Many of the computing systems programmed using Machine Learning are opaque: it is difficult to know why they do what they do or how they work. Explainable Artificial Intelligence aims to develop analytic techniques that render opaque computing systems transparent, but lacks a normative framework with which to evaluate these techniques’ explanatory successes. The aim of the present discussion is to develop such a framework, paying particular attention to different stakeholders’ distinct explanatory requirements. Building on an analysis of “opacity” from (...) philosophy of science, this framework is modeled after accounts of explanation in cognitive science. The framework distinguishes between the explanation-seeking questions that are likely to be asked by different stakeholders, and specifies the general ways in which these questions should be answered so as to allow these stakeholders to perform their roles in the Machine Learning ecosystem. By applying the normative framework to recently developed techniques such as input heatmapping, feature-detector visualization, and diagnostic classification, it is possible to determine whether and to what extent techniques from Explainable Artificial Intelligence can be used to render opaque computing systems transparent and, thus, whether they can be used to solve the Black Box Problem. (shrink)

According to the classical account, propositions are sui generis, abstract, intrinsically-representational entities and our cognitive attitudes, and the token states within us that realize those attitudes, represent as they do in virtue of their propositional objects. In light of a desire to explain how it could be that propositions represent, much of the recent literature on propositions has pressured various aspects of this account. In place of the classical account, revisionists have aimed to understand propositions in terms of more familiar (...) entities such as facts, types of mental or linguistic acts, and even properties. But we think that the metaphysical story about propositions is much simpler than either the classical theorist or the revisionist would have you believe. In what follows, we argue that a proper understanding of the nature of our cognitive relations to propositions shows that the question of whether propositions themselves represent is, at best, a distraction. We will argue that once this distraction is removed, the possibility of a very pleasing, minimalist story of propositions emerges; a story that appeals only to assumptions that are shared by all theorists in the relevant debate. (shrink)

According to standard assumptions in semantics, ordinary users of a language have implicit beliefs about the truth-conditions of sentences in that language, and they often agree on those beliefs. For example, it is assumed that if Anna and John are both competent users of English and the former utters ‘grass is green’ in conversation with the latter, they will both believe that that sentence is true if and only if grass is green. These assumptions play an important role in an (...) intuitively compelling picture of communication, according to which successful communication through literal assertoric utterances is normally effected thanks to our shared beliefs about the truth-conditions of the sentences uttered in the course of the conversation. Against these standard assumptions, this paper argues that the participants in a conversation rarely have the same beliefs about the truth-conditions of the sentences involved in a linguistic interaction. More precisely, it argues for Variance, the thesis that nearly every utterance is such that there is no proposition which more than one language user believes to be that utterance’s truth-conditional content. If Variance is true, we must reject the standard picture of communication. Towards the end of the paper I identify three ways in which ordinary conversations can be communication-like despite the truth of Variance and argue that the most natural amendments to the standard picture fail to explain them. (shrink)

Altruistic deception (or the telling of “white lies”) is common in humans. Does it also exist in non-human animals? On some definitions of deception, altruistic deception is impossible by definition, whereas others make it too easy by counting useful-but-ambiguous information as deceptive. I argue for a definition that makes altruistic deception possible in principle without trivializing it. On my proposal, deception requires the strategic exploitation of a receiver by a sender, where “exploitation” implies that the sender elicits a behaviour in (...) the receiver that is beneficial in a different type of situation and is expressed only because the signal raises the probability, from the receiver’s standpoint, of that type of situation. I then offer an example of a real signal that is deceptive in this sense, and yet potentially altruistic (and certainly cooperative): the purr call of the pied babbler. Fledglings associate purr calls with food, and adults exploit this learned association, in the absence of food, to lead fledglings away from predators following an alarm call. I conclude by considering why altruistic deception is apparently so rare in non-human animals. (shrink)

Most facts of grounding involve nonfundamental concepts, and thus must themselves be grounded. But how? The leading approaches—due to Bennett, deRosset, and Dagupta—are subject to objections. The way forward is to deny a presupposition common to the leading approaches, that there must be some simple formula governing how grounding facts are grounded. Everyone agrees that facts about cities might be grounded in some complex way about which we know little; we should say the same about the facts of grounding themselves. (...) The kinds of facts that might enter into the grounds of the facts of grounding are explored at length. (shrink)

Kriegel has revived adverbialism as a theory of consciousness. But recent attacks have shed doubt on the viability of the theory. To save adverbialism, I propose that the adverbialist take a stance on the nature of adverbial modification. On one leading theory, adverbial modification turns on the instantiation by a substance of a psychological type. But the resulting formulation of adverbialism turns out to be a mere notational variant on the relationalist approaches against which Kriegel dialectically situates adverbialism. By contrast, (...) I argue that the way to be an adverbialist is to adopt an event ontology, emphasizing the active contribution of the mind to the phenomenology of experience. My close examination of the semantics of adverbial modification throws this metaphysical distinction into sharp relief. The event-based semantics overcomes recent objections in a way superior to the methods that would have been obviously available in the absence of a sophisticated semantics. (shrink)

How can we think about things in the outside world? There is still no widely accepted theory of how mental representations get their meaning. In light of pioneering research, Nicholas Shea develops a naturalistic account of the nature of mental representation with a firm focus on the subpersonal representations that pervade the cognitive sciences.

The concept of morality underpins the moral responsibility that not only depends on the outward practices (or ‘output’, in the case of humanoid robots) of the agents but on the internal attitudes (‘input’) that rational and responsible intentioned beings generate. The primary question that has initiated extensive debate, i.e. ‘Can humanoid robots be moral?’, stems from the normative outlook where morality includes human conscience and socio-linguistic background. This paper advances the thesis that the conceptions of morality and creativity interplay with (...) linguistic human beings instead of non-linguistic humanoid robots, as humanoid robots are indeed docile automata that cannot be responsible for their actions. To eradicate human ethics in order to make way for humanoid robot ethics highlights the moral actions and adequacy that hinges the myth of creative agency and self-dependency, which a humanoid robot can scarcely express. (shrink)

One argument for the moving spotlight theory is that it better explains certain aspects of our temporal phenomenology than does any static theory of time. Call this the argument from passage phenomenology. In this paper it is argued that insofar as moving spotlight theorists take this to be a sound argument they ought embrace a new version of the moving spotlight theory according to which the moving spotlight is a cresting wave of causal efficacy. On this view it is more (...) than just presentness that is temporary: a range of other fundamental properties are also temporary because presentness synchronically changes the fundamental properties that are instantiated in the present moment, from those instantiated when that moment is future, or past. On this view having experiences as of presentness co-varies with the presence of presentness, allowing the moving spotlight theorist to provide a compelling explanation for why we have the temporal phenomenology we do, and to explain how we can know that we are present by reflecting on the nature of said phenomenology. (shrink)

I present a puzzle about belief and credence, which takes the form of three independently supported views that are mutually inconsistent. The first is the view that S has a modal belief that p (e.g., S believes that probably-p) if and only if S has a corresponding credence that p. The second is the view that S believes that p only if S has some credence that p. The third is the view that, possibly, S believes that p without a (...) modal belief that p. [Word Count: 85]. (shrink)

According to a view I’ll call Epistemic Normativism, knowledge is normative in the same sense in which paradigmatically normative properties like justification are normative. This paper argues against EN in two stages and defends a positive non-normativist alternative. After clarifying the target in §1, I consider in §2 some arguments for EN from the premise that knowledge entails justification. I first raise some worries about inferring constitution from entailment. I then rehearse the reasons why some epistemologists reject the Entailment Thesis (...) and argue that a non-normativist picture provides the best explanation of all the intuitions surrounding this thesis, favorable and unfavorable. On this picture, human knowledge is a structured non-normative complex that has as one of its parts a justification-making property, analogous in role to good-making properties like pleasurableness. After giving three arguments against EN in §3 and answering an objection in §4, I turn in §5 to further develop the positive view sketched in §2. In §6, I take stock and conclude. (shrink)

This paper argues that deciding on whether the cognitive sciences need a Representational Theory of Mind matters. Far from being merely semantic or inconsequential, the answer we give to the RTM-question makes a difference to how we conceive of minds. How we answer determines which theoretical framework the sciences of mind ought to embrace. The structure of this paper is as follows. Section 1 outlines Rowlands’s argument that the RTM-question is a bad question and that attempts to answer it, one (...) way or another, have neither practical nor theoretical import. Rowlands concludes this because, on his analysis, there is no non-arbitrary fact of the matter about which properties something must possess in order to qualify as a mental representation. By way of reply, we admit that Rowlands’s analysis succeeds in revealing why attempts to answer the RTM-question simpliciter are pointless. Nevertheless, we show that if specific formulations of the RTM-question are stipulated, then it is possible, conduct su... (shrink)

This paper engages critically with anti-representationalist arguments pressed by prominent enactivists and their allies. The arguments in question are meant to show that the “as-such” and “job-description” problems constitute insurmountable challenges to causal-informational theories of mental content. In response to these challenges, a positive account of what makes a physical or computational structure a mental representation is proposed; the positive account is inspired partly by Dretske’s views about content and partly by the role of mental representations in contemporary cognitive scientific (...) modeling. (shrink)

The Significance Argument (SA) for the irreducibility of consciousness is based on a series of new puzzle-cases that I call multiple candidate cases. In these cases, there is a multiplicity of physical-functional properties or relations that are candidates to be identified with the sensible qualities and our consciousness of them, where those candidates are not significantly different. I will argue that these cases show that reductive materialists cannot accommodate the various ways in which consciousness is significant and must allow massive (...) vagueness in conscious experience. I also will argue that a nonreductive theory of the conscious-of relation can easily provide a very satisfying, unified explanation of the ways in which this relation is significant. (shrink)

The past two decades have seen an explosion of literature on Hume's views about mental representation and intentionality. This essay gives a roadmap of this literature, while arguing for two main interpretive claims. First, Hume aims to naturalize all forms of mental representation and intentionality, that is, to explain them in terms of properties and relations that are found throughout the natural world (not just in minds) and that are not, individually, peculiar to representational or intentional things. Second, Hume holds (...) that the passions are not representational but do have intentionality extrinsically. (shrink)

Many psychologists studying lay belief attribution and behavior explanation cite Donald Davidson in support of their assumption that people construe beliefs as inner causes. But Davidson’s influential argument is unsound; there are no objective grounds for the intuition that the folk construe beliefs as inner causes that produce behavior. Indeed, recent experimental work by Ian Apperly, Bertram Malle, Henry Wellman, and Tania Lombrozo provides an empirical framework that accords well with Gilbert Ryle’s alternative thesis that the folk construe beliefs as (...) patterns of living that contextualize behavior. (shrink)

A number of recent theories of quantum gravity lack a one-dimensional structure of ordered temporal instants. Instead, according to many of these views, our world is either best represented as a single three-dimensional object, or as a configuration space composed of such three-dimensional objects, none of which bear temporal relations to one another. Such theories will be empirically self-refuting unless they can accommodate the existence of conscious beings capable of representation. For if representation itself is impossible in a timeless world, (...) then no being in such a world could entertain the thought that a timeless theory is true, let alone believe such a theory or rationally believe it. This paper investigates the options for understanding representation in a three-dimensional, timeless, world. Ultimately it concludes that the only viable option is one according to which representation is taken to be deeply non-naturalistic. Ironically then we are left with two seemingly very unattractive options. Either a very naturalistic motivation—taking seriously a live view in fundamental physics—leads us to a very non-naturalistic view of the mental, or else views in the philosophy of mind partly dictate what is an acceptable theory in physics. (shrink)

The intuition that we can think about non-existent objects seems to be in tension with philosophical concerns about the relationality of intentionality. Tim Crane’s psychologism removes this tension by proposing a psychologistic account of intentionality according to which intentionality is a purely non-relational notion. I argue that his account has counterintuitive consequences regarding our thoughts about existing objects, and as such is insufficiently plausible to convince us to reject the relationality of intentionality.

The paper presents a new theory of perceptual demonstrative thought, the property-dependent theory. It argues that the theory is superior to both the object-dependent theory (Evans, McDowell) and the object-independent theory (Burge).

Representationalism holds that a perceptual experience's qualitative character is identical with certain of its representational properties. To date, most representationalists endorse atomistic theories of perceptual content, according to which an experience's content, and thus character, does not depend on its relations to other experiences. David Rosenthal, by contrast, proposes a view that is naturally construed as a version of representationalism on which experiences’ relations to one another determine their contents and characters. I offer here a new defense of this holistic (...) representationalism, arguing that some objections to atomistic views are best interpreted as supporting it. (shrink)

A standard view in contemporary philosophy is that belief is involuntary, either as a matter of conceptual necessity or as a contingent fact of human psychology. We present seven experiments on patterns in ordinary folk-psychological judgments about belief. The results provide strong evidence that voluntary belief is conceptually possible and, granted minimal charitable assumptions about folk-psychological competence, provide some evidence that voluntary belief is psychologically possible. We also consider two hypotheses in an attempt to understand why many philosophers have been (...) tempted to view belief as involuntary: that belief is a prototype concept and that belief is a dual character concept. Altogether, our findings contribute to longstanding philosophical debates about the relationship between the will and the intellect, while also advancing scientific understanding of important social judgments. (shrink)

Among the cognitive capacities of evolved creatures is the capacity to represent. Theories in cognitive neuroscience typically explain our manifest representational capacities by positing internal representations, but there is little agreement about how these representations function, especially with the relatively recent proliferation of connectionist, dynamical, embodied, and enactive approaches to cognition. In this talk I sketch an account of the nature and function of representation in cognitive neuroscience that couples a realist construal of representational vehicles with a pragmatic account of (...) mental content. I call the resulting package a deflationary account of mental representation and I argue that it avoids the problems that afflict competing accounts. (shrink)

The mental realm seems different to the physical realm; the mental is thought to be dependent on, yet distinct from the physical. But how, exactly, are the two realms supposed to be different, and what, exactly, creates the seemingly insurmountable juxtaposition between the mental and the physical? This review identifies and discusses five marks of the mental, features that set characteristically mental phenomena apart from the characteristically physical phenomena. These five marks (intentionality, consciousness, free will, teleology, and normativity) are not (...) presented as a set of features that define mentality. Rather, each of them is something we seem to associate with phenomena we consider mental, and each of them seems to be in tension with the physical view of reality in its own particular way. It is thus suggested how there is no single mind-body problem, but a set of distinct but interconnected problems. Each of these separate problems is analyzed, and their differences, similarities and connections are identified. This provides a useful basis for future theoretical work on psychology and philosophy of mind, that until now has too often suffered from unclarities, inadequacies, and conflations. (shrink)

Recently, Kroedel and Schulz have argued that the exclusion problem—which states that certain forms of non-reductive physicalism about the mental are committed to systematic and objectionable causal overdetermination—can be solved by appealing to grounding. Specifically, they defend a principle that links the causal relations of grounded mental events to those of grounding physical events, arguing that this renders mental–physical causal overdetermination unproblematic. Here, we contest Kroedel and Schulz’s result. We argue that their causal-grounding principle is undermotivated, if not outright false. (...) In particular, we contend that the principle has plausible counterexamples, resulting from the fact that some mental states are not fully grounded by goings on ‘in our heads’ but also require external factors to be included in their full grounds. We draw the sceptical conclusion that it remains unclear whether non-reductive physicalists can plausibly respond to the exclusion argument by appealing to considerations of grounding. (shrink)

Philosophers commonly say that beliefs come in degrees. Drawing from the literature, I make precise three arguments for this claim: an argument from degrees of confidence, an argument from degrees of firmness, and an argument from natural language. I show that they all fail. I also advance three arguments that beliefs do not come in degrees: an argument from natural language, an argument from intuition, and an argument from the metaphysics of degrees. On the basis of these arguments, I conclude (...) that beliefs do not come in degrees. (shrink)

Scepticism is sometimes expressed about whether there is any interesting problem of other minds. In this paper I set out a version of the conceptual problem of other minds which turns on the way in which mental occurrences are presented to the subject and situate it in relation to debates about our knowledge of other people's mental lives. The result is a distinctive problem in the philosophy of mind concerning our relation to other people.

This paper examines the prospects for a conceptual or functional role theory of moral concepts. It is argued that such an account is well-placed to explain both the irreducibility and practicality of moral concepts. Several versions of conceptual role semantics for moral concepts are distinguished, depending on whether the concept-constitutive conceptual roles are wide or narrow normative or non-normative and purely doxastic or conative. It is argued that the most plausible version of conceptual role semantics for moral concepts involves only (...) ‘narrow’ conceptual roles, where these include connections to motivational, desire-like, states. In the penultimate section it is argued, contrary to what Wedgwood, Enoch and others have claimed, that such an account of moral concepts cannot plausibly be combined with the claim that moral concepts refer to robust properties. (Published with Open Access.). (shrink)

Jerry Fodor, by common agreement, is one of the world’s leading philosophers. At the forefront of the cognitive revolution since the 1960s, his work has determined much of the research agenda in the philosophy of mind and the philosophy of psychology for well over 40 years. This special issue dedicated to his work is intended both as a tribute to Fodor and as a contribution to the fruitful debates that his work has generated. One philosophical thesis that has dominated Fodor’s (...) work since the 1960s is realism about the mental. Are there really mental states, events and processes? From his first book, Psychological Explanation (1968), onwards, Fodor has always answered this question with a resolute yes. From his early rejection of Wittgensteinian and behaviourist conceptions of the mind, to his later disputes with philosophers of mind of the elminativist ilk, he has always been opposed to views that try to explain away mental phenomena. On his view, there are minds, and minds can change the world. (shrink)

What is the connection between being in a conscious mental state and believing that you yourself are currently in that state? On the one hand, it is natural to think that this connection is, or involves, a necessary connection of some sort. On the other hand, it is hard to know what the nature of this necessary connection is. For there are plausible arguments according to which this connection is not metaphysically necessary, not rationally necessary, and not merely naturally necessary. (...) If these arguments are correct, and if these options are exhaustive, while there is a necessary connection between a conscious state and the belief that you are in it, there is apparently no necessary connection it could be! This paper sets out this problem—the necessary connection problem, I will call it—and defends and explores a novel proposal about how to solve it. (shrink)

The question, ‘Is cognition linguistic?' divides recent cognitive theories into two antagonistic groups. Sententialists claim that we think in some language, while advocates of non linguistic views of cognition deny this claim. The Introspective Argument for Sententialism is one of the most appealing arguments for sententialism. In substance, it claims that the introspective fact of inner speech provides strong evidence that our thoughts are linguistic. This article challenges this argument. I claim that the Introspective Argument for Sententialism confuses the content (...) of our thoughts with their vehicles: while sententialism is a thesis about the vehicles of our thoughts, inner speech sentences are the content of auditory or articulatory images. The rebuttal of the introspective argument for sententialism is shown to have a general significance in cognitive science: Introspection does not tell us how we think. (shrink)

The overwhelming majority of those who theorize about implicit biases posit that these biases are caused by some sort of association. However, what exactly this claim amounts to is rarely specified. In this paper, I distinguish between different understandings of association, and I argue that the crucial senses of association for elucidating implicit bias are the cognitive structure and mental process senses. A hypothesis is subsequently derived: if associations really underpin implicit biases, then implicit biases should be modulated by counterconditioning (...) or extinction but should not be modulated by rational argumentation or logical interventions. This hypothesis is false; implicit biases are not predicated on any associative structures or associative processes but instead arise because of unconscious propositionally structured beliefs. I conclude by discussing how the case study of implicit bias illuminates problems with popular dual-process models of cognitive architecture. (shrink)

In The Elm and the Expert (1994), Fodor rejects the notion of narrow content as superfluous. He envisions a scientific intentional psychology that adverts only to broad content properties in its explanations. I show that there has been no change in Fodor's treatment of Frege cases and cases involving the so‐called deferential concepts. And for good reason: his notion of narrow content (1985‐91) couldn't explain them. The only apparent change concerns his treatment of Twin Earth cases. However, I argue that (...) the notion of broad content that his purely informational semantics delivers is, in some interesting sense, equivalent to the mapping notion of narrow content he officially gave up. I also critically reconstruct the evolution of Fodor's thinking between 1980 and 1994. (shrink)

Deviant phenomenal knowledge is knowing what it’s like to have experiences of, e.g., red without actually having had experiences of red. Such a knower is a deviant. Some physicalists have argued and some anti-physicalists have denied that the possibility of deviants undermines anti-physicalism and the Knowledge Argument. The current paper presents new arguments defending the deviant-based attacks on anti-physicalism. Central to my arguments are considerations concerning the psychosemantic underpinnings of deviant phenomenal knowledge. I argue that physicalists are in a superior (...) position to account for the conditions in virtue of which states of deviants constitute representations of phenomenal facts. (shrink)

RÉSUMÉ : Dans son récent recueil d’articles Language, Truth and History, Donald Davidson semble pencher en faveur d’une philosophie du langage mettant l’accent sur la notion de l’intention communicative du sujet parlant; en quoi il se rapproche du point de vue de Paul Grice. Si cela est juste, la pensée de Davidson se serait dégagée de l’approche sémantique formelle qu’il soutenait dans ses Inquiries into Truth and Interpretation. Dans cet article, je soutiens que, bien qu’il y ait beaucoup de similitudes (...) entre ces deux penseurs, Davidson n’a pas abandonné ses précédentes vues sur le langage. (shrink)

Der These des erweiterten Geistes zufolge befinden sich manche mentalen Repräsentationen außerhalb der körperlichen Grenzen der Wesen, zu denen sie gehören. Einer der stärksten Einwände gegen diese These stellt das Argument der Nichtabgeleitetheit von Frederick Adams, Ken Aizawa und Jerry Fodor dar. Dieses Argument setzt voraus, dass genuine mentale Repräsentationen nichtabgeleitete Gehalte haben – ihre semantischen Eigenschaften sind also nicht durch Absichten, Wünsche oder Konventionen konstituiert. Repräsentationen mit nichtabgeleitetem Gehalt finden sich jedoch, so das Argument weiter, nur innerhalb der körperlichen (...) Grenzen mentaler Wesen. Ich werde dafür argumentieren, dass das Argument der Nichtabgeleitetheit scheitert, da es insbesondere bei Tieren externe Repräsentationen gibt, deren Gehalt nichtabgeleitet ist. Dies folgt jedenfalls aus der aussichtsreichsten Theorie nichtabgeleiteter Repräsentationen, der Teleosemantik, und es gibt gute Gründe anzunehmen, dass auch andere naturalistische Gehaltstheorien dieselbe Implikation haben. (shrink)

Free will, if such there be, involves free choosing: the ability to mentally choose an outcome, where the outcome is 'free' in being, in some substantive sense, up to the agent of the choice. As such, it is clear that the questions of how to understand free will and mental causation are connected, for events of seemingly free choosing are mental events that appear to be efficacious vis-a-vis other mental events as well as physical events. Nonetheless, the free will and (...) mental causation debates have proceeded largely independently of each other. Here we aim to make progress in determining how the free will and mental causation debates bear on one another. We first argue that the problems of free will and of mental causation can be seen as special cases of a more general problem, concerning whether and how mental events of a given type may be efficacious, qua the types of event they are---qualitative, intentional, freely deliberative---given their apparent causal irrelevancy for effects of the type in question; here we generalize what Horgan 1989 identifies as "the problem of mental quausation" (S1). We then build on this result to identify fruitful parallels between hard determinism and eliminative physicalism (S2) and soft determinism and non-reductive physicalism (S3). (shrink)

In this dissertation, I argue that we should be pluralists about truth and in turn, eliminativists about the property Truth. Traditional deflationists were right to suspect that there is no such property as Truth. Yet there is a plurality of pluralities of properties which enjoy defining features that Truth would have, were it to exist. So although, in this sense, truth is plural, Truth is non-existent. The resulting account of truth is indebted to deflationism as the provenance of the suspicion (...) that Truth doesn't exist. But it would be hasty to simply classify the account as deflationary. Each of the 'truth-like' properties that it recognizes is highly substantive--that is, complex and explanatorily potent. So we should deflate Truth by recognizing that it doesn't exist, but we should also recognize that one of the most vital tasks in truth theory is to discover the essences of the many truth-like properties. My aim here is to do precisely this. (shrink)

Skyrms, building on the work of Dretske, has recently developed a novel information-theoretic account of propositional content in simple signalling systems. Information-theoretic accounts of content traditionally struggle to accommodate the possibility of misrepresentation, and I show that Skyrms’s account is no exception. I proceed to argue, however, that a modified version of Skyrms’s account can overcome this problem. On my proposed account, the propositional content of a signal is determined not by the information that it actually carries, but by the (...) information that it would carry at the nearest separating equilibrium of the underlying evolutionary dynamics. I show that this amended account yields reasonable ascriptions of false propositional content in a well-known formal model of the evolution of communication , and close with a discussion of the serious but perhaps not insuperable difficulties we face in applying the account to examples of signalling in the real world. (shrink)

What is creativity? One new idea may be creative, whereas another is merely new: What's the difference? And how is creativity possible? These questions about human creativity can be answered, at least in outline, using computational concepts. There are two broad types of creativity, improbabilist and impossibilist. Improbabilist creativity involves novel combinations of familiar ideas. A deeper type involves METCS: the mapping, exploration, and transformation of conceptual spaces. It is impossibilist, in that ideas may be generated which – with respect (...) to the particular conceptual space concerned – could not have been generated before. The more clearly conceptual spaces can be defined, the better we can identify creative ideas. Defining conceptual spaces is done by musicologists, literary critics, and historians of art and science. Humanist studies, rich in intuitive subtleties, can be complemented by the comparative rigour of a computational approach. Computational modelling can help to define a space, and to show how it may be mapped, explored, and transformed. Impossibilist creativity can be thought of in “classical” Al terms, whereas connectionism illuminates improbabilist creativity. Most Al models of creativity can only explore spaces, not transform them, because they have no self-reflexive maps enabling them to change their own rules. A few, however, can do so. A scientific understanding of creativity does not destroy our wonder at it, nor does it make creative ideas predictable. Demystification does not imply dehumanization. (shrink)