Posted: 2026-02-06

Non-Dopaminergic Antipsychotics

I have taken my dopaminergic antipsychotics continuously since 2006. I like them, they work for me, and I have no intention of stopping. The dosage is in my checkin for every essay. When feasible, I often get custom doses compounded and have dialed it in with sub-milligram accuracy at times in the past.

Reminder

I'm not a psychiatrist and I have absolutely no qualifications in neuroscience or medicine.

This article is regarding speculative trends in research for pharmacologically treating psychosis. It's intended to be food for thought first and foremost, taken as informed speculation.

"It was a brilliant cure, but we lost the patient"

The usage of dopamine antagonists is, in far too many cases, driven first and foremost by risk profile. The fear of catastrophic, on-the-news failures of treatment often forces psychiatrists to be extremely aggressive when it comes to treating psychosis — which ironically creates a much larger danger: disengagement.

In cases where excessively aggressive and rigid treatment processes are structurally harming your life for the benefit of a profession's risk tolerance, without the domain knowledge of fully understanding where it will lead, disengagement can become a rational thing to do, even if it's ultimately not in your interests. Situations where this is the case benefit nobody, so if psychiatry genuinely intends to do no harm, we should configure the field so that these scenarios are avoided.

What we have is often more of a system design failure than a compliance and clinical management failure. We optimise for relapse prevention, but often fail to account for personhood erosion.

Luckily, a new paradigm is beginning to emerge — one which doesn't necessarily involve dopamine antagonism at all.

Nuance and Subtlety

"Subtle" does not mean "small." It means "non-obvious."

What all of these new treatments have in common is that they do not function like "take this pill = less psychosis". They each change different aspects of psychosis, and creating a personalised cocktail which eliminates the most problematic psychosis symptoms while retaining enough of the person that it's a net win will not be trivial — but it will reflect the reality of the conditions much better.

We will be assembling a collection of puzzle pieces which only work when assembled together. The question is not neuroleptic potency in the sense that, say, clozapine is "more potent" than quetiapine. The question is which problem we're solving.

Mechanisms

Apart from dopamine, there are a few other psychosis-related levers known to medicine so far; four stand out (only one is approved to treat psychosis, and even then in a narrow sense). None of these are readily available in Australia, so the risk of self-experimentation (bad idea) is very low.

A hard guarantee that these drugs will help us keep more of what we want while still eliminating what we don't cannot be made. This is early stage and investigational; a lot is still unknown.

5HT2A antagonists

Validating the legitimacy of this one is extremely straightforward: it's the main site which makes atypical antipsychotics "atypical".

It's a bit cute, but we can describe this action as "anti-LSD" and be mostly right (LSD the other classical psychedelics primarily hit this site but in the opposite direction). 5HT2a is the absolutely prototypical site associated with visual hallucinations in particular, though activating it certainly has profound cognitive effects too.

5HT2a antagonists as a component have a long and well-studied history of being used to treat psychosis, but the effectiveness of targeting the site on its own has not.

Azacyclonol

This drug was developed to treat perceptual disturbances in schizophrenia in the 1950s, but was abandoned (among other reasons) because it does not possess full-spectrum antipsychotic properties on its own — which, the whole point of this topic is that we wouldn't expect it to. Because its patent is long since deceased, there is little incentive to commercialise it in the modern day, even if it turned out to be perfectly serviceable for this purpose.

Pimavanserin

Pimavanserin, a selective inverse agonist of 5HT2a, was recently approved as a treatment for psychosis secondary to Parkinson's Disease — a group which experiences drastic increases in mortality from the use of dopamine antagonists.

Despite the more targeted and selective nature of pimavanserin, there is some evidence that its profile of serious side effects might be unfavourable. Making a case for taking that risk simply to access a non-dopaminergic antipsychotic is unlikely to pass muster.

M1/M4 agonists

These are also a pretty easy sell: this site is likely a part of the secret sauce which often makes clozapine and olanzapine more effective than other atypicals.

These drugs appear to work by altering dopamine activity upstream; rather than simply blockading the receptors globally, they modulate circuits that regulate dopamine firing patterns.

The difference between M1 and M4 activation is important, but out of scope for what we're talking about.

Cobenfy (xanomeline-trospium)

This is a real drug which is actively being clinically applied (in the US) for the treatment of psychosis, but its side effect profile is non-trivial, and its effectiveness and safety profile is still being investigated.

Apparently it's considered well-tolerated by some people on the ground, but it's early days yet. One person I spoke to heard that it's an "all-natural antipsychotic" which was kind of funny to hear said about a cutting-edge (and clever!) dual-molecule piece of pharmacological engineering.

This one's particularly exciting, because it's actively being marketed as a viable antipsychotic — with the first new primary mechanism in many decades. It may potentially improve psychosis-related cognitive deficits while still being a viable alternative for treating hallucinations and delusions.

KOR antagonists

When people say "opioid", they are nearly always specifically talking about mu-opioid agonists — things like morphine, fentanyl, codeine, oxycodone etc. There are in fact five opioid receptor sites: mu, delta, kappa, zeta and nociceptin.

Kappa is a really interesting one.

When you activate it, it has roughly equal painkilling efficacy to mu agonists, but there are two fundamental differences:

• It does not produce tolerance or addiction
• It makes you feel terrible and often also hallucinate

For this reason, this class of drug never really got all that much airplay in medicine. Many people are surprised to learn that the profound effects of salvia divinorum have nothing to do with traditional psychedelics, and are in fact the result of it intensely activating the kappa opioid site.

What's actually interesting is that we can tickle the same receptor in reverse.

This is speculation: kappa opioids might provide an emotional integration effect. This is potentially an effect profile which can support equanimity and resilience under stress, which is one of the central entry points into psychosis.

They may not treat psychosis as well, but as the part of this system which prevents destabilisation in the first place, they might be helpful.

Kappa opioid antagonists have not been shown to directly improve any aspect of psychosis; their presence here is speculative.

Aticaprant

Aticaprant is a kappa-antagonist which was assessed for the possibility of treating depression. The development process was ultimately discontinued due to a failure to demonstrate compelling effectiveness, which, as aforementioned, doesn't definitively preclude it from combination therapy. The IP holders have stated they intend to explore other uses for the drug.

A7 Nicotinic Acetylcholine PAMs

Schizophrenics endlessly chain-smoking is a well-established phenomenon. In the interests of learning why, I briefly flirted with the vice (and did not end up addicted).

What I experienced was integration: this is a very difficult effect to explain in words, but my experience of sensory, memory and cognitive coherence rapidly and drastically changed. I feel like it wouldn't be unreasonable to posit that people experiencing chronic psychosis sometimes find that nicotine brings the world back together and making sense again for a few minutes.

Apparently I am not the only one. There has been academic interest in replicating this effect in a safer form, and a few candidate compounds have been identified for the purpose. Nicotine hits a bunch of receptor sites, but the one we care about here is Alpha-7 nicotinic acetylcholine. This is actually part of the addiction mechanism of cigarettes/vapes: its effect on long-term memory can more strongly encode the feeling of relief when a cigarette is consumed.

Of course I would never recommend self-medicating with nicotine: even if it doesn't work, so many as 2-3 cigarettes can create a beast of an addiction.

A7 PAMs, as a group, are not known to reduce delusions or hallucinations. They are in this list for the cognitive negative symptoms.

PNU-120,596

Absolutely a known and highly potent/selective a7 PAM. I haven't been able to find much on its activity in humans, other than that despite being a mainstay of research it was not found to be a clinically viable drug candidate.

Other Emerging Mechanisms

There are a bunch of other potential target sites for a drug of this nature.

I haven't dug into these significantly yet, but this is my hit list:

• TAAR1 — modulates dopaminergic tone without directly antagonising and quite promising — several drugs in development
• mGluR2/3
• NMDA
• Sigma-1 — more on those in another essay
• GABA-B

I've had more than enough people instantly assume that I'm working on something cannabinoid-related that I am giving those a wide berth. I have had that conversation enough times already, and "no" is the shortest possible answer which avoids that derail.

Interesting Phytochemicals

I've been investigating this at length, and have found some interesting candidates for phytochemicals which might be worth researching to create derivatives of. In the interests of responsible disclosure, I'm not going to share this list, because there is a large subset of the population which are extremely motivated to misread this as "stop taking your meds and swap them out for these (ineffective) herbal remedies instead."

There are some interesting leads here, but the herbs themselves won't help, and the discourse is too keen to distort and misrepresent this type of discussion.

Evidence Bases And Compliance

We have extensive evidence on the effectiveness of dopaminergic antipsychotics.

But we have to ask: effective at what...?

A very large source of data used to determine antipsychotic effectiveness comes from acute inpatient units. It naturally follows that this data is constrained by the definition of effectiveness that an acute inpatient unit would use: measuring compliance and risk of catastrophic deterioration. Most of the formal metrics which are not these are either directly or indirectly connected to them; these two factors are the boundary conditions which drive most other metrics.

One of the troubles with the above approach is that it is unlikely to provide that form of compliance: it is very easy to imagine a scenario where someone taking their non-dopaminergic cocktail may have their thinking clear and delusions/hallucinations taken care of, but then (often quite rightly) express hostility towards the staff treating them, which is likely to then be taken as risk and lack of insight.

It is not clear that the existing paradigms for measuring efficacy will even able to properly measure the effectiveness of a non-dopaminergic combination therapy. The body of evidence we have for these drugs may not be optimised for the correct outcomes.

Psychoinformatics For Longitudinal Evidence

Almost all of the data on drug effectiveness comes from presentation snapshots; these are assessments done at a point in time and repeated over a relatively short (weeks to months) period. The long-term trajectory of a mental health condition is often measured in years at least, so we should not be surprised when that data means decisions are made based on false premises.

There is a marvellous source of data at all of our fingertips: our online interactions. It is an incredibly rich source of information covering our mood, cognition and relationships, but it is also intensely personal.

Digital phenotyping is a concept which has seen some research airplay — tracking a person's digital interactions over time — but it is fatally flawed by the fact that other people are the ones who have access to it. Someone who knows they're being spied on will almost always change their behaviour (deliberately or otherwise) because of that fact — and they'll likely be paying a lot of attention to when the study ends.

Personal psychoinformatics only works if the system is extremely careful about what it collects and stores, and the person can fearlessly express themselves because they own their own data (and can choose to give it out or not), and when we're taking on a treatment paradigm which depends on subtle but high-significance nudges, that is the only way we can realistically gather this data continuously over a long timescale.

The Vision

Whether or not these four, or some other permutation of non-dopaminergic drugs, could come together to become greater than the sum of their parts has not been proven yet. We can also think about combining lower doses of dopaminergics with the others.

This understanding of treatment undoes the historical collapsing of a range of complex pathologies into a single clinical entity. It might help us see psychosis not just as many diseases, but in fact as many states a person's brain can be in, each of which need subtly different handling.

This whole concept is not one of a crazy new treatment; it's an entire paradigm which is in the very early stages of emerging. A lot of questions still need answering, and still more even need to be asked. But there are glimmers of something very exciting.

It is going to be extremely challenging for some psychiatrists, because they are going to have to rely on the patient's experience a lot more.

It seems that the picture we're piecing together here is one which pushes beyond the current model of psychosis treatment. We're moving from a vision of cognitive and emotional coercion — where large chunks of a person's core self are forcibly shut down "for their own good" — to one which can only work if it is deeply nuanced, personalised and respectful of autonomy. Even if this concept is not where true recovery lives, I believe the answer will look a lot more like it.

🔗 Checkin

Version: 1

Written: 2026-02-06

Written on: 7.5mg olanzapine since 2025-11-11

Mental health was: very poor - estimate 15% brain