Psychiatry - Science and Pseudoscience

I finished the first chapter in Philosophy of Pseudoscience: Reconsidering the Demarcation Problem by Pigliucci and Boudry.  I became aware of Massimo Pigliucci and his work back in 2002 when I read his book on Intelligent Design and since then have discovered his blog Rationally Speaking where he has recently posted his best papers from his careers as a scientist and a philosopher.  He writes very clearly on the philosophy of science and has such a command of the field that he can include a history lesson of relevant references.  He also does not shy away from controversy or the apparent lack of a clean solution to a problem.  One of the central concepts in his chapter is this chart of empirical knowledge versus theoretical understanding. (click to enlarge)

The purpose of this essay is to look at possible boundaries between science and pseudoscience as well as a couple of interesting observations as they apply to psychiatry.  One of his key concepts is that the lines of demarcation are not necessarily sharp and the variables are not necessarily linear.  He uses the above graph of empirical knowledge versus theoretical understanding as an example.   Starting in the upper right corner of the diagram we have hard sciences with particle physics given as the most clear cut hard science.  I like to think about my undergraduate chemistry experience as being hard science.  Even introductory chemistry exposes the student to an amazing array of facts, observations, and theories that are incredibly accurate.  From there, chemistry majors build on their ability to measure specific compounds, synthesize them and study the theory in Physical Chemistry.   I don't think that there is any doubt that chemistry as a field is not too far removed from particle physics in terms of empirical knowledge or theoretical understanding.  String physics has much theory but is low in terms of empirical support.  He refers to evolutionary psychology,  scientific history and Search for Extraterrestrial Intelligence (SETI) as a "proto-quasi science" cluster with decreased amounts of theory and empirical support.  Other fields like the so-called "soft sciences" of sociology, economics, and psychology have a fair amount of empirical knowledge but less theoretical understanding.  The true pseudosciences are in the zone with astrology, HIV denialism, and Intelligent Design.  From the history of psychiatry - Freudian psychoanalysis and Adlerian psychology would also be included here but there is also a list of theories from general medicine and surgery that would also qualify.

In psychology and psychiatry a central philosophical problem is the so-called hard problem or the explanatory gap between the neurobiology of conscious states and subjective experience.  This is exactly where psychiatry resides.  A lot of political criticism of psychiatry involves the ability to parse these states and accurately classify different conscious states.  Resolving the hard problem would move psychiatry and psychology firmly to the right in the demarcation diagram but probably not nearly as far as particle physics or maybe not even as far as molecular biology.

The relevant question for me of course is where psychiatry fits on the plane of empirical knowledge x theoretical understanding.  What about medicine in general?  Could we plot a plane of medical and surgical sub specialties on this plane instead of the hard and soft sciences?  Does medicine and surgery have theories or practices end up in the same zone as Freudian psychoanalysis.  Of course they do.  A great example from my days as a medicine intern was highlighted by Ghaemi as "The cult of the Swan-Ganz catheter."  In the places where I trained, anyone with moderately serious cardiopulmonary problems was at risk for placement of a Swan-Ganz catheter.  The actual person inserting the catheter could be a medicine resident, a cardiologist, or an anesthesiologist.  Since the intern is responsible for doing the initial history and physical exam, I witnessed the placement of a large number of these catheters.  Once placed they gave an impressive number of parameters on ICU monitors.  We were routinely grilled about the meaning of these parameters by attending physicians on rounds.  It all seemed very scientific.  The cult of the Swan-Ganz catheter was subsequently disproved by randomized clinical trials.  This standard of care from the 1980s and 1990s disappeared much faster than Freud.

The best way to plot medicine and psychiatry on Pigliucci's empirical knowledge versus theoretical understanding plane would be to consider the clinical basic sciences taught in the first two years of medical school.  In my experience that was anatomy, neuroanatomy, histology, microbiology, biochemistry/molecular biology, genetics, pathology, physiology, pharmacology, epidemiology, and statistics.  Practically all clinical specialties carry these basic sciences forward in one form or another.  The research literature in any particular specialty in full of theory and techniques from these basic sciences.  The psychiatric literature cuts across all of the basic sciences in the same way as other specialties.  At the minimum, some of psychiatry will be at the level of molecular biology on the diagram in some areas and at the level of psychology in others.  Hopefully the unscientific theories will be relegated to the lower left hand corner of the diagram as unscientific and not stand the test of time.

I think that Professor Pigliucci's conceptualization is a very useful one.  I expect that he will continue to refine these ideas.  I think that measurement precision and categorization may be important dimensions to add to these concepts.  As Merskey has pointed out both the phone book and the periodic table are much more accurate forms of categorization than any scheme of medical classification.  I think that probably says a lot about the underlying scientific dimensions and how measurement is done.


George Dawson, MD, DFAPA

Ghaemi SN.  A Clinician's Guide to Statistics and Epidemiology in Mental Health.  (2009) Cambridge University Press, Cambridge, UK.  p. 91.

Two Undergrad Experiments to Illustrate - What Is Science?

I have always been somewhat of a science nerd and had what I consider to be a first rate science education at a liberal arts college.  I had gone to this school on a football scholarship with the intention of becoming a phy ed teacher and a football coach, but the science education there was too compelling to ignore.  I ended up being a biology and chemistry major and rapidly forgot about football.  Brook's essay of what is science, what is the most compelling science and how that  makes psychiatry as far from science as possible led me to think about memorable experiments from my undergrad days.

Experiment 1:  The Limnology experiment:   For a while in my undergrad career, I considered being a limnologist or fresh water biologist.  My undergrad college was one of the first to emphasize the environment and ecology.  A lot of the work involved doing population estimates of plankton and aquatic invertebrates.  We spent hours classifying and counting thousands of organisms  that are unknown to most people. We used various sampling techniques and statistics to determine populations of these organisms and whether they seemed to be influenced by any environmental variables.  At one point I had equations from an journal article to calculate the probability that a specific species would be in contact with another one - called the "probability of inter species interaction."  This is biological science.

Experiment 2:  The PChem experiment:  Physical Chemistry was the undergrad chemist's dream course when I was in college.  You dreamed that you would be able to pass it.  We had a text that was not very accessible, but a professor who was brilliant, very accessible and an excellent lecturer.  I liked it a lot after we finished thermodynamics and moved on to other topics.  Back in the 1970s we had very primitive computing power. Our lab had an old HP calculator that was as big as a current desktop with less computing power than a modern day scientific calculator.  One of our tasks was to estimate electron densities around carbon atoms in aromatic hydrocarbons.  In an afternoon in the lab we ran the numbers.  This was the science of physical chemistry.

I have intentionally left out all of the details of the experiments because for the purpose of comparison with Brooks thesis they are unnecessary.  From his essay we learn that biology and chemistry are real sciences with a "distinctive model of credibility".  The examples I have given are from those fields.   We learn that psychiatry is a "semi-science" because "the underlying reality they describe is just not as regularized as the underlying reality of say, a solar system".  I will stop at that point because Brooks further examples rapidly degenerate.  What do we have so far?

Looking at my experiments, #2 clearly has the regularity of a solar system.  What could be more regular than the electron density for a specific molecule?  It fits Brooks definition of science to a tee.  What about experiment #1, the biological experiment?  Here we have a number of organisms.  Some have nervous systems and the others (eg. phytoplankton) do not.  I did a series of calculations to look at the probability of one species encountering another.  There were certain assumptions to those calculations about randomness to make the calculation much easier to do.  But what if I wanted for a moment to be a "behavioral limnologist" and attempt to predict the behavior of a specific stoneflies in the sample?  What if I wanted to determine the 5%  of stoneflies that exhibited behavioral characteristics, that differentiated them from the other 95%?  Suddenly we have a problem.  The source of that problem is a nervous system.  The underlying reality of most even slightly complicated nervous systems is that they will never have the regularity of a physical system.  They have evolved not to.  Regularity in a nervous system locking it into a physically predictable system is not in any way adaptive for any animal that needs to forage and reproduce.  It is the kiss of death.

But is gets complicated at additional levels.  The human brain is highly evolved to have significant processing power.  At another level, there are theoretical concerns about whether it is possible to ever to map behaviors and psychiatric symptoms directly onto some neurobiological system.  Unlike my experiment 1 above we are rarely interested in looking at only life or death as the outcome variable.  The variables that will allow us to study different populations are going to be much more complex than grossly observed behaviors.  There is a complicated nervous system between those behaviors and the environment.

Is psychiatry really not a science because it is complex and attempts to deal with the complicated phenomena associated with the human brain?   Should we ever be concerned about 1:1 mappings of psychiatric disorders onto a specific genetic or neurobiological defect?  Is it possible that a human nervous system is so complex that it is unrealistic to expect that this might happen?

Unlike Brook's theme nobody is a "Hero of Uncertainty".  Uncertainty is the expected condition and one that every psychiatrist should be comfortable with.  Psychiatry and the associated neurosciences will never be reduced to the predictable calculation of a physical system and that has nothing to do with one being a more prestigious science.  It has to do with evolution and complexity.  It has to do with what philosophers call the "demarcation problem" between what is and what is not science.  More to come on that in the near future.

George Dawson, MD, DFAPA

On the Validity of Pseudopatients

Every now and again the detractors and critics of psychiatry like to march out the results of an old study as "proof" of the lack of validity of psychiatric diagnoses.  In that study,  8 pseudopatients feigned mental illness to gain admission to 12 different psychiatric hospitals.  The conclusion of the study author was widely seen as having significant impact on the profession, but that conclusion seems to have been largely retrospective.  I started my training about a decade later and there were no residuals at that time.  I learned about the study largely through the work of antipsychiatrists and psychiatric critics.

Several obvious questions are never asked or answered by the promoters of this test as an adequate paradigm.  The first and most obvious one is why this has not been done in other fields of medicine.  It would certainly be easy to do.  I could easily walk into any emergency department in the US and get admitted to a Medicine or Surgical service with a faked diagnosis.  I know this for a fact, because one of the roles of consulting psychiatrists to Medicine and Surgery services is to confront the people who have faked illness in order to be admitted.  Kety (9) uses a more blunt example in response to the original pseudopatient experiment (1):

"If I were to drink a quart of blood and, concealing what I had done, come to the emergency room of any hospital vomiting blood, the behavior of the staff would be quite predictable. If they labeled and treated me as having a bleeding peptic ulcer, I doubt that I could argue convincingly that medical science does not know how to diagnose that condition. "(9)

I also know that this happens because of the current epidemic of prescription opiate abuse and the problem of drug seeking and being successful at it.  An estimated 39% of diverted drugs (7) come from "doctor shopping."  By definition that involves presenting yourself to a physician in a way to get additional medications.  In the case of prescription opioids that usually means either faking a pain disorder or misrepresenting pain severity.  So it is well established that medical and surgical illness well outside of the purview of psychiatry can be faked.  And yet to my knowledge, there is hardly any research on this topic and nobody is suggesting that medical diagnoses don't exist because they can be faked.  Does that mean the researchers consider the time of these other doctors too valuable to waste?  More likely it did not fit a preset research agenda.

The second obvious question has to do with conflict of interest.  It is currently in vogue to suggest that psychiatrists are swayed in their prescribing practices by incentives ranging from a free pen to a free meal.  Compensation as a company employee or to give lectures is also thought of as a compromising incentive. The free pen/free meal incentive is pretty much historical at this time.  What about intentionally misrepresenting yourself?  What is the conflict of interest involved at that level and how neutral can you stay when you are trying to escape detection in order to prove a point?  A vague script like a mono-symptomatic presentation of schizophrenia should suggest that the intent is to escape detection.  How should a person with a vague script act when they are face to face with a real clinician?  The logical conclusion is that they would be as evasive as possible even if they were adhering to that protocol.

The bottom line is that the pseudopatient experiments were seriously flawed out of the box.  Continuing to promote them as meaningful reflects a serious lack of scholarship in reading the relevant literature and a need to suspend the reality that in fact mental illness does exist, that distinctions can be made among various types of mental illness, and that those distinctions are useful to psychiatrists trying to help people with those problems.

George Dawson, MD, DFAPA

1: Rosenhan DL. On being sane in insane places. Science. 1973 Jan 19;179(4070):250-8. PubMed PMID: 4683124.

2: Fleischman PR, Israel JV, Burr WA, Hoaken PC, Thaler OF, Zucker HD, Hanley J, Ostow M, Lieberman LR, Hunter FM, Pinsker H, Blair SM, Reich W, Wiedeman GH, Pattison EM, Rosenhan DL. Psychiatric diagnosis. Science. 1973 Apr 27;180(4084):356-69. PubMed PMID: 17771687.

3: Bulmer M. Are pseudo-patient studies justified? J Med Ethics. 1982 Jun;8(2):65-71. PubMed PMID: 7108909; PubMed Central PMCID: PMC1059372.

4: Spitzer RL, Lilienfeld SO, Miller MB. Rosenhan revisited: the scientific credibility of Lauren Slater's pseudopatient diagnosis study. J Nerv Ment Dis. 2005 Nov;193(11):734-9. PubMed PMID: 1626092

5: Spitzer RL. More on pseudoscience in science and the case for psychiatric diagnosis. A critique of D.L. Rosenhan's "On Being Sane in Insane Places" and "The Contextual Nature of Psychiatric Diagnosis". Arch Gen Psychiatry. 1976 Apr;33(4):459-70. PubMed PMID: 938183.

6: Zimmerman M. Pseudopatient or pseudoscience: a reviewer's perspective. J Nerv Ment Dis. 2005 Nov;193(11):740-2. PubMed PMID: 16260928.

7: Inciardi JA, Surratt HL, Cicero TJ, Kurtz SP, Martin SS, Parrino MW. The "black box" of prescription drug diversion. J Addict Dis. 2009 Oct;28(4):332-47.  PubMed PMID: 20155603; PubMed Central PMCID: PMC2824903.

8: Millon T. Reflections on Rosenhan's "On being sane in insane places". J AbnormPsychol. 1975 Oct;84(5):456-61. PubMed PMID: 1194506.

9: Kety SS. From rationalization to reason. Am J Psychiatry. 1974 Sep;131(9):957-63. PubMed PMID: 4413516.