Welcome!

Welcome to my blog, a place to explore and learn about the experience of running a psychiatric practice. I post about things that I find useful to know or think about. So, enjoy, and let me know what you think.


Sunday, January 18, 2015

Why I'm Not A Scientist

Well, my first day at the Waldorf convinced me that it's a good thing I'm not a scientist, because I would drive myself and everyone else crazy.  I'm just referencing one of the discussion groups I went to last week, Neuroscience Perspectives on Psychoanalysis.
The way the discussion groups work is that there are the same groups every year, give or take a few, and they have the same titles, but cover different topics each year. So this year, this group was conducted as a journal club. A week or two ago,everyone registered for the group got an email from the guy running the group that included 4 articles-one main article, another written about the main article, another that was a supplement to the main article, and another that we would get to if there was time. 
The main article was, "Subliminal Unconscious Alpha Power Inhibits Supraliminal Conscious Symptom Experience", by Shevrin et al, published in Frontiers in Human Neuroscience in September 2013.

It's a bit of a fiddly article to read, and I think it's easier to understand if you read the article about it, from JAPA, first, but since that's not open source, I'll try to summarize it, before I get to my issues with it,

The idea was to find some kind of specific EEG correlate for the unconscious underpinnings of the clinically observable phenomena of repression and avoidance in social phobia. They looked at 10 subjects with social phobia, and the first thing they did was interview each subject at length. Then they picked out 21 words or short phrases from the interview, per subject. 7 reflected the interviewer's understanding of the unconscious conflict that was driving the phobia (the assumption was that there WAS such a conflict), 7 reflected the subject's conscious experience of symptoms, and 7 represented "negative" words that had nothing to do with either symptoms or unconscious conflict.  Supposedly, the underlying conflict for one subject was anxiety about a homosexual seduction. One unconscious conflict word was, "paddle me", one conscious symptom word was "public talks", and one negative word was "atomic bomb". Just to give you an idea.

They then used a tachistoscope to flash words at the subjects, in a sequence of two, a primer word, and a target word. Some were flashed subliminally, and appeared for 1ms, some supraliminally, for 30ms. They measured alpha rhythms after each word was flashed. Stop me if you already know this, but alpha rhythms are between 3 and 13Hz. They referred to them as alpha power.

What they found was that there was high alpha power after subliminal unconscious conflict words, and these corresponded with high alpha power after conscious symptom words, when the latter followed the former. And they didn't have any significant findings with any other combinations. 

The positive findings look like this:


Where each plotted point represents a subject.

What I like about the study was that it had some actual findings. One person in the group commented that people who don't make a living listening for other people's unconscious conflict would be shocked that there were any findings at all.

What I didn't like about the study was a lot of the methodology, and what the authors extrapolated from the findings.

The guy who ran the group is a neurologist, and he pointed out some problems that I would never have been aware of, since I'm not a neurologist. First, alpha rhythms are typically measured with closed eyes. In this study, of necessity, they were measured with open eyes. Second, alpha rhythms are typically thought to represent arousal, while the authors were using them as a marker of inhibition. Third, the authors didn't look at any other rhythms on EEG.

My methodological difficulty has to do with the choice of unconscious conflict words. Each point on the graph above represents a subject, so there's some kind of average or composite of the alpha power generated. There's no raw data to indicate what the alpha power was for each word. So suppose 1 of the 7 words generated very high alpha power, for a given subject, but the other 6 didn't generate much. You could still end up with the same graph, but you wouldn't be able to tell if the 6 "wrong" words were off because they weren't related to unconscious conflict, or if those 6 WERE related to unconscious conflict, but what you were measuring was something else. Since the purpose of the study was to demonstrate the unconscious conflict involved in social phobia, and since a lot of people probably don't believe the symptoms of social phobia are generated by unconscious conflict, to begin with, this matters.

Then there were the conclusions. The authors extrapolated from alpha power as an inhibitory process, to repression and avoidance. They claimed that the inhibition, as demonstrated by the alpha rhythm, following the unconscious conflict words, represented a repression of the anxiety-producing conflict, and could only be noted subliminally, i.e. unconsciously. They went further, suggesting that this is observable clinically, when a patient stumbles across an area of conflict, then quickly forgets what he was talking about, or changes the subject without realizing it.

To me, this seems like a pretty big leap. Could it be true? Sure. It could even be true if alpha power was thought of, as it normally is, as a mark of arousal-the conflictual word evokes anxiety and arousal, which is quickly followed by repression or avoidance. But they haven't demonstrated this.

Finally, the JAPA article, Research on the Relation of Psychoanalysis and NeuroscienceClinical Meaning and Empirical Science, by Barry and Fisher, goes even further. It describes a "repression index" created by the study authors, to examine the differences in individual degrees of repression. They found a lot of variation among subjects. The authors hypothesized that those who repressed the most would not benefit that much from CBT, which all of the subjects received, while those who didn't repress that much would:

The results were spectacular. Unconscious repression clearly impeded recovery, and those who exhibited the most repression benefited hardly at all from exposure treatment. It was hypothesized also that unconsciously these subjects were unable to tolerate the treatment and so could not benefit from it. Those who showed little repression not only could tolerate the treatment but greatly benefited from it. With the repression index, the researchers could explain two-thirds of the variations in treatment outcome. If these results can be replicated, the index can be used as a diagnostic test to indicate the best treatment option in a given case—who would benefit from CBT/exposure therapy and who would require psychodynamic psychotherapy. 

To me, "spectacular" seems like a bit of a stretch for a study with 10 subjects.

This is why it's a good thing I'm not a scientist. I'd never get anything published, because I'm such a stickler. I'm very disappointed in this study, because it actually had some findings that are worthwhile to try to understand, and that can be further explored experimentally. But the grandiose conclusions, or predictions, if you will, undermine the credibility of what is otherwise interesting and promising work.