According to Kuhn, there is nothing that more clearly proclaims a field of science than a paradigm (SSR, 22). In his Structure of Scientific Revolution (SSR), Kuhn refers to paradigms as a marker of a mature science (SSR, 12). A paradigm ensures that scientists have a common set of rules and standards that will direct them towards collective progress. It follows that the lack of a paradigm can hamper growth leading to an inertia resulting from the community being pulled in myriad diverging directions. As such, a paradigm is a necessity for a field to achieve any semblance of cumulative progress, and establishing a field’s first paradigm is therefore a crucial step. Though the majority of Kuhn’s examples and theories are based on physics, his framework concerns the sciences in general, including more recent Special Sciences such as neuroscience.
Given the inherent diversity that exists in the field of neuroscience, in this paper I will ask to what extent the field can be governed by a single paradigm that emerged from a single pre-paradigm phase and hence undergoes co-extensive cycles of crisis and revolution. I will argue that neuroscience is comprised of multiple paradigms, each exhibiting characteristic features of normal science, along with certain areas of inquiry that are in pre-paradigm phases. One such field is that of cognitive neuroscience, which is a subfield of neuroscience concerned with the biological processes underlying cognition. I demonstrate that this specific line of inquiry in the wider context of neuroscience lacks (i) consensus on the problems it attempts to solve (ii) a theoretical framework and (iii) consistent methodology. I conclude by claiming that a field can have sub-fields that are mature based on Kuhnian standards, while other sub-fields that are in a pre-paradigm phase and hence not mature. The absence of a paradigm in cognitive neuroscience curtails efficiency and efficacy in the scientific activity as there is no common basis for its evaluation, criticism and selection (SSR, 16-17).
Introduction
I will begin by defining the terms used in this paper to avoid confusion. I define a paradigm as an accepted model or pattern that stands for the system of beliefs, values and techniques shared by a scientific community. I define a paradigm shift as exclusively a conceptual shift likened to a gestalt switch. Paradigm shifts can only occur through a scientific revolution that drastically changes the nature of concepts in the field and is hence a non-cumulative developmental episode. Under this strict definition, tools that transform how certain puzzles in normal science can be solved do not count as revolutionary. A scientific revolution must uproot the tools and questions of normal science, leading to a new set of tools and questions that seem not only incompatible with their predecessors, but incommensurate. Following a paradigm shift, the field enters a phase of normal science where the scientific community follows the same rules and standards and engaging in a process referred to as “puzzle solving”. Puzzle solving resembles mop-up work within the framework of a paradigm that merely tries to articulate theories that the paradigm already supplies. It does not attempt to unearth new phenomena or find anomalies (SSR, 24). A mature science is one that undergoes successive transition from one paradigm to another via periods of normal science that are interrupted by revolutions (SSR, 12).
Neuroscience is an emerging multi-disciplinary field concerned with the nature and function of the nervous system. Although it is often considered a more recent sub discipline of biology with its own methods and questions, neuroscience consists of many sub fields of its own with drastically different modes of inquiry. These draw and rely on insights ranging from mathematics to psychology. Practioners from virtually any scientific field are constantly pouring into the field because of its charm and novelty, bringing with them their respective field’s presuppositions and metaphysical commitments about nature. These new perspectives edify the field, but also make consensus within subfields in neuroscience all the more difficult to reach.
To ground the discussion that follows, I will briefly mention two particularly helpful examples from physics and genetics. Given that physics is a mature science, the subfield of physics known as optics has developed through transitions from one paradigm to another via revolution (SSR, 12). The pre-paradigm phase of optics was characterized by lack of a single generally accepted view about the nature of light. There were competing schools each with their own metaphysical commitments and as a result of their own theories, had their own theory driven observations about a particular cluster of optical phenomena. Despite significant contributions that informed future work, this period was vastly different from the period that followed Newton’s Optick, which was focused, directed and unified (SSR, 12). The absence of a standard set of methods or phenomena made the activity of the pre-Newtonian scientists as not strictly science, sincescientific activity in the Kuhnian sensenecessitates a common paradigm.
A second example is the molecular biological revolution ushered by the Watson-Crick model of DNA in 1953. At first glance this may look like a classic example of a paradigm shift given the rapid and nearly universal acceptance of the model, as well as the new research program that followed suit. However, one characteristic of a paradigm shift or scientific revolution in the Kuhnian model is that it must overthrow a previous one (SSR 66-69). In other words, there must be a Gestalt Switch from something. Prior to 1953, there were no clear ideas nor consensus about gene structure or action. There was no old paradigm that was relegated to a scientific dustbin with the advent of the discovery of the Double Helix. Genetics, in the wider the context of biology, had a paradigm emerge from a pre-paradigm state. Molecular genetics is now, one could say, in normal science and molecular geneticists are engaging in the sort of puzzle solving that is characteristic of a paradigm.
These examples from optics and molecular genetics frame my discussion of cognitive neuroscience and why although its practitioners are scientists – cognitive neuroscientists – the net result of their activity is slightly different to scientific activity that characterizes Newtonian optics or modern molecular genetics. Rather, it resembles a huge mass of fact finding with some relation, but no composite direction or net scientific consensus. As such, according to Kuhn, there is no basis for crediting some facts as more or less relevant (SSR, 15). Neuroscientists congregate at conferences to share results that may all equally seem impressive but there is no clear reason for why a given experimental result matters. Granted, there are statistical standards that give a study the right of passage for publication, but no standard for assessing why a statistically significant result is, in effect, significant.
The Multiple Paradigms in Neuroscience
Neuroscience has a diversity of subareas, such as developmental, physiological, computational or psychological, and levels of analysis, such as molecular, cellular, cognitive and behavioral. As a result, it is difficult to lump their questions, techniques, approaches and overarching concepts together. Not only is it unfeasible to lump them together, but given their different aims and practical uses, it may be counterproductive to force them all to abide by the same governing paradigm. It is hence possible that some of the subfields have reached a status of maturity, whereas others have not.
Modern day neuroscience can be viewed as a cluster of paradigms including, but not limited to, electrical signaling, the neuron doctrine, chemical synaptic transmission and cortical localization[i]. Each of these radically changed the way in which various aspects of the nervous system were viewed, yet each of these independently evolved from either a pre-paradigm state or a former paradigm of its own. For example, electrical signaling arose from paradigms of animal spirits based on Aristotle’s views on the mind in De Anima and Galen’s views on animal spirits contained in brain ventriclesi. The Neuron Doctrine and chemical synaptic transmission, on the other hand, emerged from pre-paradigm states. Prior to the Neuron Doctrine there were competing school of “Reticularists” who believed in continuous elements of the nervous system, and “Neuronists” believe they were discrete entities. There was no consensus, no theoretical framework and no agreement on methods. That is, although the Golgi stain was crucial for the triumph of the Neuron Doctrine, it was being used to generate opposing results from both schools. Today it is believed that “no neuroscientific discipline could be understood without recourse to the concept of neuronal, individuality and nervous transmission at a synaptic level, as basic units of the nervous system”[ii]. Chemical synaptic transmission similarly arose from a pre-paradigm state quainter-neuron communication, though it built on the neuron doctrine. Crucial experiments from Otto Loewi and Walter Dixon grounded the first unanimously agreed upon theories for how neurons communicate chemically, which in turn gave rise to another major paradigm in neuroscience. Though we do not have enough space in this essay to go through the other paradigms I mention, suffice it to say that there is a manifold of paradigms in the field of neuroscience that is each following its own course.
Given the burgeoning nature of the field, it follows that there will likely be new paradigms and hence existing pre-paradigm states in different sub-fields. I argue that one such subfield is cognitive neuroscience. Though we have a firm grasp of some of the lower level principles such as how genes relate to cellular function, as well as the paradigms mentioned including synaptic transmission and cortical localization, as of yet there remains no clear language that translates neural activity into cognitive and behavioral output. To quote neuroscientist and Nobel laureate Professor Richard Axel, “there is no logic for the transformation of neural activity into thought and action”[iii]. Higher level brain functions such as perception, emotion, thought and action cannot be captured by examining the level of the molecule or cell, but rather the level of wide scale neural activity. There have been advancements in the tools available: electrodes used to record neural activity have been transformed to militrodes allowing for higher precision, and new optical approaches have been designed to visualize wide scale neural activity iii. Yet, the language by which these neural activity patters are interpreted, afforded meaning, or translated to appropriate behavioral or cognitive output, remains elusive. As thought and action are presumably two of the central roles of the human brain, deciphering this logic is a key frontier of neuroscience. Furthermore, just as the discovery of the Double Helix did not overthrow a previous paradigm, when applying the same reasoning, I argue that this logic of how neural activity is translated to thought will similarly emerge from a pre-paradigm state. Importantly, acknowledgingthe need for the logic itself does not yet constitute the paradigm. The logic itself does.
In SSR Kuhn does mention subspecialties and the notion of small and large revolutions. He argues against the idea that normal science is a single monolithic and unified enterprise that experiences the same crises and revolutions (SSR, 50). He states that explicit rules may govern a field more broadly, but paradigms need not. In the course of professional specialization scientists acquire different applications of the paradigm and hence revolutions may be small, on the level of the subspecialty, or large, on the level of the field. Kuhn’s example is that physicists apply the paradigm of quantum mechanics differently and as such revolutions in quantum mechanical law may not extend to all subspecialties of physics (SSR, 50). This idea holds to a certain extent in neuroscience too, though it is complicated given the significant degree to which neuroscience relies on paradigms in biology, chemistry and physics. A revolution in gene function for example will bear on neurophysiology, albeit not as consequentially as it will bear on molecular or cellular neuroscience. An analogous concept within the field of neuroscience, such as the Neuron Doctrine, however, seems to underlie and predicate all other aspects of neuroscience. Its application is basic and universal, and a revolution in this most integral feature of the nervous system would, presumably, have a ripple effect on all sub-specialties of neuroscience.
Cognitive Neuroscience is Pre-Paradigm
Given the manifold paradigm model of neuroscience discussed, I will now argue that a certain subspecialty of neuroscience, namely cognitive science, is in a pre-paradigm stage, and that much like the field of molecular genetics prior to the Central Dogma, its first paradigm will usher in a period of normal science.
In the pre-paradigmatic stage, there are competing schools with their own theories, questions, and methods. There is no consensus on any of these components and as such there is no research program that can guide the whole group’s research. In the context of cognitive neuroscience, as mentioned previously, there is no consensus on the logic that translates neural activity into thought and action. Each of the competing schools is fair game, and as such, Kuhn stipulates that all the facts that could possibly uncover this logic seem equally relevant (SSR, 15). One of the greatest challenges to discovering this logic lies in disagreement on questions being asked, and as a result on the theories and methods used to answer them. This point is two pronged, as there is both disagreement on the natureof the problems, and what problems are considered important or not. I will provide an example of both aspects of this argument from the field of cognitive neuroscience.
One recent debate in cognitive neuroscience is between the predictive processing theory and the representational framework for describing brain function in the neocortex. The representational framework focuses on bottom-up-driven sensory computations where portions of sensory space known as receptive fields can elicit neuronal responses when stimulated[iv]. The representational framework gave rise to the notion of feature detection that came to prevail cortical function. Under this framework, sensory systems in the brain extract invariant representations from sensory input, and then separate parts of the brain are tasked with deciding and acting upon that representationiv.
Though the representational framework has gained a lot of traction in recent years, there is no consensus on whether it provides an adequate description of the nature of neural processing of external information. The main qualm is with the perception-driven bottom up sensory input because it fails to explain how animals distinguish self-generated sensory feedback from externally generated input. The optokinetic reflex, for example, does not prevent self-motion of the eye, which indicates that the brain makes this distinction between self-generated sensory feedback and externally generated input. In response to this discrepancy, neuroscientists concluded that the brain generates internal models that are equivalent to a simulation of the external world iv. These internal models predict sensory input based on movements and past sensory experience. The predictive processing framework comes in different flavors such as predictive coding, hierarchical temporal memory and Bayesian inference all based on the idea that the neocortical function is comprised of a generative model of the world used to predict sensory input iv.
There is ample evidence for both the predictive processing framework and the representational framework both on a physiological level and circuit level, however there is no consensus on which framework more accurately describes the nature of cortical function. As can be expected, each framework prompts its own questions and methods, and each has its own ramifications on cognitive processes such as decision making, attention and memory. The very nature of the problems in cognitive neuroscience thus vary according to the framework adopted.
A further pre-paradigmatic feature of cognitive neuroscience is lack of consensus on what questions are considered important. To illustrate how this pertains to cognitive neuroscience, I will use the example of the study of consciousness as there is a divide on whether it constitutes a so-called “acute” problem that scientists should be focusing on (SSR, 24). This is due to a discrepancy in metaphysical assumptions and commitments as well as agreement on appropriate methods. As of now there are countless different theories of consciousness spanning a breadth of neural signs and mechanisms. Conceptions of consciousness traditionally bifurcate into access and phenomenal consciousness. Access consciousness refers to conscious states that can be reported by virtue of high-level cognitive functions such as memory and decision making[v]. Phenomenal consciousness refers to the subjective aspect of experiencing the world. In the neuroscience community, many well-regarded scientists with different metaphysical commitments will dispute over the viability of phenomenal consciousness and whether it is scientifically intelligible. Some say that it is an illusion, others that ignoring it will miss the essence of consciousness, yet other that it exists but cannot be studied with the tools of science.
To a large extent, the disagreement on whether this is an important problem for cognitive neuroscience hinges on the fact that there is no unified definition of what consciousness is and what aspects are essential for understanding it. For example, the main explanandum of different theories will vary[vi]. Some theories such as the Synchrony Theory (ST) seek to explain phenomenal consciousness and the content of consciousness. On the other hand, the Global Neuronal Workspace Theory (GNWT) completely disregards phenomenal consciousness and instead exclusively focuses on access consciousness vi. They differ vastly in the techniques used, the modalities studied, as well as the neural signatures they find. For example, ST employs animal physiology via electroencephalogram (EEG) and magnetoencephalogram (MEG), and uses vastly different behavioral markers to manipulate consciousness. Further, it regards neural synchrony as its neural signature and studies consciousness early after stimulus onset. GNWT theory on the other hand also considers neural activity through functional magnetic resonance imaging (fMRI) and computational modeling. As opposed to synchrony, the neural signatures in the GNWT theory consist of late EEG signals and activity in the dorsolateral prefrontal cortex corresponding to late onset of consciousness vi. Solely from these two theories alone out of the slew of theories that exist, one can see that the object of investigation and the methods with which it is investigated are almost incommensurable.
Various Counterarguments
Now that I have argued the pre-paradigm nature of some subfields of neuroscience, I would like to consider some counter arguments the claims presented. One counter argument is that tool development in the field of cognitive neuroscience has been revolutionary and hence the field has already undergone cycles of paradigm shifts. For example, the discovery of magnetic resonance imaging or various stimulation techniques such as optogenetics expanded the realm of possible inquiry in the field considerably. However, I argue that tool development is just a feature of puzzle solving in normal science. As tools themselves are built to produce a certain kind of result, they are functions of the frameworks they arise from. In the case of the predictive processing framework emerging tools and methods are being built to effectively access internal models and control over associated predictions with the hopes of being able to disambiguate between neural activity associated with bottom up representation and the neural activity associated with predictive processing hypotheses iv. These new tools are notscientific revolutions as they do not require a change in concepts and a degree of incommensurability between the pre and post-tool frameworks. The language being used before and after the tool development is constant, and the tools merely elucidate clearer more precise measurements and experiments within the same framework.
Another counter argument to the claim that there are non-paradigmatic sub areas of neuroscience is that it is at odds with the collective progress being made in psychopharmacology, psychiatry and neurology. This argument stems from the fact that cumulative progress of the type seen in medical advancements is a sign of a science being in the normal science phase. Granted, in recent years there have been monumental achievements in psychopharmacology and neurology. Neurological and psychiatric diseases that were previously enigmatic and fatal are now within the realm of treatment. Given these impressive feats, it seems as though there is collective progress being made in the field of neuroscience that undergirds these medical advancements. However, I argue that this progress is mainly in diagnostics, which is largely dependent on new tools such as fMRI and MRI. Furthermore, the treatments that do exist rely heavily on some of the other sub-paradigms of neuroscience. For example, Selective Serotonin Reuptake Inhibitors (SSRIs) are the product of normal science puzzle-solving within the chemical synaptic transmission paradigm. Deep Brain Stimulation likewise is a product of the normal science puzzle solving in the electrical signaling paradigm. I would argue that there is as of yet insufficient and ineffective treatment for many cognitive impairments in dementia, as well as aberrations in conscious states in various psychiatric disorders such as schizophrenia. Despite sincere efforts, the lack of a paradigm in cognitive neuroscience precludes precisely the type of collective progress that can result in ground breaking neurological treatments.
Conclusion
In this paper, I have shown that neuroscience consists of a manifold of different paradigms each operating at their own pace, with a certain sub field, namely cognitive neuroscience, still in a pre-paradigm phase. I have demonstrated this due to the absence of two crucial components of a paradigm, namely a common theoretical framework and consensus on challenging problems that scientists would tackle in the normal science phase. This complicates the issue of labeling certain fields as mature, as they may have subfields that have reached the stage of maturity, and others that have not. I argue that the logic of translating neural activity to higher level cognitive processes such as thought and action will constitute the first paradigm in the realm of cognitive science, just as the central dogma constituted the first paradigm in the realm of molecular genetics. This I believe is and should be the central goal of the field of cognitive neuroscience, as without it there will be no semblance of collective progress.
[i]Parker, D. Kuhnian revolutions in neuroscience: the role of tool development. Biol. Philos.33, 17 (2018).
[ii]López-Muñoz, F., Boya, J. & Alamo, C. Neuron theory, the cornerstone of neuroscience, on the centenary of the Nobel Prize award to Santiago Ramón y Cajal. Brain Res. Bull.70, 391–405 (2006).
[iii]Richard Axel. Neuron99, 1110–1112 (2018).
[iv]Keller, G. B. & Mrsic-Flogel, T. D. Predictive Processing: A Canonical Cortical Computation. Neuron100, 424–435 (2018).
[v]Cohen, M. A. & Dennett, D. C. Consciousness cannot be separated from function. Trends Cogn. Sci.15, 358–364 (2011).
vi Northoff, G. & Lamme, V. Neural signs and mechanisms of consciousness: Is there a potential convergence of theories of consciousness in sight? Neurosci. Biobehav. Rev.118, 568–587 (2020).