical computational system. However, they do not match common intuitions about MR. I show that MR is deeply interest-related, and for (...) this reason, difficult to pin down exactly. I claim that MR is of little importance for defending computationalism, and argue that it should rather appeal to organizational invariance or substrate neutrality of computation, which are much more intuitive but cannot support strong antireductionist arguments. (shrink)

The problem of multiple-computations discovered by Hilary Putnam presents a deep difficulty for functionalism (of all sorts, computational and causal). We describe in out- line why Putnam’s result, and likewise the more restricted result we call the Multiple- Computations Theorem, are in fact theorems of statistical mechanics. We show why the mere interaction of a computing system with its environment cannot single out a computation as the preferred one amongst the many computations implemented by the system. We explain (...) why nonreductive approaches to solving the multiple- computations problem, and in particular why computational externalism, are dualistic in the sense that they imply that nonphysical facts in the environment of a computing system single out the computation. We discuss certain attempts to dissolve Putnam’s unrestricted result by appealing to systems with certain kinds of input and output states as a special case of computational externalism, and show why this approach is not workable without collapsing to behaviorism. We conclude with some remarks about the nonphysical nature of mainstream approaches to both statistical mechanics and the quantum theory of measurement with respect to the singling out of partitions and observables. (shrink)

Saul Kripke has proposed an argument to show that there is a serious problem with many computational accounts of physical systems and with functionalist theories in the philosophy of mind. The problem with computational accounts is roughly that they provide no noncircular way to maintain that any particular function with an infinite domain is realized by any physical system, and functionalism has the similar problem because of the character of the functional systems that are supposed to be (...) realized by organisms. This paper shows that the standard account of what it is for a physical system to compute a function can avoid Kripke's criticisms without being reduced to circularity; a very minor and natural elaboration of the standard account suffices to save both functionalist theories and computational accounts generally. (shrink)

Substrate independence and mind-body functionalism claim that thinking does not depend on any particular kind of physical implementation. But real-world information processing depends on energy, and energy depends on material substrates. Biological evidence for these claims comes from ecology and neuroscience, while computational evidence comes from neuromorphic computing and deep learning. Attention to energy requirements undermines the use of substrate independence to support claims about the feasibility of artificial intelligence, the moral standing of robots, the possibility that we (...) may be living in a computer simulation, the plausibility of transferring minds into computers, and the autonomy of psychology from neuroscience. (shrink)

The article presents preliminary results of the conceptual analysis of the mechanistic profile of the computer metaphor. Mechanic reductionism is a special direction of computer metaphor rooted in various historical forms of word usage. Here we trace the stages of formation of the principles of transferring the properties of a mechanical computer to the properties of the human body and mind. We are also trying to identify the basic principles of semantic transfer, which have survived to this day in the (...) discourse of modern computationalism. The reasons are analyzed due to which the metaphor «Human (body) is machine», traditional for the Modern Age, was transformed into a more complex version of «Mind is machine». What happened to the concepts of «mind» and «machine»? How have ideas about the properties of computational procedures changed? What keeps the counterintuitive computer metaphor viable today? The answers to this series of questions and theoretical ways of solving these problems are contained in this paper. (shrink)

An intuitive view is that creativity involves bringing together what is already known and familiar in a way that produces something new. In cognitive science, this intuition is typically formalized in terms of computational processes that combine or associate internally represented information. From this computationalist perspective, it is hard to imagine how non-representational approaches in embodied cognitive science could shed light on creativity, especially when it comes to abstract conceptual reasoning of the kind scientists so often engage in. The (...) present article offers an entry point to addressing this challenge. The scientific project of embodied cognitive science is a continuation of work in the functionalist tradition in psychology developed over a century ago by William James and John Dewey, among others. The focus here is on how functionalist views on the nature of mind, thought, and experience offer an alternative starting point for cognitive science in general, and for the cognitive science of scientific creativity in particular. The result may seem paradoxical. On the one hand, the article claims that the functionalist conceptual framework motivates rejecting mainstream cognitive views of creativity as the combination or association of ideas. On the other hand, however, the strategy adopted here—namely, revisiting ideas from functionalist psychology to inform current scientific theorizing—can itself be described as a process of arriving at new, creative ideas from combinations of old ones. As is shown here, a proper understanding of cognition in light of the functionalist tradition resolves the seeming tension between these two claims. (shrink)

In this paper, I defend the metalinguistic solution to the problem of mathematical omniscience for the possible-worlds account of propositions by combining it with a computational model of knowledge and belief. The metalinguistic solution states that the objects of belief and ignorance in mathematics are relations between mathematical sentences and what they express. The most pressing problem for the metalinguistic strategy is that it still ascribes too much mathematical knowledge under the standard possible-worlds model of knowledge and belief on (...) which these are closed under entailment. I first argue that Stalnaker's fragmentation strategy is insufficient to solve this problem. I then develop an alternative, computational strategy: I propose a model of mathematical knowledge and belief adapted from the algorithmic model of Halpern et al. which, when combined with the metalinguistic strategy, entails that mathematical knowledge and belief require computational abilities to access metalinguistic information, and thus aren't closed under entailment. As I explain, the computational model generalizes beyond mathematics to a version of the functionalist theory of knowledge and belief that motivates the possible-worlds account in the first place. I conclude that the metalinguistic and computational strategies yield an attractive functionalist, possible-worlds account of mathematical content, knowledge, and inquiry. (shrink)

The “Slicing Problem” is a thought experiment that raises questions for substrate-neutral computational theories of consciousness, including those that specify a certain causal structure for the computation like Integrated Information Theory. The thought experiment uses water-based logic gates to construct a computer in a way that permits cleanly slicing each gate and connection in half, creating two identical computers each instantiating the same computation. The slicing can be reversed and repeated via an on/off switch, without changing the amount of (...) matter in the system. The question is what do different computational theories of consciousness believe is happening to the number and nature of individual conscious units as this switch is toggled. Under a token interpretation, there are now two discrete conscious entities; under a type interpretation, there may remain only one. Both interpretations lead to different implications depending on the adopted theoretical stance. Any route taken either allows mechanisms for “consciousness-multiplying exploits” or requires ambiguous boundaries between conscious entities, raising philosophical and ethical questions for theorists to consider. We discuss resolutions under different theories of consciousness for those unwilling to accept consciousness-multiplying exploits. In particular, we specify three features that may help promising physicalist theories to navigate such thought experiments. (shrink)

Architecture often relies on mathematical models, if only to anticipate the physical behavior of structures. Accordingly, mathematical modeling serves to find an optimal form given certain constraints, constraints themselves translated into a language which must be homogeneous to that of the model in order for resolution to be possible. Traditional modeling tied to design and architecture thus appears linked to a topdown vision of creation, of the modernist, voluntarist and uniformly normative type, because usually (mono)functionalist. One available instrument of calculation/representation/prescription (...) orders this conception of architecture: indeed the search for an optimal solution through mathematical calculation of a model itself mathematical, thus homogeneous and simple, is only possible when one or two functions or functional constraints are formulated, never more, and this, on a global level, therefore starting from a unique and homogenizing viewpoint. It is essential to grasp that, even applied to material and its properties or towards a particular esthetic or functional dimension, this viewpoint is thus abstractive and generalizing: disregarding singularity of context, insertion and relationship to the environment or local, social behavior. It leaves aside functional specificity and heterogeneousness – re-contextualized each time – of functions that the object or edifice are required to fulfill and optimize under diverse constraints, in their different parts. The computational turning point today’s digital design and computational architecture embody modifies these instrumental, original prescriptions, rendering them more flexible. Perhaps in light of this turnabout we should retrospectively interpret 20th century calls for modernism, functionalism and even biomorphism as being just as many ex post rationalizations in respect to techniques of strongly prescriptive modeling since our only instrument is a monolithic language, and so being, incites a top-down conception, (naturally weakly reactive to contexts), including forms whose overall appearance resembles in fine a living form. In order to liberate oneself from this and despite everything, emerge as its initiators, one has constructed from ideology and philosophy (of object, habitat, the urban) ex post, even while it is the instrument of modeling and conception that largely determines, normalizes and dictates ex ante, the possibilities and limitations of the creation of forms and living experiments4 in a given time. (shrink)

The properties of Turing’s famous ‘universal machine’ has long sustained functionalist intuitions about the nature of cognition. Here, I show that there is a logical problem with standard functionalist arguments for multiple realizability. These arguments rely essentially on Turing’s powerful insights regarding computation. In addressing a possible reply to this criticism, I further argue that functionalism is not a useful approach for understanding what it is to have a mind. In particular, I show that the difficulties involved in distinguishing (...) implementation from function make multiple realizability claims untestable and uninformative. As a result, I conclude that the role of Turing machines in philosophy of mind needs to be reconsidered. (shrink)

In this paper, I shall review the reasons that let Putnam to propose functionalism and the reasons that subsequently led him to abandon it. I would like to discuss Putnam's views belonging to early Putnam and later Putnam. First, let us focus on early Putnam. Early Putnam tries to show the possibility of robot consciousness. As a functionalist, Putnam shows that the human being is an autonomous: that is, human mind is a computing machine. Later, he changes his position (...) to demonstrate not just his intellectual confidence, but also the virtue of seeing the bigger picture. Later Putnam realized this because this sort of utopianism is an illustration of what is called ‘scientism’. It is based on speculations regarding scientific possibilities. The problem is that it is completely unclear just what possibility has been envisaged when one speaks of robotic consciousness. While arguing against Turing machine, the later Putnam said that pessimism about the success of artificial intelligence (AI) in simulating human intelligence amounts to pessimism about possibility of describing the functioning of the brain. He raised the question: what connection is there between simulating intelligence and describing the brain? Even if the computer model of the brain is correct, it does not follow that AI will succeed. Simulation is just simulation, it does not mean exact replications of human mind. (shrink)

Espousing non-reductive physicalism, how do we pick out the specific relevant physical notion(s) from physical facts, specifically in relation to phenomenal experience? Beginning with a historical review of Gilbert Ryle’s behaviorism and moving through Hilary Putnam’s machine-state functionalism and Wilfrid Sellars’ inferential framework, up to more contemporaneous computationalist- and cognitivist-functionalism (Gualtiero Piccinini), we survey accounts of mentality that countenance the emergence of mental states vide input- and output-scheme. Ultimately arriving at the conclusion that functionalism cannot account for (...) problems such as no-cognition reports, we see any robust defense of physicalism must appeal to other principles. Thus we move on to the question of emergence, not as it pertains to the hard(er) problem, but to the matter of conceptual externalization of mental properties from physical properties. Accordingly, we navigate Karen Bennett’s compatibilist solution to the exclusion argument against mental causation for the non-reductive physicalist position, according to which the physical effects of mental cases are not overdetermined, demonstrating that this backfires by offering a path for the mind-body interactionist Dualist to claim causal closure by appealing to this same schema. We conclude with a series of conceptual musings regarding rationality which take into account our challenges and findings, querying about whether phenomenal consciousness is a fundamentally private, or socially configured, notion. (shrink)

Marco Buzzoni Gödel, Searle, and the Computational Theory of the (Other) Mind According to Sergio Galvan, some of the arguments offered by Lucas and Penrose are somewhat obscure or even logically invalid, but he accepts their fundamental idea that a human mind does not work as a computational machine. His main point is that there is a qualitative difference between the principles of the logic of provability and those of the logic of evidence and belief. To evaluate this (...) suggestion, I shall first compare it with Searle’s concept of “intentionality”, and then introduce a distinction between two different senses of intentionality: a reflexive-transcendental sense and a positive (that is, historical empirical or formal logical) one. In the first of these senses, the nature of human reason is such that we have no idea how a real material system – or the corresponding formal one – could instantiate it. However, although this will turn out to be an important element of truth in Searle’s and Galvan’s conception, it does not exclude the opposite truth of Turing’s functionalism: because intentionality, intuition, vision or insight – taken in their reflexive-transcendental sense – are simply invisible to the scientific eye, a man and a machine (or a robot) that is and one that is not endowed with intentionality are de facto indistinguishable from a strictly scientific point of view. For this reason, we might eventually be entitled, or even — by the practical precautionary principle – morally obliged, to attribute minds to machines. (shrink)

The aims of this paper are threefold: To show that game-playing (GP), the discipline of Artificial Intelligence (AI) concerned with the development of automated game players, has a strong epistemological relevance within both AI and the vast area of cognitive sciences. In this context games can be seen as a way of securely reducing (segmenting) real-world complexity, thus creating the laboratory environment necessary for testing the diverse types and facets of intelligence produced by computer models. This paper aims to promote (...) the belief that games represent an excellent tool for the project of computational psychology (CP). To underline how, despite this, GP has mainly adopted an engineering-inspired methodology and in doing so has distorted the framework of cognitive functionalism. Many successes (i.e. chess, checkers) have been achieved refusing human-like reasoning. The AI has appeared to work well despite ignoring an intrinsic motivation, that of creating an explanatory link between machines and mind. To assert that substantial improvements in GP may be obtained in the future only by renewed interest in human-inspired models of reasoning and in other cognitive studies. In fact, if we increase the complexity of games (from NP-Completeness to AI-Completeness) in order to reproduce real-life problems, computer science techniques enter an impasse. Many of AI’s recent GP experiences can be shown to validate this. The lack of consistent philosophical foundations for cognitive AI and the minimal philosophical commitment of AI investigation are two of the major reasons that play an important role in explaining why CP has been overlooked. (shrink)