Felipe De Brigard and Walter Sinnott-Armstrong Edit Neuroscience and Philosophy II

What happens to philosophical theories of unified consciousness, voluntary agency, and sensory representation when every chapter must be co-written by an empirical neuroscientist and an analytic philosopher? Published by The MIT Press in late 2026 (Neuroscience and Philosophy II, ISBN 9780262055901, 464 pages), editors Felipe De Brigard and Walter Sinnott-Armstrong present 11 original collaborative chapters developed through the Summer Seminars in Neuroscience and Philosophy (SSNAP) at Duke University.

Serving as the direct sequel to their 2022 MIT Press volume Neuroscience and Philosophy, the 2026 collection avoids surface-level interdisciplinary summaries. Instead, each chapter targets a specific empirical dissociation where laboratory data and philosophical concepts collide. For readers tracking how scientists race to define and test consciousness in 2026, the book’s sections on split-brain agency, aphantasia, and the geometry of neural representations provide stringent empirical boundary conditions that any computational theory of mind must satisfy.

Structure of the Four Thematic Sections Across 11 Chapters

De Brigard and Sinnott-Armstrong organize the 464 page volume into four thematic parts, moving from normative moral psychology to the foundational architecture of conscious representation.

Split-Brain Agency and the Unity of the Conscious Workspace

In Part II, the volume revisits one of the most fiercely debated clinical cases in consciousness science, namely corpus callosotomy (split-brain) patients whose cerebral hemispheres have been surgically disconnected to control intractable epilepsy. From Roger Sperry and Michael Gazzaniga’s classic lateralized tachistoscope experiments to modern high-density neuroimaging by Yair Pinto and colleagues, split-brain studies have forced philosophers from Derek Parfit and Thomas Nagel to Tim Bayne and Elizabeth Schechter to confront what makes a conscious mind “one” rather than “two.”

The collaborative analysis in Neuroscience and Philosophy II moves past the binary stalemate between the “two-minds” view and the “single-subject” view by dissecting functional versus phenomenal unity across task regimes. While subcortical pathways through the superior colliculus, brainstem, and intact anterior or posterior commissures allow split-brain patients to coordinate bimanual locomotion and orienting attention as a single behavioral agent in everyday life, lateralized perceptual and metacognitive tasks expose independent attentional workspaces and distinct perceptual confidence thresholds in each hemisphere.

For computational theories of consciousness, this clinical dissociation proves that behavioral unity at the motor output layer does not guarantee a single unified global workspace. A system can maintain coherent external action selection through low-bandwidth subcortical or routing bottlenecks while running two non-communicating perceptual representations internally.

Aphantasia and Cognition Without Sensory Phenomenology

Part III turns to aphantasia, the condition coined by Adam Zeman in 2015 in which individuals report a complete absence of voluntary visual mental imagery (the “mind’s eye”) while maintaining intact visual perception, semantic memory, and spatial reasoning. Building on Felipe De Brigard’s research into episodic memory and counterfactual simulation, the volume’s chapters on mental imagery and perspective-taking ask a central question for Higher-Order and Global Workspace theories.

If an aphantasic individual can remember their childhood bedroom, rotate three-dimensional shapes accurately, and simulate future social interactions without generating conscious visual imagery inside early visual cortex (V1 to V4), what role does sensory phenomenology actually play in high-level cognition? The authors show that aphantasia separates structural spatial and relational simulation (mediated by hippocampal, parietal, and prefrontal networks) from top-down sensory reactivation (feedback projections into early visual cortices). This finding directly parallels clinical dissociations studied in Jonathan Huntley and Pedro Mediano’s multimodal investigation of perceptual awareness in Alzheimer’s disease and contrasts with the vivid involuntary hallucinations analyzed in Baland Jalal’s The Phantom Mind on sleep paralysis.

Representational Geometry and Neural Computation

In Part IV, the contributors tackle the foundational question of neural representation. Neuroscientists routinely speak of a population of neurons in macaque inferotemporal cortex or human prefrontal cortex “representing” a face, a value, or an intention because a linear decoder can read out the stimulus from spike counts. Philosophers of mind object that mere mutual information or linear decodability is far too cheap, since a thermometer covaries with room temperature without representing temperature to a subject.

The chapters in Part IV bridge this divide by analyzing representational geometry (such as Representational Similarity Analysis, RSA, and manifold topology). Rather than looking at single cells as isolated feature detectors, the authors argue that a neural population functions as a genuine representation when the relational distances and topological neighborhoods among its activation states are actively exploited by downstream consumer circuits to guide flexible, error-correcting action across novel contexts. Combined with the physical energy constraints explored in the SFI Press volume The Energetics of Computing in Life and Machines, this structural criterion gives mechanistic interpretability researchers a principled test for separating spurious linear probe hits from functional representations.

Comparison to The Consciousness AI

In The Consciousness AI, our multi-layer architecture separates early sensory spike encoding from recurrent workspace broadcast and self-model tracking. The analyses in Neuroscience and Philosophy II directly inform two architectural questions in our codebase.

First, the split-brain agency chapters demonstrate why measuring overall behavioral coherence is insufficient to verify global workspace integration, so internal metrics must measure cross-module mutual information and phase locking directly across the workspace bottleneck. Second, the Part IV criteria for relational representational geometry align with how our evaluation scripts test whether population state manifolds preserve topological neighborhood structure across recurrent cycles rather than relying on post-hoc linear classifiers alone.

Editorial Assessment

Because Neuroscience and Philosophy II is an edited anthology of 11 specialized pairs of authors rather than a single-voice monograph, it does not advance one unified theory of consciousness. Instead, its strength is methodological. By forcing philosophical definitions of agency, mental imagery, and representation to answer to specific clinical and neurophysiological dissociations, De Brigard and Sinnott-Armstrong provide an indispensable reference volume for anyone designing empirical indicators of awareness across biological and synthetic minds.

Researchers covered here

• FB Felipe De BrigardDuke University, Department of Philosophy and Center for Cognitive NeuroscienceMemory and Mental Time Travel (2023), episodic counterfactual thinking, co-director of Duke SSNAP, co-editor of Neuroscience and Philosophy I and II (MIT Press, 2022, 2026) 1 article
• WS Walter Sinnott-ArmstrongDuke University, Department of Philosophy and Kenan Institute for EthicsMoral psychology, neuroethics, free will and responsibility, co-editor of Neuroscience and Philosophy I and II (MIT Press, 2022, 2026), AI morality and agency 1 article