Our aim in this reply is to defend Global Workspace theory (GWT) from the challenge of Block's article. We argue that Block's article relies on an outdated and imprecise concept of access, and perpetuates a common misunderstanding of GWT that conflates the global workspace with working memory. In the light of the relevant clarifications, Block's conclusion turns out to be unwarranted, and the basic tenets of GWT emerge unscathed.

This paper proposes a brain-inspired cognitive architecture that incorporates approximations to the concepts of consciousness, imagination, and emotion. To emulate the empirically established cognitive efficacy of conscious as opposed to non-conscious information processing in the mammalian brain, the architecture adopts a model of information flow from global workspace theory. Cognitive functions such as anticipation and planning are realised through internal simulation of interaction with the environment. Action selection, in both actual and internally simulated interaction with the environment, is mediated by (...) affect. An implementation of the architecture is described which is based on weightless neurons and is used to control a simulated robot. (shrink)

It has been claimed that we need singular self-knowledge to function properly as rational agents. I argue that this is not strictly true: agents in certain relations could dispense with singular self-knowledge and instead rely on plural self-knowledge. In defending the possibility of this kind of ‘selfless agent’, I thereby defend the possibility of a certain kind of ‘seamless’ collective agency; agency in a group of agents who have no singular self-knowledge, who do not know which member of the group (...) they are. I discuss four specific functions for which singular self-knowledge has been thought indispensable: distinguishing intentional from unintentional actions, connecting non-indexical knowledge with action, reflecting on our own reasoning, and identifying which ultimate practical reasons we have. I argue in each case that by establishing certain relations between agents – relations I label ‘motor vulnerability’, ‘cognitive vulnerability’, ‘evidential unity’ and ‘moral unity’ – we would allow those agents to do everything a rational agent needs to do while relying only on plural, rather than singular, self-knowledge. Finally, I consider the objection that any agents who met the conditions I lay out for selfless agency would thereby cease to qualify as distinct agents, merging into a single agent without agential parts. Against this objection, I argue that we should recognise the possibility of simultaneous agency in whole and parts, and not regard either as disqualifying the other. (shrink)

This paper presents a computer model of cortical broadcast and competition based on spiking neurons and inspired by the hypothesis of a global neuronal workspace underlying conscious information processing in the human brain. In the model, the hypothesised workspace is realised by a collection of recurrently inter-connected regions capable of sustaining and disseminating a reverberating spatial pattern of activation. At the same time, the workspace remains susceptible to new patterns arriving from outlying cortical populations. Competition among these cortical populations for (...) influence on the workspace is effected by a combination of mutual inhibition and top-down amplification. (shrink)

Carruthers presents an interesting analysis of confabulation and a clear attack on introspection. Yet his theory-based alternative is a mechanistic view of which neglects the fact that social understanding occurs within a network of social relationships. In particular, the role of language in his model is too simple.

Unlike human soldiers, autonomous weapons systems are unaffected by psychological factors that would cause them to act outside the chain of command. This is a compelling moral justification for their development and eventual deployment in war. To achieve this level of sophistication, the software that runs AWS will have to first solve two problems: the frame problem and the representation problem. Solutions to these problems will inevitably involve complex software. Complex software will create security risks and will make AWS critically (...) vulnerable to hacking. I claim that the political and tactical consequences of hacked AWS far outweigh the purported advantages of AWS not being affected by psychological factors and always following orders. Therefore, one of the moral justifications for the deployment of AWS is undermined. (shrink)

The target article demonstrates how neurocognitive modellers should not be intimidated by challenges such as Jackendoff's and should explore neurally plausible implementations of linguistic constructs. The next step is to take seriously insights offlered by neuroscience, including the robustness allowed by analogue computation with distributed representations and the power of attractor dynamics in turning analogue into nearly discrete operations.

The claims that “The brain processes information” or “Cognition is information processing” are accepted as truisms in cognitive science. However, it is unclear how to evaluate such claims absent a specification of “information” as it is used by neurocognitive theories. The aim of this article is, thus, to identify the key features of information that information-based neurocognitive theories posit. A systematic identification of these features can reveal the explanatory role that information plays in specific neurocognitive theories, and can, therefore, be (...) both theoretically and practically important. These features can be used, in turn, as desiderata against which candidate theories of information may be evaluated. After discussing some characteristics of explanation in cognitive science and their implications for “information”, three notions are briefly introduced: natural, sensory, and endogenous information. Subsequently, six desiderata are identified and defended based on cognitive scientific practices. The global workspace theory of consciousness is then used as a specific case study that arguably posits either five or six corresponding features of information. (shrink)

Carruthers considers and rejects a mixed position according to which we have interpretative access to unconscious thoughts, but introspective access to conscious ones. I argue that this is too hasty. Given a two-level view of the mind, we can, and should, accept the mixed position, and we can do so without positing additional introspective mechanisms beyond those Carruthers already recognizes.

We address the claim that nonhuman animals do not represent unobservable states, based on studies of physical cognition by rooks and social cognition by scrub-jays. In both cases, the most parsimonious explanation for the results is counter to the reinterpretation hypothesis. We suggest that imagination and prospection can be investigated in animals and included in models of cognitive architecture.

This response defends the view that human metacognition results from us turning our mindreading capacities upon ourselves, and that our access to our own propositional attitudes is through interpretation rather than introspection. Relevant evidence is considered, including that deriving from studies of childhood development and other animal species. Also discussed are data suggesting dissociations between metacognitive and mindreading capacities, especially in autism and schizophrenia.

How can we disentangle the neural basis of phenomenal consciousness from the neural machinery of the cognitive access that underlies reports of phenomenal consciousness? We can see the problem in stark form if we ask how we could tell whether representations inside a Fodorian module are phenomenally conscious. The methodology would seem straightforward: find the neural natural kinds that are the basis of phenomenal consciousness in clear cases when subjects are completely confident and we have no reason to doubt their (...) authority, and look to see whether those neural natural kinds exist within Fodorian modules. But a puzzle arises: do we include the machinery underlying reportability within the neural natural kinds of the clear cases? If the answer is ‘Yes’, then there can be no phenomenally conscious representations in Fodorian modules. But how can we know if the answer is ‘Yes’? The suggested methodology requires an answer to the question it was supposed to answer! The paper argues for an abstract solution to the problem and exhibits a source of empirical data that is relevant, data that show that in a certain sense phenomenal consciousness overflows cognitive accessibility. The paper argues that we can find a neural realizer of this overflow if assume that the neural basis of phenomenal consciousness does not include the neural basis of cognitive accessibility and that this assumption is justified by the explanations it allows. (shrink)

The currently leading cognitive theory of consciousness, Global Workspace Theory,1,2 postulates that the primary functions of consciousness include a global broadcast serving to recruit internal resources with which to deal with the current situation and to modulate several types of learning. In addition, conscious experiences present current conditions and problems to a "self" system, an executive interpreter that is identifiable with brain structures like the frontal lobes and precuneus.1Be it human, animal or artificial, an autonomous agent3 is said to be (...) functionally conscious if its control structure (mind) implements Global Workspace Theory and the LIDA Cognitive Cycle, which includes unconscious memory and control functions needed to integrate the conscious component of the system. We would therefore consider humans, many animals4 and even some virtual or robotic agents5,6 to be functionally conscious. Such entities may approach phenomenal consciousness, as found in human and other biological brains, as additional brain-like features are added.Here we argue that adding mechanisms to produce a stable, coherent perceptual field7 in a LIDA controlled mobile robot might provide a small but significant step toward phenomenal consciousness in machines.>8 We also propose that implementing several of the various notions of self in such a LIDA controlled robot may well prove another step toward phenomenal consciousness in machines. (shrink)

Analogical cognition refers to the ability to detect, process, and learn from relational similarities. The study of analogical and similarity cognition is widely considered one of the ‘success stories’ of cognitive science, exhibiting convergence across many disciplines on foundational questions. Given the centrality of analogy to mind and knowledge, it would benefit philosophers investigating topics in epistemology and the philosophies of mind and language to become familiar with empirical models of analogical cognition. The goal of this essay is to describe (...) recent empirical work on analogical cognition as well as model applications to philosophical topics. Topics to be discussed include the epistemological distinction between implicit knowledge and explicit knowledge, the debate between empiricists and nativists, the frame problem, expertise, creativity and autism, cognitive architecture, and relational knowledge. Particular attention is given to Dedre Gentner and colleague’s structure-mapping theory – the most developed and widely accepted model of analogical cognition. (shrink)

The Frame Problem originated as an obstacle for classical, symbolic A.I. that was adopted, expanded, and reformulated by philosophers. The version of the problem that I focus on, the Holism Problem, points out the difficulty in programming systems to recognize and consider mostly relevant information, given that relevance is context-sensitive. My goal in this thesis is to determine whether the Global Workspace Theory (GWT) can solve the holism problem. GWT proposes that distributed parallel processing, global broadcast, and chaotic itinerancy can (...) solve the problem by providing a system with 1) the speed to search through information, 2) access to the information it needs to compute relevance, 3) the ability to synthesize information. I argue that GWT fails to enable a system to recognize any relevant information because it inadequately responds to the Epistemological Holism Problem, which requires successfully determining the norms by which a system can recognize relevance. (shrink)

I review three problems that have historically motivated pessimism about artificial intelligence: (1) the Problem of Consciousness, according to which artificial systems function without the right sort of conscious oversight; (2) The Problem of Global Relevance, according to which artificial systems cannot solve fully general theoretical and practical problems; (3) The Problem of Semantic Irrelevance, according to which artificial systems cannot be guided by semantic comprehension. I connect the dots between all three problems by drawing attention to non-syntactic inferences—inferences that (...) are made on the basis of insight into the rational relationships among thought-contents. Consciousness alone affords such insight, I argue, and such insight alone confers positive epistemic status on the execution of the inferences. Only when artificial systems can execute inferences that are rationally guided by phenomenally conscious states will such systems count as intelligent in a literal sense. (shrink)

Purpose: Constructivism postulates that the perceived reality is a complex construct formed during development. Depending on the particular school, these inner constructs take on different forms and structures and affect cognition in different ways. The purpose of this article is to address the questions of how and, even more importantly, why we form such inner constructs. Approach: This article proposes that brain development is controlled by an inherent anticipatory drive, which biases learning towards the formation of forward predictive structures and (...) inverse goal-oriented control structures. This drive, in combination with increasingly complex environmental interactions during cognitive development, enforces the structuring of our conscious self, which is embedded in a constructed inner reality. Essentially, the following questions are addressed: Which basic mechanisms lead us to the construction of inner realities? How are these emergent inner realities structured? How is the self represented within the inner realities? And consequently, which cognitive structures constitute the media for conscious thought and selfconsciousness? Findings: Due to the anticipatory drive, representations in the brain shape themselves predominantly purposefully or intentionally. Taking a developmental, evolutionary perspective, we show how the brain is forced to develop progressively complex and abstract representations of the self embedded in the constructed inner realities. These self representations can evoke different stages of self-consciousness. Implications: The anticipatory drive shapes brain structures and cognition during the development of progressively more complex, competent, and flexible goal-oriented bodyenvironment interactions. Self-consciousness develops because increasingly abstract, individualizing self representations are necessary to realize these progressively more challenging environmental interactions. (shrink)

This thesis contributes to a better conceptual understanding of how self-organized control works. I begin by analyzing the control problem and its solution space. I argue that the two prominent solutions offered by classical cognitive science (centralized control with rich commands, e.g., the Fodorian central systems) and embodied cognitive science (distributed control with simple commands, such as the subsumption architecture by Rodney Brooks) are merely two positions in a two-dimensional solution space. I outline two alternative positions: one is distributed control (...) with rich commands, defended by proponents of massive modularity hypothesis; the other is centralized control with simple commands. My goal is to develop a hybrid account that combines aspects of the second alternative position and that of the embodied cognitive science (i.e., centralized and distributed controls with simple commands). Before developing my account, I discuss the virtues and challenges of the first three. This discussion results in a set of criteria for successful neural control mechanisms. Then, I develop my account through analyzing neuroscientific models of decision-making and control with the theoretical lenses provided by formal decision and social choice theories. I contend that neural processes can be productively modeled as a collective of agents, and neural self-organization is analogous to democratic self-governance. In particular, I show that the basal ganglia, a set of subcortical structures, contribute to the production of coherent and intelligent behaviors through implementing “democratic" procedures. Unlike the Fodorian central system—which is a micro-managing “neural commander-in-chief”—the basal ganglia are a “central election commission.” They delegate control of habitual behaviors to other distributed control mechanisms. Yet, when novel problems arise, they engage and determine the result on the basis of simple information (the votes) from across the system with the principles of neurodemocracy, and control with simple commands of inhibition and disinhibition. By actively managing and taking advantage of the wisdom-of-the-crowd effect, these democratic processes enhance the intelligence and coherence of the mind’s final "collective" decisions. I end by defending this account from both philosophical and empirical criticisms and showing that it meets the criteria for successful solution. (shrink)

The objectives of this article are twofold. First, by denying the dualism inherent in attempts to load metaphysical significance on the inner/outer distinction, it defends the view that scientific investigation can approach consciousness in itself, and is not somehow restricted in scope to the outward manifestations of a private and hidden realm. Second, it provisionally endorses the central tenets of global workspace theory, and recommends them as a possible basis for the sort of scientific understanding of consciousness thus legitimised. However, (...) the article goes on to argue that global workspace theory alone does not constitute a fully worked-out objective account of the conscious subject. This requires additional attention to be paid to the issue of embodiment, and to the possibility of indexicality that arises when an instantiation of the global workspace architecture inhabits a spatially localised body. (shrink)

How can we explain the intentional nature of an expert’s actions, performed without immediate and conscious control, relying instead on automatic cognitive processes? How can we account for the differences and similarities with a novice’s performance of the same actions? Can a naturalist explanation of intentional expert action be in line with a philosophical concept of intentional action? Answering these and related questions in a positive sense, this dissertation develops a three-step argument. Part I considers different methods of explanations in (...) cognitive neuroscience (Bennett & Hacker’s philosophical, conceptual analysis; Marr’s three levels of explanation; Neural Correlates of Consciousness research; mechanistic explanation), defending ‘mechanistic explanation’ as a method that provides the necessary tools for integrating interdisciplinary insights into human action. Furthermore, a dynamic, explanatory mechanism allows the assessment of the impact of learning and development on expert action in a valuable way that other methods don’t. Part II continues by scrutinizing several cognitive neuroscientific theories of learning and development (neuroconstructivism; dual-processing theories; simulation theory; extended mind/cognition hypothesis), arguing for the complex interactions between different types of processing and different action representations involved in expert action performances. Moreover, according to our discussion of a particular ‘simulation theory’ these interactions can be influenced in several ways with the use of language, allowing an agent to configure a specific action representation for performance at a later stage. The results of Parts I and II are then applied in Part III to a parallel discussion of philosophical analyses of intentional action (discussing i.a. Frankfurt, Bratman, Pacherie and Ricoeur) and cognitive neuroscientific insights in it. Both approaches are found to converge in emphasizing the importance for an expert to develop stable patterns of actions that comply maximally not only with his intentions, but also with his motor expertise and with situational conditions. Consequently, his actions – automatic, or not – rely on this ‘sculpted space of actions’. (shrink)