Brain-Mind: From Neurons to Consciousness and Creativity is the first part of the Treatise on Mind and Society by Canadian philosopher and cognitive scientist Paul Thagard, which also includes Mind-Society: From Brains to Social Sciences and Professions and Natural Philosophy: From Social Brains to Knowledge, Reality, Morality, and Beauty. The books build on the author’s large previous body of work and together aim to provide a unified account of the cognitive and social sciences, professions, and humanities. 

The books may be read separately but the original order provides a logical thread taking the reader from (roughly) the most local and specific questions related to the neural organization in the brain to the general and abstract questions of “knowledge, reality, morality, and meaning” (Thagard, 2019, p. xvi). Brain-Mind establishes a brain-based theory of cognition and emotion, Mind-Society further explores the relationship between individual minds and social sciences, while Natural Philosophy uses theories of mind and society to answer larger ontological, epistemological, and ethical questions. 

Brain-Mind, specifically, aims to present a brain-based multilevel mechanistic account of human cognition and behavior—one of very few of such a scale—while also setting a high standard of accessibility and elegance. To construct his account, Thagard systematically applies Eliasmith’s (2013) semantic pointer architecture (SPA) to the mind-related phenomena of increasing complexity, starting from sensory perception (Chapter 3) and going all the way to emotion (Chapter 7), consciousness, (Chapter 8) creativity (Chapter 11), and self (Chapter 12). These cognitive phenomena are explained through binding of representations (understood as firing patterns) by means of convolution to form higher-order multi-domain representations, i.e. semantic pointers, and the competition among them. The account directly connects mental phenomena with neural implementations. Il also provides an up-to-date treatment for a large variety of cognitive processes. For example, it accommodates partial innateness of representations/concepts and the dual (top-down/bottom-up) nature of perception, supports (and relies on) emergent properties, ties emotion to cognition, emphasizes the role of embodiment, and allows for a degree of separation between linguistic ability and reasoning.  

Many details of the account, however, remain to be provided. These details are crucial for deciding whether SPA architecture is consistent with and is able to fully accommodate the properties of the phenomena under investigation. Some of the specific questions that remain unaddressed are: How does the framework deal with the potential ‘combinatorial explosion’ of the features relevant to a concept and their selective activation/manifestation at the conscious level? How does it deal with hypothetical scenarios and counterfactuals in reasoning? What determines the threshold for conscious access? Does conscious access have a functional role? If yes, what is it? Why do we have access to knowledge but not inference itself if the inference is readily available? What is the mechanism of attention involved in semantic pointer selection? How does the framework allow for the seemingly conscious deliberate navigation of our mental space?

In the absence of the ability to directly test the SPA architecture, only once a more fleshed out picture emerges may Thagard’s SPA-based account be considered a falsifiable theory (the author acknowledges as much in the final chapter of the book, p. 274). That said, although Thagard relies on a specific neural architecture and emphasizes the plausibility of his account in terms of neural implementation, most of the implications of SPA are also “supported by and have lessons for other computational models of cognition” (Ritter in Thagard, 2019, p. xiv). The insights of Thagard’s approach at large may be used to form a basis for a high-level framework providing guidelines on a) the scope and the kinds of phenomena to be accommodated by a potential unifying theory of the mind, and b) the requisite properties of neural representations. The emerging picture is potentially consistent with different cognitive architectures and can be augmented by a variety of computational strategies, including the widely popular Bayesian ones, as Thagard’s mechanistic approach leaves plenty of room for such specifications. The explanatory limitations of the account at its current state are similar to those of other popular contenders for the place of a unifying theory of mind such as, for example, predictive processing (PP). While Thagard’s view is mechanistic (formulated in terms of neural structures/representations and their interactions) and PP is ‘procedural’ (formulated in terms of processes and process strategies), neither allows for making specific predictions regarding the content and interaction of representations in the absence of extensive specification of implementations (and to a certain degree even after that). However, it is exciting to see multiple competing (or complementary) alternatives emerge in the sparse field of the grand unifying frameworks to be compared and contrasted against each other. 

The book further presents multiple helpful ideas about methodology in cognitive science. First, it advocates for social cognitivism (combining cognitive neuroscience with social processes of communication) and mechanistic multilevel non-reductionist accounts of the mind. Second, it emphasizes the importance of theoretical coherence as a normative requirement for a good theory. The three-analysis of definitions implemented throughout the book, i.e. the specification of definitions through exemplars, typical features, and explanations (as opposed to necessary and sufficient conditions), helps to avoid the pitfalls of traditional dictionary-type definitions and nicely lines up with the account of concepts outlined by Thagard in chapter 4. 

The manuscript itself is very accessible. The overview of cognitive science in the first chapter can be read as a separate piece, for example, as one of the introductory readings for a lower-year undergraduate course. Each chapter provides references for further exploration and a ‘project’ for the reader. The projects may be too involved for some casual readers or undergraduate students and not involved enough to be thought of as potential research directions for the more advanced audience. This, however, does not take away from the enjoyment of reading the book. Overall, Thagard’s Brain-Mind is an engaging read suitable for a wide range of audiences, from lay-people interested in the mind to the researchers specializing in cognition-related disciplines (including humanities, social, and computer sciences). The book successfully accomplishes the task of outlining a large-scale brain-based framework of human cognition, sketching out an exciting new direction in cognitive science. 

 

References 

Eliasmith, C. (2013). How to build a brain: A neural architecture for biological cognition. Oxford University Press.

Thagard, P. (2019). Brain-Mind: From Neurons to Consciousness and Creativity. Oxford University Press. 

 

Ⓒ 2020 Sofiia Rappe

 

Sofiia Rappe, Neurophilosophy PhD Candidate, Cognition, Value & Behaviour Research Group, Ludwig-Maximilians-Universität München

Categories: Philosophical, Psychology

Keywords: philosophy, psychology