Anecdotal Evidence and Inductive Conclusions in Psychiatry Part I

“Can a valid diagnosis be arrived at if laboratory tests are not available? Of course it can. Total reliance on on lab tests to the exclusion of other factors when seeking validity is not science but scientism” ~ Hannah Decker, Ph.D

In my post of April 1, 2010, Is “Data” the Plural of “Anecdote?” I pointed out that in most hard sciences, stories about a scientist or practitioner’s personal experience can be contrasted with “empirical” data from studies, and that the latter usually has more validity that the former.  

As I also pointed out, however, “…many of the randomized, placebo-controlled, double-blind drug studies can be every bit as biased as the worst anecdotal evidence. We are now living in an age of 'marketing-based medicine,' as opposed to evidence–based medicine" (randomized = the use of chance alone to assign the participants in an experiment or trial to different groups; placebo = inert sugar pill; double-blind = neither the person evaluating the research subject nor the subject knows during the experiment if the subject received the placebo or the active medication).

In psychiatry, the situation is even more complex, since there are almost no objective lab tests and we can't read minds, so we must rely on the experimenter's personal observations, or answers by subjects to psychological tests in which we do not know how honest subjects are being or even if they really understand the questions, or patient self reports. All of these sources of data, while important, are highly subjective and can hardly be considered "empirical" in the same sense as, say, a chemical reaction.

Another big point about anecdotes is that they are not worthless in science. Some of the most important discoveries in all of medicine were based initially on anecdotal evidence. In Part II of this post, I will describe how to determine the validity and generalizability of conclusions generated from a group of “anecdotal” clinical observations. That is, how can we tell if the anecdotes are the typical situation seen with a particular combination of clinical variables and which therefore can lead to a conclusion that will probably apply to most patients with a given disorder? Are the conclusions probably true, or probably false? 

In this post, however, I want to contrast conclusions based on a “clinical anecdote” with inductive reasoning. Remember, the first step in the scientific method is observation.

According to Wikipedia, inductive reasoning (as opposed to deductive reasoning) is reasoning in which the premises seek to supply strong evidence for, but not absolute proof of, the truth of the conclusion. It is usually based on seeing the same phenomena over and over again without any exceptions.

The truth of an inductive argument is supposed to be probable, based upon the evidence given. The same phenomena involving two variables are observed repeatedly, and other variables which correlate with the phenomena are shown to be not necessarily connected it, and a tentative conclusion or assumption is drawn regarding the relationship between the two variables.

Deductive reasoning, whose conclusions are more certain, is the process of reasoning from one or more general statements (premises) to reach a logically certain conclusion. Deductive reasoning links premises with conclusions. If all premises are true, if the terms are clear, and if the rules of deductive logic are followed, then the conclusion reached is necessarily true.

To anyone familiar with formal logic, deductive reasoning takes the form of a syllogism:

All men are mortal

Socrates is a man

Therefore, Socrates in mortal

If the first two statements are true, then the third statement must be true. Deductive reasoning is considered the entire basis of science by some scientifically illiterate individuals. 

The problem is that both premises in the above examples are inductive conclusions! Just because all observed men have turned out to be mortal does not in any way prove that the next man you come across cannot be immortal!

The first thing you should have learned from your Geometry 1 class in high school is that there is absolutely no way to prove that the shortest distance between two points on a flat two-dimensional surface is a straight line. It is just that every time you measure it, it is. You just have to assume it.

I have a pen in my hand which I am holding about three feet off the ground. Every time I have dropped it under controlled conditions, it fell to the ground. I can never prove that the next time I let it go, it will fall. So let’s see. Well, I’ll be darned. It did it again!

In other words, ALL scientific deductions are based on premises which are unprovable inductive conclusions. Therefore, presto change-o, if no inductive conclusions are scientific, then no deductive conclusions are either. Or in other words, science would not really exist.

Syllogistically:

              All scientific conclusions are based on inductive reasoning

              Inductive reasoning never proves anything

      Therefore, no scientific conclusions are proven.

Bullcrap.

Debating Tricks in Arguments about Psychiatry

“Arguing with some true believers is like playing chess with a pigeon.  No matter how good you are at chess, the pigeon is just going to knock over the pieces, crap on the board, and strut around as if it were victorious.” ~ Anonymous

When arguing points in psychology and psychiatry with folks from all ends of the academic political spectrum, it’s easy to see how people become emotionally attached to certain ideas (for whatever internal reasons) and take the position that, “My mind is made up; don’t confuse with me with facts."

Two of their tricks in arguing are:

• To use a word that has several different senses or referents, and then subtly shift how they are using the word as the argument progresses without admitting that that is what they are doing.
• Living their lives as if they really don’t believe a word they are saying. 

In today’s post, I would like to illustrate this by discussing two questions:  “What is depression?” and “Is our behavior pre-determined by our past, or do we have free will?”

First, what is depression?  In his great new book, A New Unified Theory of Psychology, my 

colleague from the Unified Psychotherapy Project, Gregg Henriques, quotes Ingram and Siegle, “The label depression has been used to discuss a mood state, a symptom, a syndrome…a mood disorder, or a disease associated with structural abnormalities.” 

Gregg Henriques

[A syndrome is a group of symptoms and/or behaviors that seem to cluster together in many individuals, creates distress or dysfunction, does not include certain other symptoms or behaviors, and has a definitive epidemiology (the incidence and distribution of a disorder in a population, and the sum of the factors controlling its presence or absence].  

These alternate definitions of depression are highly relevant to the question currently being batted around so fervently by all sides, “Do antidepressants work for depression?”  


The answer to the question depends on which definition is being used in the debates.  Especially online, it’s very difficult to tell.  Obviously, if by “depression” we mean a mood state, then of course antidepressants do not always “work.”  If you are merely unhappy about living in a bad situation, taking an antidepressant will in most cases be a colossal waste of your time and money.

Well, what if it’s a syndrome?  Well, at least according to the DSM, which many people seem to use when it suits their purpose but disparage when it does not, there are, at the very minimum, at least four completely different “syndromes” of depression:  Major Depression (with or without the “melancholic” specifier), dysthymia, adjustment disorder with depression, and depression caused by a medical condition or an outside substance like a drug.  So in a “debate,” which one are we talking about here?


In discussions of studies of the effects of antidepressants, this question is often fudged.  Rather than using the names for the two different syndromes major depression and dysthymia, debaters tend to use the terms "mild to moderate depression" and "severe depression."  The "mild to moderate" depression seen in studies is actually much more likely to be dysthymia, while the severe ones are more likely to be major depression.

Clinically, major depression is more likely to respond to antidepressants, while dysthymia is less likely to.  This is complicated by the fact that the definitions of the syndromes overlap to some degree, but making a clinical decision is what medical education is supposed to be about – or at least theoretically. 


The anti-antidepressant crowd loves to point out that recent "meta-studies" show that in mild to moderate "depression," antidepressants are often no better than placebos.  They also conveniently love to ignore the fact that in the exact same meta-studies, the drugs are shown to be highly effective for severe "depression."  

Actually, the use of the word depression is being used here to describe two very different syndromes, but no one mentions that!  

Even the president of the American Psychiatric Association, John Oldham, was guilty of this confusion of terms in a recent editorial in Psychiatric News.

By the way, if you don’t think defining syndromes based on symptomatic and other variables is a valid way to identify diseases, then ask yourself  how people could tell infectious diseases apart before we knew about the existence of germs.

These debates also revolve around another question - whether the syndromes that do tend to respond to drugs are loosely defined mental disorders, or are actually a “disease associated with structural [brain] abnormalities.”

When this question comes up, the people who think that psychiatric drugs have to either be “all good for all people” or “all bad for all people” with nothing in between (you “splitters” know who you are), tend to presume that we have a scientific understanding of the neurochemical microarchitecture of the brain that we are not even close to actually having.  Since that is the relevant science needed to answer this question, at present we are unable to definitively answer it, so we can only make judgments based on a variety of indirect evidence.

The scientific task is made particularly complicated by the existence of neural plasticity and epigenetics.  


Neural plasticity means that the structure and function of the brain changes significantly in response to persistent environmental demands.  This happens in normals.  For example, the part of the brain that controls finger movement is significantly larger in concert violinists than it is in average controls.  So does this mean that being a concert violinist is a disease?  I don’t think so.

Epigenetic influences, loosely translated, means that most genes in any given cell have a switch that turns them off or on – a switch which is also activated in response to environmental contingencies. So, if the average brain of someone with a given psychiatric syndrome seems to be significantly different than the average brain of someone who does not, does this mean that the syndrome is a normal variant caused by a conditioned response, or evidence of a diseased brain?  With our current knowledge of neurochemical brain microarchitecture being so minimal, we cannot answer that question for certain. 

If there are a very large number of brain differences that are unique to a disorder, and some of the differences do not seem to be directly related to the syndrome in question (such as significant differences in eye movements in schizophrenia), then I lean heavily towards the “disease” model for that syndrome. If not, then I lean towards the “normal variant” model.  But I am at least willing to admit that we can’t possibly know for sure given our current knowledge base.  Major Depression is somewhere near schizophrenia in this regard, whereas dysthymia is not.

The epigenetic issue leads us to my second question: is our behavior predetermined or do we have free will?  Of course, people who take the determinist position do not agree among themselves as to exactly what factors are making the determination.  “Determinism” when used by an psychoanalyst means determined by prior psychological events; when used by a radical behaviorist it means determined by prior external events like rewards and punishments. For family systems folk, it means determined by collective forces, and for biological psychiatrists, it means determined by genes.

That so-called doctors and scientists could each completely discount three out of four variables which clearly and undoubtedly affect human behavior is bizarre in and by itself. 

But do any of these determinism types actually live their lives in a way consistent with their professed belief that when we think we are freely and willfully making a choice as to how we are going to behave in a given environment, it is nothing but an illusion?    If so, then their belief in this proposition itself would have been predetermined by one or more of the four factors, and not by the actual scientific merits of their position!  They do not seem to be saying that, do they?

In fact, they live their lives weighing information and anticipating the future in order to make behavioral choices in order to meet their goals.  When they stop doing that, then we’ll talk.

Of course, this admittedly still leaves open the question of the existence of free will.  What do I think?

Well, I believe in free will, although there is no way to actually prove it one way or another.  The idea that may answer the question is actually provided by the radical Skinnerian behaviorists.  They start from the proposition that when a rat is learning to run a maze, if it is rewarded (operantly given positive reinforcement), then the probability that it will behave in the way that was reinforced increases.  (Of course, if food is used as a reinforcer the rat has to be hungry or else this does not work, but that’s a different issue).

In no case does the provision of positive reinforcement lead to a 100% chance that the rat will do anything.

In fact, when we look at all four factors described above, each one increases the oddsthat certain behavior will ensue.  We could express this probability mathematically as P = p + e + c + g, where P = the probability of a certain human behavior taking place at a given point in time, p = prior psychological experiences, e = external reinforcement by environmental factors, c = collective or social forces impinging on the person, and g = their genome.

I submit that when we add together all four of these factors, P will still not ever reach 100% - there is no certainty or absolute predictability.  If this is true, there would have to be a fifth factor.  Lets call this unknown factor f.w.