A Model of Trauma


Pain and grief

“In principle, all the ancestral history implicitly represented in cognitive adaptations can be converted to explicitly represented knowledge, running on a core consequentialist.”1

What is pain for? From the point of view of evolution it’s a way to make us maneuvre away from certain situations. Pain tells you not to keep your hand on the stovetop; pain tells you to not aggravate an existing wound. It compels certain actions.

Pain can be thought of as a sense, like sight, but sight doesn’t have a feeling because sight isn’t there to compel action on its own like pain does. Why do emotions feel like something? Because they compel action. (Pain, though, doesn’t depend on our personal knowledge of the context for a situation, and so it’s a sensation instead of an emotion.)

What about grief? Humans have relied on things other than physical health to survive, so losing eg. a strong social tie is a threat to fitness as much as injury is, so loss is painful (aversive).

But loss is more complex and contextual than physical injury; it causing pain alone isn’t a great strategy. The Kübler-Ross model’s five stages of grief (denial, anger, bargaining, depression and acceptance) can be seen as a process of learning an effective strategy for tackling a loss: attempting increasingly costly/unprofitable methods to avoid, soften, or learn from the blow.2 3

Trauma as learning process failure

If the stages of grief are a learning process, stressors which end up being traumatic can be seen as being so because they were antecedent to a learning process which doesn’t correctly resolve- an inability to move through the stages of grief to acceptance, or getting stuck with maladaptive coping mechanisms.

To rephrase, trauma happens when the feeling of ability to affect or cope with one’s environment doesn’t reach acceptable levels (or doesn’t reach it through acceptable coping mechanisms).

That said, trauma in the form of complex PTSD is noted for disturbing identity- can that mesh with this idea? I see two options here.

The first is that disturbed identity is a result of the learning process, developed to increase ability to handle current trauma or future traumatic events. If the mind develops dissociation because it relieves pain, it would fall under this category.

The second option is that disturbed identity it’s not a result, but instead a byproduct or just a different view on the process itself.

Identity (in the sense of who a person thinks of themself as) doesn’t exist in a void- a person thinks of themselves as an angry person if they noticed they’re angrier than other people, etc.4 Identity is, in this sense, learned (through observation of the self and others). Inability to resolve the process of fitting the traumatic event (event=an observation) into the mental model of self and world leave the self and world in doubt.5

The first might seem simpler, and therefore better by Occam’s Razor, but the second seems to me to fit in with other things. Here I said trauma is low confidence in identity or in one’s relationship to the world- depression might also be because of low confidence, and of course, trauma and depression are related. Another thing in this cluster is self-harm. SSC again, with regards to extremes of stimming in autism: “If the brain is in some sense minimizing predictive error, and there’s no reasonable way to minimize prediction error because your predictive system is messed up and registering everything as a dangerous error – then sometimes you have to take things into your own hands, bang your head against a metal wall, and say “I totally predicted all that pain”.”

Learning from the environment

Trauma is mediated by personal self-efficacy.6 Example: imagine a student going to an oral exam. He doesnt believe in himself so he stutters and answers without confidence. He would cope better purely by having stronger belief in his ability to cope.

Personal self-efficacy, or confidence, though seeming like things “belonging to” a person, are influenced by the environment, as people infer facts about themselves based on their environment. If the student knows that someone less prepared than him passed, he’d feel more confident in his own abilities being sufficient.

In the context of confidence in ability to cope with failure (rather than just being confidence in ability to pass), imagine the student was part of a group of students taking the exam one by one. If he thinks of his skills as comparable to the others’, and the members of his group who went before him failed and took it in stride, it seems like it’d feel like less of a blow than if he’d been the only one to fail. Like before, others’ successes can increase his own confidence.

(Of course, “infering that he can cope if his peers can” isn’t the only possible mechanism for environmental influence on trauma, it’s just a simple one. Another possibility is that all of them failing reduces uncertainty about why they failed because there are more data points to compare, or reduces uncertainty by making it a question of why they all failed instead of both why he failed and why his self-assessment of his skill relative to peers was wrong. In the context of grief/trauma as learning, all of these examples lead him towards being able to resolve the process of answering the question of whether he is equipped to cope, in the first case by making the answer more confidently a yes, and in the other cases by making it easier to answer by removing uncertainty about capabilities. As an analogy, consider the changes in confidence of a coin coming up heads by 1. observing coins you know to be similar come up heads, and 2. infering specifics about the coin in question.)

Learning what and how?

So trauma is about learning, and we can learn from the environment. But despite the reasonable-sounding examples above, learning isn’t always unbiased or truthful.

(1.) Firstly, we don’t have direct access to the truth and don’t know when evidence is misleading us. Even if its purpose was learning truths, and it was a perfect process, that wouldn’t guarantee true results if the input was biased.

(2.) Secondly, a “biased” process can on average lead to beliefs that are more accurate. We use social proof because if we think others reason as we do then them believing something should make us believe it more likely to be true7. If the consensus is biased, we’ll end up more biased, but avoiding this bias would mean throwing away information that’s usually useful. Science can afford to not admit social proof as evidence because science is purely about knowledge and doesn’t need to inform action. Once knowledge has to inform action, throwing out information can be too costly. Admitting low-quality evidence isn’t strictly right or wrong.

1 & 2 together show how an abusive environment can cause someone to believe that they’re worthless even if they know logically that they’re not.

(3.) Thirdly, attention is limited, which also introduces bias.

From The Centrality of Event Scale: A Measure of Integrating a Trauma into One’s Identity and its Relation to Post-Traumatic Stress Disorder Symptoms (which I liked and will therefore pick on):

Because a traumatic or highly stressful event often causes profound changes in a person’s outlook (Janoff-Bulman, 1998), often stays highly accessible for years, and often comes to mind spontaneously in response to internal and external cues (Berntsen, 2001), it may be perceived as a major causal agent and thus as a highly salient turning point in the person’s life. However, at the same time, having a trauma as a salient turning point in the life story may lead to oversimplifications. It may cause the person to focus on those aspects of his or her current life conditions that can be explained by reference to the trauma, and to ignore aspects that defy such causal attributions. It can therefore be seen as a way of optimizing the internal consistency of the life story at the expense of the multiplicity of meaning that normally characterizes our life narratives (Linde, 1993; Robinson, 1996).

This reads to me a lot like “traumatic events cause changes in outlook, which causes changes in outlook.” Optimization of the internal consistency of the life story still seems like a connected phenomenon, but makes less sense as an effect than as a cause. Consider: If an event showed itself to be unexpectedly threatening, it’s worth allocating more attention to keeping an eye out for warning signs. If attention is finite, multiplicity of meaning is deprioritized to make really sure another instance of event isn’t on its way. This compounds: a more intensive search can return more false positives. This isn’t necessarily bad: it leads to ever-increasing vigilance until a high enough ratio of positives being confirmed as false stabilizes it (at the correct amount of vigilance). However, standards for what kind of evidence can confirm positives as false rise, and there’s not necessarily a way to act which both feels duly vigilant and which doesn’t cause damage to the person’s life and relationships.

(4.) Fourthly, beliefs themselves don’t have to be true, they just have to lead to the right action. For example, we don’t like to eat things that smell/taste bad, even though “it tastes bad so it is bad” doesn’t explain/prove why things that taste bad shouldn’t be eaten. Following the advice of the senses leads to avoiding eating rotten food despite a lack of understanding; a genetically- instead of culturally-embedded superstition. Just as useful genes flourish disproportionately and reflect truths about a relationship between self and environment, so would memes.

And to be more specific, the criteria for the right action isn’t the action one’d take given knowing the truth, but rather, the one that’s most useful. If we value knowing the truth it’s because us valuing knowing the truth turns out to be useful on average; underlying mental processes don’t have to agree. A way of looking at it is to think of it as a mechanism like a hedonic treadmill. Happiness and unhappiness don’t exist to measure how good things are, they’re to motivate productive action.

In this case it’s a sort of “reverse” hedonic treadmill- if something is bad and can’t be changed, what good would constant misery be?

maturity is just having all the bad things beaten out of you and developing Stockholm syndrome w/ reality

— Biggidy Boss (@turrible_tao) September 10, 2017


How do we assign blame? If someone knowingly leaves their umbrella inside and walks into the rain, do we blame the rain for them getting wet? It seems to me like the key is the degree you forsee each party to be able to change. Rain can’t be reasoned with, but the person knew that rain would get them wet, knew the umbrella would protect them, and had the choice to use it.8 If someone said “the rain is equally to blame” it might not necessarily be wrong, but it’d be missing some important point. We blame the person because again we have a reasoning process which is about producing useful action. If someone continues to try to place blame on the rain, we’d see it as trying to dodge responsibility- refusing to learn.

So again we have something about learning. How does this fit together with trauma? Well, we might look at the five stages of grief and think to ourselves that blaming others would lie in the anger stage, and blaming the self in the depression stage- but that seems a bit confused, because phrasing it that way doesn’t make clear what is cause and what is effect. Blame seems more deeply related than that.

Let’s look at the idea of “trauma as a (maladaptive) learning process which continues until confidence is gained in the ability to cope”. What does coping require? It requires knowing thy enemy and knowing thyself.9 Knowing thy enemy seems a lot like determining blame.

Blame being integral to the troubleshooting process explains why trauma (where this process is pushed past its limits) causes malfunctions of blame, such as irrational self-blame, survivor’s guilt, and Stockholm syndrome.

Another thing this explains is why people are more likely to be traumatized by situations involving other people.10 People are complex, so a situation involving people means that there’s more question of whether it would’ve been better to have acted differently than if the situation was something more easily understandable, like a natural disaster. How and to what degree to blame the self- to change behavior- is in question. In this framework of trauma as a learning process failure, more uncertainty would lead to the learning process not resolving as easily or as well.

Change in cultural beliefs surrounding a trauma

To recap, blame is about guiding action, and is based on the perceived characteristics of the environment/actor.

Again, continuing to actively blame forces that can’t be changed isn’t useful; we wouldn’t want someone to get stuck there. Transhumanists today can recoup some of the cost of giving up feel-good beliefs about death by the feeling there’s some purpose to seeing that truth because death can be fought. Destroying an average medieval peasant’s comforting illusions about death would be useless and therefore cruel. The two ideal strategies are the opposite ends of the scale: knowing something can be changed and changing it, and not knowing something can be changed when there’s no power to change it.

The painful middle ground is knowing something is unjust and being powerless to change it. Once something is known to be possible to interact with- blameable, unlike rain in the earlier example- the learning process takes longer to resolve because there are more possible actions to consider.

That said, it’s not useless. If circumstances change and the problem becomes tractable11, then those who successfully minimized the problem might not notice that taking action is possible and correct.

On a larger scale, stubborn outliers help trigger information cascades towards a new stable state. Eg., if the information environment is completely shared, then necessary for a cascade (instead of a clean flip) is different weightings of social vs. personal information for belief.

Of course, if the system was truly stable to start with, something outside of preexisting beliefs had to to trigger the cascade in the first place. So what could do it?

A sign that the thing doesn’t have to be that way might do it. For instance, if the harmful event starts to happen less often than it did previously, maybe that would 1. increase motivation/decrease compassion fatigue, 2. increase compassion because the victim is more traumatized (if trauma is tied to unexpectedness) or 3. increase compassion by the fact that increased unexpectedness makes victims seem less blameworthy (see: previous umbrella/rain example.)12

  1. http://lesswrong.com/lw/l3/thou_art_godshatter/
  2. Denial is making sure the loss exists before taking costly actions. Anger and bargaining attempt to make a deal to avoid the loss. Depression encourages thought about what went wrong (rumination) so it can be avoided in the future (and is a fire sale on friendship, now that the loss is confirmed)
  3. “If the point is learning, then why do we have to deal with emotions even when we know that there was nothing to be done?” Maybe without the pressure of grief, it could be relatively easy to trick oneself into believing there was nothing to be done, leading to greater losses than otherwise if there really was something to be done. Compared to this, grief is relatively cheap. Compare that pain can get in the way in dangerous situations, but that being unable to feel pain is dangerous.
  4. That description of non-material characteristics of a thing is relative goes in general, eg. I would describe a person as tall if they’re taller than most, but would call a tree small if it were that height.
  5. (Above) More of the idea of of human functioning relying on narrow confidence intervals.
  6. https://www.scribd.com/document/120718364/Bandura-Social-cognitive-theory-of-posttraumatic-recovery-The-role-of-perceived-self-efficacy
  7. see also: Auman’s agreement theorum.
  8. This means that the victim knowing more makes them more “rightly to blame”. Spreading knowledge of safety procedures (which also raises awareness of the issue, that is, makes the crime more known to happen) can be seen as an information hazard causing victim blaming.
  9. I’m being a bit flowery here; another way of putting it is “knowing the criteria for productive action and knowing the set of possible actions.”
  10. “People who experience interpersonal trauma (for example rape or child abuse) are more likely to develop PTSD, as compared to people who experience non-assault based trauma such as accidents and natural disasters.”
  11. In the case of anti-deathism, through technological change; in the case of culturally-enabled harms, through, uh, a lot of complicated ways.
  12. Partially related: Kindness Against the Grain