science – Glitter Resist

Flaws in the Philosophy of medicine

Posted on Sep 9, 2025Sep 9, 2025 by Glitter Resist by David Scott

Medicine makes philosophical and epistemological assumptions about nonlinear systems that are only always true for linear systems. In this blog, this collection of assumptions are referred to as the Classical Scientific Method and includes that diseases are always caused by a direct identifiable malfunction and that classical scientific and linear methods are always valid to investigate all phenomena. These assumptions are not always true for nonlinear systems (e.g. Bertalannfy, 1968), and medicine otherwise fails to fully recognise that these systems are qualitatively different to linear systems, and particularly that nonlinear systems may be counterintuitive. These fail patients by oversimplifying the complexities of human biologies. The first step is to define a linear system and a nonlinear system, and to state their qualitative philosophical and epistemological characteristic and differences.

A linear system is defined as a system that complies with the superposition principle, namely the properties of additivity and homogeneity, and thus the variables are governed by linear equations, and the input directly proportionate to the output. This generally means that systems that produce regular predictable patterns (like an analogue clock) are linear, whilst systems that are chaotic (erratic with less predictable behaviours like the weather) are not. For linear systems, understanding their component parts means (intuitively) understanding the whole, a view called Reductionism (Descartes, 1637), and forms part of the Classical Scientific Method. Reductionism means that system failures are always caused by component failures. This definition of linear systems for real-world systems, and particularly biological systems, is not always helpful because all real-world systems are nonlinear. Even if systems appear to or act linearly normally, real-world systems are always nonlinear because of real-world physical, biochemical, and mathematical constraints, e.g. up- and downregulation of morphine receptors (i.e. MOR) may be assumed as linear ordinarily, but if morphine is prescribed (a MOR agonist) then receptors cannot downregulate indefinitely thus leading to a discontinuity, which is nonlinear. The characteristics of a linear system are thus that failures can only occur from component failures, system interactions can be easily understood, systems can always be easily combined, systems are always deterministic, and systems are always intuitive.

In contrast, a nonlinear system is defined as one that does not always comply with the superposition principle, which means that the variables are governed by at least one nonlinear equation and the input is thus not directly proportionate to the output, e.g. the morphine receptor example earlier and likely explaining morphine’s physical dependency due to slower pathways attempting to restore homeostasis. The characteristics of a linear system are therefore that component failures are not needed for system failures, system interactions can be difficult to predict, systems cannot necessarily be easily or always combined, systems may not be deterministic, and systems can be counterintuitive. The study of nonlinear systems in motion is referred to as nonlinear dynamics. A comparison between the philosophical characteristics of a linear system compared to a nonlinear system is presented in Table 1 below. These twin concepts of linear and nonlinear systems are considered for medicine in the next paragraphs to identify flaws in philosophical thinking.

Table 1: Linear System vs Nonlinear System

Linear System	Nonlinear System
Direct malfunction for system failures e.g. a pathogen or injury.	Direct malfunction for system failures not necessary e.g. emergent behaviours from feedback loops.
System interactions can be easily understood e.g. drug and/or comorbidities interact simply.	System interactions difficult to predict e.g. drug and/or comorbidities interactions complex.
Systems can always be combined meaningfully e.g. combining patient results statistically for studies or meta-analysis is valid.	Systems cannot always or easily be combined e.g. combining patients statistically in studies or in meta-analysis may not always be easy or valid.
Systems are deterministic e.g. variations between identical experiments are from participant differences.	Systems may not be deterministic e.g. variations between identical experiments may be due to real-world nonlinearities not participants.
Systems are intuitive e.g. improving symptoms improves pathology.	Systems may be counterintuitive e.g. improving symptoms may worsen the underlying pathology.

Medicine began to become much more scientific from the 17th Century onwards with the discipline recognisable as a science from the early 19th Century (Tweel 2010). In the 1980s, attempts were being made to unite medical research, training and clinical practice with what was later to be termed Evidence-Based Medicine (EBM). EBM (Sackett, 2000) is concerned with “the conscientious, explicit and judicious use of current best evidence in making decisions about the care of individual patients” (Sackett, 1996, p71) with due regards to the specifics of the clinical case, and the values and wishes of the patient. This paradigm is now the dominant form of medicine practiced in the West.

EBM requires consideration of the certainty of evidence, taking account of all factors, but particularly bias; thus (like other scientific disciplines) considers deductive experiments as having the highest certainty with double-blinded randomised clinical trials having the most. This is principally because these methods more clearly demonstrate causation over correlation (ecological validity notwithstanding), and because blinding and randomisation reduces bias. These trials are principally used to determine efficacy of treatments. Proponents of EBM believe that truth can only be found in these randomised trials (Sackett, 1996).

Figure 1 – Hierarchy of Evidence Pyramid taken from Murad (2016)

On that basis, EBM includes a hierarchy of evidence relating levels of certainty to research methods (Murad, 2016), with systematic reviews and meta-analysis of high certainty Randomised Controlled Trials (RCTs) at the top, followed by RCTs with definitive results, and then everything else from observational to qualitative studies coming afterwards. Systematic reviews consider all the evidence on a medical research question and may include meta-analysis, whereby the results of similar studies are statistically combined. Proponents of EBM also believe that these secondary research methods are “one of the greatest achievements of evidence-based medicine” (Masic, 2015, p219). However, this hierarchy of evidence only includes basic research at the bottom and this research often does not even appear in lists or figures (e.g. Murad, 2016 – shown as Figure 1 above).

Basic research includes bench research, modelling, and other theoretical research. As this research is at the bottom then it is considered to have the least certainty and thus least value with clinicians and other medical researchers, with such research often denigrated by scientists, funding agencies, and the general public (e.g. La Caze, 2011). This affects funding, research priorities, and the ability to publish in mainstream journals. This is despite basic research being key to the advancement of medical research, and particularly for diseases with nonlinear emergent mechanisms. EBM therefore prioritises RCTs, sidelining basic research (e.g. bench or computational studies such as in systems biology) that is essential for studying nonlinear diseases because of their need to be modelled prior to human empirical study. So although EBM works for simple linear cases (e.g. vaccines), its approach fails patients where biological complexity is paramount.

These classical methods used by medicine are similar to those carried out on the other sciences, in that on the basis of assumptions that systems can be considered philosophically as linear, and therefore the superposition principle can be assumed to apply, measurements on those with a disease and those without are taken, which for deductive research is used to test hypotheses. RCTs’ include the primary outcome(s), which are principally related to the efficacy of diagnostics or treatments, and secondary outcome(s), which are principally related to reported side effects and harms. If the potential harms are unknown then this latter part of the study is inductive, i.e. a concurrent observational study. This means that RCTs are normally part deductive and part inductive, despite being considered as having high certainty in the hierarchy. Other confounders are controlled as per other sciences to ensure the relationships detected between variables are due to the effect studied and not some unrelated cause. However, there is generally no consideration of nonlinear confounders, which may lead to inconsistent results between studies.

Medicine adapts the scientific method principally by collecting data from multiple participants during research to attempt to average out differences caused by variability amongst patients, effectively carrying out multiple studies at the same time and averaging the results. For biomarker research as an example, there is an assumed relationship between the biological failure, the biomarker, and the signs and symptoms of the disease, e.g. high fasting blood glucose is usually indicative of Type 2 diabetes. In considering whether systems are linear or nonlinear then it is these biological relationships that matter, which is related to the relationships investigated in studies, though often indirectly. For these studies to be valid then these all require Classical Scientific Method assumptions, that is the relationships must be able to be assumed to be linear philosophically. The reasons for this are because biomarkers may not be deterministic, fluctuating seemingly erratically, or fluctuating deterministically but periodically. These mean that research carried out on one day may draw one conclusion, whilst on another draw the opposite conclusion, and that these are real-world nonlinear effects not merely due to variations in participants and chance. These also mean that classical statistical methods may average away these nonlinear biological phenomena, that is order may be disguised as randomness (Lorenz, 1963).

In these circumstances, biomarkers from participants cannot be statistically combined in studies or in meta-analyses without careful study design or further analysis and/or modelling. For example, common diseases like Type II diabetes or epilepsy for which linear assumptions are unlikely to apply, are still principally studied as linear failures (e.g. faulty genes), ignoring complex biological dynamical interactions and potential emergent failures; phenomena that is difficult (if not impossible) to study using the Classical Scientific Method. There is thus no consideration of emergent failure modes (in the engineering sense) and no consideration as to whether the patient otherwise must be considered philosophically as a nonlinear system. No hypotheses for these or any other similar diseases have considered these points (e.g. Cui, 2023). These issues likely apply to many other quantitative medical studies.

Medicine is therefore assuming Reductionism and using the Classical Scientific Method, and thus practicing a limited and out of date form of the scientific method, which by definition assumes patients philosophically and epistemologically are linear systems. These implicit assumptions have never been stated in any study (e.g. Meyer, 2006) and are not always true. This means that medicine may be failing to understand many common diseases, failing to demonstrate the validity of diagnostics on all patients, and failing to fully confirm the efficacy and safety of treatments, because it’s failing to understand the assumptions and limitations of its own methods. Worse, medicine is deprioritising basic research including modelling that is essential for understanding nonlinear systems. These philosophical flaws drive systemic failures, necessitating reform of medical research and clinical practice to protect and meet the needs of all. Unlike medicine, other sciences like engineering have addressed these philosophical and epistemological flaws and these are discussed in the next blog.

References:

von Bertalanffy L. (1968). General system theory: foundations, development, applications. New York: George Braziller.

Descartes R. (1637). Discourse on method. Translated by Cress DA (1998). 4th ed. Indianapolis: Hackett.

La Caze A. (2011). The role of basic science in evidence-based medicine. Biol Philos. 26, 81–98. https://doi.org/10.1007/s10539-010-9231-5

Lorenz EN. (1963). Deterministic nonperiodic flow. J Atmos Sci. 20(2):130-141.

Masic I, Miokovic M, Muhamedagic B. (2008). Evidence based medicine – new approaches and challenges. Acta Inform Med. 16(4):219-225. https://doi.org/10.5455/aim.2008.16.219-225

Meyer JH, Ginovart N, Boovariwala A, et al. (2006). Elevated monoamine oxidase A levels in the brain: an explanation for the monoamine imbalance of major depression. Arch Gen Psychiatry. 63(11):1209-1216. https://doi.org/10.1001/archpsyc.63.11.1209

Murad MH, Asi N, Alsawas M, Alahdab F. (2016). New evidence pyramid. BMJ Evid Based Med. 21(4):125-127. https://doi.org/10.1136/ebmed-2016-110401

Sackett DL, Rosenberg WM, Gray JA, Haynes RB, Richardson WS. (1996). Evidence based medicine: what it is and what it isn’t. BMJ. 312(7023):71-72. https://doi.org/10.1136/bmj.312.7023.71

Sackett DL, Straus SE, Richardson WS, Rosenberg W, Haynes RB. (2000). Evidence-based medicine: how to practice and teach EBM. Edinburgh: Churchill Livingstone.

van den Tweel JG, Taylor CR. (2010). A brief history of pathology: preface to a forthcoming series that highlights milestones in the evolution of pathology as a discipline. Virchows Arch. 457(1):3-10. https://doi.org/10.1007/s00428-010-0934-4

Why Medicine is doing science wrong

Posted on Mar 31, 2025Apr 3, 2025 by Glitter Resist by David Scott

Medicine assumes that diseases are always caused by some identifiable biological malfunction, i.e. system failures are always caused by component failures. Despite any protestations to the contrary, this is an assertion of Cartesian Reductionism^[1], namely that any system can be divided into its component parts and the sum of those parts thus explains the whole. Further, methods that medicine prioritises and/or considers epistemologically valid to investigate diseases, also assume Cartesian Reductionism. These mean that medicine is practicing a limited form of the scientific method, which by definition requires systems to be assumed philosophically and epistemologically to be linear; medicine therefore assumes philosophically and epistemologically that patients are linear systems in all circumstances. Linear systems in this context being ones to which the superposition principle applies^[2]^,[3], whilst nonlinear systems are those that do not.

This implicit assumption has never been stated in any study and is not always true; patients in a substantial minority of cases must be considered philosophically and epistemologically as nonlinear systems^[4]. This philosophical and epistemological mistake was identified and corrected in engineering decades ago, largely following Three Mile Island. This was the worst nuclear accident in North America and forced engineers to identify and fix their discipline’s philosophical and epistemological problems^[5].

Of principle relevance to medicine philosophically is that nonlinear systems may have emergent failures for which there are no identifiable biological malfunctions, rather system failures may emerge from components interacting adversely and dynamically. An example of an emergent failure mode is provided in the first Johnny and Susie example below.

Of secondary relevance is that nonlinear systems can be counterintuitive, which means that what someone thinks is the right thing to do is in fact the opposite of the right thing, that is if the system is assumed to be a linear but is in fact nonlinear, then attempts to correct failures in the system can result in making the problem worse.

Of principle relevance epistemologically is that nonlinear systems may not be deterministic and/or may be oscillating abnormally as emergent failure modes and self-reinforcing pathological dynamical steady-states. These mean that when investigating such systems using linear methods, then results obtained may appear random but are in fact ordered nonlinearly, meaning that statistics cannot always be easily applied to these results or at all. The second Johnny and Susie example below illustrates how statistics can mislead when used on nonlinear systems, whilst the third Johnny and Susie example below provides an example of an emergent failure mode in the form of a self-reinforcing pathological dynamical steady-state discussed above. (Note that the transition from linear to oscillatory behaviour in the third example is usually referred to as a dynamical phase transition, analogous in hydrodynamics to the transition from laminar to turbulent flow.)

Failure to assert in studies that the patient can be assumed philosophically and epistemologically to be a linear system has rendered all quantitative medical studies as scientifically invalid, and where the assumption is not true, rendered studies both invalid and wrong. Multiple other obvious ordinary mistakes have also been identified.

As the differences between autistic people and neurotypicals are due to the former having higher neuronal densities^[6] than the latter, as higher neuronal density results in qualitative or nonlinear differences in responses to stimuli both external and internal, normal or pathological, and as medicine has failed to account for these qualitative or nonlinear differences because patients are wrongly assumed philosophically and epistemologically to be linear systems; medical knowledge for diagnostics and treatments is consequently dangerously wrong for autistic people, leading to our harms and deaths. Nonlinear differences between groups of people were of course the root cause of the thalidomide defects (the between groups being men and pregnant women in this case), and it’s disappointing that the wider lessons of this scandal were not learnt. The fourth and final Johnny and Susie example below shows how network behaviour is nonlinear pursuant to Network Theory.

The philosophical, epistemological, and other mistakes that directly affect autistic people include:

Failure to recognise that patients cannot always be assumed philosophically and epistemologically to be a linear system and thus consequent failure to assert in studies that this assumption is valid or to amend operationalisations when not.
Failure to recognise the existence of emergent failure modes, which are unique to nonlinear systems. Unlike every other similar scientific discipline like engineering, there is no concept of emergent failure modes or emergent pathologies in medicine, only linear pathologies.
Failure to recognise that nonlinear systems can be counterintuitive.
Failure to account for nonlinear confounders or even recognise their existence.
Failure to recognise that nonlinear systems may not be deterministic and/or may be oscillating periodically (occurring in cycles) or chaotically (occurring in episodes) as emergent failure modes, and thus that ordinary scientific and statistical methods as only used in medicine may give meaningless results.
Failure to recognise that increased neuronal density in autistic people is the reason for our differences in cognition and behaviours, pursuant to Information Theory and Network Theory.
Failure to recognise that these differences are qualitative or nonlinear compared to neurotypicals and that thus medical knowledge is dangerously wrong for autistic people. This includes diagnostics and treatments, particularly medications’ side effects which may be atypical or dangerously worse.
Failure to recognise that consequently there are unique pathophysiologies for autistic people.
Failure to recognise that drugs’ effects may be nonlinear and that these effects may increase not decrease over time. This is more likely to be true for autistic people and can be fatal.
Failure to recognise that side effects may be nonlinear and thus may be somewhat unpredictable with new side effects appearing and existing ones worsening over periods of time. This is more likely to be true for autistic people and can be fatal.
Failure to recognise that the liver enzyme CYP2D6^[7] is also present in the brain in most people where it deactivates dopamine. Autistic people are more likely to be high or ultra-high metabolisers leading to dangerously nonlinear effects from drugs that affect this enzyme.
Failure to recognise that because of this most antidepressants are little more than modern day cocaine as almost all inhibit CYP2D6, with associated effects on most people. These effects will be worse on autistic people.
Failure to recognise that most antipsychotics induce CYP2D6 thus in most people such drugs would be dysphoric, and with some painfully so. The latter would be more likely autistic.
Failure to recognise that the effects on CYP2D6 may cause akathisia in some people, and these people are more likely to be autistic.
Failure to recognise that abruptly stopping most antidepressants may cause catatonia or neuroleptic malignant syndrome (which can be fatal) on some people, and these people are more likely to be autistic.
Failure to recognise that antipsychotics that induce CYP2D6 may cause catatonia or neuroleptic malignant syndrome on some people, and these people are more likely to be autistic.
Similar failures for a host of non-psychiatric drugs that inhibit or induce CYP2D6.
Failure to recognise that normal and abnormal blood chemistry levels and vitals are not the same for autistic people because of nonlinear differences compared to neurotypicals pursuant to Network Theory, and because autistic people have on average larger brains with consequent greater metabolic and other related needs.
Failure to recognise the existence of post-partem ME/CFS, an illness autistic women are more likely to suffer from.
Failure to recognise that even mild kidney damage has the potential to harm and kill autistic people because our blood chemistry needs to be controlled more precisely for neurological reasons pursuant to Network Theory. No medication that may harm the kidneys states that such should be avoided by autistic people and/or only taken under medical supervision where kidneys can then be closely monitored.
Failure to recognise that fortifying foods with calcium may kill autistic people who have only mild kidney damage, because increases in blood calcium may cause an autistic brain to overexcite leading to seizures, heart attacks, strokes, or death.
Failure to recognise a similar problem for calcium containing supplements or medications with no warning that autistic people should avoid these or only take them under medical supervision.
Failure to recognise that triage as practiced by medicine is dangerous for autistic people because of potentially fatal rapid neurological transients including seizures that are more likely to occur on autistic people, leading to inappropriate prioritisation and consequent autistic harms and deaths.
Failure to recognise nutrition advice that resistant starches are healthy has doomed autistic people to pain and suffering because such starches are not very tolerable by us. Medicine is aware that some people cannot tolerate resistant starches, but those people are mostly autistic.
Failure to recognise that statutory requirements for medical treatment authorisation are woefully inadequate because they fail to account for nonlinear treatment effects and side effects, fail to account for interactions with nonlinear diseases like ME/CFS and epilepsy, fail to account for other nonlinear differences like autism, fail to consider whether the treatment in improving symptoms worsens the underlying pathology, and fails to attempt to explain the biological reason for most side effects, consequently failing to make (obvious) predictions of potentially harmful effects on large groups of people, including autistic people.
Failure to recognise that in using statistics to attempt to provide treatments that are safe and effective on most people, medicine is harming or killing or allowing to die those at genetic extremes (which includes autistic people). This is a form of eugenics and could eventually lead to humanities extinction.

The four problem areas that I have identified leading to autistic harms and deaths by medicine include the following:

Autistic people have different pathophysiologies to diseases that may affect the brain, either directly or indirectly, leading to different signs, symptoms, and risks.
Medication that is known or suspected to affect the brain, either as treatment or side effect, may have an atypical and/or exaggerated effect or side effect on autistic people.
Vitals and blood chemistry that are considered normal and abnormal are different for autistic people. These include (non-exhaustive) minimum and normal BP, body temperature, O₂, CO₂, blood glucose, sodium, potassium, calcium, magnesium, bicarbonates, and blood pH.
Signs and symptoms for diseases otherwise are different for autistic because of differences in sensory processing.

A report has thus been filed at the International Criminal Court alleging that the NHS is committing crimes against humanity against autistic people. This report can be considered medicine’s Three Mile Island moment.

[1] “Discourse on the Method of Rightly Conducting One’s Reason and of Seeking Truth in the Sciences” by René Descartes (1637).

[2] “Penguin Dictionary of Physics” by J Cullerne (2009).

[3] “Penguin Dictionary of Mathematics” by D Nelson (2008).

[4] Medicine uses the phrase ‘nonlinear system’ to refer to any system with nonlinear behaviour (which is not the definition in engineering) but then makes philosophical and epistemological assumptions about the system that only always apply to linear systems.

[5] “Normal Accidents” by Charles Perrow (1984).

[6] “Autism spectrum disorders pathogenesis: Toward a comprehensive model based on neuroanatomic and neurodevelopment considerations” by A Beopoulos et al in Frontiers in Neuroscience(2022).

[7] “The neuroprotective enzyme CYP2D6 increases in the brain with age and is lower in Parkinson’s disease patients” by A Mann et al in Neurobiology of Aging (2012).