Nicholas Shea

Just how far can externalism go? In this exciting new book Ruth Millikan explores a radically externalist treatment of empirical concepts (Millikan 2000). For the last thirty years philosophy of mind’s ties to meaning internalism have been loosened. The theory of content has swung uncomfortably on its moorings in a fickle current, straining against opposing ties to mind and world. In this book Millikan casts conceptual content adrift from the thinker: what determines the content of a concept is not cognitively accessible. She has only the stanchion of the world to hold her theory fast. She hopes that the tide will turn, and the theory of meaning will come stably to rest downstream of this anchor. This book is a bold exploration of how that might be achieved.

This paper advocates explicitness about the type of entity to be considered as content- bearing in connectionist systems; it makes a positive proposal about how vehicles of content should be individuated; and it deploys that proposal to argue in favour of representation in connectionist systems. The proposal is that the vehicles of content in some connectionist systems are clusters in the state space of a hidden layer. Attributing content to such vehicles is required to vindicate the standard explanation for some classificatory networks’ ability to generalise to novel samples their correct classification of the samples on which they were trained

A step towards a theory of consciousness would be to characterise the effect of consciousness on information processing. One set of results suggests that the effect of consciousness is to interfere with computations that are optimally performed non-consciously. Another set of results suggests that conscious, system 2 processing is the home of norm-compliant computation. This is contrasted with system 1 processing, thought to be typically unconscious, which operates with useful but error-prone heuristics. These results can be reconciled by separating out two different distinctions: between conscious and non-conscious representations, on the one hand, and between automatic and deliberate processes, on the other. This pair of distinctions is used to illuminate some existing experimental results and to resolve the puzzle about whether consciousness helps or hinders accurate information processing. This way of resolving the puzzle shows the importance of another category, which we label ‘type 0 cognition’, characterised by automatic computational processes operating on non-conscious representations.

This paper sets out a view about the explanatory role of representational content and advocates one approach to naturalising content – to giving a naturalistic account of what makes an entity a representation and in virtue of what it has the content it does. It argues for pluralism about the metaphysics of content and suggests that a good strategy is to ask the content question with respect to a variety of predictively successful information processing models in experimental psychology and cognitive neuroscience; and hence that data from psychology and cognitive neuroscience should play a greater role in theorising about the nature of content. Finally, the contours of the view are illustrated by drawing out and defending a surprising consequence: that individuation of vehicles of content is partly externalist.

An interesting feature of some sets of representations is that their structure mirrors the structure of the items they represent. Founding an account of representational content on isomorphism, homomorphism or structural resemblance has proven elusive, however, largely because these relations are too liberal when the candidate structure over representational vehicles is unconstrained. Furthermore, in many cases where there is a clear isomorphism, it is not relied on in the way the representations are used. That points to a potential resolution: that an isomorphism must be used, hence usable, if it is to be an ingredient in a theory of content. This paper argues that the class of exploitable isomorphisms can indeed play a content-constituting role

The success of a piece of behaviour is often explained by its being caused by a true representation (similarly, failure falsity). In some simple organisms, success is just survival and reproduction. Scientists explain why a piece of behaviour helped the organism to survive and reproduce by adverting to the behaviour’s having been caused by a true representation. That usage should, if possible, be vindicated by an adequate naturalistic theory of content. Teleosemantics cannot do so, when it is applied to simple representing systems (Godfrey-Smith 1996). Here it is argued that the teleosemantic approach to content should therefore be modified, not abandoned, at least for simple representing systems. The new ‘infotel-semantics’ adds an input condition to the output condition offered by teleosemantics, recognising that it is constitutive of content in a simple representing system that the tokening of a representation should correlate probabilistically with the obtaining of its specific evolutionary success condition

Humans can think about their conscious experiences using a special class of ?phenomenal? concepts. Psychophysical identity statements formulated using phenomenal concepts appear to be contingent. Kripke argued that this intuited contingency could not be explained away, in contrast to ordinary theoretical identities where it can. If the contingency is real, property dualism follows. Physicalists have attempted to answer this challenge by pointing to special features of phenomenal concepts that explain the intuition of contingency. However no physicalist account of their distinguishing features has proven to be satisfactory. Leading accounts rely on there being a phenomenological difference between tokening a physical-functional concept and tokening a phenomenal concept. This paper shows that existing psychological data undermine that claim. The paper goes on to suggest that the recalcitrance of the intuition of contingency may instead by explained by the limited means people typically have for applying their phenomenal concepts. Ways of testing that suggestion empirically are proposed

Fodor’s radical concept nativism flowed from his view that hypothesis testing is the only route to concept acquisition. Many have successfully objected to the overly-narrow restriction to learning by hypothesis testing. Existing representations can be connected to a new representational vehicle so as to constitute a sustaining mechanism for a new representation, without the new representation thereby being constituted by or structured out of the old. This paper argues that there is also a deeper objection. Connectionism shows that a more fundamental assumption underpinning the debate can also be rejected: the assumption that the development of a new representation must be explained in content-involving terms if innateness is to be avoided. Fodor has argued that connectionism offers no new resources to explain concept acquisition: unless it is merely an uninteresting claim about neural implementation, connectionism’s defining commitment to distributed representations reduces to the claim that some representations are structured out of others (which is the old, problematic research programme). Examination of examples of representational development in connectionist networks shows, however, that some such models explain the development of new representational capacities in non-representational terms. They illustrate the possibility of representational development that is not explicable-by-content. Connectionist representations can be distributed in an important sense, which is incompatible with the assumption of explanation-by-content: they can be distributed over non-representational resources that account for their development. Rejecting the assumption of explanation-by-content thereby opens up a more radical way of rejecting Fodor’s argument for radical concept nativism.

Millikan’s theory of content purports to rely heavily on the existence of isomorphisms between a system of representations and the things in the world which they represent — “the mapping requirement for being intentional signs” (Millikan 2004, p. 106). This paper asks whether those isomorphisms are doing any substantive explanatory work. Millikan’s isomorphism requirement is deployed for two main purposes. First, she claims that the existence of an isomorphism is the basic representing relation, with teleology playing a subsidiary role — to account for misrepresentation (the possibility of error). Second, Millikan relies on an isomorphism requirement in order to guarantee that a system of representations displays a kind of productivity. This seemingly strong reliance on isomorphism has prompted the objection that isomorphism is too liberal to be the basic representing relation: there are isomorphisms between any system of putative representations and any set, of the same cardinality, of items putatively represented. This paper argues that all the work in fixing content is in fact done by the teleology. Deploying Millikan’s teleology-based conditions to ascribe contents will ensure that there is an isomorphism between representations and the things they represent, but the isomorphism ‘requirement’ is playing no substantive explanatory role in Millikan’s account of content determination. So an objection to her theory based on the liberality of isomorphism is misplaced. The second role for isomorphism is to account for productivity. If some kind of productivity is indeed necessary for representation, then functional isomorphism will again be too liberal a constraint to account for that feature. The paper suggests an alternative way of specifying the relation between a system of representations and that which they represent which is capable of playing an explanatory role in accounting for Millikan’s type of productivity. In short, the liberality of isomorphism is no objection to Millikan’s teleosemantics, since the isomorphism ‘requirement’ need play no independent substantive role in Millikan’s account of representation.

The concept of innateness is used to make inferences between various better-understood properties, like developmental canalization, evolutionary adaptation, heritability, species-typicality, and so on (‘innateness-related properties’). This article uses a recently-developed account of the representational content carried by inheritance systems like the genome to explain why innateness-related properties cluster together, especially in non-human organisms. Although inferences between innateness-related properties are deductively invalid, and lead to false conclusions in many actual cases, where some aspect of a phenotypic trait develops in reliance on a genetic representation it will tend, better than chance, to have many of the innateness-related properties. The account also shows why inferences between innateness-related properties sometimes fail and argues that such inferences are especially misleading when applied to human psychology and behaviour because human psychological development is especially reliant on non-genetic inherited representations

Emerging evidence suggests that the long-established distinction between habit-based and goal-directed decision-making mechanisms can also be sustained in humans. Although the habit-based system has been extensively studied in humans, the goal-directed system is less well characterized. This review brings to that task the distinction between conceptual and nonconceptual representational mechanisms. Conceptual representations are structured out of semantic consituents - the use of which requires an ability to perform some language-like syntactic processing. Decision making - as investigated by neuroscience and psychology - is normally studied in isolation from questions about concepts as studied in philosophy and cognitive psychology. We ask what role concepts play in the "goal-directed" decision-making systems. We argue that one fruitful way of studying this system in humans is to investigate the extent to which it deploys conceptual representations

The distinction between top-down and bottom-up effects is widely relied on in experimental psychology. However, there is an important problem with the way it is normally defined. Top-down effects are effects of previously-stored information on processing the current input. But on the face of it that includes the information that is implicit in the operation of any psychological process – in its dispositions to transition from some types of representational state to others. This paper suggests a way to distinguish information stored in that way from the kind of influence of prior information that psychologists are concerned to classify as a top-down effect. So-drawn, the distinction is not just of service to theoretical psychology. Asking about the extent of top-down processing is one way to pose some of the questions at issue in philosophical debates about cognitive penetrability – about the extent of the influence of cognitive states on perception. The existence of a theoretically-useful perception-cognition distinction has come under pressure, but even if it has to be abandoned, some of the concerns addressed in the cognitive penetrability literature can be recaptured by asking about the extent of top-down influences on any given psychological process. That formulation is more general, since it can be applied to any psychological process, not just those that are paradigmatically perceptual.

Recent work combining cognitive neuroscience with computational modelling suggests that distributed patterns of neural firing may represent probability distributions. This paper asks: what makes it the case that distributed patterns of firing, as well as carrying information about (correlating with) probability distributions over worldly parameters, represent such distributions? In examples of probabilistic population coding, it is the way information is used in downstream processing so as to lead to successful behaviour. In these cases content depends on factors beyond bare information, contra Brian Skyrms’ (2010) view that representational content can be fully characterised in information-theoretic terms.

We discuss a recent approach to investigating cognitive control, which has the potential to deal with some of the challenges inherent in this endeavour. In a model-based approach, the researcher defines a formal, computational model that performs the task at hand and whose performance matches that of a research participant. The internal variables in such a model might then be taken as proxies for latent variables computed in the brain. We discuss the potential advantages of such an approach for the study of the neural underpinnings of cognitive control and its pitfalls, and we make explicit the assumptions underlying the interpretation of data obtained using this approach.

This thesis concerns the content of mental representations. It draws lessons for philosophical theories of content from some empirical findings about brains and behaviour drawn from experimental psychology (cognitive, developmental, comparative), cognitive neuroscience and cognitive science (computational modelling). Chapter 1 motivates a naturalist and realist approach to mental representation. Chapter 2 sets out and defends a theory of content for static feedforward connectionist networks, and explains how the theory can be extended to other supervised networks. The theory takes forward Churchland’s state space semantics by making a new and clearer proposal about the syntax of connectionist networks − one which nicely accounts for representational development. Chapter 3 argues that the same theoretical approach can be extended to unsupervised connectionist networks, and to some of the representational systems found in real brains. The approach can also show why connectionist systems sometimes show typicality effects, explaining them without relying upon prototype structure. That is discussed in chapter 4, which also argues that prototype structure, where it does exist, does not determine content. The thesis goes on to defend some unorthodox features of the foregoing theory: that a role is assigned to external samples in specifying syntax, that both inputs to and outputs from the system have a role in determining content, and that the content of a representation is partly determined by the circumstances in which it developed. Each, it is argued, may also be a fruitful way of thinking about mental content more generally. Reliance on developmental factors prompts a swampman-type objection. This is rebutted by reference to three possible reasons why content is attributed at all. Two of these motivations support the idea that content is partly determined by historical factors, and the third is consistent with it. The result: some empirical lessons for philosophical theories of mental content.

Recent work on inheritance systems can be divided into inclusive conceptions, according to which genetic and non-genetic inheritance are both involved in the development and transmission of nearly all animal behavioral traits, and more demanding conceptions of what it takes for non-genetic resources involved in development to qualify as a distinct inheritance system. It might be thought that, if a more stringent conception is adopted, homologies could not subsist across two distinct inheritance systems. Indeed, it is commonly assumed that homology relations cannot survive a shift between genetic and cultural inheritance systems, and substantial reliance has been placed on that assumption in debates over the phylogenetic origins of hominin behavioral traits, such as male-initiated intergroup aggression. However, in the homology literature it is widely accepted that a trait can be homologous—that is, inherited continuously in two different lineages from a single common ancestor—despite divergence in the mechanisms involved in the trait’s development in the two lineages. In this paper, we argue that even on an extremely stringent understanding of what it takes for developmental resources to form a separate inheritance system, homologies can nonetheless subsist across shifts between distinct inheritance systems. We argue that this result is a merit of this way of characterizing what it is to be an inheritance system, that it has implications for adjudicating between alternative accounts of homology, and that it offers an important cautionary lesson about how to reason with the homology concept, particularly in the context of cultural species

Commentary on Bergstrom and Rosvall, ‘The transmission sense of information’, Biology and Philosophy. In response to worries that uses of the concept of information in biology are metaphorical or insubstantial, Bergstrom and Rosvall have identified a sense in which DNA transmits information down the generations. Their ‘transmission view of information’ is founded on a claim about DNA’s teleofunction. Bergstrom and Rosvall see their transmission view of information as a rival to semantic accounts. This commentary argues that it is complementary. The idea that DNA is transmitting information down the generations only makes sense if it is carrying a message, that is to say if it has semantic content

There is ongoing controversy as to whether the genome is a representing system. Although it is widely recognised that DNA carries information, both correlating with and coding for various outcomes, neither of these implies that the genome has semantic properties like correctness or satisfaction conditions, In the Scope of Logic, Methodology, and the Philosophy of Sciences, Vol. II. Kluwer, Dordrecht, pp. 387–400). Here a modified version of teleosemantics is applied to the genome to show that it does indeed have semantic properties – there is representation in the genome. The account differs in three respects from previous attempts to apply teleosemantics to genes. It emphasises the role of the consumer of representations. It rejects the standard assumption that genetic representation can be used to explain the course of an organism’s development. And it identifies the explanatory role played by representational properties of the genome. A striking consequence of this account is that other inheritance systems could also be representational. Thus, a version of the parity thesis is accepted. However, the criteria for being an inheritance system are demanding, so semantic properties are not ubiquitous

Many have doubted whether the transition to genuinely new representational resources is susceptible to psychological explanation. In The Origin of Concepts (O.U.P. 2009), Susan Carey makes a strong empirical case for the existence of discontinuities in conceptual development. Carey also offers a plausible psychological explanation of some of these transitions, in particular of the child’s acquisition of the ability to represent natural numbers. The combination amounts to a forceful answer to puzzles about the learnability of new representations.

What is the evolutionary significance of the various mechanisms of imitation, emulation and social learning found in humans and other animals? This paper presents an advance in the theoretical resources for addressing that question, in the light of which standard approaches from the cultural evolution literature should be refocused. The central question is whether humans have an imitationbased inheritance system—a mechanism that has the evolutionary function of transmitting behavioural phenotypes reliably down the generations. To have the evolutionary power of an inheritance system, an imitiation-based mechanism must meet a range of demanding requirements. The paper goes on to review the evidence for and against the hypothesis that there is indeed an imitation-based inheritance system in humans.

There is increasing evidence for epigenetically mediated transgenerational inheritance across taxa. However, the evolutionary implications of such alternative mechanisms of inheritance remain unclear. Herein, we show that epigenetic mechanisms can serve two fundamentally different functions in transgenerational inheritance: (i) selection-based effects, which carry adaptive information in virtue of selection over many generations of reliable transmission; and (ii) detection-based effects, which are a transgenerational form of adaptive phenotypic plasticity. The two functions interact differently with a third form of epigenetic information transmission, namely information about cell state transmitted for somatic cell heredity in multicellular organisms. Selection-based epigenetic information is more likely to conflict with somatic cell inheritance than is detection-based epigenetic information. Consequently, the evolutionary implications of epigenetic mechanisms are different for unicellular and multicellular organisms, which underscores the conceptual and empirical importance of distinguishing between these two different forms of transgenerational epigenetic effect.

The New Thinking contained in this volume rejects an Evolutionary Psychology that is committed to innate domain-specific psychological mechanisms: gene-based adaptations that are unlearnt, developmentally fixed and culturally universal. But the New Thinking does not simply deny the importance of innate psychological traits. The problem runs deeper: the concept of innateness is not suited to distinguishing between the two positions. That points to a more serious problem with the concept of innateness as it is applied to human psychological phenotypes. This paper argues that the features of recent human evolution highlighted by the New Thinking imply that the concept of inherited representation, set out here, is a better tool for theorising about human cognitive evolution.

One of the great outstanding problems in materialist philosophy of mind is the problem of how there can be space in the material world for intentionality. In the 1980s Ruth Millikan formulated a detailed theory according to which representations are physical particulars and their contents are complex relational properties of those particulars which can be specified in terms of respectable properties drawn from the natural sciences. In particular, she relied on the biological concept of the function of a trait, and the existence of historical conditions which enter into an evolutionary explanation of the operation of that trait. The present article is an introduction to this influential theory of intentionality.