Keith Frankish

This paper reflects on and replies to the fifteen contributions responding to our target article 'What Forms Could Introspective Systems Take? A Research Programme' (all found in this issue). We focus first on contributions that criticize our research programme, then turn to ones that test our framework against various views and models of human introspection, and finally consider contributions that explore possible variations of introspection in humans, non-human animals, current AI systems, and imaginary minds. We conclude by drawing some lessons for our research programme and making some suggestions for future research on possible forms of introspection.

Subjectivity presents a challenging puzzle at the intersection of neuroscience and philosophy. While minds can be understood as biological systems evolved to regulate bodily processes and guide adaptive action, their subjective nature remains elusive. Perconti highlights that, although aspects of subjectivity are functionally characterizable, our minds appear to possess intrinsic qualities, qualia, which defy third-person description. His cognitive social theory posits that self-consciousness evolved from social cognition, emphasizing the role of introspective illusions. This theory suggests that our inner worlds, though seemingly metaphysically unique, are products of complex brain functions. A rethinking of subjectivity, which sheds assumptions of intrinsic human specialness, could lead to a comprehensive understanding of consciousness.

This paper presents an extended thought experiment conducted by Daniel Dennett and the author as a way of exploring how consciousness might develop without a central “Cartesian Theater.” The starting point is a system with a real Cartesian Theater – a drone remotely piloted by a human sitting at a monitor. The imaginative exercise involves the removal of the human operator and their gradual replacement by onboard systems, emancipating the drone from external control. Following Braitenberg’s “law of uphill analysis and downhill invention,” the rebuilding proceeds by bottom-up engineering rather than top-down design and makes use of various tricks and shortcuts. Six iterations of the drone are described, in which it acquires basic capacities for recognition and response, learning, imagination, metacognition, communication, and, finally, autostimulation – the ability to process its own verbal reports. This final development allows the drone to access its own internal response patterns, giving it the sense of having an inner life with qualities that resist explanation. The resulting sketch thus not only illustrates how consciousness could be achieved without a Cartesian Theater but also shows how the process might naturally create the illusion of such a theater.

This chapter proposes a framework for thinking about implicit bias and how we can control it. It argues that systematic implicit bias is best thought of as a form of belief, which can conflict with the subject’s explicit beliefs. Explicit belief itself is identified with a kind of active acceptance, yielding a dual-level view of the human mind, in which a passive implicit level supports an active explicit one—a view that has close links with dual-process theory. The chapter then turns to self-control. If we are systematically biased, how can we form unbiased explicit beliefs and make them effective in action? The dual-level view offers novel answers, assigning a crucial role to certain metacognitive attitudes, which constitute what we call strength of will. The chapter sets out the conditions for deliberate override of implicit bias and outlines some predictions of the proposed account.

The illusionist theory of consciousness denies the existence of the ‘phenomenal’ properties widely thought to render experience conscious. Central to the illusionist approach is what we call _the distortion thesis_ — the claim that introspective mechanisms systematically misrepresent neural processes in a way that strongly inclines us to judge that they have phenomenal properties, creating the ‘illusion’ of phenomenality. However, there is no consensus among illusionists as to the exact nature of this distortion and there has been relatively little exploration of the options. This article seeks to make progress by mapping some patterns of distortion compatible with an illusionist approach. We propose a preliminary classification scheme based on the three dimensions of _positiveness_, _conceptuality_, and _dispellability_, and use this scheme to describe eight possible distortion patterns, each corresponding to a different hypothesis about how the illusion of phenomenality is created. We offer tentative assessments of the plausibility of these patterns from the perspective of the distortion thesis, but we do not argue for a specific view. Our aim here is to map possibilities, stimulate further theorizing, and encourage empirical research on the nature of introspective distortion. Both illusionists and their opponents should welcome this increased scrutiny of a key element of the illusionist approach.

This article presents the case for an approach to consciousness that I call illusionism. This is the view that phenomenal consciousness, as usually conceived, is illusory. According to illusionists, our sense that it is like something to undergo conscious experiences is due to the fact that we systematically misrepresent them as having phenomenal properties. Thus, the task for a theory of consciousness is to explain our illusory representations of phenomenality, not phenomenality itself, and the hard problem is replaced by the illusion problem. Although it has had powerful defenders, illusionism remains a minority position, and it is often dismissed as failing to take consciousness seriously. This article seeks to rebut this accusation. It defines the illusionist programme, outlines its attractions, and defends it against some common objections. It concludes that illusionism is a coherent and attractive approach, which deserves serious consideration.

Illusionism is a revisionary view of consciousness, which denies the existence of the phenomenal properties traditionally thought to render experience conscious. The view has theoretical attractions, but some think it also has objectionable ethical implications. They take illusionists to be denying the existence of consciousness itself, or at least of the thing that gives consciousness its ethical value, and thus as undermining our established ethical attitudes. This article responds to this objection. I argue that, properly understood, illusionism neither denies the existence of consciousness nor entails that consciousness does not ground ethical value. It merely offers a different account of what consciousness is and why it grounds ethical value. The article goes on to argue that the theoretical revision proposed by illusionists does have some indirect implications for our ethical attitudes but that these are wholly attractive and progressive ones. The illusionist perspective on consciousness promises to make ethical decision making easier and to extend the scope of our ethical concern. Illusionism is good news.

This paper takes another look at Davidson's paratactic theory of indirect discourse and evaluates some revisions to it, proposed recently by Ian Rumfitt (Mind, 1993). Davidson's original version of the theory – according to which indirect speech reports refer to token utterances – has a problem dealing with ambiguity. Rumfitt suggests that we can solve this problem by supposing that the immediate objects of verbs in indirect speech are token representations of disambiguated LF tree-structures. I argue that this proposal is inadequate and suggest that it is better to think of indirect speech as relating speakers to utterance types.

We propose a new approach to the study of introspection. Instead of asking what form introspection actually takes in humans or other animals, we ask what forms it could take, in natural or artificial minds. What are the dimensions along which forms of introspection could vary? This is a relatively unexplored question, but it is one that has the potential to open new avenues of study and reveal new connections between existing ones. It may, for example, focus attention on possible forms of introspection radically different from the human one and help to integrate competing theories of human introspection in a non-adversarial manner. We introduce and motivate the project, provide a preliminary mapping of the space of possible forms of introspection, and sketch a programme for interdisciplinary research on the topic.

It is plausible to regard inner speech as an activity, whose functions are continuous with those of outer speech. Yet there is also a case for thinking that inner speech has a cognitive role, as a vehicle for conscious thought. This chapter reconciles these apparently conflicting claims by showing how outer speech can itself play a cognitive role. Drawing on dual-process theories of reasoning, it develops a view of conscious (‘Type 2’) thinking as an activity, initially performed in outer speech. By engaging in self-directed speech, we can break down complex problems into subproblems that can be solved by nonconscious (‘Type 1’) processes, thereby vastly extending our reasoning powers. It is as the subsequent internalization of such ‘outer thinking’ that inner speech has a cognitive function. The chapter also extends this account to show how acts of inner speech can function as judgements and decisions.

According to dual-process theory, human cognition is supported by two distinct types of processing, one fast, automatic, and unconscious, the other slower, controlled, and conscious. These processes are sometimes said to constitute two minds – an intuitive old mind, which is evolutionarily ancient and composed of specialized subsystems, and a reflective new mind, which is distinctively human and the source of general intelligence. This theory has far-reaching consequences, and it means that research on enhancing and replicating human intelligence will need to take different paths, depending on whether it is the old mind or the new mind that is the target. This chapter examines these issues in depth. It argues first for a reinterpretation of dual-process theory, which pictures the new mind as a virtual system, formed by culturally transmitted habits of autostimulation. It then explores the implications of this reinterpreted dual-process theory for the projects of cognitive enhancement and artificial intelligence, including the creation of artificial general intelligence. The chapter concludes with a brief assessment of the risks of those projects as they appear in this new light.

Goff argues that Galileo erred in denying that sensory qualities are present in the physical world and that we should correct his error by supposing that all matter has an intrinsic conscious aspect. This paper argues that we should be open to another theoretical option. Galileo's real error, I argue, was not about the location of sensory qualities, but about their very existence. Like most people, Galileo assumed that sensory qualities are instantiated somewhere. I argue that this is a theoretical assumption which can and should be questioned. If we drop it, we can give a natural account of the function of sensory quality talk and explain how our puzzlement about consciousness arises.

The problem of consciousness arises when we depsychologize consciousness—that is, conceptualize it in terms of phenomenal feel rather than psychological function. Panpsychism offers an elegant solution to the problem, which takes depsychologization seriously. In doing so, however, it also illustrates the perils of depsychologization. Nagasawa highlights one dead end for panpsychism, and I shall argue that there are more. Panpsychism consigns consciousness to a metaphysical limbo where it is beyond the reach of science and lacks ethical and personal significance. The moral is that we should retrace our steps and question the depsychologized conception itself.

The meta-problem of consciousness prompts the metaquestion: is it the only problem consciousness poses? If we could explain all our phenomenal intuitions in topic-neutral terms, would anything remain to be explained? Realists say yes, illusionists no. In this paper I defend the illusionist answer. While it may seem obvious that there is something further to be explained -- consciousness itself -- this seemingly innocuous claim immediately raises a further problem -- the hard meta-problem. What could justify our continued confidence in the existence of consciousness once all our intuitions about it have been explained away? The answer would involve heavy-duty metaphysical theorizing, probably including a commitment either to substance dualism or to the existence of a mysterious intrinsic subjectivity. A far less extravagant option is to endorse the illusionist response and conclude that the meta-problem is not a meta-problem at all but the problem of consciousness.

This chapter outlines a new argument for the view that language has a cognitive role. I suggest that humans exhibit two distinct kinds of belief state, one passively formed, the other actively formed. I argue that actively formed beliefs (_virtual beliefs_, as I call them) can be identified with _premising policies_, and that forming them typically involves certain linguistic operations. I conclude that natural language has at least a limited cognitive role in the formation and manipulation of virtual beliefs

It can be argued that dual-system theorists should adopt an action - based view of System 2 (S2), on which S2 reasoning is an intentional activity. It can also be argued that they should adopt a dual - attitude theory, on which the two systems have distinct sets of propositional attitudes. However, Peter Carruthers has argued that on the action-based view there are no S2 attitudes. This paper replies to Carruthers, proposing a view of S2 attitudes as virtual ones, which are partially realized in S1 attitudes. This view is compatible with, and a natural extension of, the action-based view.

This book deals with the nature of consciousness. Many philosophers and psychologists today believe that the mind is a physical phenomenon, whose processes can be explained in scientific terms. Consciousness presents the biggest challenge to this view. Can the physical sciences really explain the nature of conscious experience—the way it feels to have a throbbing headache, or see a sunset, or smell freshly ground coffee? Or is there more to these experiences than a physical account can ever capture? If consciousness is non-physical, then it is hard to see how it can have effects within the physical world. But if it is physical, then why does it seem so different from other physical phenomena? And what physical processes does it involve? Is the feel of a conscious experience just a matter of what it represents? Does consciousness involve a form of inner awareness? Finally, could it be that our view of consciousness is mistaken? Do we need to rethink some of our fundamental assumptions about it? These questions go to the heart of our conception of ourselves and our place in the universe, and are the subject of vigorous debate among contemporary philosophers

Artificial intelligence, or AI, is a cross-disciplinary approach to understanding, modeling, and creating intelligence of various forms. It is a critical branch of cognitive science, and its influence is increasingly being felt in other areas, including the humanities. AI applications are transforming the way we interact with each other and with our environment, and work in artificially modeling intelligence is offering new insights into the human mind and revealing new forms mentality can take. This volume of original essays presents the state of the art in AI, surveying the foundations of the discipline, major theories of mental architecture, the principal areas of research, and extensions of AI such as artificial life. With a focus on theory rather than technical and applied issues, the volume will be valuable not only to people working in AI, but also to those in other disciplines wanting an authoritative and up-to-date introduction to the field.