Chapter 5
Conclusion and Further Directions
In this thesis, I tried to adopt a synthetic approach. By this, I mean an effort to integrate philosophical stances that the researcher independently advocate, allowing these views to constrain the space of acceptable answers to a given problem and, as a result, seeking and, if possible, presenting a holistic and unified solution. For instance, at the outset, I observed that theoretical debates in cognitive science often focus on the reality of various entities, which led me to engage with the debate on scientific realism. This discussion, for independent reasons, pushed us toward a scientific and metaphysical anti-realist position. Similarly, I outlined the new mechanistic philosophy of science, which I endorse independently of the realism debate. I concluded that there is no categorical incompatibility between the two. I included that discussion to provide a coherent and unified meta-constraint for the philosophical stance we adopt toward a particular scientific discipline, here cognitive science in particular. In light of these, I considered the relevant implications for cognitive science. In doing so, I gave attention to respect the meta-constraints established through our previous discussion. As a result, I have attempted to present as comprehensive and consistent a framework as possible.
In this context, I argued that anti-realism is inevitable when considering any non-trivial definition of mind-independent reality and/or truth. I contended that the arguments put forward in favor of realism are inadequate to justify it. I critiqued Putnam’s early notion of natural kind semantics and pointed out that approaches like entity realism, structural realism, or explanationist realism, which often appeal to a certain kind of historical continuity, can only merely highlight a continuation of historical consensus and tell us nothing substantive about mind-independence. Similarly, I argued that views like perspectival realism, which reject the objectivist assumptions of realism, are ad hoc, as it remains unclear what purpose a non-objectivist conceptualization of reality would serve beyond attempting to resolve the realism debate itself. In other words, I suggested that abandoning the criterion of mind-independence requires a non-ad-hoc justification independent of this debate, and that such a move trivializes realism. In contrast, I emphasized that we cannot escape Quinian holism and socio-historical and cultural theses regarding theory change in science, which remain relevant. Additionally, unlike some philosophers of science, I argued that the distinction between observable and unobservable phenomena is untenable. Furthermore, I maintained that the anti-realist arguments in question also apply to folk ontologies, making it arbitrary to adopt a scientific anti-realist stance while refraining from metaphysical anti-realism. Consequently, I defended a form of radical anti-realism.
Then, considering that the anti-realism left us without any answer to the questions of reality in the theoretical and philosophical conflicts of cognitive science, I supposed that looking at how science works might give us a lead. In this context, New Mechanistic philosophy became relevant. I outlined the framework of new mechanistic philosophy, focusing on mechanistic inter-level integration and demonstrating how it overcomes many of the limitations traditionally attributed to classical mechanistic approaches. I also addressed objections that new mechanistic philosophy trivializes the concept of mechanisms, arguing that, unlike causal-functional analysis, it provides a cohesive framework on science and this is its important and characteristic aspect. Whether one chooses to call it mechanistic or something else is only a stylistic dispute. Finally, while many mechanists are realists, I noted that mechanistic philosophy is not, in principle, incompatible with anti-realism. I later concluded that coherence could serve as a criterion instead of mind-independent reality. However, I argued that empirical coherence (EC) alone is not enough and that a criterion of conceptual coherence (CC) is also necessary.
I claimed that schools in cognitive science are fundamentally divided based on their responses to the metaphysics of mind. I argued that this division stems from functionalism, reviewed its connections to multiple realizability, the mental causation problem, the autonomy of special sciences, and the hard problem of content. I explained the relationships between these issues, showing how they reinforce one another, all rooted in functionalism. Specifically, I argued that the autonomy of special sciences leads cognitive science to be treated as a special science, hence requiring a distinct and autonomous ontology which seem to have received considerable support to this point. I then highlighted how the very nature of the so-called autonomy of special sciences gives rise to certain problems, particularly those regarding the concept of scientific progress. I argued that the distinction between special and natural sciences on the grounds how laws work differently in each of them is problematic and ungrounded. Accordingly, I further claimed that to a priori assume some scientific constructs are irreducible and unexplainable, and even unrelatable in any way to lower levels is at odds with the goals of scientific inquiry and scientific explanation. In response, I introduced the concepts of CC and NC. While conceptual coherence can be applied to any conceptual framework, naturalistic coherence is specifically applicable to naturalistic domain. I suggested that, when thinking in terms of mechanistic levels, we can use a variety of constructs according to scale, phenomena, levels of abstraction, or degrees of concreteness/vagueness depending on the phenomena and explanations at hand. However, I argued that choosing a construct which is already defined “irreducible,” “unexplainable,” or in a way incoherent with the existing naturalistic framework over an alternative is ultimately a theoretical choice of the particular theorist or scientific community. This, as such, might severely hinder genuine explanatory efforts. To illustrate this, I provided an example from the history of developmental biology and a thought experiment. Lastly, I noted that people are intuitively inclined to attend to coherence, but like all intuitions, this can sometimes fail. To prevent this as much as we can, I suggested making this implicit intuition explicit. In this context, I proposed the Natural Coherence Criterion (NCC) as a constraint for the appraisal of various constructs and conceptual frameworks in science or for proposing new ones.
Misunderstandings are inevitable given that I might have undertaken a task far larger than my reach in the attempt to address these complicated issues within the scope of a master’s thesis. Therefore, I should now attempt to answer the question of what have I not claimed as opposed to what I have claimed?
First and foremost, I have not claimed that functionalism is wrong. On this matter, I am agnostic. Functionalism may very well be correct, but that is an issue for the philosophers of mind. My argument is simply that cognitive science, as a scientific discipline, does not need to rely on any particular metaphysics of mind to establish itself or ground its theories. In principle, one could even be a functionalist and still support the thesis presented here. There is no inherent contradiction in doing so. Although admittedly I would not understand why someone would even bother to do that.
I have not argued that concepts like content, intentionality, normativity, or mindedness cannot be grounded and should no longer be discussed even though I because of the reasons I have outlined above I am skeptical towards such attempts. On the other hand, philosophers of mind are, of course, welcome to continue debating these topics, and in doing so, they can very well draw upon a non-mentalistic cognitive science just as they draw upon a non-vitalist biology in trying to answer the question of "what is life."
I do not claim that anti-realism entails New Mechanism, nor that New Mechanism entails anti-realism. I defended each independently, giving distinct justifications for both, and argued that they are compatible with one another. I further suggest that bringing these two together, particularly when considering the contextualist stance of the kind of anti-realism defended here, can provide a foundation for a normative criterion to guide the creation and appraisal of new concepts in science.
I think that the framework presented here differs significantly from most, perhaps all, forms of reductionism. The primary reason for this is that our radical anti-realism prevents us from considering any explanatory level as more "real" or “fundamental” than another. In this framework, atoms or neurons are no more "real" or “fundamental” than human behaviors or vice versa. All levels are interdependent fictions. However, this topic warrants its own separate treatment.
The proposal here differs from most, perhaps all, forms of eliminativism. Unlike common forms of eliminativism, it does not make claims about the existence or non-existence of particular entities. Rather, it argues that certain constructs or purported entities should be eliminated from the naturalistic domain if they are incoherent with it. Furthermore, due to the commitment to anti-realism and contextualism, "existence vs. non-existence" questions become meaningless. When these questions are meaningless, it is possible that many and even most conceptual problems in science could be addressed through conceptual and linguistic hygiene. This is because, under this framework, only the concepts and terms we use as tools to "work with" remain relevant. However, this issue also deserves a separate treatment.
I have not claimed that one must be an anti-realist to defend the Naturalistic Coherence Criterion (NCC). In principle, someone could defend NCC while being a realist of some sort. However, in that case, they would need to either "naturalize" or reduce everything they consider real. This is the common practice currently observed in the field. While it is difficult to make a definitive statement, my pessimism on this strategy is evident. Notice also that for this strategy, even for the strong emergentist, there is a necessity to show how and why the (strongly) emerged new entities or properties correspond to the ones we take to be real in our folk-ontologies, e.g. mind, life, memory, etc.
Similarly, I have not claimed that one must adopt Mew Mechanism or a mechanistic framework of any kind to defend NCC. In principle, someone could oppose mechanism and still defend NCC. However, in doing so, assuming that they are not anti-realist, it would lead to functionalism and hence to facing all the problems that functionalism entails. For instance, they would face nearly all the issues we have found problematic about autonomy of special sciences. Without mechanistic levels and their interrelations, we lack constraints for proposing, connecting, explaining, or grounding concepts. In such a case, anything goes. This, as noted above, ultimately leads to the kinds of crises related to scientific progress that we have already highlighted. But an anti-realist and non-mechanistic defense of NCC is possible. Though probably it would still lead to problems related to scientific progress and the formulation of non-ad-hoc criteria for theoretical and philosophical conflicts between conceptual frameworks.
I have not claimed that defending the NCC requires accepting the broad interpretation of New Mechanism, i.e. the view that all of science is mechanistic. For the purposes of the present discussion, it is adequate to accept that only relevant parts of science, say, biology, cognitive science, neuroscience, are mechanistic. I am also not claiming that it is currently possible to explain or formulate all the phenomena that cognitive science considers and formulates as "mental" in a non-mentalistic way. Whether this is possible cannot be known a priori. The only way to determine this is to follow the progress of a non-mentalistic cognitive science with cautious and deliberate steps.
From where can we begin to develop such a conceptual framework then? What are the relevant clues at hand? How might further research proceed, and how could the foundations for a non-mentalistic cognitive science be established? Two potential approaches come to mind. The first is to start with the brain, following something in the vicinity of Buzsáki (2019). The second is to begin with evolution, drawing on something akin to Cisek’s phylogenetic refinement (Cisek, 2020). Note that, though, neither of these perspectives explicitly aims to offer a non-mentalistic ontology and, in fact, they both accept certain mentalistic concepts. However, both are skeptical of some of the categorizations inherited from folk psychology and target them through two distinct routes. Of course, there is no space for a thorough review of these views in this chapter or even in this thesis. However, the first challenge faced by approaches like Buzsáki's, which start with the brain, is simply the vast complexity of the brain itself. The brain is incredibly complex so much so that without a solid conceptual framework, we can find correlations for virtually anything we choose to investigate (Meijer, 2021). Thus, it is unclear where to even begin. What is more, the purpose of a conceptual framework is precisely to provide this kind of constraint. Moreover, our experimental studies of the brain must rely on behavioral taxonomies. Yet, to my knowledge, there is no existing non-mentalistic behavioral taxonomy. (For similar critiques of Buzsáki, see: Poeppel & Adolfi, 2020.) Note that, that means without using mentalistic concepts, there is currently no foundation to categorize brain activity. In this context, relying on behavioral taxonomies raises two additional issues: First, behavior is as complex as the brain. Without a framework like folk psychology, it is unclear how we could carve up the behavioral domain, particularly at the human scale, in a way that would both provide a non-mentalistic ontology and avoid arbitrary and ad-hoc frameworks. Second, complex behaviors are observed in organisms without brains (Gomez-Marin, 2021). This undermines the motivation for starting cognitive science with the brain, calling into question the plausibility of a brain-centric approach.
On the other hand, this final point might suggest a novel and interesting direction. We could begin constructing our conceptual framework and taxonomy by focusing on simpler phenomena, using the taxonomies we develop for these behavioral phenomena as clues and constraints for approaching more complex phenomena, building incrementally. This approach parallels Cisek's strategy of relying on evolutionary and comparative studies for classification of functions and mechanisms in neuroanatomu rather than experimental research grounded in folk psychology. Admittedly, as I previously noted, Cisek is not categorically opposed to mentalistic concepts and actively employs them. Nevertheless, his strategy could provide important insights for developing a non-mentalistic cognitive science.
Notice the parallelisms between this view and the approaches like enactivism (Di Paolo & Thompson, 2014; Froese & Di Paolo, 2011; Gallagher, 2017; Noë, 2021; Steward, Gapenne & Di Paolo, 2010; Varela, Thıompson & Rosch, 1991; Ward, Silverman & Villalobos, 2017), autopoietic theory (Maturana & Varela, 1973/1980; Villalobos, 2015; Villalobos, 2022; Villalobos & Silverman, 2018; Villalobos & Ward, 2015), ecological psychology (Gibson, 1979; Lobo, Heras-Escribano & Travieso, 2018; Segundo-Ortin & Raja, 2024), the Free Energy Principle (FEP) (Friston, 2010; Friston, Fortier & Friedman, 2018; Parr, Pezzulo & Friston, 2022), and Technological Approach to Mind Everywhere (TAME) (Levin, 2022) in advocating a form of life-mind continuity[1] even if significant differences remain. The question of how the framework proposed here overlaps or diverges from these schools of thought in cognitive science certainly warrants a dedicated treatment. For the purposes of this discussion, suffice it to say that starting with the behavior of so-called simpler organisms or, more precisely, organism-environment interactions and progressively developing our conceptual framework toward more complex phenomena appears to be an epistemological ground-zero that most of us can agree upon. Approaches like enactivism, TAME, FEP, and ecological psychology are far from being non-mentalistic. However, autopoietic theory, with its anti-realist commitments rooted in second-order cybernetics and its proximity to a mechanistic framework (Lee, 2023), represents the closest view to the framework proposed here. That being said, significant differences remain. Most fundamentally, in autopoietic theory, mind is treated as a conventional kind as opposed to a natural kind (Villalobos & Palacios, 2021). On this point, I remain entirely agnostic in the present thesis, as I make no claims about the nature of the mind. Rejecting such a stance also complicates autopoietic theory’s proposal to use mental concepts instrumentally, making it problematic within the framework I have outlined here.
Of course, the program I just outlined covers actually a vast amount of scientific area that will inevitably bring with it numerous questions, challenges, and puzzles. While some speculations about potential issues and how they might be addressed should have become obvious after my discussions in this thesis, I cannot explore them here due to space limits. Moreover, it is impossible at this stage to see beforehand where such a research program might lead, or how many researchers, master’s theses, doctoral dissertations, articles, books, institutions, and years or decades it will require to achieve its goals. Nevertheless, this is science. We have always owed our success in science to its hard ways. Once again, we should bite the bullet and take the hard way.
I want to end this thesis with a quote by Lichtenberg, who reminds us something we all seem to have forgotten long ago: “There is no greater impediment to progress in the sciences than the desire to see it take place too quickly.”
[1] Of course, within the framework discussed here, referring to the proposed continuity as life-mind is no longer possible for obvious reasons. Terms like evolutionary scaffolding, evolutionary coherence or evolutionary continuity seem more fitting.