It is accepted that “primary awareness” may emerge from the integration of two classes of information. It is unclear, however, why this cannot take place within the comparator rather than in conjunction with feedback to the perceptual systems. The model has plausibility in relation to the continuity of conscious experience in the normal waking state and may be extended to encompass certain aspects of the “sense of self” which are frequently disrupted in psychotic patients.

: Stich, Nichols et al. assert that the process of deriving predictions by simulation must be cognitively impenetrable. Hence, they claim, the occurrence of certain errors in prediction provides empirical evidence against simulation theory. But it is false that simulation‐derived prediction must be cognitively impenetrable. Moreover the errors they cite, which are instances of irrationality, are not evidence against the version of simulation theory that takes the central domain of simulation to be intelligible transitions between states with content.

Stich, Nichols et al. assert that the process of deriving predictions by simulation must be cognitively impenetrable. Hence, they claim, the occurrence of certain errors in prediction provides empirical evidence against simulation theory. But it is false that simulation‐derived prediction must be cognitively impenetrable. Moreover the errors they cite, which are instances of irrationality, are not evidence against the (very defensible) version of simulation theory that takes the central domain of simulation to be intelligible transitions between states with content.

In recent years the cognitive science community has witnessed the rise of a new, dynamical approach to cognition. This approach entails a framework in which cognition and behavior are taken to result from complex dynamical interactions between brain, body, and environment. The advent of the dynamical approach is grounded in a dissatisfaction with the classical computational view of cognition. A particularly strong claim has been that cognitive systems do not rely on internal representations and computations. Focusing on this claim, we (...) take as a starting point a question recently raised by Cliff and Noble: " if evolution did produce a design that used internal representations, how would we recognize it?" (Knowledge-based vision and simple visual machines, Philosophical Transactions of the Royal Society: Biological Sciences, 352 , 1165-1175, 1997). We will argue that cognitive science lacks a proper operationalization of the notion of representation, and therefore is unable to fruitfully discuss whether a particular system has representations or not. A basic method to detect representations in a physical system, grounded in isomorphism, turns out to be quite unconstrained. We will look at a practical example of this problem by examining the debate on whether or not van Gelder's (What might cognition be, if not computation? Journal of Philosophy, 92 , 345-381, 1995) controversial example of the Watt Governor is representational. We will conclude that cognitive science, as of yet, has no empirically applicable means to answer Cliff and Noble's question unequivocally. This makes the recent representationalism vs. anti-representationalism debate a debate for the sake of appearance. (shrink)

In trying to characterize the relationship between psychology and neuroscience, the trend has been to argue that reductionism does not work without suggesting a suitable substitute. I offer explanatory extension as a good model for elucidating the complex relationship among disciplines which are obviously connected but which do not share pragmatic explanatory features. Explanatory extension rests on the idea that one field can "illuminate" issues that were incompletely treated in another. In this paper, I explain how this "illumination" would work (...) between psychology and neuroscience. (shrink)

Turing's celebrated 1950 paper proposes a very general methodological criterion for modelling mental function: total functional equivalence and indistinguishability. His criterion gives rise to a hierarchy of Turing Tests, from subtotal ("toy") fragments of our functions (t1), to total symbolic (pen-pal) function (T2 -- the standard Turing Test), to total external sensorimotor (robotic) function (T3), to total internal microfunction (T4), to total indistinguishability in every empirically discernible respect (T5). This is a "reverse-engineering" hierarchy of (decreasing) empirical underdetermination of the theory (...) by the data. Level t1 is clearly too underdetermined, T2 is vulnerable to a counterexample (Searle's Chinese Room Argument), and T4 and T5 are arbitrarily overdetermined. Hence T3 is the appropriate target level for cognitive science. When it is reached, however, there will still remain more unanswerable questions than when Physics reaches its Grand Unified Theory of Everything (GUTE), because of the mind/body problem and the other-minds problem, both of which are inherent in this empirical domain, even though Turing hardly mentions them. (shrink)

Within AI and the cognitively related disciplines, there exist a multiplicity of uses of belief. On the face of it, these differing uses reflect differing views about the nature of an objective phenomenon called belief. In this paper I distinguish six distinct ways in which belief is used in AI. I shall argue that not all these uses reflect a difference of opinion about an objective feature of reality. Rather, in some cases, the differing uses reflect differing concerns with special (...) AI applications. In other cases, however, genuine differences exist about the nature of what we pre-theoretically call belief. To an extent the multiplicity of opinions about, and uses of belief, echoes the discrepant motivations of AI researchers. The relevance of this discussion for cognitive scientists and philosophers arises from the fact that (a) many regard theoretical research within AI as a branch of cognitive science, and (b) even if theoretical AI is not cognitive science, trends within AI influence theories developed within cognitive science. It should be beneficial, therefore, to unravel the distinct uses and motivations surrounding belief, in order to discover which usages merely reflect differing pragmatic concerns, and which usages genuinely reflect divergent views about reality. (shrink)

The present paper looks at one of the most thorough articles on the intelligence of GPT, research conducted by engineers at Microsoft. Although there is a great deal of value in their work, I will argue that, for familiar philosophical reasons, their methodology, ‘Black-box Interpretability’ is wrongheaded. But there is a better way. There is an exciting and emerging discipline of ‘Inner Interpretability’ (also sometimes called ‘White-box Interpretability’) that aims to uncover the internal activations and weights of models in order (...) to understand what they represent and the algorithms they implement. In my view, a crucial mistake in Black-box Interpretability is the failure to appreciate that how processes are carried out matters when it comes to intelligence and understanding. I can’t pretend to have a full story that provides both necessary and sufficient conditions for being intelligent, but I do think that Inner Interpretability dovetails nicely with plausible philosophical views of what intelligence requires. So the conclusion is modest, but the important point in my view is seeing how to get the research on the right track. Towards the end of the paper, I will show how some of the philosophical concepts can be used to further refine how Inner Interpretability is approached, so the paper helps draw out a profitable, future two-way exchange between Philosophers and Computer Scientists. (shrink)

Alan M. Turing’s last published work and some posthumously published manuscripts were dedicated to the development of his theory of organic pattern formation. In “The Chemical Basis of Morphogenesis” (1952), he provided an elaborated mathematical formulation of the theory of the origins of biological form that had been first proposed by Sir D’Arcy Wendworth Thompson in On Growth and Form (1917/1942). While arguably his most mathematically detailed and his systematically most ambitious effort, Turing’s morphogenetical writings also form the most thematically (...) self-contained and least philosophically explored part of his work. We dedicate our inquiry to the reasons and the implications of Turing’s choice of biological topic and viewpoint. We will probe for possible factors in Turing’s choice that go beyond availability and acquaintance with On Growth and Form. On these grounds, we will explore how and to what extent his theory of morphogenesis actually ties in with his concept of mechanistic computation. Notably, Thompson’s pioneering work in biological ‘structuralism’ was organicist in outlook and explicitly critical of the Darwinian approaches that were popular with Turing’s cyberneticist contemporaries—and partly used by Turing himself in his proto-connectionist models of learning. Resolving this apparent dichotomy, we demonstrate how Turing’s quest for mechanistic explanations of how organisation emerges in nature leaves room for a non-mechanist view of nature. (shrink)

The aim of this paper is to grasp the relevant distinctions between various ways in which models and simulations in Artificial Intelligence (AI) relate to cognitive phenomena. In order to get a systematic picture, a taxonomy is developed that is based on the coordinates of formal versus material analogies and theory-guided versus pre-theoretic models in science. These distinctions have parallels in the computational versus mimetic aspects and in analytic versus exploratory types of computer simulation. The proposed taxonomy cuts across the (...) traditional dichotomies between symbolic and embodied AI, general intelligence and symbol and intelligence and cognitive simulation and human/non-human-like AI. -/- According to the taxonomy proposed here, one can distinguish between four distinct general approaches that figured prominently in early and classical AI, and that have partly developed into distinct research programs: first, phenomenal simulations (e.g., Turing’s “imitation game”); second, simulations that explore general-level formal isomorphisms in pursuit of a general theory of intelligence (e.g., logic-based AI); third, simulations as exploratory material models that serve to develop theoretical accounts of cognitive processes (e.g., Marr’s stages of visual processing and classical connectionism); and fourth, simulations as strictly formal models of a theory of computation that postulates cognitive processes to be isomorphic with computational processes (strong symbolic AI). -/- In continuation of pragmatic views of the modes of modeling and simulating world affairs, this taxonomy of approaches to modeling in AI helps to elucidate how available computational concepts and simulational resources contribute to the modes of representation and theory development in AI research—and what made that research program uniquely dependent on them. (shrink)

-/- The aim of this paper is to grasp the relevant distinctions between various ways in which models and simulations in Artificial Intelligence (AI) relate to cognitive phenomena. In order to get a systematic picture, a taxonomy is developed that is based on the coordinates of formal versus material analogies and theory-guided versus pre-theoretic models in science. These distinctions have parallels in the computational versus mimetic aspects and in analytic versus exploratory types of computer simulation. The proposed taxonomy cuts across (...) the traditional dichotomies between symbolic and embodied AI, general intelligence and symbol and intelligence and cognitive simulation and human/non-human-like AI. -/- According to the taxonomy proposed here, one can distinguish between four distinct general approaches that figured prominently in early and classical AI, and that have partly developed into distinct research programs: first, phenomenal simulations (e.g., Turing’s “imitation game”); second, simulations that explore general-level formal isomorphisms in pursuit of a general theory of intelligence (e.g., logic-based AI); third, simulations as exploratory material models that serve to develop theoretical accounts of cognitive processes (e.g., Marr’s stages of visual processing and classical connectionism); and fourth, simulations as strictly formal models of a theory of computation that postulates cognitive processes to be isomorphic with computational processes (strong symbolic AI). -/- In continuation of pragmatic views of the modes of modeling and simulating world affairs, this taxonomy of approaches to modeling in AI helps to elucidate how available computational concepts and simulational resources contribute to the modes of representation and theory development in AI research—and what made that research program uniquely dependent on them. (shrink)

The problem of epistemic opacity in Artificial Intelligence is often characterised as a problem of intransparent algorithms that give rise to intransparent models. However, the degrees of transparency of an AI model should not be taken as an absolute measure of the properties of its algorithms but of the model’s degree of intelligibility to human users. Its epistemically relevant elements are to be specified on various levels above and beyond the computational one. In order to elucidate this claim, I first (...) contrast computer models and their claims to algorithm-based universality with cybernetics-style analogue models and their claims to structural isomorphism between elements of model and target system. While analogue models aim at perceptually or conceptually accessible model-target relations, computer models give rise to a specific kind of underdetermination in these relations that needs to be addressed in specific ways. I then undertake a comparison between two contemporary AI approaches that, although related, distinctly align with the above modelling paradigms and represent distinct strategies towards model intelligibility: Deep Neural Networks and Predictive Processing. I conclude that their respective degrees of epistemic transparency primarily depend on the underlying purposes of modelling, not on their computational properties. (shrink)

Drawing on previous models of anxiety, intermediate memory, the positive symptoms of schizophrenia, and goal-directed behaviour, a neuropsychological hypothesis is proposed for the generation of the contents of consciousness. It is suggested that these correspond to the outputs of a comparator that, on a moment-by-moment basis, compares the current state of the organism's perceptual world with a predicted state. An outline is given of the information-processing functions of the comparator system and of the neural systems which mediate them. The hypothesis (...) appears to be able to account for a number of key features of the contents of consciousness. However, it is argued that neitherthis nor any existing comparable hypothesis is yet able to explain why the brain should generate conscious experience of any kind at all. (shrink)

The first claim in the target article was that there is as yet no transparent, causal account of the relations between consciousness and brain-and-behaviour. That claim remains firm. The second claim was that the contents of consciousness consist, psychologically, of the outputs of a comparator system; the third consisted of a description of the brain mechanisms proposed to instantiate the comparator. In order to defend these claims against criticism, it has been necessary to clarify the distinction between consciousness-as-such and the (...) contents of consciousness, to widen the description of the neural machinery instantiating the comparator system, and to clarify the relationship between the contents of consciousness in the here-and-now and episodic memory. (shrink)