Book Symposium on The Brain Abstracted in Philosophy and the Mind Sciences
Philipp Haueis, Department of Philosophy and Institute for Studies of Science (ISOS), Bielefeld University, Germany
The journal Philosophy of the Mind Sciences has just published a book symposium (LINK) of The Brain Abstracted: Simplification in the History and Philosophy of Neuroscience, by Mazviita Chirimuuta. The book has received both the 2024 Nayef Al-Rodhan International Prize in Transdisciplinary Philosophy and the 2025 Lakatos Award for Philosophy of Science and has been discussed in previous posts on the Brains Blog.
The book offers a novel and groundbreaking philosophical study of the brain sciences. Starting from the assumption that the brain is massively heterogeneous and dynamically changing, Chirimuuta argues that neuroscientific practice always constructs simplifications of the brain. This is obviously true for mathematical models, but also holds for experimental practice, in which neuroscientists manipulate and intervene into reduced preparations such as model organisms. The book combines a form of haptic realism, which holds that we learn about the brain by manipulating simplified such material and mathematical models, with formal idealism, warning us that the patterns we discover in the models should not be attributed an ontological reality.
The commentaries in the new special issue fall into three thematic clusters. The first cluster focuses on questions about scientific realism raised by The Brain Abstracted. In her commentary Zoe Drayson raises challenges for two of Chirimuuta’s arguments against realism. The first argument is that the brain is too complex, dynamic and irregular for our simplified, regularity-focused models of it to be true. Against this Drayson argues that that the “scientific realist can accommodate highly convoluted theories and irregular models as long as they provide us with knowledge of the mind-independent world.” The second argument against realism is that neuroscientific knowledge is the product of active interaction with the brain as an object of research, a process which is bound up with practical goals. Drayson comments that this may be true of the process of acquiring knowledge (context of discovery), but does not speak against a realist interpretation of its epistemic status (context of justification).
The commentary by Alexandros Constantinou and colleagues points to a challenge in the notion of ideal pattern in Chirimuuta’s haptic realism. Ideal patterns, such as the “simple cell” in visual cortex are simplified analogies to brain processes constructed by scientists active interaction with brains in the lab. We should be agnostic about whether they represent anything in reality itself. Constantinou et al. challenge this metaphysical neutrality: haptic realism employs causal notions like interaction with the brain, which require metaphysical interpretation: either they are mind-dependent or they are not. This highlights a tension between the claim that brains (literally) have neurons in one sense, but that neurons are ideal patterns in another.
The second cluster concerns questions about computation in biological and artificial systems. Chirimuuta argues that a literal interpretation of computational models poses two metaphysical issues: how to implement computational descriptions in brains in a non-trivial manner, and whether computation is multiply realizable in biological and artificial systems. Danielle Williams argues for a version of computationalism (the claim that part of cognition is computation) that avoids the two challenges. Mechanistic or analogical accounts claim that cognitive systems compute without the computational descriptions of models being literally implemented in the brain. One may also accept multiple realizability but reject that AI computes and realizes cognition in a strong sense.
The Brain Abstracted also defends biologism – the claim that only biological systems have minds. Zed Adams and Jake Browning explicate two ways of arguing for this position. The too messy argument points out that biological complexity is required for minds. For example: subneuronal processes, such as neurochemical modulation, make a difference to information processing, attention, and action guidance. These processes are intertwined in ways that makes it impossible to emulate them in digital systems. The too neat strategy points out that the energy efficiency of biological systems massively outperforms our best artificial computers. The authors ask which of the strategies does Chirimuuta endorse, and whether they are compatible with one another.
The third cluster of commentaries explores practical implications of The Brain Abstracted for practices of neuroscience and adjacent fields. The commentary by Nedah Nemati asks what distinguishes productive and unproductive simplifications, and suggests that answers will depend on the experimental structure and psychological affordances of the research project at hand. Nemati’s constructive proposal is that material properties make simplifications like Skinner boxes or data processing tools productive, since they allow neuroscientists to proceed efficiently.
The commentary by Ida Momennejad argues that Chirimuuta underestimates the entanglement of philosophy and science. Paradigmatic topics of philosophy, such as personhood, are equally studied by neuroscience, including Momennejad’s own work on memory and intersubjectivity. Conversely ideas about control in neuroscience stem from philosophers like Hobbes and Bentham, whose ideas shaped science over the longue durée of history of Western thought. Computational neuroscience thus risks ontological reversal, where optimization-based models of intelligence are taken as biological facts rather than simplifications of the actual processes the brain as a homeostatic dynamic system is engaged in.
This risk of “misplaced concreteness” is addressed by Nicolás Hinrichs, who applies critical ideas of The Brain Abstracted to the practice of hyperscanning in psychotherapy. Hyperscanning involves simultaneous recording of a patient’s and therapist’s brain activity, and synchrony is taken as indicator for therapeutic success. Hinrichs warns other practitioners in this field to not mistake a seemingly simple metric as answer, while avoiding harder interpretative questions, such as “what happened during the synchronous event in the therapeutic session?” His positive proposal proposes several guardrails to safeguard against the fallacy of misplaced concreteness.
Together, the commentaries demonstrate that The Brain Abstracted not only offers fresh perspectives in philosophy of the mind sciences, but also provides inspiration of how to change and improve scientific and therapeutic practices surrounding the brain. For the full articles and Mazviita Chirimuuta’s response to the commentaries, please go to https://philosophymindscience.org/index.php/phimisci/issue/view/385. All texts of Philosophy of the Mind Sciences are diamond open access.
Dimensional Logic and the “Abstracted Brain”
In this commentary I want to sketch a perspective I call dimensional logic and relate it to the central theses of The Brain Abstracted and the accompanying commentaries. My aim is not to refute Chirimuuta’s approach but to reconfigure it: away from the alternative realism vs. anti-realism, toward a multi-dimensional framework in which models, brains, abstractions, and practices are understood as points or trajectories in a shared space of description.
1. Starting Point: From the Brain as Object to the Brain as Descriptive Space
Chirimuuta emphasizes the heterogeneity and complexity of the brain and argues that every neuroscientific representation necessarily abstracts. The brain is too “messy” to be captured in simple, smooth models. I share this diagnosis entirely.
My proposal is to formalize this finding logically. Rather than treating the brain as an object we represent more or less adequately, we should understand it as a space of descriptive dimensions in which biological processes, cognitive states, mathematical models, experimental practices, and theoretical concepts move as points or trajectories. Crucially, this space is not ontologically plural. There is one ontology, namely the physical-biological processes that constitute the brain. What varies are the languages in which we describe these processes and the levels of abstraction at which we operate.
The question then is no longer “Is this model realistic?” but rather “In which region of this descriptive space does this model move, and which dimensions does it leave explicitly or implicitly aside?” This shifts the focus from a logic of representation to a logic of descriptive space.
2. Dimensional Logic: Sketch of a Framework
By dimensional logic I mean a form of thinking in which concepts, models, and systems are treated not as one-dimensional entities but as positions in a space of descriptive dimensions. These dimensions are epistemic and semantic in nature; they are not ontological layers. They differ according to the degree of reflexivity with which a system or model thematizes its own presuppositions.
In my developed framework, four such dimensions can be distinguished: a reactive dimension, on which systems respond to inputs without representing their own mode of operation; a systematic dimension, on which regularities are recognized and modeled; a reflexive dimension, on which models thematize their own abstraction decisions; and a contextual dimension, on which the embedding of models in practices, institutions, and interests becomes explicit.
Neuroscientific models occupy different positions in this space. A spiking network model operates primarily on the systematic dimension. A philosophical critique such as Chirimuuta’s moves on the reflexive dimension. The question of which social and practical interests inform a model belongs to the contextual dimension. Abstraction in this framework is not distortion but structured movement through the space.
3. Haptic Realism as Navigation in Descriptive Space
Chirimuuta’s concept of haptic realism, the idea that we learn about the brain through active interaction with models, can be precisely reconstructed in dimensional logic.
Rather than saying “we feel our way toward reality,” I would say: “we move through a descriptive space by manipulating models, varying parameters, changing levels of abstraction, and shifting between material configurations.” Haptic realism then describes not merely an epistemic attitude but a trajectory, understood here strictly as directed movement through descriptive levels, not through ontologically distinct regions. Scientists traverse the space of possibilities by moving from biological preparations to computer simulations, from single-cell recordings to network analyses, from animal models to clinical studies, from qualitative descriptions to formal theories. These movements are not arbitrary but logically structurable, and this is precisely where dimensional logic intervenes.
4. Formal Idealism and the Confusion of Descriptive Levels
Chirimuuta’s formal idealism warns against reifying patterns found in models as properties of the brain itself. Models are tools, not ontological maps.
In dimensional language this means: the abstraction dimension must not be confused with the ontological structure of the brain itself. A mathematical model that describes a neural population as a smooth function is a projection of a high-dimensional biological state into a low-dimensional formal space. The danger Chirimuuta identifies, “misplaced concreteness”, is in my terminology a confusion of descriptive levels: one treats a projection as though it were the object itself.
This is not an ontological problem but a semantic one. The brain exists at one level. Models exist at another level, namely that of descriptive language. Dimensional logic makes this separation explicit and compels us, for every model, to ask: at which descriptive level is this model defined? Which dimensions are silently ignored? Which trajectories through the space thereby become invisible?
5. Realism as a Debate about Descriptive Levels, Not as a Position in Space
A central point of contention in the symposium concerns whether Chirimuuta’s position is too anti-realist. From the standpoint of dimensional logic, this debate is itself undercomplex, because it conflates an ontological question with an epistemic one.
The ontological question is: does a brain with causal structures exist independently of our descriptions? Within the framework of ontological monism, I answer this clearly in the affirmative. There is a domain of physical-biological processes, and it does not depend on our descriptions. This is not a negotiable coordinate but a foundational commitment of the framework.
The epistemic question is: do our models represent these processes, or are they tools that allow us to interact with them? This question concerns not ontology but the relation between descriptive language and object. Chirimuuta’s instrumentalism addresses this second question, not the first. To label it “anti-realism” smuggles in the assumption that the choice of descriptive language constitutes an ontological commitment, and that is the actual error driving the debate.
Dimensional logic allows us to identify these apparent tensions not as contradictions but as confusions of levels: those who criticize Chirimuuta as an anti-realist operate on the systematic dimension and thereby miss the reflexive dimension on which her actual argument is situated.
6. Computationalism and Biologism as Descriptive Strategies, Not Ontologies
Another strand of the symposium concerns the tension between computationalism and biologism. Are mental states primarily to be understood computationally, or is the biological irreducible?
In a one-dimensional framework these positions appear as opposites. In dimensional logic they are descriptive strategies operating on different axes, which do not exclude each other because they do not answer the same question. Computationalism operates on the abstraction dimension: it asks which functional regularities can be identified in a system, independently of its substrate. Biologism operates on the causal dimension: it asks which concrete physical-biological processes are constitutive for certain properties.
A biological brain is therefore not an argument for biologism against computationalism, but a system that occupies a particular position on both axes. The ontologically relevant question is not “Is cognition computation?” but “Which biological processes are causally constitutive for what we epistemically describe as cognition?” This is an empirical and causal question, not a definitional one about substrates.
Chirimuuta’s biologism can be read, within this framework, as the claim that the causally constitutive processes on the biological axis require a minimal complexity threshold that digital systems do not reach. That is an empirically testable thesis, not an ontological demarcation.
7. Scientific Practice as Trajectories in Descriptive Space
The contributions in the symposium dealing with experimental practices, hyperscanning, therapy, and model construction can be understood in my framework as trajectories, directed movements through descriptive levels rather than through ontologically distinct domains.
An experiment moving from animal models to clinical studies describes movement from a region of high controllability and low ecological validity to a region with the reverse profile. A hyperscanning setup measuring inter-brain synchrony moves from the individual neural dimension toward the interactive social dimension, but remains on a very narrow abstraction axis, correlation metrics, and neglects the reflexive and contextual dimensions entirely. The danger Hinrichs identifies, premature ontological conclusions drawn from simple metrics, is, in dimensional language, an overinterpretation of local trajectories: a movement through a very narrow region of the descriptive space is taken to reveal the global structure of the phenomenon.
Dimensional logic demands epistemic modesty here: every trajectory is local, every projection partial, every metric dimensionally constrained.
8. What Dimensional Logic Adds to the Symposium
I see the role of dimensional logic not as replacing Chirimuuta’s approach but as framing and sharpening it. It provides a meta-language for reconstructing realism, abstraction, modeling, and practice not as opposites but as positions in a descriptive space. It supplies a structure in which haptic realism can be understood as navigation through that space. It offers an instrument for systematically identifying confusions between descriptive levels and ontological claims. And it makes it possible to reconstruct debates such as computationalism vs. biologism as descriptive strategies operating on different, non-exclusive axes rather than as competing ontologies.
In this sense one can say in conclusion: The Brain Abstracted shows that we cannot understand the brain without abstraction. Dimensional logic shows that without an explicit space of descriptive levels we cannot understand what we are actually doing in that process, or which ontological assumptions we are silently carrying along.
9. Outlook: From Metaphor to Formal Structure
The dimensional logic sketched here is more than a metaphor. It can be formalized: as a vector space model in which models are vectors and practices are operators; as a topology in which regions of the space possess particular epistemic profiles; or as a dynamical system modeling trajectories, attractors, and transitions between descriptive regimes. Such a formalization could classify neuroscientific model classes as positions in the space, describe the translatability between models as projections, and make the limits of particular descriptive strategies visible not only qualitatively but structurally.
In this sense I understand dimensional logic as a proposal to not merely accept the abstraction Chirimuuta diagnoses, but to map it systematically, without losing sight of the ontological unity of the object in question.
Stegemann, W. (2026). Dimensional Logic as a Formal Inference System. Zenodo. https://doi.org/10.5281/zenodo.19101525
Sorry, that was the wrong link; here is the correct one: Stegemann, W. (2026). Dimensional Logic as a Formal Inference System. Zenodo. https://doi.org/10.5281/zenodo.19225035