Chapter 1
Introduction
Someone looking at cognitive science from a distance today would likely see a fragmented field. When trying to understand its theoretical structure, they might notice a rough division: on one side are computationalist approaches inspired by computer science, and on the other are embodied cognition frameworks rooted in biology. Some schools of thought would seem to fall somewhere in between, attempting to integrate both perspectives, while others appear to be in sharp contrast, positioned far apart from one another.
Many disputes between different schools of thought revolve around if a theoretical construct, an entity or a posit is real or not. For example, philosophers and cognitive scientists have been discussing if mental or neural representations are real (Fodor & Pylyshyn, 1981; Thompson & Piccinini, 2018; Piccinini, 2022; Ramsey, 2007; Şahin & Deveci, 2023; Van Gelder, 1995). If they are real, what is the nature of them? Is representational content a real entity (Schulte, 2023)? Is mind or/and brain really a computer (Chalmers, 2010; Richards et al. 2019)? Do our mind/brain really encode and store the things outside of it (Bickhard & Richie, 1983; Brette, 2019)? Are really there modules in our mind (Carruthers, 2006; Fodor, 1983)? How much of computation assumed or proposed in our models is actually taking place? For example, is there really activation function(s) in brain? Do our neurons really calculate probabilities or really take derivatives (Haefner et al. 2024; Marblestone, Wayne & Kording, 2016; Son et al. 2020; Richards et al. 2019)? Do all organisms really have agency or only animals, or humans (Moreno & Etxeberria, 2005; Walsh, 2015)? Is autopoiesis real (Maturana & Varela, 1973/1980; Varela, 1979)? Is free will real (Mitchell, 2023; Sapolsky, 2023)? Is consciousness real (Dennett, 1991; Frankish, 2016; Seth, 2021)? These are the questions of the sort that distinguish varying schools of thought, different frameworks, different ontologies. Cognitive scientists and philosophers can be said to be ‘from’ such and such a school of thought or framework depending on the answers they give to such questions. People usually choose some framework over another depending on what they take to be real and not.
Among these questions, one stands out as the most significant and influential: whether mental representations are real, and if they are, what exactly their nature is. Since Brentano, representations have been regarded as intrinsic to the nature of mentality, what distinguishes the mental from the non-mental. Given that cognitive science claims to study the mind, the answers offered to this question have shaped the broader assumptions and methods of various schools of thought, often leading to deep and fundamental differences.
Recently, though, Wheeler (2017) claims that the real dispute is not between representationalism versus anti-representationalism but rather between internalism and externalism, as externalism taken as the thesis that “the material realizers (or vehicles) of cognitive states and processes are distributed over brain, body and world, and so are located partly outside the skin.” Expanding on Wheeler (2017), Villalobos and Silverman (2018) identifies the root cause here as systemic asymmetry. In their view, even in the anti-representationalist approaches like radical enactivism (Hutto & Myin, 2012), it has been the common practice to propose something as ‘mark of the mental/cognitive’ to distinguish it from the non-cognitive, i.e. to answer the question of what constitutes cognition. Only then, for the common practice, we could know what the relevant phenomena are and how cognition produces them whether it is embodied or computational. However, Villalobos and Silverman claims, this only resulted in the systemic asymmetry in which cognitively constitutive parts carry the explanatory burden and become primary against the non-constitutive causes, hence opens up a space for possible internalist interpretations where only the boundary shifts but the division itself remains. That undermines the coherence of embodied approaches and make them potentially vulnerable to pitfalls of other internalist accounts which they are critique of. To avoid this, a secure, a complete externalism is needed. Even though anti-representationalism is a prerequisite for such an account, it is not adequate by itself.
In this thesis, following the sequence of the discussion outlined above, I will address the question of whether it is meaningful to debate the “reality” of a given theoretical entity in science and, more specifically, in cognitive science. In this context, the second chapter will be dedicated to the debate between scientific realism and anti-realism. I will propose a form of radical anti-realism, one that applies not only to scientific theories but also to everyday or folk ideas about what is real. Of course, rejecting realism comes with challenges. If we give up the idea that science tells us what is truly real, how do we evaluate theories? How do we decide which concepts are worth using? To address these questions, I will turn to the New Mechanistic philosophy of science.
In the third chapter, I will examine how the new mechanist approach conceptualizes science. In doing so, I will distinguish between its broad and narrow interpretations. According to the broad interpretation, New Mechanism aims to account for science as a whole and provides a normative framework for how science ought to be conducted. In contrast, the narrow interpretation treats it as one of several alternative approaches, more suitable for specific domains of science rather than universally applicable. This thesis adopts a neutral stance with respect to this distinction. Following this, I will address the question of whether New Mechanism is compatible with anti-realism. Although many mechanist philosophers are realists, I will argue that no such commitment necessarily follows from the core assumptions of the new mechanist framework. On this view, what matters is not whether mechanisms are real in some metaphysical sense, but whether they enable us to give explanations that are both conceptually and empirically coherent.
In the fourth chapter, I will examine prevailing practices in cognitive science. In doing so, I will focus on functionalism and argue that, due to the influence of the ideas of the autonomy of special sciences and multiple realizability, nearly all existing schools of thought in cognitive science, despite their many differences, ultimately diverge in terms of their commitments to the metaphysics of mind. I will offer a critique of both the autonomy of special sciences and multiple realizability, contending that these views lack sufficient grounding. Building on the discussions developed in the second and third chapters, I will then introduce the concepts of Conceptual Coherence (CC) and Naturalistic Coherence (NC) as the core proposal of this thesis. Drawing from these, I will propose the Naturalistic Coherence Criterion (NCC) as a normative standard to be employed both in the introduction of new concepts and in the evaluation of existing conceptual frameworks within naturalistic sciences. I will argue that, when combined with the commitments of New Mechanism, this criterion leads us toward an approach to cognitive science that does not center metaphysical commitments about the mind, but instead treats them as secondary and a posteriori concerns. To support this claim, I will draw on both a historical case from developmental biology and a thought experiment.
In the final chapter, I will clarify what this thesis does and does not claim, in order to avoid potential misunderstandings. I will briefly highlight how the approach developed here diverges from existing schools of thought within cognitive science, and I will outline possible directions for further research.
With this thesis, I aim to outline both the necessity and the contours of what may be called a radically externalist cognitive science. By this, I refer to an approach that makes no intrinsic attribution and reference to the mind, remains entirely agnostic with respect to the metaphysics of mind, and is thus “externalist” in this broader sense. Although at such a radical level of externalism there may be no room left for any kind of “internalism” to the extent that the very distinction between internal and external collapses, though the label remains, in my view, a convenient and useful one. In this context, it may be helpful to provide an example of an internalism–externalism spectrum in order to clarify the position of the view defended here. On one end of this spectrum, we find phenomenology as the most internalist approach, while on the other end lies the radically externalist position. This is not because phenomenology is committed to locating the mind “in the head” or treating it as something exclusively tied to the brain. In fact, phenomenology explicitly avoids such commitments. Rather, its position on the internalist pole stems from its methodological orientation. Phenomenology centers the mental, intentionality, and first-person experience as the primary focus of analysis, and resists reducing these to spatiotemporal processes. This might be called the intensionalist strategy of phenomenology. By contrast, what makes radical externalism “radical” is not a straightforward reduction of the mental to spatiotemporal processes either, it also avoids grounding the mental in such terms. However, it does so by adopting an extensionalist methodology that shifts the focus entirely away from mental phenomena. Accordingly, when I refer to radical externalism, I am also taking into account a methodological criterion — intensional–extensional distinction — for evaluating and positioning views along the internalism–externalism spectrum. This axis distinguishes approaches not merely by where they locate the mind, but by how they treat the status and explanatory role of mental concepts within scientific inquiry.