John A Barnden

This chapter speculatively addresses the nature and effects of metaphorical views that a mind can intermittently use in thinking about itself and other minds, such as the view of mind as a physical space in which ideas have physical locations. Although such views are subjective, it is argued in this chapter that they are nevertheless part of the real nature of the conscious and unconscious mind. In particular, it is conjectured that if a mind entertains a particular (metaphorical) view at a given time, then this activity could of itself cause that mind to become more similar in the short term to how it is portrayed by the view. Hence, the views are, to an extent, self-fulfilling prophecies. In these ways, metaphorical self-reflection, even when distorting and inaccurate, is speculatively an important aspect of the true nature of mind. The chapter also outlines a theoretical approach and related implemented system (ATT-Meta) that were designed for the understanding of metaphorical discourse but that have principles that could be at the core of metaphorical self-reflection in people or future artificial agents.

How, if at all, consciousness can be part of the physical universe remains a baffling problem. This article outlines a new, developing philosophical theory of how it could do so, and offers a preliminary mathematical formulation of a physical grounding for key aspects of the theory. Because the philosophical side has radical elements, so does the physical-theory side. The philosophical side is radical, first, in proposing that the productivity or dynamism in the universe that many believe to be responsible for its systematic regularities is actually itself a physical constituent of the universe, along with more familiar entities. Indeed, it proposes that instances of dynamism can themselves take part in physical interactions with other entities, this interaction then being “meta-dynamism” (a type of meta-causation). Secondly, the theory is radical, and unique, in arguing that consciousness is necessarily partly constituted of meta-dynamic auto-sensitivity, in other words it must react via meta-dynamism to its own dynamism, and also in conjecturing that some specific form of this sensitivity is sufficient for and indeed constitutive of consciousness. The article proposes a way for physical laws to be modified to accommodate meta-dynamism, via the radical step of including elements that explicitly refer to dynamism itself. Additionally, laws become, explicitly, temporally non-local in referring directly to quantity values holding at times prior to a given instant of application of the law. The approach therefore implicitly brings in considerations about what information determines states. Because of the temporal non-locality, and also because of the deep connections between dynamism and time-flow, the approach also implicitly connects to the topic of entropy insofar as this is related to time.

The notion of Simulative Reasoning in the study of propositional attitudes within Artificial Intelligence (AI) is strongly related to the Simulation Theory of mental ascription in Philosophy. Roughly speaking, when an AI system engages in Simulative Reasoning about a target agent, it reasons with that agent’s beliefs as temporary hypotheses of its own, thereby coming to conclusions about what the agent might conclude or might have concluded. The contrast is with non-simulative meta-reasoning, where the AI system reasons within a detailed theory about the agent’s (conjectured) reasoning acts. The motive within AI for preferring Simulative Reasoning is that it is more convenient and efficient, because of a simplification of the representations and reasoning processes. The chapter discusses this advantage in detail. It also sketches the use of Simulative Reasoning in an AI natural language processing system, ATT-Meta, that is currently being implemented. This system is directed at the understanding of propositional attitude reports. In ATT-Meta, Simulative Reasoning is yoked to a somewhat independent set of ideas about how attitude reports should be treated. Central here are the claims that (a) speakers often employ commonsense (and largely metaphorical) models of mind in describing agents’ attitudes, (b) the listener accordingly needs often to reason within the terms of such models, rather than on the basis of any objectively justifiable characterization of the mind, and (c) the commonsense models filter the suggestions that Simulative Reasoning comes up with concerning target agents’ reasoning conclusions. There is a yet tighter connection between the commonsense models and the Simulative Reasoning. It turns out that Simulative Reasoning can be rationally reconstructed in terms of a more general type of reasoning about the possibly-counterfactual “world” that the target agent believes in, together with an assumption that that agent has a faithful representation of the world. In the ATT-Meta approach, the reasoner adopts that assumption when it views the target agent through a particular commonsense model (called IDEAS-AS-MODELS).

At a previous conference (CONTEXT’99), the author described the ATT-Meta context-based AI system for (a) reasoning uncertainly about agents’ beliefs and (b) performing some of the uncertain reasoning needed for the understanding of metaphorical language. ATT-Meta’s handling of uncertainty is qualitative, and includes provisions for adjudicating conflicts between different lines of reasoning. But most of the detail on conflict-handling given in the earlier paper concerned conflicts arising for the special requirements of (a). Furthermore, there have been major recent changes in the conflict-handling approach. The present paper provides a detailed account of major parts of the current approach, stressing how it operates in metaphorical reasoning. In concentrating on uncertainty-handling, this paper does not seek to repeat the justifications given elsewhere for ATT-Meta’s approach to metaphor.

Jackendoff comes to some appealing overall conclusions, but several of his assumptions and arguments are questionable. The present commentary points out the following problems: oversimplifications in the translation-based argument for the independence of language and thought; a lack of consideration of the possibility of unconscious use of internalized natural languages; insufficient consideration of possible characteristics of languages of thought ; neglect of the possibility of thinking in example-oriented and metaphorical ways; unfair bias in contrasting visual to linguistic imagery; neglect of other types of imagery; and neglect of the possibility of unconscious attentional processes.

It is proposed that conscious qualia arise when and only when the 'running' of physical processes takes a special, complex form. Running in general is the unified unfolding of processes through time, and is claimed to be an additional quality of physical processes beyond their state trajectories. The type of running needed for conscious qualia is reflexive in physically affecting and responding to itself. Intuitively, running is essentially the flow of causation, and the self-affecting/responding is a matter of causation bearing a causal relationship to itself: that is, causation can itself be reflexive. The proposal potentially makes temporal qualia the most fundamental qualia, with others derivatively arising. The proposal is neutral about whether a conscious process occurs in a natural organism, in a physically implemented computational system, or in some other physical substrate, as long as the substrate involves the special reflexive runningness

This paper is a commentary on the target article by Lokendra Shastri & Venkat Ajjanagadde [S&A]: “From simple associations to systematic reasoning: A connectionist representation of rules, variables and dynamic bindings using temporal synchrony” in same issue of the journal, pp.417–451. It puts S&A's temporal-synchrony binding method in a broader context, comments on notions of pointing and other ways of associating information - in both computers and connectionist systems - and mentions types of reasoning that are a challenge to S&A's system (amongst many other connectionist systems, recognizing that S&A's is much more capable as regards reasoning than most other contemporary systems).

Reasoning about mental states and processes is important in various subareas of the legal domain. A trial lawyer might need to reason and the beliefs, reasoning and other mental states and processes of members of a jury; a police officer might need to reason about the conjectured beliefs and reasoning of perpetrators; a judge may need to consider a defendant's mental states and processes for the purposes of sentencing and so on. Further, the mental states in question may themselves be about the mental states and processes of other people. Therefore, if AI systems are to assist with reasoning tasks in law, they may need to be able to reason about mental states and processes. Such reasoning is riddled with uncertainty, and this is true in particular in the legal domain. The article discusses how various different types of uncertainty arise, and shows how they greatly complicate the task of reasoning about mental states and processes. The article concentrates on the special case of states of belief and processes of reasoning, and sketches an implemented, prototype computer program (ATT-Meta) that copes with the various types of uncertainty in reasoning about beliefs and reasoning. In particular, the article outlines the system's facilities for handling conflict between different lines of argument, especially when these lie within the reasoning of different people. The system's approach is illustrated by application to a real-life mugging example. [NB: The date on the archived preprint avail via this Phil* page has an incorrect date, 2011, on it. The correct date is 2001.]

  Connectionist attention to variables has been too restricted in two ways. First, it has not exploited certain ways of doing without variables in the symbolic arena. One variable-avoidance method, that of logical combinators, is particularly well established there. Secondly, the attention has been largely restricted to variables in long-term rules embodied in connection weight patterns. However, short-lived bodies of information, such as sentence interpretations or inference products, may involve quantification. Therefore short-lived activation patterns may need to achieve the effect of variables. The paper is mainly a theoretical analysis of some benefits and drawbacks of using logical combinators to avoid variables in short-lived connectionist encodings without loss of expressive power. The paper also includes a brief survey of some possible methods for avoiding variables other than by using combinators

Jackendoff comes to some appealing overall conclusions, but several of his assumptions and arguments are questionable. The present commentary points out the following problems: oversimplifications in the translation-based argument for the independence of language and thought; a lack of consideration of the possibility of unconscious use of internalized natural languages; insufficient consideration of possible characteristics of languages of thought ; neglect of the possibility of thinking in example-oriented and metaphorical ways; unfair bias in contrasting visual to linguistic imagery; neglect of other types of imagery; and neglect of the possibility of unconscious attentional processes.

We consider the varieties and directions of influence that the source and target domains involved in a conceptual metaphor can have on each other during the course of understanding metaphorical utterances based on the metaphor. Previous studies have been restricted both as to direction of influence and as to type of influence. They have been largely confined to the “forward” (source to target) direction of influence, and they have concentrated on the transfer of features or propositions and (to some extent) the highlighting of aspects of a domain. By contrast, this article stresses the importance both of other varieties of influence (e.g., transfer of queries and uncertainty effects) and of “reverse” influence (target to source). We seek to curb the natural tendency to think that, because metaphor involves an overall move from source to target, therefore all inter-domain influences in the course of understanding go in that direction. The bulk of the article explores the theoretical issues involved. These issues have arisen out of a reasoning-based approach to metaphor, seeking to make best use of information from metaphorical utterance, context, source and target domains, and known metaphorical mappings, by applying a complex overall reasoning process. The article briefly explains the thorough implementation of reverse influence in a computer program for metaphorical reasoning called ATT-Meta, although the theoretical considerations are relevant to all disciplines concerned with metaphor. We make some suggestions for further computational and psychological research on metaphor.

This article arises indirectly out of the development of a particular approach, called ATT-Meta, to the understanding of some types of metaphorical utterance. However, the specifics of the approach are not the focus of the present article, which concentrates on some general issues that have informed, or arisen from, the development of the approach. The article connects those issues to the questions of metaphorical meaning and truth. A large part of the exploration of metaphor in fields such as Cognitive Linguistics and natural language Pragmatics takes metaphor to rest on complex mappings between a target subject matter and a source subject matter (see, e.g., Lakoff 1993, 2008). This is one main way in which some aspects of the source, generally including some structural aspects, can have parallels in the target. It is typical for much, and perhaps the whole, of the meaning of a metaphorical expression to be explained by means of these parallels. We will talk about “parallelism” rather than “isomorphism,” or even “analogy,” because “isomorphism” implies a strict one-to-one correspondence between items in the source scenario and items in the target scenario, and analogy theorists often hold that analogy rests on isomorphism. We wish not to prejudge the question of whether looser, messier forms of parallelism are sometimes needed. Also, we will talk about parallels rather than mappings, because our discussion will embrace the question of parallelism that may be discernible, albeit implicitly, in accounts of metaphor that are not presented as being based on mappings. Three main cases of such accounts are Relevance Theory (RT: e.g., Sperber & Wilson 2008, Wilson & Carston 2006), the categorization-based or "class-inclusion“ approach (Glucksberg 2008), and Ritchie’s (2006) CLST account (Context-Limited Simulation Theory). One main issue we will address is that of substantial non-parallelism that can exist between source and target subject-matters in metaphor: more precisely, the issue that there may be source aspects that are exploited by the utterance and have a deep effect on metaphorical interpretation but that do not themselves have a natural parallel in the target subject-matter (although we will point out that artificial parallels can generally be created). This is part of a broader point that the content of metaphorical discourse should often be seen as being derived in a rather holistic way from several metaphorical bits of the discourse, which conspire to describe some source-domain scenario, rather than being derived by putting together metaphorical meanings of each metaphorical bit (even if each bit could in principle be assigned its own separate metaphorical meaning). As a result, we suggest that it is misguided to think of the propositions making up the content (or, if you like, "meaning“) about the target scenario that is being described is a matter of metaphorical meaning of specific grammatical units such as sentences, clauses or other constituents that a traditional semantic theory would assign propositional meaning to. Rather, grammatical units (that are to be taken metaphorically) have meanings in source-domain terms; content in target-domain terms is derived from the source-domain scenario depicted by those units and fleshed out through inference; and the target-domain content is only (in general) fuzzily relatable to particular grammatical units. This view is on a spectrum at whose extreme point we could place Davidson’s (1984) view that metaphors only have literal (i.e., source-domain) meanings, with other effects on the understander not being a matter of propositional content. However, on our view it is proper to take metaphorical discourse as having non-literal meaning, couched as a collection of propositions, among other things possibly; it is just that the propositions are not to be thought of (in general) as meaning of specific sentences, clauses or other grammatical units rather than of a possibly only fuzzily delineated piece of discourse. A second main concern of the article is the question of how much, and what types of, inference are involved in the derivation of metaphorical meaning. While our own ATT-Meta approach accords with, amongst other accounts, Relevance Theory (RT) as to the centrality of inference in metaphor interpretation, we disagree with RT on its claim that metaphor interpretation is a matter of (a relatively high degree of) concept "broadening“ and (often) "narrowing“. We argue that almost all inference can in fact be theoretically re-described as concept broadening and/or narrowing, and what is left over should be included in the metaphor interpretation process anyway. Thus, the RT claim about broadening and narrowing really just says that inference, of unrestricted type, is involved in metaphor understanding, in accord with what ATT-Meta claims.

I present considerations surrounding pre-reflective self-consciousness, arising in work I am conducting on a new physicalist, process-based account of [phenomenal] consciousness. The account is called the meta-causal account because it identifies consciousness with a certain type of arrangement of meta-causation. Meta-causation is causation where a cause or effect is itself an instance of causation. The proposed type of arrangement involves a sort of time-spanning, internal reflexivity of the overall meta-causation. I argue that, as a result of the account, any conscious process has PRSC. Hence, PRSC does not need to be taken as a stipulation or argued for on purely phenomenological grounds or as a necessary support for reflective consciousness. I also show how it is natural to the account that PRSC is not an additional, peripheral, sort of consciousness, but is intrinsic to all consciousness, thereby fitting claims about self-intimation and co-constitution by various authors, and also being amenable to an adverbial account. As part of this, consciousness of an external object is just the form that current self-consciousness takes, the meta-causal constitution of it being inextricably modulated by the causal relationship with the object. The article also discusses how MCA helps explore issues of for-me-ness, transparency of perceptual consciousness, and possible immediacy and non-relationality of self-consciousness.

The representation of propositional attitudes (beliefs, desires, etc.) and the analysis of natural-language, propositional-attitude reports presents difficult problems for cognitive science and artificial intelligence. In particular, various representational approaches to attitudes involve the incorrect “imputation,” to cognitive agents, of the use of artificial theory-laden notions. Interesting cases of this problem are shown to occur in several approaches to attitudes. The imputation problem is shown to arise from the way that representational approaches explicate properties and relationships, and in particular from the way they explicate propositional attitudes themselves. Another factor contributing to imputation is the compositional nature of typical semantic approaches to propositional-attitude reports. Some strategies for avoiding undesirable imputation are examined. One of the main conclusions is that the importance of imputations that arise in a representation scheme depends strongly on the use to which the scheme is put -- on whether, for instance, the scheme is used as part of a formal, objective account of natural language, or is used rather as a representational tool within an agent.

This article discusses the extension of ViewGen, an algorithm derived for belief ascription, to the areas of intensional object identification and metaphor. ViewGen represents the beliefs of agents as explicit, partitioned proposition sets known as environments. Environments are convenient, even essential, for addressing important pragmatic issues of reasoning. The article concentrates on showing that the transformation of information in metaphors, intensional object identification, and ordinary, nonmetaphorical belief ascription can all be seen as different manifestations of a single environment-amalgamation process. The article also briefly discusses the extension of ViewGen to speech-act processing and the addition of a heuristic-based relevance-determination procedure, and justifies the partitioning approach to belief ascription.

The science of the mind, and of consciousness in particular, needs carefully to consider people's common-sense views of the mind, not just what the mind really is. Such views are themselves an aspect of the nature of (conscious) mind, and therefore part of the object of study for a science of mind. Also, since the common-sense views allow broadly successful social interaction, it is reasonable to look to the common-sense views for some rough guidance as to the real nature of the mind. On the other hand, to the extent that common-sense views are inaccurate, and perhaps even in gross conflict with the true nature of the mind, one interesting scientific question is: why do we hold such views, given our access to our own minds? Why should introspection be limited in a way that allows inaccurate views to hold sway? Now, common-sense views of the mind are revealed in natural language discourse that describes mental states, and such descriptions are largely metaphorical. The metaphors are used within thinking about the mind as well as in language. Therefore the study of metaphor is central to the study of mind. The present article is a preliminary study of the importance of metaphor in the scientific study of consciousness. It concentrates on analysing the nature of a range of important metaphors of mind, briefly discussing the extent to which they can be used to describe or qualify states of consciousness, and pointing to important questions about the nature of consciousness that the study of the metaphors raises. The article further conjectures that the reason people use metaphors in describing themselves is often not (just) that they have intellectually worked out some structural analogy --- e.g., between interactions of physical objects in physical space and interactions of ideas in the mind --- but rather that they {\it feel} their own minds to be as described by the metaphor --- e.g., they feel that the ideas in their minds are interacting physical entities. This matter of how a mind feels to itself is an aspect of the central issue of consciousness, namely its phenomenal quality. (NB: The archived file is a preprint. However, it is substantially the same as the published version.)

This paper is a commentary on the target article by Dana H. Ballard, “Cortical connections and parallel processing: Structure and function”, in the same issue of the journal, pp. 67–120. I raise some issues about the connectionist or neural-network implementation of information and information processing. Issues include the sharing of information by different parts of a connectionist/neural network, the copying of complex information from one place to another in a network, the possibility of connection weights not being synaptic weights, and the possibility of fast communication within single neurons as complex systems in their own right.

The book is thought-provoking and informative, wide in scope while also being technically detailed, and still relevant to modem AI [at least as of 1998, the time of writing this review, but probably also at the time of posting this entry here, 2023] even though it was published in 1989. This relevance lies mainly in the book’s advocacy of distributed computation at multiple levels of description, its combining of neural networks and other techniques, its emphasis on the interplay between action and perception, and its particular approach to natural language processing. The criticisms I have occasionally made above are relatively minor, and one of the major shortcomings of schema theory at least at the time of the book’s publication -- namely the lack of a detailed, preferably neural, mechanism for schema instantiation -- is one of which Arbib is well aware. The book is well produced, for the most part clearly written, and has useful summaries at the beginning of each chapter. Both general readers and AI researchers (mainly but not exclusively those working on robotics, vision and neural networks) could benefit from reading the book.

As the author has previously argued, a statement of form “Y is X” can often be taken as hyperbolic for a notably high degree of likeness between Y and X, or, instead, as hyperbolically stating how important Y is as a part of X. The present article goes further and argues that these types of hyperbole, as well as various others, are just special cases of reflexive hyperbole, a style that appears not previously to have been explored in its own right. The article therefore serves to introduce this style and to unify under it various more specific, disparate-seeming types of hyperbole, revealing their deep similarity. They all rest in a uniform way on a special property that some relationships have, namely of being reflexive in a broadened sense. The relationship of likeness is reflexive in the standard sense that any entity bears the relationship to itself with maximum possible strength. But some other relationships of interest in this article are only reflexive in a broader way that relaxes these universality and maximality requirements to an extent. The article also explains how reflexive hyperbole about likeness is a novel addition to the theory of metaphor, involving a distinctive interpretative process with special effects, and how reflexive hyperbole about part importance is not reducible to whole-for-part metonymy, because the latter does not systematically access sufficiently important parts. In addition, the article briefly considers the defaultness or otherwise of reflexive hyperbolic interpretations.

This article reveals an overlooked way of interpreting sentences like “The Internet is crack [cocaine]” or “Libraries are supermarkets.” Many existing theories of metaphor could apply here. However, they can instead be interpreted in a likeness-exaggerating way, under which “Libraries are supermarkets” is simply an exaggerated way of saying that libraries are like supermarkets to a very high degree. This interpretation option follows from simple, general considerations about exaggeration and likeness scales. In this way it is preferable to the abbreviated-simile view of metaphor, and in any case it can be added to any existing metaphor account. It has broad significance for the theory of metaphor and simile, but also provides a new, straightforward explanation of the special, likeness-strengthening effect in utterances such as “Libraries aren’t merely like supermarkets, they are supermarkets.” This effect exists despite evidence that X-is-Y metaphors do not generally convey more likeness than correspondin.

This paper continues the debate about how to distinguish metaphor from metonymy, and whether this can be done. It examines some of the differences that have been alleged to exist, and augments the already existing doubt about them. The main differences addressed are the similarity/contiguity distinction and the issue of whether source-target links are part of the message in metonymy or metaphor. In particular, the paper argues that metaphorical links can always be used metonymically and regarded as contiguities, and conversely that two particular, central types of metonymic contiguity essentially involve similarity. The paper also touches briefly on how metaphor and metonymy interact with domains, frames, etc. and on the role of imaginary identification/categorization of target as/under source items. With the possible exception of this last issue, the paper suggests that no combination of the alleged differences addressed can serve cleanly to categorize source/target associations into metaphorical ones and metonymic ones. It also suggests that it can be more profitable to analyse utterances at the level of the dimensions involved in the differences than at the higher level of metaphor and metonymy as such.

[Edited from Conclusion section:] We have looked at various challenging issues to do with getting connectionism to cope with high-level cognitive activities such a reasoning and natural language understanding. The issues are to do with various facets of generalization that are not commonly noted. We have been concerned in particular with the special forms these issues take in the arena of propositional attitude processing. The main problems we have looked at are: (1) The need to construct explicit representations of generalizations, not just generalize correctly to individual cases; (2) The need to be able to match two or more complex short-term information structures, to enable rapid generalization from recent examples rather than from long-term memories; (3) The need to represent and reason with anomalous combinations of concepts; (4) The need to perform embedded reasoning. This presents special problems for systems using non-concatenative representations (as in mainstream connectionist approaches). We also touched on vague quantification in attitude report complements. Neither this topic nor that of analogies between short-term structures (point 2) has been adequately addressed in the symbolic framework, let alone in connectionism. The opportunities and problems covered are put forward as things worth being optimistic about or pessimistic about, respectively. They are not put forward as decisive arguments for or against connectionism. The hope is that this chapter contributes to a greater understanding of the connectionist/symbolist gap by presenting some unusual issues and by throwing new light on some well known ones.