Pluralism in the Cognitive Sciences: Theoretical, Methodological or Explanatory?

Sushchin Mikhail Aleksandrovich ORCID: 0000-0002-8805-6716 PhD in Philosophy Senior Researcher, Federal Budgetary Institution of Science Institute of Scientific Information on Social Sciences, Russian Academy of Sciences 117418, Russia, Moscow, Nakhimovsky Prospekt, 51/21 sushchin@bk.ru Other publications by this author

From theoretical complexes to the defense of pluralismThis work continues a number of recent publications ^{[1, 2]}, in which we proposed the idea of theoretical complexes to answer the question of what are major theoretical areas, or groups (aggregates) of theories close to each other, in the cognitive sciences: classical computational cognitivism, connectionism, embodied cognition, etc.

One of the consequences of the normative plan of the concept of theoretical complexes in cognitive sciences was the opposition expressed by T. Kuhn in the book "The Structure of Scientific Revolutions" ^[3] to the ideas of theoretical monism and explicit defense of theoretical pluralism and the principle of theoretical proliferation. Meanwhile, the confrontation between pluralism and monism (as well as the principles of proliferation and unification associated with them, respectively) is by no means exhausted by the theoretical context alone – this confrontation can also be considered at a more general level, which also includes, at least, methodological and explanatory contexts (i.e., at the level involving consideration of issues about theoretical and experimental research methods, as well as explanations formulated within its framework). Accordingly, in this article we are going to consider the question of the need for pluralism or monism for the cognitive sciences at this more general level, which includes theoretical, methodological and explanatory contexts.

Let us now reveal in more detail the trajectory that leads us from the most important provisions of our concept of theoretical complexes to the question of the relationship between pluralism and monism in the cognitive sciences. First of all, as we have already said, the idea of theoretical complexes (or theory complexes) was proposed by us in order to properly characterize major theoretical trends in cognitive sciences, such as connectionism, etc. It is widely recognized that these trends have played a significant role in modern cognitive sciences since their emergence in the 1950s. But what exactly are they? Paradoxically, no significant attention is paid to this issue – they are often referred to simply as certain "approaches", "paradigms", etc. When it is necessary to characterize these areas more specifically, they most often resort to one of the three well-known postpositivist concepts of the philosophy of science, in which the emphasis was not on the analysis of individual scientific theories and models, and whole groups of such theories are considered: the concepts of scientific revolutions of the already mentioned T. Kuhn and her ideas of the paradigm, the concepts of I. Lakatos' research programs ^[4], as well as the concepts of L. Laudan and the most important idea of research traditions for her ^[5].

It should be noted that the very attempt to appeal to the concepts of Kuhn, Lakatos or Laudan seems quite reasonable, since, obviously, in the case of large cognitive theoretical trends, we are not dealing with individual scientific theories and models, but with groups of theories and models that are close to each other in content. Accordingly, the traditional methods of analyzing scientific theories developed in philosophy and methodology of science, similar to those used by logical positivists at one time, are not applicable here. In addition, there is no doubt that major theoretical trends in cognitive sciences have a number of features that are well grasped by one or another of these concepts.

However, as we stated earlier, first of all, due to a number of descriptive problems, none of these concepts can be taken as such as a basis for understanding large sets of theories in the cognitive sciences. Thus, when faced with difficulties even in the field for the analysis of which they were primarily intended (i.e., in the field of philosophy of physics), none of these concepts is able to properly characterize the structure and functions of major theoretical directions in the cognitive sciences. (A detailed analysis of the difficulties of the concepts of Kuhn, Lakatos and Laudan is given in our other work ^[1].)

In an attempt to overcome the limitations of the ideas of Kuhn, Lakatos and Laudan, the American philosopher of science B. von Eckardt came up with an alternative concept of research conceptual frameworks in the early 1990s ^[6]. Unlike the ideas of these authors, von Eckardt's concept has no universalist claims, i.e. claims of universal applicability for the analysis of any field of scientific knowledge, being specially developed to describe two historically first cognitive theoretical trends: classical computational cognitivism and connectionism. Accordingly, the concept of a research conceptual framework turns out to be free from a number of shortcomings inherent in the ideas of Kuhn, Lakatos and Laudan due to their universalist orientation. Nevertheless, the von Eckardt concept also faces significant difficulties – first of all, it did not envisage the emergence of radically different directions from classical cognitivism and connectionism, whose supporters in no way rely on the concepts of representation and computation to explain cognitive processes. Such a direction is the so-called radical embodied cognition, and the concept of a research conceptual framework is completely unsuitable for its description.

The problems of the concepts of Kuhn, Lakatos, Laudan and von Eckardt have put us in front of the need to look for a different perspective to describe large cognitive complexes of theories. Thus, we have proposed our own concept of theoretical complexes in cognitive sciences for this role. Like von Eckardt's concept, the perspective of theoretical complexes has no universalist claims. Its main task is to give an adequate description of large sets of theories in the cognitive sciences. To this end, it provides answers to questions about the structure and priority functions of cognitive groups of theories.

The distinction we make between the more abstract and more concrete components of the cognitive theoretical directions of individual scientific theories and models is important for answering these two questions. Thus, the more abstract components of cognitive theories and models, often implied only implicitly, determine the general ontological concepts of the relevant subject area (for example, "cognition is a calculation based on mental representations", "cognition is an embodied action", etc.), as well as the methodology of research. Whereas more specific components are statements subject to direct empirical verification like "on a time scale of about 1/3 of a second, orienting body movements play a key role in cognition and form a useful computational level.... At this "level of physicality", the limitations of the physical system determine the nature of cognitive operations" ^{[7, p. 723]}. It is the more abstract ontological and methodological components of individual theories and models that make it possible to consider a particular group of theories as a theoretical complex – they form the conceptual core of such a complex.

Thus, taking into account this distinction, we can give the following answer to the first question posed above – the question of the structure of theoretical directions in the cognitive sciences (meaning by the structure how the individual components of their theories and models directly relate to each other): firstly, most often there is a fundamental opportunity to identify any one similarity or a number of similarities that distinguish all the theories and models included in the complex. In this case, we can say that the theoretical complex has a homogeneous structure. These are, for example, the same complexes of classical computational cognitivism or connectionism mentioned above. Secondly, a situation is possible, much rarer and more interesting, when complex theories are combined not on the basis of one general position or a number of such positions, but on the basis of components that overlap only partially. Such theoretical complexes are characterized by a heterogeneous structure, for the description of which the idea of family similarities according to L. Wittgenstein is best suited. In our first article devoted to the perspective of theoretical complexes, we put forward and tried to substantiate the thesis that it is in the manner of Wittgenstein family similarities that theories and models of the movement of embodied cognition relate to each other in its moderate version, which does not reject the representational nature of the processes of cognition ^{[1, p. 14-15]}.

Similarly, the distinction made allows us to answer the second key question in the characterization of large cognitive theoretical trends – the question of their primary functions. To these we refer only two functions. One of them is certainly constructive, and the other is negatively constructive.

In the first function, the complex acts as a kind of reference point for creating new or changing existing scientific theories and models. More specifically, in this function, theoretical complexes help: (1) to create new theories and models, (2) to develop, refine and modify existing theories and models by defining their (2.1.) conceptual apparatus, as well as, at least in some respects (2.2.) methodology and (2.3.) the problem field. Earlier, as an example of complexes in this function, we mentioned images of cognition as symbol manipulation by a Turing machine (for supporters of classical computational cognitivism), an image of a working neural network (for adherents of connectionism), etc.

The second key function of theoretical complexes has a mostly negative (or, more precisely, negative-constructive) coloring. In this function, the complex plays the role of a negative example in the construction and modification of theories and models (or just a target for criticism) for supporters of alternative theoretical directions. We noted earlier that it was in this way that the supporters of the complexes of connectionism and embodied cognition that emerged at the turn of the 1980s and 1990s considered the direction of classical computational cognitivism that prevailed up to that time. According to the adherents of connectionism and embodied cognition, classical cognitivism carried with it a greatly simplified, inadequate understanding of cognitive processes taking into account a number of neurophysiological and socio-cultural factors. However, there is no silver lining. Because even as a negative example, the complex shows researchers who support another complex what was missed, where they need to direct their efforts, what problems to solve, what factors should be taken into account, what methods to use, what concepts to rely on. Thus, even in this role, the complex partly contributes to the creation of new theories and models, but within the framework of a different theoretical complex.

All that has just been said about functions can be summarized as follows: theoretical complexes contribute to the proliferation of scientific theories (after all, in fact, in both main roles they somehow help to create new and/or improve existing theories and models). In this connection, a fairly obvious conclusion of a methodological normative nature suggests itself, namely that theoretical complexes should simply contribute to the creation of scientific theories and models and that theoretical pluralism, not monism, represents the desired state of affairs in the cognitive sciences. But this conclusion runs counter to one of the main methodological normative conclusions that can be drawn based on the analysis of the "image of science" created by Kuhn in his book "The Structure of Scientific Revolutions".

It is well known that Kuhn was a proponent of distinguishing between the so-called mature and immature sciences and that he saw the criterion of maturity in the acquisition of a particular scientific field of its paradigm. And this means that in order for this discipline to become mature (having launched the so–called normal phase of scientific research), theoretical unification must take place within its framework, i.e. the replacement of many theories and schools of the "pre-paradigm period" with one single dominant point of view - a paradigm. From this point of view, it is theoretical monism (according to Kuhn, interrupted after the establishment of a certain paradigm only during periods of crisis) that represents the desired state of affairs in science.

In our other works being prepared for publication, we, appealing to the well-known expressions of J.-S. Mill, P. Feyerabend, etc. arguments in favor of pluralism, opposed the basically monistic "image of science" created by Kuhn and its necessity for modern cognitive sciences. However, as we already noted at the beginning of this article, the confrontation between pluralism and monism in the philosophy of science is not limited to a theoretical aspect alone – the question of pluralism/monism can also be considered, at least in methodological and explanatory terms.

For example, in connection with the well-known concept of explanation as unification, the leading exponent of which in the modern philosophy of science was the British-American philosopher Philip Kitcher, the question may arise about the need for this kind of unification for the cognitive sciences. (Referring partly to Kitcher, the philosopher of cognitive sciences Marcin Milkovsky and his co-author recently spoke in favor of the possibility of explanatory unification for cognitive research ^[8].) A similar question can be raised with regard to methodological unification.

Accordingly, in the remainder of the work we are going to consider the opposition of pluralism and monism within the framework of all three of these contexts: theoretical, methodological and explanatory. We will defend the need for theoretical and methodological pluralism for cognitive sciences. At the same time, we cannot draw a similar conclusion on the question of the preferred nature of explanations for the cognitive sciences – in our opinion, only future cognitive studies can show whether explanatory unification or a certain kind of explanatory pluralism will be preferable for them.

Pluralism vs. monism: theoretical, methodological and explanatory contextsWe will begin our consideration of the confrontation between pluralism and monism from the theoretical context.

The main problem that arises in this context, we can formulate as follows: since Kuhn became the main exponent of the ideas of theoretical monism in the modern philosophy of science, should the sciences be guided by the monistic scheme proposed by him, i.e. should they strive to create a paradigm that would occupy a monopoly position within a certain discipline Or should the sciences (including cognitive sciences) strive for theoretical pluralism, assuming the possibility of coexistence, interaction, struggle, etc. of different theories incompatible with each other both in relation to the basic ontological and methodological premises mentioned above, and in terms of more specific provisions aimed at empirical verification? Do the sciences need theoretical unification in the manner of Kuhn, or do they need theoretical proliferation, i.e. the spread and development of different points of view?

Let's start with arguments in defense of theoretical monism. As is well known, Kuhn's main arguments in favor of theoretical monism are historical in nature. As a matter of fact, as Kuhn himself noted at the beginning of The Structure of Scientific Revolutions, the purpose of his book is "to give an outline of a very different idea (concept) about science arising from the historical description of the research activity itself" ^{[3, p. 1]}. In his opinion, "historiographical studies speak of the possibility of a new image of science" ^{[3, p. 3]} – new in comparison with the image of the development of science through the consistent accumulation of individual discoveries. Let's list the main characteristics of this image. According to Kuhn, the early stage of development of any field of science (the so-called pre-paradigm science) is characterized by the presence of many competing scientific schools. Their adherents, when faced with similar facts, tend to give them different descriptions and interpretations. They also differ in their understanding of the key provisions of their field of research. An important example for Kuhn here is the field of electricity research in the first half of the XVIII century . "During this period," Kuhn writes, "there were almost as many different points of view on the nature of electricity as there were significant experimenters in this field, including Hawksby, Gray, Desaguliers, Dufay, Nollet, Watson, Franklin, etc. All their numerous concepts of electricity had something in common – they were partially derived from one or another version of the mechanical-corpuscular philosophy that guided all scientific research of that time. In addition, they were all components of truly scientific theories, theories derived partly from experiments and observation and partly determining the choice and interpretation of additional research problems. Despite the fact that all the experiments were devoted to the study of electricity, as well as the fact that the experimenters read each other's works, their theories had only family similarities among themselves" ^{[3, p. 13-14]}.

However, at some point, the discrepancies of the initial period "... disappear to a large extent and, apparently, once and for all. Moreover, their disappearance usually turns out to be the result of the triumph of one of the schools of the pre-paradigm period, which, due to its characteristic positions and prejudices, focuses only on a special part of the too large and inchoate volume of information" ^{[3, p. 17]}. Thus, the completion of the pre-paradigm period brings with it theoretical unification, i.e. the replacement of many competing theories with a single generally accepted paradigm. According to Kuhn, the paradigm acts as some kind of "achievement" that serves in the eyes of the scientific community as the basis for its further activities – scientists starting from it can rely on "the same rules and standards of scientific practice" ^{[3, p. 11]}. An important consequence of the adoption of the Kuhn paradigm is the actual absence of criticism of its principles by its adherents – these fundamental propositions are taken by scientists as a matter of course. The formation of the paradigm marks the beginning of a new more effective phase of scientific activity – the so-called normal science. "The latter is, by and large, esoteric and aimed at solving puzzles, since a group can work only when its members take the foundations of their field for granted" ^{[3, p. 178]}.

This phase of "normal science" stops when specialists working within its framework encounter a number of anomalies – unforeseen phenomena taking into account the accepted paradigm. During the crisis, as in the pre-paradigm era, there is a proliferation of a number of competing points of view. After scientists choose one of these alternatives, the phase of "normal science" resumes, etc.

Meanwhile, the image of science compiled by Kuhn in the "Structure of Scientific Revolutions" has attracted considerable and, in our opinion, well–founded criticism - criticism primarily of a historiographical nature. Thus, Kuhn's thesis about the typicality of the situation of the unconditional dominance of paradigms in science, and his statement about the uncritical acceptance of the foundations of the paradigm by its supporters, were criticized. For example, in relation to the first thesis, I. Lakatos wrote: "The history of science was and should be the history of competing research programs (or, if you like, 'paradigms'), but it was not and should not become a sequence of periods of normal science: the earlier the competition begins, the better for progress (author's italics – MS). 'Theoretical pluralism' is better than 'theoretical monism': in this question Popper and Feyerabend are right, but Kuhn is not" ^{[4, p. 78]}. Convincing criticism of the second thesis was given, in particular, by L. Laudan, who wrote about Newtonian mechanics (Kuhn's favorite example) that "From the moment of its perception by Huygens and Leibniz, its key provisions were under incessant critical consideration, even by those many physicists who willingly recognized its mathematical virtuosity and its empirical achievements.7 George Berkeley, some of the early Bernoulli, Maupertuis, Hutchinson's followers, Boskovich, early Kant, and even Euler posed a number of fundamental problems in relation to ontological (hereafter the author's italics are M.S.) the foundations of Newtonian mechanics. At the same time, other scientists (Hartley, Lesage, Lambert) challenged the methodological assumptions of the Newtonian tradition" ^{[5, p. 135]}. Thus, contrary to Kuhn's claims, the history of science does not at all unequivocally testify in favor of theoretical monism.

However, of course, there may be other arguments in favor of theoretical monism. Thus, the mentioned authors from Poland express concern that interdisciplinary research in cognitive sciences can lead to extreme fragmentation and, as a result, "doubling of efforts and lack of understanding" ^{[8, p. 2]}. The answer to this argument will be given at the end of the article, but for now we turn to the main arguments in favor of theoretical pluralism in the cognitive sciences.

So, as you know, many authors were supporters of pluralism, including J.-S. Mill, K. Popper, I. Lakatos and, in particular, P. Feyerabend. The latter gave the following formulation of the principle of proliferation and the position of methodological pluralism: "Invent and develop theories incompatible with the currently generally accepted point of view.... Any methodology that accepts this principle will be called a pluralistic methodology" ^{[9, p. 107-108]}. We can correct this definition from the perspective of theoretical complexes outlined above. Thus, speaking in favor of the proliferation of incompatible theories, Feyerabend meant only theories incompatible with each other in a strict logical sense (i.e., he meant only theories that contradict each other). From a logical point of view, theories belonging to the same theoretical complex can also be incompatible. Nevertheless, being logically incompatible, these theories can be based on similar ontological and methodological premises, due to which they can be attributed to the same theoretical complex (they may differ in relation to predictions of specific experimental effects). Therefore, the idea of theoretical complexes calls for the proliferation of both theories that differ in predictions about specific experimental effects (but are based on similar ontological and methodological assumptions) and theories based on completely different ontological and methodological assumptions and, accordingly, related to different complexes. The development of science requires not only an "external" critique of ideas (i.e., criticism from the standpoint of alternative ideas and complexes), but also the derivation of all possible consequences from a particular idea and their study within a certain set of theories.

We believe that there are two key arguments in favor of theoretical pluralism and proliferation. They can be formulated as follows: (1) new theories are necessary for meaningful clarification and/or criticism of existing theories, as well as for the possibility of their refutation; (2) the theory that is proposed to be rejected may still be promising and useful in the future. Let's discuss these two arguments in more detail.

The first argument suggests that new theories are necessary to clarify and criticize existing theories. P. Feyerabend made this thesis more radical: according to him, the relevance and refuting nature of decisive facts can be established solely with the help of theories that diverge from the currently accepted point of view ^{[10, p. 20-21]}. It is impossible to agree with this thesis. In the strict sense, to refute some theory T ₁, another theory T _{2 is not logically required}. From a logical point of view, in order to refute the theory of T ₁, it is only necessary to deduce from it, as well as a number of auxiliary hypotheses associated with it, the prediction of P and observe whether it is confirmed in the experiment or not. If there is a non-P and it is known that all the auxiliary hypotheses involved in deducing the prediction are true, then it is safe to say that the theory of T ₁ has been refuted in the experiment.

However, in our opinion, Feyerabend's argument can be adjusted. So, (i) as follows from his works, as well as from the works of a number of other authors (such as N. Hanson), it is impossible to conduct research without theories at all – one way or another we always start from theories; (ii) therefore, if we abandon some theory, we need another theory (in the power of "(i)"); (iii) in addition, the development of alternatives helps to better understand the shortcomings of the current point of view; (iv) in the same way, if some theory T ₁ predicts the result of P, and according to the results of the experiment there is a non-P, then instead of T ₁ we need another theory T ₂ to explain why non-P is the case. In a nutshell, criticism and significant refinements of a particular theory suggest the existence of an alternative point of view (even if it is often implied only implicitly).

The second methodological argument in favor of proliferation is based again on research on the history of science. As Feyerabend writes: "The philosophy of atomism is an excellent illustration. It was formulated in antiquity with the aim of ‘saving’ macrophenomenes like the movement. It gave way to the dynamically more sophisticated philosophy of the Aristotelians, returned during the scientific revolution, was pushed aside with the development of continuity theories, returned again at the end of the nineteenth century and was again limited by [the idea of] complementarity" ^{[11, p. 139-140]}. (We believe that very similar metamorphoses can manifest themselves in the development of cognitive sciences – of course, on an incomparably more modest scale. In particular, there is a possibility of "reviving" the main hypothesis of classical computational cognitivism, namely that cognition is a calculation based on mental representations, and the mind can be organized in the manner of a Turing machine, in combination with probabilistic ideas about the structure and operation of unconscious cognitive mechanisms ^[12].)

All this, coupled with the more general historical arguments mentioned above in favor of theoretical pluralism and the principle of proliferation, allows us to draw the following conclusion: cognitive sciences do not need a monopolist paradigm in the sense of Kuhn - the establishment of such a paradigm for a long time can significantly hinder progress in cognitive research. If we are talking about the need for a theory capable of covering a wide range of cognitive phenomena, then it would be more appropriate to talk about the need for competing theories of the great unification ^{[8, p. 4]}.

Is it possible to draw a similar conclusion on the need for methodological pluralism for cognitive sciences? It is definitely worth giving an affirmative answer to this question. After all, if we advocate the need for pluralism of different ontological and methodological points of view, then the pluralism of methods follows most obviously from here.

However, on this issue, we can say that the practice of cognitive research, in general, is in agreement with methodological pluralism. An example is research from a wide variety of fields of cognitive sciences. For example, in modern neuroscience, a combination of different neuroimaging methods is common due to the limitations of each method separately. Thus, neuroscientists are often forced to combine methods of functional magnetic resonance imaging and magnetoenciphalography, since the former has a good spatial but low temporal resolution, which is compensated by the features of the latter, and this sometimes "requires the participation of an interdisciplinary team consisting of physicists, psychologists, neurophysiologists and physicians...." ^{[13, p. 150]}. Undoubtedly, it has become a common place for cognitive research and the addition of psychological methods for studying cognitive processes (such as visual perception, speech, thinking, etc.) by determining their brain substrate (brain correlates, neural correlates ^[14]).

Here it will be appropriate to recall the methodological concept of the Anglo-American vision researcher of the second half of the XX century, which is classic for cognitive sciences. David Marr. In his book "Vision", Marr presented a methodological concept for studying complex information processing systems, both natural (say, the human brain) and artificial (for example, computers) ^{[15, pp. 36-44]}. The meaning of his concept is that any such information processing system can be analyzed at three levels. The first level according to Marr is associated with the definition of the general computational task that the system performs. For example, the computational task of a cash register is to calculate the amount to be paid, subject to a mathematically defined addition operation. At the second level, the system of representation of input and output information (for example, binary or decimal) and the algorithm by which the task is implemented are investigated. Finally, at the third level, we study the implementation of a computational task and a system of representation and algorithms in a specific physical substrate (microcircuits in the case of a modern cash register, the brain in the case of a person, etc.).

In our opinion, Marr's concept is in good agreement with the pluralism of methods in cognitive research and can even serve as an additional basis for it. At the same time, a brief discussion of the Marr concept brings us to the question of explanatory and ontological pluralism. Indeed, when presenting his methodological concept, Marr wrote that "It is almost never possible to understand any complex system based solely on extrapolation of the properties of its elementary components.<...> Descriptions of such phenomena are given at the appropriate level, i.e. at the level representing a huge set of elements as a whole, while it is necessary to indicate that the descriptions related to the micro-level and macro-level, respectively, are combined. If you want to achieve a complete understanding of a system as complex as the nervous system, a developing embryo, a set of metabolic pathways, a bottle filled with gas, or even a large computer program, then you should be prepared to consider various scientific explanations at various levels of description, linked at least into a single whole, and regardless of the practical meaninglessness of tracing links between levels in all details" ^{[15, p. 36]}. In connection with this passage, the question may arise: should cognitive sciences also strive for explanatory pluralism, based on the hierarchy and heterogeneity of cognitive mechanisms and systems?

However, let's first analyze the arguments of the supporters of the opposite point of view – the point of view of explanatory unification. As we have already said, the leading exponent of explanatory unificationism in the modern philosophy of science was F. Kitcher. In a nutshell, his unificationist concept of explanation, presented in a number of publications in the 1980s ^{[16, 17]}, is based on the use of a minimum of explanatory constructions ("general patterns of arguments") to generate as many explanatory consequences as possible (i.e., specific explanations of phenomena). As Kitcher himself writes: "Science advances our understanding of nature by showing us how to derive descriptions of many phenomena using the same patterns of conclusions over and over again, and by demonstrating this, it teaches us to reduce the number of types of facts that it must accept as final (rough) (italics of the author – M.S.Therefore, the unification criterion that I will try to clarify will be based on the idea that E(K) is a series of conclusions that establish the best tradeoff between minimizing the number of output patterns involved and maximizing the number of consequences generated" ^{[17, p. 432]}.

Kitcher's concept contains a number of very difficult to understand technical terms, without at least a brief description of which, however, it is impossible to properly explain its meaning. The most important concept of this concept is the concept of a general pattern of arguments. As Kitcher noted, an explanation from an ontological point of view is a special type of argument. So, the general pattern of arguments includes (1) a schematic argument, (2) a series of filling instructions, (3) a classification for a schematic argument. (1) A schematic argument is a series of schematic sentences that result from replacing illogical terms with dummy letters (A, O, P, X, etc.). For example: if you replace illogical terms with dummy letters in the sentence "Organisms homozygous for A develop P" ^{[17, p. 432]}, then you will get a sentence like those used in mathematical logic: "For all x, if x is O and A, then x is P" ^{[17, p. 432]}. (2) A number of filling instructions suggest how to replace the substitute letters in sentences – for example, in the sentence "Organisms homozygous for A develop P" filling instructions may indicate that A should be replaced by the name of an allele, P by the name of a phenotypic trait, etc. A schematic argument is a sequence of schematic sentences. Finally, (3) classification of a schematic argument there are a number of sentences that say which sentences should be deduced from which, what are the rules of inference, etc.

In addition, the concept of rigor plays an important role in this concept – the rigor of the pattern of arguments depends partly on the classification that characterizes the logical structure of this pattern, and partly on the nature of schematic sentences and filling instructions that are jointly responsible for ensuring that sentences have similar illogical terms in certain places. If the strictness requirements in both senses are completely relaxed, then the concept of a pattern degrades to such an extent that it can include any argument. In the opposite situation, there is a "pattern" that is its only implementation. Softening the requirements for an illogical dictionary, but maintaining a strict approach to classification (in order to ensure the exact derivational status of each term in a schematic argument) gives logicians the concept of a pattern ^{[17, p. 433]}.

As a result, as Kitcher notes, the unifying power of a certain generating set of patterns of arguments is "directly proportional to the number of consequences derived from it, directly proportional to the rigor of patterns in the set and inversely proportional to the number of patterns in the set" ^{[17, p. 433]}.

Meanwhile, the unificationist concept of Kitcher's explanation faced a number of significant difficulties. First of all, as many authors pointed out, Kitcher never substantiated his thesis that explanation and unification are one and the same thing. Moreover, there are good reasons to believe that they are actually different things. Further, Kitcher never gave a convincing answer to the objection raised, in particular, by the philosopher David Lewis, namely that if the world is heterogeneous, then why should its explanation be unified ^{[17, p. 494-495]}? There are other objections. Philosopher Carl Craver, among others, cites the following: an explanation of a more common phenomenon (such as Parkinson's disease) is not necessarily better than an explanation of a rarer phenomenon (such as Kuru) ^{[18, p. 48]} – it is not at all clear how comparisons can be made on this principle, which explanation is "better", and what is "worse".

In short, all this does not allow us to agree with Kitcher's statement that explanation is unification. Nevertheless, unification in science is often regarded as a significant cognitive value ^[19]. For example, the unification of subject areas by many refers to the most important events in the history of science – elementary examples here are Newton's unification of celestial and terrestrial mechanics, Maxwell's unification of studies of electricity and magnetism. Is this kind of unification possible in cognitive sciences? Is it possible to find a theory that would provide a unified explanation for disparate cognitive phenomena and processes?

We believe that the answer to this question can be obtained only in the course of further research in the sciences of cognition and the brain. It is quite possible that to explain the work of the mind and brain in the cognitive sciences for a certain time, the most successful will be an eclectic combination of hypotheses referring to principles taken from different branches of mathematics – like the above-mentioned combination of hypotheses of classical cognitivism, based on mathematical logic and the theory of computability, and predictive coding, referring to the principles of probability theory. After all, if it turns out that the mechanisms of reason and cognition themselves are not homogeneous, then why should the corresponding cognitive theories claiming to explain them be unified? In our opinion, the "disorder of things" ^[20] would also require a pluralistic approach to explaining the mind and its mechanisms. (An objection might arise here: does not the theoretical pluralism, which is defended in the article, also presuppose a pluralism of explanations? No, it doesn't. Theoretical pluralism is a position that emphasizes only the need for at least a few alternative theories. At the same time, the explanations offered within the framework of these theories can be both pluralistic in nature and monistic. Accordingly, theoretical pluralism does not necessarily entail explanatory pluralism.)

Conclusion

So, in this paper, the confrontation of pluralism and monism in research in the cognitive sciences was considered within the framework of three important contexts: theoretical, methodological and explanatory. Developing the arguments of our previous works, we defended the need for theoretical and methodological pluralism in cognitive research. Theoretical pluralism and methodological pluralism inevitably follow from our concept of theoretical complexes in cognitive sciences and the principle of proliferation postulated within its framework, which calls for the creation/improvement of compatible and incompatible scientific theories and models with respect to basic ontological and methodological assumptions. At the same time, we cannot draw a similar conclusion with regard to the explanatory context: whether the strategy of explanatory monism or explanatory pluralism will be preferable for the cognitive sciences will become clear in the course of further research.

In conclusion, it is also worth noting that pluralism and the principle of proliferation in the form in which we formulate them do not call for the creation of the maximum number of new theories and models by any possible means. On their basis, it is impossible to draw primitivistic conclusions that all this can lead to the creation of many useless theories (it is impossible to know in advance whether the theory will eventually be useless or not ^{[11, p. 33]}) or that the more theories there are within a certain set of theories, the better. We turn to the doctrine of pluralism and the principle of proliferation not in order to create tools for an accurate quantitative assessment of sets of theories or individual theories in order to be able to draw certain conclusions. Their purpose is to ensure progress in cognitive sciences through the availability of alternatives and mutual criticism/discussions between proponents of different theories. Excessive fragmentation and dispersion of efforts can only lead to immoderate pluralism, which has no other purpose than the creation of new theories. Constructive pluralism, in our opinion, should contribute to progress in relation to a number of so-called epistemic values (cognitive values, the merits of scientific theories), both associated with the idea of approaching the truth (predictive success, the ability to give unplanned explanations to existing facts, the ability to unite disparate subject areas, the degree of reinforcement by successfully passed experimental tests), and and the advantages of scientific theories that are not related to this group (empirical adequacy, simplicity in a number of interpretations, the breadth of the proposed explanations).