Edited by Gualtiero Piccinini, Neurocognitive Foundations of the Mind is a remarkable volume comprising fourteen chapters that reflect the growing, varied, and rich area of research lying at the intersection between philosophy of mind and neuroscience. The first chapter is Piccinini’s helpful introduction to the volume, where he begins by telling us that the common thread throughout the chapters is that ‘neuroscience makes a profound difference to psychology and the philosophy of mind’ (p. 1). To us—neuroscientifically inclined philosophers—this general claim may not be surprising, but the truth is that there is a sizable group of philosophers of mind who do believe that neuroscience makes no difference to the philosophical study of the mind. Some, still empirically minded, tend to accept results from psychology and cognitive science, but are less bothered by neuroscientific research. Others dismiss empirical sciences altogether. As a result, the claim that neuroscience makes a difference to philosophy of mind is, for some, a claim that ought to be argued for.

Chapter 2, by Luis H. Favela, addresses the challenge of incorporating neuroscience into ecological approaches to cognition. Spearheaded by James J. Gibson, ecological psychology treats perception as direct, continuous with action, and centred on affordances within the organism–environment system, rather than on either subject or object alone. This emphasis has sometimes been taken to leave little room for neuroscience, particularly given Gibson’s own dismissive remarks about the brain. Favela disagrees, arguing that neuroscience can enrich ecological psychology through the neuro-ecological nexus theory (NExT), which takes the dynamic, multiscale organism–environment system as its subject, neuronal populations as its primary physiological units, and topological spaces as its preferred modelling framework. When neural population activity is projected onto low-dimensional manifolds, these latent structures (essentially, a simpler geometric-space hidden within a larger one), or ‘neural modes’, are taken to capture the principal dynamics underlying behaviour. Although compelling, the proposal raises several questions. Why focus on neuronal populations rather than broader physiological systems, given evidence that non-neuronal cells such as astrocytes and oligodendrocytes contribute to cognition? More fundamentally, how does NExT extend beyond perception and action to ‘non-peripheral’ cognition such as mind-wandering or imagination? Sitting motionless on a bus, eyes closed and wearing noise-cancelling headphones, I can vividly imagine my Dungeons and Dragons character escaping a dragon—but what, in such a case, is the relevant organism–environment interaction? Invoking affordances within an imagined environment risks reintroducing representations under a different label, as does the question of why latent variables identified through manifold learning should not themselves count as representations. These are difficult questions, and Favela’s (2024) treatment is necessarily brief; readers interested in a fuller defence should consult his recent book.

The third chapter, by Marcin Miłkowski, tackles the difficult issue of the relationship between functional and mechanistic explanations in neuroscience. In particular, he argues that those who claim that functional and mechanistic explanations are independent of each other and that mechanistic details are, at best, confirmatory of functional theories, face a dilemma: they either have to accept that mechanistic details are merely consistent with the functional account, but not confirmatory; or that if they are confirmatory, then functional explanations are not independent of mechanistic ones. This chapter is careful, well argued, and very informative, but also reveals something that is becoming increasingly evident in the philosophy of neuroscience: that there is a constant negotiation between the computational, algorithmic, and implementational levels of explanation that are typically traced back to Marr (1982). Indeed, it is time for philosophers of neuroscience to go back and read Marr, not the one from 1982, but the one who built computational theories of the cerebellar, archi- and neo-cortices (Marr 1969, 1970), for he was very keen on paying careful attention to the known neuroanatomy in order to constrain computational and algorithmic explanations. It is simply false that Marr thought of those levels as completely independent of each other, and this chapter by Miłkowski helps us to further understand why.

Chapter 4, by Beate Krickel and Mariel K. Goddu, is a must-read for those interested in cognitive ontologies across species. The chapter starts with apt criticisms of ‘top-down’ approaches to cognitive ontologies—that is, those that start from categories derived from cognitive theories and try to identify corresponding brain regions or processes (for more criticisms of the top-down approach, see De Brigard and Gessell 2027). Instead, they suggest that cognitive ontologies should be based on the notion of cognitive homology, which in turn is based on the notion of cognitive ‘character identity mechanism’ or ‘cognitive ChIM’. Roughly, a cognitive character identity mechanism is a causally necessary and non-redundant developmental mechanism that determines a particular cognitive trait, and when one finds the same character identity mechanism in two organisms sharing a phylogenetic story that can be traced back to a common ancestor, then the relevant character identity mechanism are homologous. In their chapter, they carefully distinguish their view from two alternatives in the offing (García 2010; Bergeron 2021) and end up with a concrete example of their proposal: the character identity mechanism for episodic memory. According to their view, hippocampal-dependent episodic memory likely develops through self-locomotion, as infantile amnesia is present in altrician mammals whose babies are carried for a while after being born. Unfortunately, as happens with many of these proposals, one can quickly find counterexamples: several precocial species, such as wildebeest, giraffes, and even horses, have babies that start walking as soon as they are born, and yet some of them share common ancestors with us that are closer in time relative to some altrician mammals that seem to develop infantile amnesia. If, as the authors say, ‘the development of self-locomotion tunes their hippocampal place and grid cells to regularities in the environment’ (p. 77), which in turn is needed for the development of episodic memory, then infantile amnesia should not occur in precocial species—a question that is still much debated (for example, Ramsaran et al. 2019).

Chapters 5 (by Adina Roskies), 6 (by Dimitri Coelho Mollo and Alfredo Vernazzani) and 7 (by Manolo Martinez) could be read almost as a package deal, as they all have to do with the thorny question of neural representation. Roskies’s chapter starts with the traditional distinction between representational content and representational vehicle, and reminds us that although scepticism about the former has risen in contemporary philosophy of neuroscience, the latter tends to still play a significant role as it helps to anchor causal explanations (Egan 2025). However, a critical (perhaps minimal?) criterion for us to make sense of a representational vehicle is that it should be identifiable and re-identifiable, an idea that harks back to Quine’s no entity without identity. Then she goes on to explore whether our best multi-unit recording technology and analytic approaches can offer this much sought-out strategy for re-identification, and to that end explores ground-breaking work by Doris Tsao (for example, Chang and Tsao 2017). Unfortunately, the verdict isn’t positive, for her careful analysis ends up demonstrating that even the best recording techniques and analytic tools available today are not only insufficient to ‘clearly define what subset of neurons acts as a vehicle for a given content’ (p. 91), but also lack the power to tell us when an alleged vehicle is the same as it was before, thus falling short of meeting that minimal criterion.

Coelho Mollo and Vernazzani, in turn, deal with the notion of representational format and explore the shortcomings of using public representational formats (for example, paintings, photographs, sentences) as a frame to talk about representational formats in cognitive psychology and neuroscience. Specifically, they discuss two cases: one having to do with the compositionality debate between symbolists and connectionists, and another pertaining to visual demonstratives. Their extremely careful and useful discussion shows how using public representation frames—linguistic structure in one case and pointers in the other—obscures differences that either hide alternative representational formats that could also fit the target computational role, or suggest instead the existence of cognitive entities that need not be postulated at all. At the end of the chapter, they offer a notion of format that is not based on putatively analogous public representations but that is rather defined in computational terms. Then, immediately after, in chapter 8, Martinez explores the notion of structural representation in terms of complexity management and shows how, from that perspective, they need not be seen as a kind of representation that is essentially distinct from the more traditional Shannon informational or correlational signals. This chapter is a bit complex (no pun intended), in part because each sub-section glosses over a lot of material that could easily require its own book to fully explain, but I think it pairs nicely with recent work on representational pluralism according to which structural and correlational or informational representation are two different kinds of exploitable representations in the brain (Shea 2018).

Chapter 8, by Corey Maley and Oron Shagrir, moves away from the question about representation to the no less thorny issue of computation. When the computational theory of the mind took hold—something that happened steadily from the 1960s to the 1990s or so—it was grounded, according to the authors, on two conceptual assumptions or ‘dogmas’. The logical dogma holds that the notion of computation should be restricted to certain formal operations whose semantic contents are amenable to be regimented by the resources of classical logic and computability theory; Turing machines are the paradigmatic example here. Likewise, the architectural dogma holds that only certain systems with the right syntactic structure to support such formal operations can instantiate computations. However, as the authors demonstrate (in my opinion, quite convincingly), the more we know about neuroscience, the less likely both dogmas seem to be. Different accounts of computation are also explored, and they end up suggesting that before coming up with a unified notion of neural computation—that may or may not exist—it is best to start by abandoning the dogmas and instead focus our attention on better understanding the physical and causal details of the mechanisms that are likely to implement them.

Chapter 9 must be one of my favourites. It is on inference, which is a term that has been widely used in computational or functionalist accounts of the mind in philosophy. Indeed, recent work at the intersection of philosophy of mind and cognitive science has tried to rescue a notion of inference by assuming that it is conscious, contrasted with a-rational association, involves language-like symbolic representations, and comprises an indirect relationship between the mind and the world. But Laukaityte and Colombo argue—quite successfully, I think—that these assumptions are completely out of touch with actual scientific practice in cognitive psychology and neuroscience, and that, instead, a different, broader account of inference should be adopted in order to make sense of this notion in the sciences of the mind. Contra the traditional account of inference, and largely inspired by the work of Siegel (2017), the authors argue in favour of a view of inference ‘understood as a form of rationally evaluable transition from some inputs and current representations to some subsequent or output representation’ (p. 184) that need not be conscious, need not involve language-like symbolic representation, explores a more direct ecological interaction between mind and world, and that has a more nuanced relationship to other associative transitions, all while being faithful to how cognitive psychologists and neuroscientists use the term.

Next, Millière and Buckner, in chapter 10, explore both behaviourist and interventionist methods for interpreting deep neural networks. They begin by discussing two non-interventionist methods—benchmarking and probing—and explore their many interpretative limitations. Then they move on to discuss how interventionist methods, such as activation patching (or causal tracing), offer better insights into the causal structure of artificial neural networks. The chapter goes on to demonstrate how, paired with mechanistic interpretations, the insights afforded by interventionist methods offer increasingly promising strategies to interpret the workings of deep neural networks. By the end of the chapter, the reader leaves with the hopeful thought that, contrary to the popularized belief that artificial neural networks are uninterpretable ‘black boxes’, interventionist methods are very promising strategies to help us understand their structure—and this is notwithstanding some of their limitations.

Chapter 11 and chapter 12 could easily be read in tandem, as both offer strong reasons to reject the hypothesis of the language of thought. The first one, authored by Fabrizio Calzavarini, explores the influence of neuroscientific research on what we know about how we process word meanings. The author argues that evidence from cognitive neuroscience of language processing supports two empirical generalizations. The first one is the sensorimotor hypothesis, that is, that ‘the cognitive representation of word meaning is partly distributed in perceptual and motor systems’ (p. 228). The second generalization is an amodal direction: ‘The cognitive representation of word meaning is partly distributed in a network of high-order associative, amodal neural systems, or amodal “hubs”’ (p. 230). As a result, most contemporary neuroscientists endorse a sort of hybrid approach to understanding word meaning, involving both modal and amodal components. The nature of this semantic neuroarchitecture—Calzavarini argues—should put an end to the idea that we need an additional system of propositional representations, such as Fodor’s ‘mentalese’, to understand language. He goes on to show, very much in the spirit of the whole volume, that neuroscientific findings can impact longstanding theories in the philosophy of mind.

A similar lesson is explored in the following chapter. Here, Wade Munroe argues that the empirical evidence coming from research in neuroscience overwhelmingly supports the claim that ‘inner speech modulates ongoing mnemonic, attentional, and sensorimotor processing and, in certain cases, serves as the necessary vehicle for conceptual processing’ (p. 246). Moreover, and consistent with the claims from the previous chapter, the author also argues that neuroscientific evidence strongly suggests that inner speech is embodied and grounded in sensorimotor, affective, and interoceptive systems. Taken together, both claims cast serious doubt on the idea that we need a language of thought at all. Insofar as there is any explanatory need for which a mentalese was supposed to be required, Munroe argues, our natural language is perfectly suited to fulfil that role. And insofar as our natural language speech is embodied and grounded in modality-specific processes, then there is further counterevidence that anything resembling the classical language of thought supports any of our cognitive life.

In keeping with the general theme of the book—namely, that neuroscientific research does impact philosophy of mind—Wayne Wu takes on the much-discussed notion of ‘intention’ in chapter 13. He starts off with what may be one of the earliest uses of the notion of working ‘memory’, by Miller et al. (1960), in which the connection with intention is evident, as its stated function is ‘to store an agent’s plan so that “it can be remembered while its being executed”’ (pp. 275–76). He then goes on to explore more recent work in the cognitive psychology and neuroscience of working memory, and argues that the empirical evidence favours two claims about the notion of working memory and its relation to intentions: first, that intentions are not static states but, rather, are dynamic and, second, that they are at least partially constituted by sensory information. This is a fun chapter to read because the notion of intention isn’t typically one of those concepts in the philosophy of mind one would think neuroscience could make an obvious difference for. And yet it does.

Finally, in chapter 14, Marco Viola and Fausto Caruana explore recent debates on the cognitive ontology of emotion. In particular, they are concerned with recent disagreements on how to interpret neuroscientific data pertaining to Ekman’s (1992) model of the six basic emotions. For some, the data clearly support the model; while for others, the data clearly debunk it. The authors argue that at the basis of the disagreement lies a more fundamental difference in how neural data should impact cognitive ontologies. Following Anderson’s (2015) typology, the idea is that depending on whether one is a conservative, a radical, or a reformist, one’s views on how neural data should impact cognitive ontology are going to vary. But that does not mean that all views are equally valid. There are a number of reasons to prefer conservative, radical, or reformist approaches in light of the neural evidence. And, as they carefully demonstrate in their chapter, there are many reasons to reject the claim that neural data support Ekman’s six-emotions model, and much reason to believe that some kind of reformist approach is needed.

In sum, Piccinini has put together a fantastic volume. The fourteen chapters that comprise it cover a wide range of topics and explore different ways that neuroscientific research can impact foundational issues in the philosophy of mind and psychology. I was already sold on the idea that the philosophical study of the mind could not be conducted in total ignorance of scientific developments in cognitive psychology and neuroscience. I believe the chapters of this book could help bring some remaining sceptics to the bright side.

Felipe De Brigard
Duke University
felipe.debrigard@duke.edu

References

Anderson, M. L. (2015). ‘Mining the Brain for a New Taxonomy of the Mind’, Philosophy Compass, 10, pp. 68–77.

Bergeron, V. (2021). ‘Carving the Mind at Its Homologous Joints’, Biology and Philosophy, 36, available at .

Chang, L. and Tsao, D. Y. (2017). ‘The Code for Facial Identity in the Primate Brain’, Cell, 169, pp. 1013­–28.

De Brigard, F. and Gessell, B. (2027). ‘The Mirage of Big-Data Phrenology’, British Journal for the Philosophy of Science, 78, available at .

Egan, F. (2025). Deflating Mental Representation, MIT Press.

Ekman, P. (1992). ‘An Argument for Basic Emotions’, Cognition and Emotion, 6, pp. 169–200.

Favela, L. H. (2024). The Ecological Brain: Unifying the Sciences of Brain, Body, and Environment, Taylor and Francis.

García, C. L. (2010). ‘Functional Homology and Functional Variation in Evolutionary Cognitive Science’, Biological Theory, 5, pp. 124­–35.

Marr, D. (1969). ‘A Theory of Cerebellar Cortex’, The Journal of Physiology, 202, pp. 437–70.

Marr, D. (1970). ‘A Theory for Cerebral Neocortex’, Proceedings of the Royal Society of London B, 176, pp. 161–234.

Marr, D. (1982). Vision: A Computational Investigation into the Human Representation and Processing of Visual Information, MIT Press.

Miller, G. A., Galanter, E. and Pribram, K. H. (1960). Plans and the Structure of Behavior, Holt, Rinehart and Winston.

Ramsaran, A. I., Schlichting, M. L. and Frankland, P. W. (2019). ‘The Ontogeny of Memory Persistence and Specificity’, Developmental Cognitive Neuroscience, 36, available at .

Shea, N. (2018). Representation in Cognitive Science, Oxford University Press.

Siegel, S. (2017). The Rationality of Perception, Oxford University Press.