Book summary: Unlocking the Emotional Brain

post by Kaj_Sotala · 2019-10-08T19:11:23.578Z · score: 144 (48 votes) · LW · GW · 23 comments


  Emotional learning
  Memory reconsolidation: updating the emotional learning
  Applied reconsolidation: an example of the schema update process
  Applied reconsolidation: the schema update process in general
    Accessing sequence
    Erasure sequence
    Verification step
    Further examples
  Evaluating the book’s plausibility
  UtEB and the subagent model

If the thesis in Unlocking the Emotional Brain (UtEB) is even half-right, it may be one of the most important books that I have read. Written by the psychotherapists Bruce Ecker, Robin Ticic and Laurel Hulley, it claims to offer a neuroscience-grounded, comprehensive model of how effective therapy works. In so doing, it also happens to formulate its theory in terms of belief updating, helping explain how the brain models the world and what kinds of techniques allow us to actually change our minds. Furthermore, if UtEB is correct, it also explains why rationalist techniques such as Internal Double Crux [1 [LW · GW] 2 [LW · GW] 3 [LW · GW]] work.

UtEB’s premise is that much if not most of our behavior is driven by emotional learning. Intense emotions generate unconscious predictive models of how the world functions and what caused those emotions to occur. The brain then uses those models to guide our future behavior. Emotional issues and seemingly irrational behaviors are generated from implicit world-models (schemas) which have been formed in response to various external challenges. Each schema contains memories relating to times when the challenge has been encountered and mental structures describing both the problem and a solution to it.

According to the authors, the key for updating such schemas involves a process of memory reconsolidation, originally identified in neuroscience. The emotional brain’s learnings are usually locked and not modifiable. However, once an emotional schema is activated, it is possible to simultaneously bring into awareness knowledge contradicting the active schema. When this happens, the information contained in the schema can be overwritten by the new knowledge.

While I am not convinced that the authors are entirely right, many of the book’s claims definitely feel like they are pointing in the right direction. I will discuss some of my caveats and reservations after summarizing some of the book’s claims in general. I also consider its model in the light of an issue of a psychology/cognitive science journal devoted to discussing a very similar hypothesis.

Emotional learning

In UtEB’s model, emotional learning forms the foundation of much of our behavior. It sets our basic understanding about what situations are safe or unsafe, desirable or undesirable. The authors do not quite say it explicitly, but the general feeling I get is that the subcortical emotional processes set many of the priorities for what we want to achieve, with higher cognitive functions then trying to figure out how to achieve it - often remaining unaware of what exactly they are doing.

UtEB’s first detailed example of an emotional schema comes from the case study of a man in his thirties they call Richard. He had been consistently successful and admired at work, but still suffered from serious self-doubt and low confidence at his job. On occasions such as daily technical meetings, when he considered saying something, he experienced thoughts including “Who am I to think I know what’s right?”, “This could be wrong” and “Watch out - don’t go out on a limb”. These prevented him from expressing any opinions.

From the point of view of the authors, these thoughts have a definite cause - Richard has “emotional learnings according to which it is adaptively necessary to go into negative thoughts and feelings towards [himself].” The self-doubts are a strategy which his emotional brain has generated for solving some particular problem.

Richard’s therapist guided Richard to imagine what it would feel like if he was at one of his work meetings, made useful comments, and felt confident in his knowledge while doing so. This was intended to elicit information about what Richard’s emotional brain predicted would happen if it failed to maintain the strategy of self-doubt. The book includes the following transcript of what happened after Richard started imagining the scene as instructed:

Richard: Now I’m feeling really uncomfortable, but-it’s in a different way.
Therapist: OK, let yourself feel it - this different discomfort. [Pause.] See if any words come along with this uncomfortable feeling.
Richard: [Pause.] Now they hate me.
Therapist: “Now they hate me.” Good. Keep going: See if this really uncomfortable feeling can also tell you why they hate you now.
Richard: [Pause.] Hnh. Wow. It’s because… now I’m… an arrogant asshole… like my father… a totally self-centered, totally insensitive know-it-all.
Therapist: Do you mean that having a feeling of confidence as you speak turns you into an arrogant asshole, like Dad?
Richard: Yeah, exactly. Wow.
Therapist: And how do you feel about being like him in this way?
Richard: It’s horrible! It’s what I’ve always vowed not to be!

Richard had experienced his father as being assertive as well as obnoxious and hated. His emotional brain had identified this as a failure mode to be avoided: if you are assertive, then you are obnoxious and will be hated. The solution was to generate feelings of doubt so as to stop him from being too confident. This caused him suffering, but the prediction of his emotional brain was that acting otherwise would produce even worse suffering, as being hated would be a terrible fate.

UtEB describes Richard as having had the following kind of unconscious schema:

Perceptual, emotional and somatic memory of original experiences: his suffering from his father’s heavily dominating, hyper-confident self-expression, plus related suffering from unmet needs for fatherly expressions of love, acceptance, understanding, validation. (This is the “raw data”; matching features in current situations are triggers of the whole schema.)
A mental model or set of linked, learned constructs operating as living knowledge of a problem and a solution:
The problem: knowledge of a vulnerability to a specific suffering. Confident assertiveness in any degree inflicts crushing oppression on others and is hated by them. I would be horrible like Dad and hated by others, as he is, if I asserted my own knowledge or wishes confidently. (This is a model of how the world is, and current situations that appear relevant to this model are triggers of the whole schema.)
The solution: knowledge of an urgent broad strategy and concrete tactic(s) for avoiding that suffering. Never express any confident assertiveness, to avoid being horrible and hated (general strategy and pro-symptom purpose), by vigilantly noticing any definite knowledge or opinions forming in myself and blocking them from expression by generating potently self-doubting, self-invalidating thoughts (concrete tactic and manifested symptom).

Emotional schemas can be brought to light during a variety of ways, including Focusing, IFS [LW · GW], and imagining yourself doing something and seeing what you expect to happen as a result.

But suppose that you do manage to bring up a schema which seems wrong to you. What do you do then?

Memory reconsolidation: updating the emotional learning

The formation of memory traces involves consolidation, when the memory is first laid out in the brain; deconsolidation, when an established memory is “opened” and becomes available for changes; and reconsolidation, when a deconsolidated memory (along with possible changes) is stored and becomes frozen again. The term “reconsolidation” is also used to refer to the general process from deconsolidation to reconsolidation; UtEB generally applies the term to mean the entire process. Unless the context indicates otherwise, I do the same.

UtEB reviews some of the history of memory research. Until 1997, neuroscientists believed that past emotional learning became permanently locked in the brain, so that memories could only consolidate, never de- or reconsolidate. More recent research has indicated that once a memory becomes activated, it is temporarily unlocked, allowing it to be changed or erased.

Starting from 2004, new studies suggested that activation alone is not sufficient to deconsolidate the memory. The memories are used to predict that things will occur in a similar fashion as they did previously. Besides just activation, there has to be a significant mismatch between what one experiences and what the memory suggests is about to happen. The violation of expectation can be qualitative (the predicted outcome not occurring at all) or quantitative (the magnitude of the outcome not being fully predicted). In either case, it is this prediction error which triggers the deconsolidation and subsequent reconsolidation.

The memory erasure seems to be specific to the interpretation from which the prediction was produced. For example, someone who has had an experience of being disliked may later experience being liked. This may erase the emotional generalization “I am inherently dislikable”, but it will not erase the memory of the person also having been disliked.

Applied reconsolidation: an example of the schema update process

So, assuming that the model outlined above is correct, how does one apply it in practice?

From what we have discussed so far, the essential steps of erasing a learned belief (including an emotional schema) involves identifying it, activating it, and then finding a mismatch between its prediction and reality.

The first difficulty is that the beliefs involved with the schema are not necessarily consciously available at first. Richard knew that he suffered from a lack of self-esteem, but he was not aware of its reason. The process started from him describing in concrete details how this manifested: as skeptical self-talk during a daily meeting.

As he was guided to imagine what would happen if he didn’t have those thoughts and acted confidently, his therapist was seeking to retrieve the implicit schema and bring it into consciousness so that its contents would become available for access. Once it had been retrieved, the therapist and Richard worked together to express the belief in the schema in maximally emotional language:

“Feeling any confidence means I’m arrogant, self-centered, and totally insensitive like Dad, and people will hate me for it, so I’ve got to never feel confident, ever.”

The authors have developed a therapeutic approach called Coherence Therapy, whose steps closely follow the steps of the memory reconsolidation process. The example of Richard is from this school of therapy.

In Coherence Therapy (as well as related approaches, such as Internal Family Systems [LW · GW]), one initially avoids any impulse to argue or disprove with the retrieved schema. This would risk it being pushed away before it has become sufficiently activated to allow for reconsolidation.

Instead, one stays with it. Richard was given a card with the above phrase and instructed to review it every day until the next therapy session, just feeling the extent to which it felt true to him. This served to further integrate access to the schema in question, making it better consciously available.

Two weeks later, Richard had frequently noticed his self-doubt, used it as a prompt for reading the card, and experienced its description ringing true as a reason for his thoughts. When speaking with his therapist, he mentioned a particular event which had stuck in his memory. In a recent meeting, he had thought of a solution to a particular problem, but then kept quiet about it. A moment later, another person had spoken up and suggested the same solution in a confident manner. Looking around, Richard had seen the person’s solution and confidence being received positively by the others. Richard had been struck by how that reaction differed from what his schema predicted would happen if he had made the same suggestion in that tone.

Because Richard had made the implicit assumptions in his schema explicit, he was able to consciously notice a situation which seemed to violate those assumptions: a prediction mismatch. His therapist recognized this as a piece of contradictory knowledge which could be used to update the old schema. The therapist then guided Richard through a process intended to activate the old schema while bringing the contradictory information into awareness, triggering a reconsolidation process.

The therapist first instructed Richard to mentally bring himself back to the situation where he had just thought of the solution, but held it back. To properly activate the schema, the therapist guided Richard’s attention to the purpose behind his reluctance and Richard’s certainty of any confidence making him disliked. Next, the therapist told Richard to re-live what happened next: the other person making the same suggestion, and the other people in the room looking pleased rather than angry.

The book then has a transcript of the therapist guiding Richard to repeat this juxtaposition of the old schema and the disconfirming experience (italicized brackets in the original):

Therapist: Stay with that. Stay with being surprised at what you’re seeing—surprised because in your life, you’ve had such a definite knowing that saying something confidently to people will always come across like Dad, like an obnoxious know-it-all, and people will hate that. That’s what you know, yet at the same time, here you’re seeing that saying something confidently isn’t always like Dad, and then people are fine with it. And it’s quite a surprise to know that. [That was an explicit prompting of another side-by-side experiencing of the two incompatible knowings, with the therapist expressing empathy for both, with no indication of any favoring of one knowing over the other. The therapist paused for several seconds, then asked:] Does it feel true to describe it like that? Your old knowing right alongside this other new knowing that’s so different?
Richard: [Quietly, seeming absorbed in the experience.] Yeah.
Therapist: [Softly.] All along, it seemed to you that saying something confidently could be done only in Dad’s dominating way of doing it, and now suddenly you’re seeing that saying something confidently can be done very differently, and it feels fine to people. [This was another deliberate repetition of the same juxtaposition experience.]
Richard: Yeah.
Therapist: Mm-hm. [Silence for about 20 seconds.] So, how is it for you be in touch with both of these knowings, the old one telling you that anything said with confidence means being like Dad, and the new one that knows you can be confident in a way that feels okay to people? [Asking this question repeated the juxtaposition experience yet again, and, in addition, the “how is it” portion of the question prompted Richard to view the experience with mindful or metacognitive awareness, while remaining in the experience.]
Richard: It’s sort of weird. It’s like there’s this part of the world that I didn’t notice before, even though it’s been right there.
Therapist: I’m intrigued by how you put that. Sounds like a significant shift for you.
Richard: Yeah, it is. Huh.
Therapist: You’re seeing both now, the old part of the world and this other part of the world that’s new, even though it was right there all along. [That cued the juxtaposition experience for a fourth time, followed by silence for about 30 seconds.] So, keep seeing both, the old part and the new part, when you open your eyes in a few seconds and come back into the room with me. [Richard soon opens his eyes and blinks a few times.] Can you keep seeing both?
Richard: Yeah.
Therapist: What’s it like to see both and feel both now? [With the transformation sequence complete, this question begins the next step of verification— Step V—because it probes for whether the target learning still exists as an emotional experience.]
Richard: [Pause, then sudden, gleeful laughter.] It’s kind of funny! Like, what? How could I think that? [This is an initial marker indicating that the pro-symptom schema may have been successfully disconfirmed, depotentiated, and dissolved by the transformation sequence.]
Therapist: Do you mean, how could you think that simply saying what you know, or mentioning some good idea that you’ve had, would make you seem arrogant, insensitive and dominating like Dad and be hated for it?
Richard: [Laughing again.] Yeah!

Afterwards, the therapist and Richard wrote a new card together, which Richard was told to review daily:

All along it’s been so clear that if I confidently say what I know, I will always come across as arrogant, insensitive, and dominating like Dad, and be hated for it. And it’s so weird, looking around the room and seeing that it doesn’t come across like that.

The purpose of the card was to provide additional juxtaposition experiences between the old schema and the new knowledge. While the original transformation sequence might have been enough to eliminate the old schema, the schema might also have been stored in the context of many different situations and contained in several memory systems. In such a situation, further juxtapositions would have helped deal with it.

In a follow-up meeting, Richard reported having lost the feelings of self-doubt, and that speaking up no longer felt like it was any big deal. To verify that the old schema really had lost its power, the therapist tried deliberately provoking his old fears again:

Dropping his voice to a quieter tone, the therapist added, “But tell me, when you have something to say and just say it, what about the danger of coming across as a know-it-all, like Dad, and being hated for that? What about your fear of that and how urgent it is to protect yourself from that?” [...]
Richard took in the question, gazed at the therapist in silence for a few seconds, and then replied, “Well, I don’t know what to tell you. All I can say is, that doesn’t trouble me any more. And hearing you say it, it seems a little strange that it ever did—like, what was my problem?”

Applied reconsolidation: the schema update process in general

Now that we have looked at a specific example, we can look at a more general version of the process.

Accessing sequence

In Coherence Therapy, the accessing sequence is the preliminary phase of making both a person’s implicit schema and some disconfirming knowledge accessible, so that they can be used in the juxtaposition process:

  1. Symptom identification. Establishing which specific symptoms the person regards as problematic, and when and where they manifest. In Richard’s case, the general symptom was a lack of confidence, which specifically manifested as negative self-talk in meetings.
  2. Retrieval of target learning. Bringing into explicit awareness the purpose behind the symptoms. This can then be used to guide the search for disconfirming knowledge, as well as accessing the original schema in order to reconsolidate it. In Richard’s case, the purpose was to avoid expressing confidence in a way that would make people hate him.
  3. Identification of disconfirming knowledge. Identifying some past or present experience which directly contradicts the original learning. This knowledge does not necessarily need to feel “better” or “more positive” than the old one, just as long as it is mutually exclusive with the old one. In Richard’s case, the disconfirming knowledge was the experience of his co-worker confidently proposing a solution and being well-received.

Erasure sequence

Once both the target schema and the disconfirming knowledge are known, the erasure steps can be applied to update the learning:

  1. Reactivation of the target schema. Tapping into the felt truth of the original learning, experiencing it as vividly as possible.
  2. Activation of disconfirming knowledge, mismatching the target schema. Activating, at the same time, the contradictory belief and having the experience of simultaneously believing in two different things which cannot both be true.
  3. Repetitions of the target-disconfirmation pairing.

Something that the authors emphasize is that when the target schema is activated, there should be no attempt to explicitly argue against it or disprove it, as this risks pushing it down. Rather, the belief update happens when one experiences their old schema as vividly true, while also experiencing an entirely opposite belief as vividly true. It is the juxtaposition of believing X and not-X at the same time, which triggers an inbuilt contradiction-detection mechanism in the brain and forces a restructuring of one’s belief system to eliminate the inconsistency.

The book notes that this distinguishes Coherence Therapy from approaches such as Cognitive Behavioral Therapy, which is premised on treating some beliefs as intrinsically irrational and then seeking to disprove them. While UtEB does not go further into the comparison, I note that this is a common complaint that I have heard of CBT: that by defaulting to negative emotions being caused by belief distortions, CBT risks belittling those negative emotions which are actually produced by correct evaluations of the world.

I would personally add that not only does treating all of your beliefs - including emotional ones - as provisionally valid seem to be a requirement [LW · GW] for actually updating them, this approach is also good rationality practice. After all, you can only seek evidence to test a theory, not confirm it [LW · GW].

If you notice different parts of your mind having conflicting models of how the world works, the correct epistemic stance should be that you are trying to figure out which one is true - not privileging one of them as “more rational” and trying to disprove the other. Otherwise it will be unavoidable that your preconception will cause you to dismiss as false beliefs which are actually true. (Of course, you can still reasonably anticipate the belief update going a particular way - but you need to take seriously at least the possibility that you will be shown wrong.)

Similar to what Eliezer previously suggested [LW · GW], this can actually be a relief. Not only would it be an error as a matter of probability theory to try to stack the deck towards receiving favorable evidence, doing so would also sabotage the brain’s belief update process. So you might as well give up trying to do so, relax, and just let the evidence come in.

I speculate that this limitation might also be in place in part to help avoid the error where you decide which one of two models is more correct, and then discard the other model entirely. Simultaneously running two contradictory schemas at the same time achieves good communication [LW · GW] within the brain, as it allows both of them to be properly evaluated and merged rather than one of them being thrown away outright. I suspect that in Richard’s case, the resulting process didn’t cause him to entirely discard the notion that some behaviors will make him hated like his dad was - it just removed the overgeneralization which had been produced by having too little training data as the basis of the schema.

Of course, this means that there does need to be some contradictory information available which could be used to disprove the original schema. One might have a schema for which no disconfirmation is available because it is correct, or a schema which might or might not be correct but which is making things worse and cannot easily be disconfirmed. UtEB mentions the example of a man, "Tómas", who had a desire to be understood and validated by someone important in his life. Tómas remarked that a professional therapist who was being paid for his empathy could never fulfill that role. The update contradicting the schema that nobody in his life really understood him, would have to come from someone actually in his life.

Another issue that may pop up with the erasure sequence is that there is another schema which predicts that, for whatever reason, running this transformation may produce adverse effects [LW · GW]. In that case, one needs to address the objecting schema first, essentially be carrying out the entire process on it before returning to the original steps. (This is similar to the phenomenon in e.g. Internal Family Systems [LW · GW], where objecting parts may show up and have their concerns addressed before work on the original part can proceed.)

Verification step

Finally, after the erasure sequence has been run, one seeks to verify that lasting change has indeed happened and that the target schema has been transformed. UtEB offers the following behavioral markers as signs that a learning which has previously generated emotional responses has in fact been erased:

The authors interpret current neuroscience to say that only memory reconsolidation can produce these kinds of markers. They cannot be produced by counteractive or competitive processes, such as trying to learn an opposite habit to replace a neurotic behavior. Counteractive processes are generally fragile and susceptible to relapse. When these markers are observed in clinical work, UtEB argues that one may infer that reconsolidation has led to the original learning being replaced.

Further examples

For additional examples of the schema update process, I recommend reading the book, which contains several more case studies of issues which were dealt with using this approach. Here’s a brief summary of the most detailed ones (note that some of these examples are actually more detailed and include additional complications, such as more than one symptom-producing schema; I have only summarized the most prominent ones to give a taste of them):

As the last item suggests, sometimes just making a schema explicit is enough to start to dismantle it. The authors suggest that the brain has a built-in detection system which compares any consciously experienced beliefs for inconsistencies with other things that a person knows, and can spontaneously create juxtaposition experiences by bringing up such inconsistent information. They suggest that therapies which are based on digging up previously unconscious material, but which do not have an explicit juxtaposition step, work to the extent that the uncovered material happens to trigger this spontaneous mismatch detection. (We already saw this happening with Richard - once his underlying schema had been made conscious, he was startled to later notice what seemed like a contradiction.)

One may note the connection to the model in Consciousness and the Brain that when some subsystem in the brain manages to elevate a mental object into the content of consciousness, multiple subsystems will synchronize their processing around that object. If the object is an explicit belief, then any subsystem which is paying attention to that object may presumably detect inconsistencies with that subsystem’s own models.

Besides these case studies from Coherence Therapy, the authors also analyze published case studies from Accelerated Experiental Dynamic Psychotherapy, Emotion-Focused Therapy, Eye-Movement Desensitization and Reprocessing, and Interpersonal Neurobiology. They try to show how these cases also carried out a juxtaposition process, even if the theoretical frameworks of those therapies did not explicitly realize it. It is the claim of the authors that any therapy which causes lasting emotional change does it through reconsolidation. Finally, the book contains four essays from other therapists (using Coherence Therapy and EMDR), who analyze some of their own case studies.

Evaluating the book’s plausibility

Now that we have looked at the book’s claims, let’s look at whether we should believe in them.

It is unclear to me how reliable the neuroscience results are; the authors cite a number of studies, but each individual claim only references a relatively small number of them.

On a brief look, I could not find any reviews or papers that would have directly made a critical assessment of the book’s model. However, I found something that might be even better.

Behavioral and Brain Sciences is a respected journal covering subject areas across the cognitive sciences. BBS publishes “target articles” which present some kind of a thesis or review about a particular topic, together with tens of brief commentaries which respond to the target article, and a final response by the target article’s authors to the commentaries.

In 2015, four prestigious (with a total of 500 published research articles between them) psychologists published a BBS target article, Memory reconsolidation, emotional arousal, and the process of change in psychotherapy: New insights from brain science (Lane et al. 2015). While the exact model that they outline has a number of differences from the UtEB model, the core idea is the same: that therapeutic change from a wide variety of therapeutic approaches, “including behavioral therapy, cognitive-behavioral therapy, emotion-focused therapy, and psychodynamic psychotherapy, results from the updating of prior emotional memories through a process of reconsolidation that incorporates new emotional experiences.”

One interesting difference was that Lane et al. describe emotional schemas somewhat differently. In their model, the schemas form memory structures with three mutually integrated components: emotional responses, episodic/autobiographical memories, and semantic structures (e.g. abstract beliefs which generalize over the various incidents, such as the claim that “people are untrustworthy”). Any of these components can be used as an entry point to the memory structure, and can potentially update the other components through reconsolidation. They hypothesize that different forms of therapy work by accessing different types of components: e.g. behavior therapy and emotion-focused therapy access emotional responses, conventional psychoanalysis uses access to biographical memories, and cognitive behavioral therapy accesses semantic structures.

I read the target article, all the commentaries, and the responses. Given the similarities between Lane et al.’s model and the UtEB model, I think we can consider the responses to Lane et al. to generally offer a useful evaluation of the UtEB model as well.

One significant difference which needs to be noted is that Lane et al.’s model of memory reconsolidation does not mention the requirement for a prediction mismatch before reconsolidation can happen. This was remarked on in the response from UtEB’s authors. In their counter-response, Lane et al. noted UtEB’s model to be highly compatible with theirs, and remarked that further research is needed to nail down the conditions which make reconsolidation the most effective.

The other responses to Lane et al. were mostly from psychologists, psychiatrists, and neuroscientists, but also included the occasional economist, philosopher, philologist and folklorist. Several of the responses were generally positive and mostly wanted to contribute additional details or point out future research directions.

However, there were also a number of skeptical responses. A common theme which emerged from several concerned the limitations of the current neuroscience research on memory reconsolidation. In particular, most of the studies so far have been carried out on rats, and specifically testing the elimination of a fear response to electric shocks. As one of the responses points out, “neither the stimuli nor the subjects are generalizable to the kind of rich autobiographical memories involved in therapy.” A number also raised the question whether all therapeutic change really involves reconsolidation, as opposed to some related mechanism, such as creating new memory structures which compete with the original as opposed to replacing it.

My non-expert reading is that the critical responses are right in that a gap remains between the clinical and behavioral findings on the other hand, and the neuroscience findings on the other. There are various patterns which can be derived from psychological research and clinical therapy experience, and a small number of neuroscience findings which establish the existence of something that could explain those patterns. However, the neuroscience findings have only been established in a rather narrow and limited context; the connection between them and the higher-level patterns is a plausible link, but it remains speculative nonetheless.

Personally I consider the book’s model tentatively promising, because it seems to explain many observations which I had independently arrived at before reading it. For example, I had noticed an interesting thing with anxieties, where I let e.g. a sensation of social anxiety stay active in my mind, neither accepting it as truth nor pushing it away while I went to do social things. This would then cause the anxiety to update [LW · GW], making me feel less anxious if it was indeed the case that the social interaction was harmless. This fits nicely together with the framework of an activated memory structure becoming open to reconsolidation and then being updated by a prediction mismatch (the situation not being as bad as expected).

Likewise, in my post Integrating disagreeing subagents [LW · GW], I reviewed a variety of rationality and therapeutic techniques, and suggested that they mostly worked either by merging or combining two existing models that a person’s brain already had, or augmenting the existing models by collecting additional information.

In particular, I considered an example from cognitive behavioral therapy, where a man named Walter was feeling like he was impossible to be in a relationship with after he had broken up with his boyfriend. At the same time, he did not think that someone else breaking up with their partner was an indication of them being impossible to be in a relationship with. He and his therapist role-played an interaction where the therapist pretended to be a friend who had recently broken up, and Walter explained why this did not make the friend a relationship failure. In the process of doing so, Walter suddenly realized that he wasn’t a failure, either.

I commented:

Walter was asked whether he’d say something harsh to a friend, and he said no, but that alone wasn’t enough to improve his condition. What did help was putting him in a position where he had to really think through the arguments for why this is irrational in order to convince his friend, and then, after having formulated the arguments once himself, get convinced by them himself.
In terms of our framework, we might say that a part of Walter’s mind contained a model which output a harsh judgment of himself, while another part contained a model which would output a much less harsher judgment of someone else who was in otherwise identical circumstances. Just bringing up the existence of this contradiction wasn’t enough to change it: it caused the contradiction to be noticed, but didn’t activate the relevant models extensively enough for their contents to be reprocessed.
But when Walter had to role-play a situation where he thought of himself as actually talking with a depressed friend, that required him to more fully activate the non-judgmental model and apply it to the relevant situation. This caused him to blend with the model, taking its perspective as the truth. When that perspective was then propagated to the self-critical model, the easiest way for the mind to resolve the conflict was simply to alter the model producing the self-critical thoughts.

This seems like a straightforward instance of belief juxtaposition, and one where I ended up independently deriving something like UtEB’s memory reconsolidation model: I too noted that the relevant belief structures need to be simultaneously activated in the right way to allow for the brain to revise one of them after noticing the contradiction. In general, UtEB’s model of how things work rings true in my experience, making me inclined to believe that its description of how therapy works is correct, and that its model of how it is connected to neuroscience might also be.

UtEB and the subagent model

As many readers know, I have been writing a sequence of posts [LW · GW] on multi-agent models of mind. In Building up to an Internal Family Systems model [LW · GW], I suggested that the human mind might contain something like subagents which try to ensure that past catastrophes do not repeat. In subagents, coherence, and akrasia in humans [LW · GW], I suggested that behaviors such as procrastination, indecision, and seemingly inconsistent behavior result from different subagents having disagreements over what to do.

As I already mentioned, my post on integrating disagreeing subagents [LW · GW] took the model in the direction of interpreting disagreeing subagents as conflicting beliefs or models within a person’s brain. Subagents, trauma and rationality [LW · GW] further suggested that the appearance of drastically different personalities within a single person might result from unintegrated memory networks, which resist integration due to various traumatic experiences.

This post has discussed UtEB’s model of conflicting emotional schemas in a way which further equates “subagents” with beliefs - in this case, the various schemas seem closely related to what e.g. Internal Family Systems calls “parts”. In many situations, it is probably fair to say that this is what subagents are.

That said, I think that while this covers a very important subset of subagents, not everything which I have been referring to as a subagent falls straightforwardly under the belief-schema model. In subagents and neural Turing machines [LW · GW] as well as Against “System 1” and “System 2” [LW · GW], I also covered subagents in a more general way, as also including e.g. the kinds of subsystems which carry out object recognition and are used to carry out tasks like arithmetic. This was also the lens through which I looked at subagents in my summary of Consciousness and the Brain [LW · GW]. Which kind of view is the most useful, depends on exactly what phenomenon we are trying to understand.


Comments sorted by top scores.

comment by pjeby · 2019-10-13T01:46:04.591Z · score: 21 (6 votes) · LW · GW

Richard took in the question, gazed at the therapist in silence for a few seconds, and then replied, “Well, I don’t know what to tell you. All I can say is, that doesn’t trouble me any more. And hearing you say it, it seems a little strange that it ever did—like, what was my problem?”

Hah. I've had that experience with clients (and myself) so many times I coined the term "change amnesia" -- it's extremely common following successful reconsolidation, and a good sign that the specific change made is going to "stick".

I was going to say that my approach doesn't use side-by-side comparison, but then on further thought the critical stage in my SAMMSA model involves the contradiction between someone else's view of them as lazy, incompetent, stupid, selfish, etc., and their actually not being those things at the time the judgment was made.

But that in and of itself isn't sufficient to make a change, AFAICT. It might open things up to allow the reconsolidation that follows, but it's necessary also to make new predictions in light of the thing that's been opened for reconsolidation.

Huh. This is interesting. It explains why certain things I've been doing work, that I knew you had to do in order to get results, but which I didn't know why they needed to be done. Metaphorically, I've been telling people they need to open the right "file" in their brain before they can make a change, but it seems that I've also been having them flag errors in the file first, before adding new data to the file.

The Work of Byron Katie also follows this general pattern, i.e. : He should listen to me -- is that true? (reflect on experience) Can I absolutely know that's true? (generate alternative possibilities) How do I react when I think this thought? (predict negative consequences for continuing with this pattern) Who would I be without that thought? (predict positive consequences of thinking differently) And finally turnarounds (e.g. "I should listen to him", "he shouldn't listen to me", "I should listen to me"), to generate alternative schemata that might apply to the trigger situation.

comment by pjeby · 2019-10-14T19:47:31.028Z · score: 17 (4 votes) · LW · GW

Of course, this means that there does need to be some contradictory information available which could be used to disprove the original schema. One might have a schema for which no disconfirmation is available because it is correct, or a schema which might or might not be correct but which is making things worse and cannot easily be disconfirmed.

This view is ignoring the distinction between denotation and connotation, or as I like to think of it, between prediction and evaluation. Our memories don't just create factual prediction, they are also tagged with evaluations: meaning, feelings, etc.

So, it's quite possible to reconsolidate different evaluations for the same factual predictions. For example:

UtEB mentions the example of a man, "Tómas", who had a desire to be understood and validated by someone important in his life. Tómas remarked that a professional therapist who was being paid for his empathy could never fulfill that role. The update contradicting the schema that nobody in his life really understood him, would have to come from someone actually in his life.

The evaluation Tómas is making is itself based in some other memory that can be reconsolidated, so that it is no longer required for somebody else to understand him. The experience of "feeling understood" is not something that actually comes from outside, it is something your brain generates according to learned rules. In this case, Tómas has learned that only certain specific people's understanding counts or is meaningful... and this learning is just as subject to reconsolidation as anything else!

Another issue that may pop up with the erasure sequence is that there is another schema which predicts that, for whatever reason, running this transformation may produce adverse effects. In that case, one needs to address the objecting schema first, essentially be carrying out the entire process on it before returning to the original steps. (This is similar to the phenomenon in e.g. Internal Family Systems, where objecting parts may show up and have their concerns addressed before work on the original part can proceed.)

Yes, checking for objections is of critical importance, because if you don't, the thing you think you fixed can come back in a few days or weeks. But this isn't because there's an agent that "objects", it's just that the thing you were working on is reinforced by another prediction/evaluation.

For example, let's say that Joe is having trouble promoting himself or his work, because he's learned never to brag and that bragging is bad. He learned this because his mother always punished him for bragging and said "Pride goeth before a fall". We do some work and get rid of that immediate response, but don't check for objections, so we miss the part where the implicit, unspoken part of the interaction was, "If I don't punish you for bragging, you'll grow up to be an obnoxious selfish person who nobody will like".

So, because of that, we've removed Joe's semi-explicit belief that bragging is prideful and will lead to a disastrous "fall", but not his more-implicit belief that he needs to punish himself for bragging. In the high of having changed the first belief, Joe will go out and start promoting himself, but feeling weirdly bad about it, until he stops again.

IOW, the "objecting" schema isn't really objecting per se. The schema is rather reinforcing the previous schema, with a need or desire to punish himself for violating it, leading to a return of the old behavior and extinguishing the new behavior we tried to establish.

These reinforcing schema don't always show up with an obvious objection at the time you're making a change, and people who are eager to get the change done will often report over-optimistic predictions when they're doing the reconsolidation part. Sometimes, the "objection" is nothing more than a vague feeling that the new scenario being projected isn't realistic in some way, or "isn't quite right". When that is the case, I always dig deeper immediately to uncover what other predictions are being made.

(Of course, for this specific pattern of "if I don't punish X in way Y, I/they will become bad type of person Z", I have a standard format for finding it even before getting to the reconsolidation part, as it's super-common in issues of self-sabotage, procrastination, perfectionism, etc.)

comment by orthonormal · 2019-10-10T04:49:39.398Z · score: 13 (7 votes) · LW · GW

Well, damn. I started out reading this really skeptically, but the process fit my own past experiences of changing core beliefs frighteningly well. (The only thing that's ever noticeably worked for me is something like showing that one of my IFS parts is a hypocrite, which results in it more or less self-destructing and allowing change to happen.) I just emailed a Coherence Therapy practitioner, since I'm looking for a new therapist; this seems worth a try.

comment by Kaj_Sotala · 2019-10-10T10:57:46.894Z · score: 5 (2 votes) · LW · GW

Good luck with the therapist! Curious to hear how that goes, if there's anything about your experience that you feel like sharing.

comment by DonyChristie · 2019-10-13T00:00:36.552Z · score: 11 (5 votes) · LW · GW

I'm so glad someone did a writeup of this! Part of me has wanted to, I think I have a draft... I remember going through severe depression over four years ago and one of my reprieves was joyfully reading the papers written about coherence psychology. I will definitely be linking this post as a reference.

There are many times I am talking with people and want to reference from the conceptual structure of coherence psychology, but there is way too much inferential distance especially with aspiring rationalists who are not therapy geeks, so I end up mentally flailing my arms in frustration. The theory seems like a better candidate for The One True Psychotherapy than almost any other and it pains me to see people go about solving their problems without it in their toolkit, and not being able to communicate this to them. e.g. It's frustrating to see people trying to correct the output of emotional schemas without accessing the generating model for disconfirmation. e.g. A person may feel uncomfortable with someone else who has low self-esteem so they will try to correct it verbally without engaging in a process that will change the underlying 'pro-symptom position'.

There's the related problem that there are very few coherence therapists. I don't think most psychologists have heard of this and I find that confusing.

Oh, there's also the fact that I tried a coherence therapist and didn't find it that helpful the way it was done. They were fine to talk to but it seems retrospectively like they were cargo-culting the motions of coherence therapy as outlined by Ecker et al. I haven't had other therapists but I suspect the inefficacy is only very weak evidence pointing against the modality vs other modalities and more a problem with cramming an attempt at powerful introspection into expensive 1-hour blocks. i.e. I think psychotherapeutic structure across the board is broken and when the singularity happens it won't be a problem anymore.

My hope is that we can develop new delivery structures into which we can import psychological techniques and have them deployed at scale while being better than 1-hour weeklies, 8-hour shamanic trips, or that annoying app with the emotionally saccharine bird.

See also: The Method of Levels

comment by pjeby · 2019-10-13T02:10:58.209Z · score: 28 (8 votes) · LW · GW

a problem with cramming an attempt at powerful introspection into expensive 1-hour blocks.

The hard part of implementing this isn't the reconsolidation part. That part is like 10-20 minutes, especially with practice. The hard part is identifying the things that need to be reconsolidated in the first place, and because that is basically a debugging process, it can take a fair amount more time.

I've seen a lot of "one theory fits all" therapeutic methods in the past (like Method Of Levels), but in practice for the type of work I do with people, none of them are very good at quickly identifying things because they're too good at describing everything, and the brain isn't just one thing.

So now I work with what I call "search grids" -- common patterns of bugs that I can go through and check, is it this? is it that? is it more like X or Y? -- and it saves boatloads of time.

(To be fair, though, I work with only a select audience with selected problems, so it's quite possible that the applicability of my grids isn't that good outside that audience, even though I think I've a fair shot at an evolutionary justification for drawing the lines on my grids where they are.)

Still, I can't even imagine trying to really solve somebody's problems in an hour a week, unless they were already trained and had been coached through the methods once or thrice. And if they're at that point, I 'd work with them via email anyway. The only part of these processes that actually requires real-time interaction is getting people over what I call their "meta-issue" -- the schema they have that gets in the way of being able to reflect on their issues.

For example, I've had clients who had what you might call a "be a good student" schema that keeps them from accurately reporting their emotions, responses, or progress in applying a reconsolidation technique. Others who would deflect and deny ever having any negative experiences or even any problems, despite having just asked me for help with same. These kinds of meta-issues are the hardest and most time-consuming part of getting someone ready to change.

Ten or twelve years ago when I thought I'd unlocked the secrets of the universe and that memory reconsolidation was going to change everything and everyone, I didn't realize yet that hard part 1 (needing to identify the things to change) and hard part 2 (needing to get past meta issues), meant that it is impossible to mass-produce change techniques.

That is, you can't write a single document, record a single video, etc. that will convey to all its consumers what they need in order to actually implement effective change.

I don't mean that you can't successfully communicate the ideas or the steps. I just mean that implementing those steps is not a simple matter of following procedure, because of the aforementioned Hard Parts. It's like expecting someone to learn to bike, drive, or debug programs from a manual.

This doesn't mean it's impossible for someone to teach themselves from such material, but it's not trivial, and my dreams of revolutionizing personal development by mass-producing books or workshops died an ignoble death almost a decade ago. (Incidentally, this limtation is also why for almost any given school of therapy, you will find that people who advertise doing therapy in that school, aren't always capable of doing more than giving it lip service or cargo cultery.)

It's a bit like the Interdict of Merlin in HPMOR: successful techniques can only be passed from one living mind to another, or independently discovered. You can write down your notes and share the story of your discovery, and then people either discover it again for themselves, learn it from interacting with someone who knows, or go through the motions and cargo-cult it.

comment by mr-hire · 2019-10-13T13:32:38.783Z · score: 13 (3 votes) · LW · GW

hard part 1 (needing to identify the things to change) and hard part 2 (needing to get past meta issues).

Yes, all the people who are really good at this stuff have really finely honed reflexes related to: Sniffing out common issues and, immediately going meta when they encounter resistance. I do think this can be more systematized (have you shared your grids anywhere?) but don't think people can always be debugged quickly.

comment by Raemon · 2019-10-14T00:42:04.470Z · score: 7 (3 votes) · LW · GW
It's a bit like the Interdict of Merlin in HPMOR: successful techniques can only be passed from one living mind to another, or independently discovered. You can write down your notes and share the story of your discovery, and then people either discover it again for themselves, learn it from interacting with someone who knows, or go through the motions and cargo-cult it.

Yeah, "Interdict of Merlin" has been a helpful (to me) handle on why a lot of rationality technique sharing is hard.

comment by Qiaochu_Yuan · 2019-10-13T21:20:26.654Z · score: 6 (3 votes) · LW · GW
The only part of these processes that actually requires real-time interaction is getting people over what I call their "meta-issue" -- the schema they have that gets in the way of being able to reflect on their issues.
For example, I've had clients who had what you might call a "be a good student" schema that keeps them from accurately reporting their emotions, responses, or progress in applying a reconsolidation technique. Others who would deflect and deny ever having any negative experiences or even any problems, despite having just asked me for help with same. These kinds of meta-issues are the hardest and most time-consuming part of getting someone ready to change.

Oof, yeah, this resonates a lot with experiences I've been having with myself and others the last few months, since coming out of a workshop on the bio-emotive framework. There are towers of meta-issues, meta-issues that prevent themselves from being looked at... what a mess.

In retrospect this illuminates something for me about the CFAR workshop and its techniques - a pattern I ran into for years was that I gradually became averse to every single CFAR technique I tried, so I never used them on my own, and I don't think I'm alone in that. I think - and this is deeply ironic - that the CFAR techniques as a whole never went meta enough to catch "meta-issues," not in any really systematic way.

comment by pjeby · 2019-10-13T23:37:10.490Z · score: 36 (6 votes) · LW · GW

the CFAR techniques as a whole never went meta enough to catch "meta-issues," not in any really systematic way.

There is no level of meta systemization that can overrule a person's meta issues, because their meta issues are always "one level higher than you". ;-) To put it more precisely, no passive information-passing process can bypass a person's meta issues, any more than you can turn a shredder into a fax machine by feeding it a copier manual. The incoming information gets processed through an existing filter that deletes any information that doesn't fit the paradigm, or mangles the information until it does fit.

As I was joking above, self-help really is governed by the Interdict of Merlin: you can discover powerful "spells", or you can pass them mind-to-mind, but you can't learn them from a book, except insofar as the book might give you enough clues to rediscover the spell for yourself.

There are towers of meta-issues, meta-issues that prevent themselves from being looked at... what a mess.

Actually, meta-issues don't have meta-levels, they only simulate recursion via iteration. A meta-issue might be something like, "when you are learning something from a teacher, then try to suck up by being super-successful really fast". This is only one meta-level, and while it can look recursive in effect, it's just an illusion.

Let's say I tell the person with this meta-issue they need to inquire about some emotional state. If we're at a part of the process where there's a possibility for them to have succeeded in fixing the problem we're working on, they will under-report issues and problems (like a lingering negative feeling) or describe hyper-optimistic scenarios that don't match what they're really feeling.

Then, if I point this out, they may now apply the pattern again, by trying to prove even harder that they've already learned what I'm telling them, even though they haven't.

(This isn't really recursion or a new meta-level, it's just the same pattern being applied to a new stimulus or situation that only incidentally happens to be at a different meta-level, if that makes sense.)

So the only escape from this iterated pseudo-recursion is for them to catch themselves in the act of this automatic response, which requires either a lot of iteration (like it did for me when I was learning my own meta-issues), or an outside party who can spot it and say, "stop that! you're doing it again..." until the person can recognize themselves doing it.

comment by Kaj_Sotala · 2019-10-14T11:23:34.749Z · score: 14 (3 votes) · LW · GW

The incoming information gets processed through an existing filter that deletes any information that doesn't fit the paradigm, or mangles the information until it does fit.

By the way, I believe this to be related to the kind of thing that Valentine was trying to point at in the Kensho post [LW · GW]:

Imagine you’re in a world where people have literally forgotten how to look up from their cell phones. They use maps and camera functions to navigate, and they use chat programs to communicate with one another. They’re so focused on their phones that they don’t notice most stimuli coming in by other means.

Somehow, by a miracle we’ll just inject mysteriously into this thought experiment, you look up, and suddenly you remember that you can actually just see the world directly. You realize you had forgotten you were holding a cell phone.

In your excitement, you try texting your friend Alex:

YOU: Hey! Look up!
ALEX: Hi! Look up what?
YOU: No, I mean, you’re holding a cell phone. Look up from it!
ALEX: Yeah, I know I have a cell phone.
ALEX: <alex_cell_phone.jpg>
ALEX: If I look up from my phone, I just see our conversation.
YOU: No, that’s a picture of your cell phone. You’re still looking at the phone.
YOU: Seriously, try looking up!
ALEX: Okay…
ALEX: looks up
YOU: No, you just typed the characters “looks up”. Use your eyes!
ALEX: Um… I AM using my eyes. How else could I read this?
YOU: Exactly! Look above the text!
ALEX: Above the text is just the menu for the chat program.
YOU: Look above that!
ALEX: There isn’t anything above that. That’s the top.
ALEX: Are you okay?

You now realize you have a perplexing challenge made of two apparent facts.

First, Alex doesn’t have a place in their mind where the idea of “look up” can land in the way you intend. They are going to keep misunderstanding you.

Second, your only familiar way of interacting with Alex is through text, which seems to require somehow explaining what you mean.

But it’s so obvious! How can it be this hard to convey? And clearly some part of Alex already knows it and they just forgot like you had; otherwise they wouldn’t be able to walk around and use their phone. Maybe you can find some way of describing it to Alex that will help them notice that they already know…?

Or… maybe if you rendezvous with them, you can somehow figure out how to reach forward and just pull their head up? But you’re not sure you can do that; you’ve never used your hands that way before. And you might hurt them. And it seems kind of violating to try.

And later in this comment:

... my personal impression had been that it’s actually quite easy to see what Looking is and how one might translate it into reductionist third-person perspectives. But my personal experience had been that whenever I tried to share that translation, I’d bounce off of weird walls of misunderstanding. After a while I noticed that the nature of the bounce had a structure to it, and that that structure has self-reference. (Once again, analogies along the lines of “get out of the car” and “look up from your phone” come to mind.) After watching this over several months and running some informal tests, and comparing it to things CFAR has been doing (successfully, strugglingly, or failing to do) over the six years I’ve been there, it became obvious to me that there are some mental structures people run by default that actively block the process of Looking. And for many people, those structures have a pretty strong hold on what they say and consciously think. I’ve learned to expect that explaining Looking to those structures simply will never work. [...]

If I try to argue with a paperclip maximizer about how maximizing paperclips isn’t all there is to life, it will care to listen only to the extent that listening will help it maximize paperclips. I claim that by default, human mind design generates something analogous to a bunch of paperclip maximizers. If I’m stuck talking to one of someone’s paperclip maximizers, then even if I see that there are other parts of their mind that would like to engage with what I’m saying, I’m stuck talking to a chunk of their mind that will never understand what I’m saying.

... with the particular difference that he was pointing to a case where the meta-issue involves one's basic ontology, and where (if I interpret him correctly) he thinks that most people have that meta-issue by default.

comment by pjeby · 2019-10-14T18:46:00.664Z · score: 21 (6 votes) · LW · GW

Yep, that's it exactly. I metaphorically yank people's heads up from their phones for pay. ;-)

Actually, it's more like I keep sticking my hand in front of the phone until they learn to notice the difference and look up themselves. Fortunately, this doesn't take too long for most people, but how fast it goes depends on how good they are at fooling me into thinking they're actually looking up when they're really still looking at the phone.

The easiest way to detect "looking at the phone" is to ask someone a yes or no or question, and see how long of an answer you get. If somebody starts talking about the past or future, they're not actually paying attention to their inner experience, because inner experience is always present tense. e.g. "I see my mom yelling at me" is an experience, while "my mom used to yell at me" is a commentary on experience. Causal chains (x happened because y) are also commentary, as are generalizations.

(Incidentally, this is another reason I dislike IFS' model: it encourages adding commentary like "a part of me X" instead of just saying "X", which makes it more difficult to know if what you're hearing is describing experience, or commentary/abstraction on experience.)

I imagine it would be possible to create a training program for people to recognize these verbal patterns and to then verify their own spoken or written statements using them, but it would be harder to get people to go through such a program vs. paying me to help fix a problem of theirs, and learning it by doing it along the way.

comment by mr-hire · 2019-10-13T13:35:29.519Z · score: 4 (2 votes) · LW · GW

So now I work with what I call "search grids" -- common patterns of bugs that I can go through and check, is it this? is it that? is it more like X or Y? -- and it saves boatloads of time.

I'm not sure if you're familiar with NLP meta patterns, but this is how they were originally developed.

comment by pjeby · 2019-10-13T22:56:43.329Z · score: 30 (6 votes) · LW · GW

My search grids don't look much like anything from NLP, least of all metaprograms. Instead, they're patterns of common bugs in the brain that I believe are evolutionarily defined, and likely to be universal.

For example, working with people on self-image problems, I've found that there appear to be only three critical "flavors" of self-judgment that create life-long low self-esteem in some area, and associated compulsive or avoidant behaviors:

  • Belief that one is bad, defective, or malicious (i.e. lacking in care/altruism for friends or family)
  • Belief that one is foolish, incapable, incompetent, unworthy, etc. (i.e. lacking in ability to learn/improve/perform)
  • Belief that one is selfish, irresponsible, careless, etc. (i.e. not respecting what the family or community values or believes important)

(Notice that these are things that, if you were bad enough at them in the ancestral environment, or if people only thought you were, you would lose reproductive opportunities and/or your life due to ostracism. So it's reasonable to assume that we have wiring biased to treat these as high-priority long-term drivers of compensatory signaling behavior.)

Anyway, when somebody gets taught that some behavior (e.g. showing off, not working hard, forgetting things) equates to one of these morality-like judgments as a persistent quality of themselves, they often develop a compulsive need to prove otherwise, which makes them choose their goals, not based on the goal's actual utility to themself or others, but rather based on the goal's perceived value as a means of virtue-signalling. (Which then leads to a pattern of continually trying to achieve similar goals and either failing, or feeling as though the goal was unsatisfactory despite succeeding at it.)

Simply knowing this fact is hugely helpful in narrowing down the search space for the memories needing reconsolidation. All you have to do is look for emotionally salient instances or patterns of learning that problematic behavior X equated to judgment flavor Y. If you've ever done something like Method of Levels or similarly undirected "theories of everything", you'll know you can wander through somebody's conscious understanding of the problem for ages without getting anywhere or even being sure you are getting somewhere.

In contrast, if somebody tells me that they've been pursuing X goal for years and keep failing at it, or even when they do achieve it, it feels awful, or if they have impostor syndrome, I can go after it immediately by looking at what virtue they're trying to signal (or flipping it and asking what bad judgment they would have of themselves if they had to give up on the goal or it were impossible for them), and then we're off to the races of tracking down examples of where they learned that judgment from, the various implicit learnings that went with it, and the underlying social values and assumptions they absorbed in the process.

Along the way, this also helps pinpoint self-destructive and self-undermining behavior and self-talk (and gets rid of them), without having to first dig around in someone's self-talk to get at the beliefs. (Which is a big win, because people are rarely aware of the ways they treat themselves badly, and often think they are helping or "motivating" themselves by being pessimistic or self-critical or having overly high expectations. So if you ask people what they think is their problem, they will often insist they need more of precisely the thing that is causing the problem in the first place!)

What I found interesting about this article was that it highlights why I've needed to pinpoint which "flavor" of low self-esteem was involved in a memory in order to fix the relevant behavior or belief: without that key piece of information, you can't generate a correct contradiction for it!

In my early use and development of the implicit beliefs framework (SAMMSA) that I now use this grid in, I was just asking the client what positive quality the other person in the memory thought the client lacked, and then helping them discover how they in fact possessed that quality at the time. But this process was still a bit hit or miss until I worked out that there were really only those three kinds or flavors of quality, regardless of how the thing was named or presented, which made things easier because I could point to the three things and ask "which of these is it more like?"

Sometimes people hesitate a bit, and think it's between two of them, and we might have to try it both ways in such a case. But that's still a lot faster than wandering around with no idea where to start or if you're getting anywhere. Discovering the three "EPIC failures of trust" (flavors of moral judgment) made the process reliable because it adds a well-formedness condition that can be checked before completing the technique, thus allowing earlier error correction and early trimming of dead-ends from the search tree, so to speak.

And now, thanks to this article, I have a better understanding of why this was needed, in a higher-order sense, which might lead to the development of new techniques, or at least a filter for better understanding or improving other techniques that appear to work via memory reconsolidation.

comment by Hazard · 2019-10-10T02:15:02.224Z · score: 8 (4 votes) · LW · GW

I like the idea of aiming to just hold the contradiction in your head and boggle at it for a while. Make me think of hold onto the curiosity [LW · GW]. Something something if the brain is layers and layers of predictive processing, focusing on the prediction error is the quickest way to get it to change.

As always, I've been enjoying this series and am finding it to be a useful compilation of a lot of ideas. Thanks!

comment by G Gordon Worley III (gworley) · 2019-10-18T01:12:14.788Z · score: 7 (3 votes) · LW · GW

My experience of personal development and models of how it works also suggest that, whether or not this model provides a correct etiology for various methods of therapy and self-help, it is describing the correct pattern, so is at least useful in that vein. I think what is potentially most interesting about this theory is not what it suggests in terms of therapy (lots of people have already semi-independently figured out that what's described here is how the process works and they've called it by lots of different names) but that it begins to point towards a specific mechanism we can investigate to explain the phenomena. My guess is that it will be partly wrong, but my hope is that it points us towards directions of neuroscience research that may allow us to develop better, successor theories that will result in us understanding the mechanism of action at play in these many therapy and self-help techniques.

comment by habryka (habryka4) · 2019-10-14T18:10:17.923Z · score: 7 (3 votes) · LW · GW

Promoted to curated: This is one of the best book reviews I've read in a long time. Here is why:

  • A very clear summary of the core claims in the book (at least from what I can tell without having read the book)
  • Putting the material in the context of other models discussed both on LW, in your own sequence, and in the literature
  • A broader explanation of the empirical context of the book, including new papers that cover highly related material somewhat independently
  • A reasonable summary of critiques, both by yourself and others

I also think the overall topic of the book is pretty important and worth covering. In some sense the core question of rationality is "how do we update our beliefs and how we can do that better?", and I found the ideas covered in this post to tackle that question pretty straightforwardly.

Obviously, building good models in this space is hard, so I don't think we should canonize these models immediately, and they likely do still have major flaws, but I agree with you that they resonate a good amount with my personal experience.

comment by vin · 2019-10-10T18:37:07.784Z · score: 6 (3 votes) · LW · GW

Thanks a lot for this post! It was a very enjoyable read. Coherence Therapy seems really promising to me.

When you compare CT to CBT you write

I note that this is a common complaint that I have heard of CBT: that by defaulting to negative emotions being caused by belief distortions, CBT risks belittling those negative emotions which are actually produced by correct evaluations of the world.

What would be an example of a negative emotion produced by a correct evaluation of the world? Being depressed about how many beings are suffering in this moment?
Although this is an accurate belief, it is still mostly unhelpful to be depressed about it all the time. Would CBT and CT come to different conclusions as to what to do about it?

comment by Kaj_Sotala · 2019-10-11T13:55:50.421Z · score: 3 (1 votes) · LW · GW

Being depressed about how many beings are suffering in this moment?

That could be one example, though I was thinking more about examples such as Tomas in the text - you can correctly believe in there being nobody in your life who really understands your problems, in which case it's also valid to want it.

Whether depression is a useful reaction to that situation is another matter, and I'm under the impression that CT can sometimes alter one's schemas about how to respond to a bad circumstance. E.g. Ted's case changed his behavior after he made the update that his strategy wasn't ever going to work. But that doesn't make the original evaluation of the badness of the circumstances any less correct. (and at least some interpretations of CBT are premised on that evaluation being irrational)

comment by paragonal · 2019-10-10T18:36:50.699Z · score: 3 (2 votes) · LW · GW

Does the book talk about schizophrenia? I'm a bit skeptical that coherence therapy and IFS can be used to heal it but I'm quite interested in hearing your thoughts about schizophrenia in relation to subagent models.

comment by Kaj_Sotala · 2019-10-10T10:57:03.029Z · score: 8 (4 votes) · LW · GW

Schizophrenia is not listed in the book's example list of conditions that Coherence Therapy might work for; there is a case study of a woman who hears hallucinatory voices, though the report states that "She did not fit the typical pattern of schizophrenia, which was the diagnosis she had been given". The general impression I get is that the writer treats them as a psychotic symptom related to her depression rather than her being schizophrenic in general.

I don't feel like I know enough about schizophrenia to put it in a subagent context.

comment by paragonal · 2019-10-10T11:32:41.093Z · score: 1 (1 votes) · LW · GW


comment by G Gordon Worley III (gworley) · 2019-10-18T01:05:38.846Z · score: 4 (2 votes) · LW · GW

My expectation is that it wouldn't work because my model of psychotic disorders suggests that they are primarily caused by sensory processing issues caused by overly strong ontology that causes the psychotic brain's model of the world to become uncorrelated with direct experience, so psychotic disorders need special treatment to deal with the unique problems they create that prevent conventional therapy techniques from working on them because those techniques never get a chance to start working before they have already been warped into providing evidence that further confirms psychotic beliefs and behaviors as adaptive.