The Sorrows of Psychiatry
nature.com
nature.com
Not only does this line up squarely with my own personal experience with addiction and other psychosis I have, but the lack of these issues in my children as well, who had a radically different early childhood then I and have no addiction issues whatsoever.
Case in point...I recently saw the documentary about the triplets that were separated at birth and given to 3 different socio-economic and emotionally-connected parents and who, by total accident, met at college and came to be national celebrities in the 70's and 80's.
One of the three eventually killed himself, due in part I'm sure to his much deeper addiction issues then that of his 2 siblings. The interesting part was that he was given to the "wealthy, but emotionally distant" family.
By far the most well balanced and successful of the 3 was the one given to the gregarious and emotionally-connected father, who showered the child with love and support throughout his whole life.
I could go on for 30 minutes discussing all the takeaways of this documentary, called Three Identical Strangers...it turns out the children where all part of an ethically-questionable experiment where, in my mind at least, the end results basically proves that 95% of the nature/nurture argument goes to the "nature" side.
It seems that all "nurture" can do is fuck you up and give you lifelong issues that are incredibly difficult to overcome.
https://www.nytimes.com/1984/01/24/science/freud-secret-docu...
(search for "real rape" in that article to read one account of Freud expending a lot of effort to hide obvious cases of rape)
Here's a fun story from Pennsylvania:
https://www.washingtonpost.com/news/morning-mix/wp/2018/01/0...
I wonder say that's a blow against the profession, but that is a blow against the individuals involved. The police are complicit in that whole thing as well, but we aren't trying to throw all of law enforcement out because of it.
“Though he may not be a household name, Sarno is probably America’s most famous back pain doctor. Before his death on June 22 2017, a day shy of his 94th birthday, he published four books and built a cult-like following of thousands of patients — including Howard Stern and Larry David. Many of them claim to have been healed by Sarno, who essentially argued back pain was all in people’s heads. And Sarno himself often said that some 80 percent of his patients got better.”
“After digging a little deeper, I learned that some of Sarno’s theories are now even being validated by science — specifically, that there can sometimes be an emotional basis for chronic back pain. And that’s an important truth mainstream pain medicine still hasn’t quite figured out what to do with.”
“More specifically, he believed that the brain distracts us from experiencing negative emotions by creating pain. We may not want to accept the uncomfortable truths that we are angry with our children, or that we hate our job, so instead of thinking those thoughts, we focus on the pain.
He also thought that pain was created by reduced oxygen and blood flow to the muscles and nerves of the body. So our brains unconsciously [sub consciously] direct blood away from certain areas of our body, and that creates pain.” - https://www.google.com/amp/s/www.vox.com/platform/amp/scienc...
The problem is that Sarno was at his peak when western medicine wasn’t willing to look at studies that show that emotions can be the root cause of physical conditions. Which is crazy because the placebo effect is well documented. Thus there is obvious evidence that our mind has the ability to effect our body in various ways (including our immune system).
serotonin, pain, phantom
...you'll find a fair amount of literature on serotonin as playing a key role in the experience of pain, in some cases seeming to prolonging the perception of pain where the underlying physical damage has completely healed.
And sorry for the ultra fuzzy anecdote
Psychiatry is like fixing a software bug by telling the customer to simply avoid doing things that causes the bug rather than correcting the actual code and fixing the software.
I know some people don't like reductionist reasoning, but the more I think about it, the more I'm inclined to believe that all psychological problems are ultimately a brain problem. We just don't know enough about the brain yet to fix many of those problems.
Example: let’s say you had a moderately abusive parent. The result is low self esteem but nothing diagnostic worthy. While I’m sure there’s a concrete neurological effect that might be measurable, but I seriously doubt that neuroscientists would have the tools today to interdict and make a difference in the way that psychology might.
Now I do agree that neurology is replacing psychology, I just don’t think it has replaced it yet.
When the actual biological cause for a condition is found, it moves over to being treated by other medical specialties and loses its psychiatric significance. Parkinson's, now treated by neurologists is a one such example. Hepatic encephalopathy from liver failure, which can cause dramatic personality changes, is another.
Sometimes its changing social conditions that help recategorize a diagnosis. Case in point, the vote to remove homosexuality as a psychiatric diagnosis from the DSM in 1973.
Psychiatry is the catch-all until we actually figure out how to fix something. You can extrapolate this: if something is a psychiatric diagnosis, theories as to cause and treatment are actually hypotheses, sometimes uncomfortably out of date. If the treatment works for you, that's excellent. But if it doesn't, realize you're dealing with a bug in the most complex object in the universe. Hitting it with a hammer might not work.
It does no such thing. This drug, and most antipsychotics, simply shuts down the forebrain. It "treats" psychosis by shutting off your mental function. As soon as you remove the drug (go "off your meds" because you want to think and feel again), psychosis resumes. This tells us nothing about the nature of psychosis.
> > I disagree.
Saved you the time of reading this fancy opinion piece.
Apart from the truly severe conditions that prevent individuals from even participating in society, IMHO what psychiatry seems to fail to account for is context. For example, it is said that many people suffer from depression, but instead of looking into the core causes, they would rather just pump patients with meds and hope it gets better with time. Often times though, all that happens is that the patient builds tolerance to medication, so larger and larger doses are needed. It makes me think: what if those people cannot escape their condition to begin with? Can't even know if you don't at least try.
So if you're being abused, there's absolutely nothing that psychiatry can do for you, because they can do nothing about the abuse (other than taking you in and throwing you back into the abuse after 2 months). Same with ancillary fields: social workers will NEVER do anything about that teacher that's abusing your kid, they will only "treat" the child (maybe with force, maybe with internment, maybe against the will of both child and parent, but only the child).
Likewise, many issues are caused by poverty, or other effectively environmental factors that just won't change with psychiatric treatment. There is absolutely nothing that can be done.
You're also forgetting, with those medications, what the patients will build tolerance for and how that happens. Opiates, Xanax, "Benzos" and SSRIs all have different mechanisms, but it boils down to the following: you will NOT like the long term effects. You can responsibly take these medications for periods from hours to a few weeks (to end what they call a manic episode or a huge panic). More will have permanent consequences. Eventually you will become permanently depressed, to the point that "zombie" will be a word used to refer to you. The way your body adapts is by raising the threshold of dopamine it takes to reward you, to a point that no non-medication-induced (and eventually any non-overdose amount) will make you happy. You won't get out of bed, you won't learn, go for a walk, wash yourself, ... hell Miss/Mister World could walk into your bedroom, offer anything goes sex, and you still wouldn't be able to find the motivation to do anything. Once that threshold is raised to a point that only medication can provide rewards, you're not coming back anymore.
Are there any statistics or studies regarding the becoming of these zombies? I thought that the body, if long enough under such medication, becomes incapable of producing its own dopamine, not that the threshold becomes higher.
Not saying it won't help out some people, but at least what I tried didn't help me.
If one oversimplifies neural networks and thinks of them as pattern-matching machines, it would make sense that our brain's neural network would have trouble if they were fed a lifetime of bad training data (abusive childhood/relationships) and therefore had bad/nonconstructive reactions to normal stimuli.
For the same reason that people who have changed a tire on a bike cannot repair a jumbojet. The statistics accepted in psychological papers ...
https://en.wikipedia.org/wiki/Replication_crisis
Or, more dramatically:
https://phys.org/news/2018-10-real-fake-hoodwinks-journals.h...
If you're saying "most science is bullshit" you can point to the replication crisis. If you're saying "only psychology is bullshit", well, that shows your bias pretty clearly.
It is often explicitly mentioned in articles, currently mostly for reinforcement learning (or it might be that that's just the vast majority of papers I see these days), that, for example "this network converges about 3 times out of every 10 attempts". Often with graphs leaving a selection of the failed attempts.
This is very different from the explicit lying you find in psychology papers and theories.
That's actually an interesting question; I'm probably not the person to answer it, since my experience with NNs is admittedly on the "still learning" side of things, and that learning has not been via any credentialed sources...but I'm willing to take a stab at it.
NN models - that is, the thing that results from training a neural network, are typically "fixed", in that once trained, the model doesn't change - at least classically. I imagine that somewhere out there, there may be NN models which can change as data they are processing is run through them, learning on the fly so to speak. I'd have to research it; I'm sure it's something that's done or been done?
But normally, once a NN has been trained, it's model is "fixed" and doesn't change when inputs are presented to it. The training phase is - or appears to be - fairly stochastic. I'm not sure I'd want to call it random, though, because I don't recall any kind of random numbers being used during back-propagation.
But the data that is presented to the neural network is usually randomized in presentation - and sometimes content. That is, say you're training the NN on recognizing horses. You might have several thousand images of horses, but you don't want to show just those. You might want to generate many more - rotate each one just a bit, skew it just a smidge, maybe change the color and/or contrast/brightness, etc - to in effect generate a bunch more positive (and just as many negative - so you don't get bias or overfitting happening) examples to train the NN on what is a "horse" vs what is a "not horse".
So that data is somewhat "random" - but from what I recall, the actual mechanics of training - the algorithms of forward passes and backwards passes (backpropagation) don't have any randomness to them; just to be sure, I checked this - which is a great explanation (not the simplest, but not impossible to follow):
https://mattmazur.com/2015/03/17/a-step-by-step-backpropagat...
I don't know what your math skill level is here, so don't let any of the calculus and "chain rule" stuff get to you if you aren't familiar with it (tbh - I suck at it), just look at the equations and explanations. It's plain that there is no random number generator to be found in the process.
So - in theory - if the model, after it has been trained and "baked in place" so to speak - is presented with the exact same inputs, it should generate the exact same output.
But the input has to be exactly the same; in the case of an image, the neural network's input layer is usually an "unrolled" 1-dimensional array representing the pixel values of the image (left-to-right, top-to-bottom - as a 1D array - usually). Those values are usually grayscale or color values, presented either as integer data or floating-point values.
As long as that image data is presented exactly the same to the NN model, the output should be the same; for instance, if shown a set of pixels values that represent something, the output of the model will always be the same if shown those exact same pixel values.
But usually, these systems aren't built to take in "exact data" but rather data from sensors of the real world. So - the data that would probably be fed into the NN model likely comes from, say, a camera - and that sensor will not always present the exact same data to the NN model, even if purposefully set up to do so - because all such sensors have noise and aren't perfect (different pixels from the camera's sensor can and will return different values, even if shown a calibrated blank image in a fixed mode, with consistent fixed lighting, with the best camera sensor available, etc - it's just a fact of the real world).
So, because of this - the output of the NN model will in effect be "random" - but only because the input is effectively "random". That's actually ok, because what the model outputs (even if set up as a classifier) are values of probability - percentage values, where (ideally) the "spike" in the overall set of values represents the actual identification for the network, and that any inherent randomness in the system (sensors and whatnot) is filtered out and (hopefully) doesn't effect the outcome.
Though as we know, this too can be exploited; because the model is "baked in", you can show a series of images to the network, and get an identification on the other end, and probably with some statistical analysis you can work out what the layers in between might actually represent (probably not exactly though) - and identify flaws that could cause misidentifications on the output. More or less "hacking the NN model middle layers" and using that information to craft an "exploit" for the input to cause a particular mis-identification on the output side.
As we know - this is possible; at least, we know it's possible to get a network to output a wrong result by simply changing the pixel values of the input image slightly (even subtly that can't be seen by a person looking at the image - basically a steganographic style attack?). Such a "hack" could be used for any number of purposes, but mostly they show how fragile such NN models can be.
Does this mean they are random, or deterministic? I'd argue for the latter at this point, but again, I'm not really the person to ask. Hopefully someone else can answer (or has, by the time I post this).