This essay was originally published using the framework’s earlier terminology (CFM, with mechanisms SCT, IH, and RDIS). The current nomenclature is the Signal Loss Model (SLM = UC + ND + PRD). See On Renaming: From Constraint Failure to Signal Loss for the terminology map.
This is the seventh essay in our ten-part series on the Signal Loss Model. The first six built the mechanism and include the full citation trail. This one asks the harder question: given everything you now understand about what’s actually failing, how do you choose an intervention without repeating the same sequencing errors that brought you here?
If you’ve followed the series, you know the architecture: a nervous system stuck in high alert, an inflammatory baseline that quietly blocks change, a reward system running too flat to register progress when it arrives. Three self-reinforcing mechanisms that produce a state stable in its dysfunction and resistant to most of what gets tried.
So when the last intervention didn’t get you out of that state, the problem wasn’t effort and it wasn’t the tools. You were applying pattern-level tools to a system that wasn’t in a learning state.
Which raises the next question: how do you choose what to do about it?
Most people who arrive here, educated, resourceful, genuinely motivated, answer by swapping “try harder” for “try different.” They accumulate interventions the way they once accumulated credentials, on the assumption that effort in the right direction eventually pays off. It often doesn’t, because they’ve skipped a prior question.
That question is epistemological. It asks what would actually convince you that an intervention can work for your nervous system, in your current biological state. Not in a trial population. Not for the colleague who swears by it. For you, now. Most people have never asked it.
I. What Magical Thinking Actually Looks Like Here
When most people hear “magical thinking,” they picture crystals, chakras, and supplements with proprietary blends. That’s not the problem. Or rather, that’s not your problem.
The magical thinking that traps sophisticated people is evidence-adjacent. It borrows the language of science and self-knowledge while committing the same underlying error: choosing an intervention for reasons that have nothing to do with whether it can work in the biological conditions you’re actually in.
It shows up in recognizable forms.
Intensity as efficacy. If it was hard, it must have worked. The “ordeal model” of transformation (heroic doses, extreme environments, punishing physical regimens) rests on the assumption that difficulty is a proxy for depth. Sometimes it is. Often it isn’t. Intensity without the biological conditions for encoding change produces powerful experiences that don’t hold.
Insight as change. Understanding the problem is not the same as solving it. The reader who has made it to essay seven of a ten-part scientific series on their own psychological collapse is, by definition, someone who has invested heavily in understanding. The investment is not wasted. But insight requires a substrate to land on. When that substrate has been suppressed, when the tissue-level capacity for encoding new patterns has been compromised, insight metabolizes into self-narration rather than change. You get better at explaining yourself. But you still don’t move.
Sequencing by availability. I can do this Tuesday, so I’ll start here. This is perhaps the most invisible error, because it masquerades as pragmatism. The intervention that happens to be accessible (the therapist with an opening, the retreat with a spot, the medication the prescriber is comfortable with) becomes the intervention attempted, regardless of whether the biological moment is right for it. Availability is not a sequencing principle. It just functions as one, by default.
Modality identity. I’m a therapy person. I’m skeptical of medication. I don’t do group work. These are not clinical assessments. They are identity commitments that constrain the solution space before the problem has been analyzed. They’re understandable. We form relationships with the approaches that have helped us, and we develop reasonable skepticism toward the ones that haven’t. But modality loyalty is a prior that can override the biology, and when it does, sequencing fails before it begins.
The “tried everything” conclusion. This one is the most consequential, because it forecloses. When someone says they’ve tried everything, they almost always mean they’ve tried many things, in an order determined by availability and identity rather than biology, at moments when the system may not have been capable of the change those interventions require. “Tried everything” is rarely accurate. “Applied many interventions to a system that wasn’t in a learnable state” is more often what happened. These are not the same conclusion.
II. What Kind of Evidence Is This, Actually?
Before you can choose an intervention wisely, you need to understand what kind of claim the evidence for it is making.
There are three categories of evidence in this space, and they are not interchangeable.
Randomized controlled trial (RCT) evidence tells you that something worked, on average, for a population that isn’t you, under conditions that probably won’t match yours. That’s useful. It’s not a prescription.
To be more precise: RCT evidence is population-level by design. It tells you that across a group, an intervention produced a statistically meaningful outcome. But averages hide variance. The people in that trial had a specific distribution of biological states: some with active inflammation, some without; some with depleted reward systems, some intact. The effect size is an average across all of them. When you apply it to a specific individual, you are not applying a finding. You are making a probabilistic bet, without necessarily knowing which part of the distribution you’re in.
And the conditions matter as much as the population. RCT conditions are structured in ways that real-world application almost never replicates: specific timing, dosing protocols, therapeutic support, integration structures, exclusion criteria. The evidence is inseparable from those conditions. Strip them (which is what happens when an approach moves from a research setting into clinical practice and then into general availability) and you have a different intervention, with unknown efficacy.
This is not an argument against RCT evidence. It is an argument for holding it correctly. What a well-conducted trial reliably gives you is base-rate guidance: a reasonable prior about what’s unlikely to be useless or harmful, and what’s worth serious consideration. That’s valuable. The reasoning breaks down when you extrapolate from population average to individual prescription without assessing the biological state of the individual.
This problem has a name in clinical research methodology: the gap between efficacy (what works under controlled conditions) and effectiveness (what works in actual practice). The consistent finding is that: 1) efficacy overpredicts effectiveness, often substantially; and 2) the gap widens the more a trial population differs from the individual being treated (Singal, et al., 2014; Kraemer, et al., 2002).
Mechanistic evidence tells you that something could work, given what we understand about biology. This is where most SLM-adjacent interventions live: the approaches that are too recent, too individualized, or too contextually specific to have robust trial data, but that are well-grounded in plausible mechanisms. This kind of evidence is promising. It tells you the intervention has a coherent theory of action. It does not tell you whether the biological conditions for that action are present in your system, at this moment.
A word of caution here. SLM is itself a mechanistic argument; plausible mechanisms fail in complex systems all the time. The history of medicine is littered with interventions that were theoretically elegant, biologically coherent, and clinically ineffective. Mechanistic plausibility raises the prior. It doesn’t close the question. The additional step (does the mechanism have a reasonable chance of operating in this person’s current biological state?) is not optional.
Narrative evidence, meaning what worked for someone you know, what worked for you once before, is not useless. But it carries hidden variables that are almost impossible to untangle. What was their inflammatory state at the time? Where was their reward system? Had they just come through a period of acute stress that had paradoxically opened a plasticity window? Narrative evidence travels without its context. The result transfers. The biology that made it possible doesn’t.
Most intervention selection combines all three in ways that treat them as equivalent. A friend’s transformation story, a clinical trial headline, and a mechanistically plausible theory get weighted together into a decision that feels evidence-based. It isn’t, quite. Each piece of evidence is making a different kind of claim. Conflating them is not irrational. It’s what everyone does. But it produces sequencing errors with predictable results.
III. Biological State First
Here is what everything above has been building toward: The biological state of the system at the moment of intervention determines whether the intervention can do what it is theoretically capable of doing.
This is not a minor nuance. It is the central principle of intervention selection in the SLM framework, and it is almost entirely absent from how interventions are typically chosen.
Three biological conditions govern whether an intervention can work as intended.
Neuroplasticity availability. Is the system currently capable of forming new patterns? Chronic inflammation, the mechanism detailed in SLM 4, suppresses BDNF production, inhibits neurogenesis, and reduces the synaptic density that learning and pattern change require. An intervention aimed at creating new patterns in a state of suppressed plasticity cannot reliably do what it promises. This is not a failure of the intervention. It is a mismatch between what the intervention requires and what the system can currently provide.
Reward system responsivity. Is the dopamine signaling system in a state that can register and reinforce behavioral change? Downregulated D2 receptors (the consequence of sustained cortisol elevation detailed in SLM 5) mean that rewarding behaviors don’t feel rewarding. Positive reinforcement loops don’t close. Behavioral activation approaches (goal-setting, habit formation, structured engagement with meaningful activity) depend on a reward system capable of registering progress. When that system is depleted, the signal doesn’t arrive. The behavior doesn’t stick. The approach isn’t wrong. The substrate for reinforcement has been compromised.
Autonomic balance. Is the nervous system in a state that allows it to learn from experience? Chronic sympathetic activation keeps the system in defensive processing mode. Threat-detection circuitry is upregulated. Cortical integration is compromised. The prefrontal processing that most therapeutic modalities implicitly rely upon, the capacity to receive new information, hold it in working memory, connect it to existing patterns, and update accordingly, is running below capacity. The therapy happens. The update doesn’t install.
The relationship between autonomic state and therapeutic receptivity is well-documented: chronic sympathetic activation shifts processing away from the prefrontal integration that most talk-based and insight-based modalities require, and toward subcortical threat-response circuits that evolved for immediate action, not pattern updating (Porges, 2011; Arnsten, 2009).
Together, these three conditions constitute what we might call the system’s learnable state: the biological configuration required for interventions aimed at pattern change to work as intended.
In practice, the learnable state has observable signatures: sleep that actually restores, baseline agitation that has dropped to something manageable, the occasional return of anticipatory interest (wanting something before you have it). Stress that lands and then passes, rather than accumulating. The ability to sustain a small change without white-knuckling it. These are not the end state. They are evidence that the system has enough biological slack to begin working with.
Most interventions assume the learnable state is present. They are designed for it, tested in populations where it was at least partially intact, and prescribed without assessment of whether it exists in the person sitting across from the clinician.
When it isn’t present, the intervention isn’t wrong. The sequencing is.
See also: SLM7 Addendum—The Vagus Nerve, the Blood-Brain Barrier, and the Inflammaging Off-Switch, extending the inflammation mechanism described above with recent vagus nerve stimulation data.
IV. The Sequencing Principle
From biological state primacy, a practical principle follows:
Interventions aimed at the substrate must precede interventions aimed at the pattern.
You cannot think your way into a new biological state. This is not a philosophical claim. It reflects a structural feature of brain organization: top-down cognitive processes operate on the same neural tissue that bottom-up biological states either enable or suppress. The substrate is shared. When the biology is dysregulated, cognition about that biology cannot repair it (van der Kolk, 2015; LeDoux, 2015).
Insight, however accurate, does not reduce neuroinflammation. Understanding your patterns does not restore depleted dopamine signaling. Commitment to change does not open a plasticity window that chronic stress has closed.
This produces a sequence:
First, biological state restoration: creating the autonomic and neurochemical conditions under which the system is capable of change. Second, disruption: targeting pattern-level change within the restored window. Third, installation:encoding new constraints while the system is in a learnable state. Fourth, integration: sustaining the pattern as ordinary biological conditions resume.
Most people attempt steps two through four without step one. This is the primary mechanism of treatment resistance in otherwise capable people. Not pathology. Not insufficient motivation. Not the wrong intervention. Wrong sequence.
In practice, substrate restoration and pattern work often happen in parallel. Light behavioral structure, supportive therapy, and routine-building during the restoration phase are not contraindicated and may help stabilize the system. The principle isn’t that pattern-level work is forbidden until every gate is open. It’s that you don’t bet the therapeutic farm on pattern-level change until the biological conditions for encoding it are present. The distinction matters: scaffolding while you restore is sensible. Expecting transformation while the plasticity window is closed is the error.
There is a single question that cuts through most of the complexity:
The SLM 7 Filter: Does this intervention require learning, reinforcement, or integration? If so, it only works when the system is in a learnable state.
Apply it to the three biological gates:
Does my system currently have the plasticity to form new patterns?
Does my reward system have the responsivity to reinforce them?
Is my autonomic state one that allows integration?
If the answer to any of these is no, the intervention isn’t contraindicated; it may be valuable later. But it isn’t the first move. Something has to restore the learnable state before the pattern-level work can take hold.
V. Three Failures, Briefly
Abstract principles land differently when you recognize them.
The insight failure. Years of high-quality therapy. A good therapist, real engagement, genuine self-understanding. The work was real. The insight was accurate. The change didn’t come. Or came partially, then faded. The explanation that usually follows is that the person wasn’t ready, or the approach wasn’t deep enough, or more time was needed. The SLM explanation is different: the biological substrate for insight-driven change had been suppressed before they walked through the door. They were trying to renovate a house with the construction equipment locked in a shed they didn’t have the key to. I can explain myself perfectly, but I can’t move. The understanding was real. The plasticity required to act on it wasn’t.
The intensity failure. A powerful experience: retreat, ceremony, extreme environment, heroic dose, physical ordeal. Something genuinely happened. The disruption was real. A window opened. And then, over days or weeks, it closed. The experience becomes a reference point rather than a turning point. The explanation usually offered is that the person didn’t do enough integration afterward, which is often true, but it’s downstream of a prior problem. What was missing was the structured installation of new patterns during the window, in a biological state capable of encoding them. The experience was real. The change wasn’t. Disruption without installation is an open window in an empty room.
The behavioral activation failure. The goals made sense. The habits were reasonable. The accountability structure was solid. Early progress was encouraging. Then the system reverted. Not dramatically. Steadily. Until the new behaviors felt effortful in a way that made them unsustainable. The explanation that’s usually reached for is lack of motivation or discipline. The SLM explanation: a reward system running below the threshold required to register and reinforce behavioral progress cannot close the positive reinforcement loop that makes new behaviors self-sustaining. The plan worked on paper. It didn’t pay in dopamine. The system wasn’t broken. It was running below the threshold required to register its own progress.
These are not failures of intelligence, effort, or the interventions themselves. They are sequencing failures. Recognizing them as such is not consolation. It is the beginning of a different approach.
VI. What This Framework Does and Doesn’t Do
This framework will not tell you which specific intervention to use. That depends on individual biology, history, available resources, and clinical factors that no essay can assess.
What it does is eliminate the largest single class of reasoning errors in intervention selection. It provides a testable hypothesis for why prior interventions may have failed: the interventions were sound but they were applied at the wrong biological moment. It also provides a criterion for evaluating any proposed next step: does this address the biological state, or does it assume the biological state is already capable?
That criterion won’t choose for you.
But it will stop you from choosing incoherently.
What Comes Next
The sequencing logic raises an obvious question: if biological state restoration is the prerequisite, what reliably does it?
This is the question SLM 8 takes up.
Psychedelics are not the only answer. But they have a specific and increasingly well-documented action on each of the three biological conditions this essay has described. The evidence (and the important limits of that evidence) is the subject of SLM 8. The short version: a temporary but reliable shift in each of the three gates appears possible. That shift is a window, not a solution. What you do inside it is.
That distinction is what makes SLM 8 necessary. And what SLM 9 resolves.
References
Arnsten, Amy F. T. 2009. “Stress Signalling Pathways That Impair Prefrontal Cortex Structure and Function.” Nature Reviews Neuroscience 10 (6): 410–22.
Kraemer, Helena Chmura, G. Terence Wilson, Christopher G. Fairburn, and W. Stewart Agras. 2002. “Mediators and Moderators of Treatment Effects in Randomized Clinical Trials.” Archives of General Psychiatry 59 (10): 877.
LeDoux, Joseph E. 2016. Anxious: Using the Brain to Understand and Treat Fear and Anxiety. Penguin Books.
Porges, Stephen W. 2011. The Polyvagal Theory: Neurophysiological Foundations of Emotions, Attachment, Communication, and Self-Regulation. W. W. Norton & Company.
Singal, Amit G., Peter D. R. Higgins, and Akbar K. Waljee. 2014. “A Primer on Effectiveness and Efficacy Trials.” Clinical and Translational Gastroenterology 5 (1): e45.
van der Kolk, Bessel A. 2015. The Body Keeps the Score: Brain, Mind and Body in the Healing of Trauma. Penguin Books.
