What published research reports on Semax, cognition and attention

Published August 26, 2026 · Artemis Labs

Semax — answer capsule: Semax is a seven-amino-acid peptide (PubChem CID 9811102) studied since the 1990s, almost entirely by Russian groups working with a Russian intranasal pharmaceutical. The human work closest to cognition is small and indirect: two Moscow imaging studies in healthy volunteers measured brain-network activity rather than test scores, and one Russian follow-up report on patients recovering from oxygen deprivation described an effect on memory symptoms with no control group. Rodent studies report memory-protective and behavioural effects in particular laboratory models, and in one the effect appeared in only one of the two mouse strains tested. No study conducted outside Russia has given Semax to people, PubMed indexes no double-blind Semax study of any kind, and no group has independently repeated the Russian results. The published record contains no test of whether Semax improves thinking in a healthy person.

Key findings

• The human studies closest to this question measured brain scans, not thinking. In 24 healthy volunteers at the Research Center of Mental Health in Moscow, Russia, resting-state fMRI showed “A greater volume of the default mode network rostral (medial frontal cortex) subcomponent … in the Semax group in comparison with controls” (PMID 30225715).
• The attention and ADHD association traces to a 2007 paper in Medical Hypotheses from Taipei Veterans General Hospital in Taiwan that contains no data: “it is proposed in this paper that Semax may have good therapeutic potential in ADHD” (PMID 16996699).
• Rodent behaviour results are tied to the animal used. One mouse study found the compounds it tested, Semax among them, “clearly improved the exploratory behavior in BALB/c mice only” — not in the C57BL/6 line tested alongside (PMID 23025044).
• The growth-factor mechanism usually offered as the explanation is not a clean increase. In rats the authors’ own word was “multidirectional”: expression “decreased in rat hippocampus and retina … and … increased in the frontal cortex” (PMID 19662538).

What did the human cognition studies actually measure?

Two of them measured brain images in healthy people, and both came from the same Moscow institution. The first, published in 2018 by the Research Center of Mental Health in Moscow, Russia, scanned 24 healthy volunteers, fourteen given the Russian intranasal preparation and ten given placebo. The reported result was a difference in the size of one part of the default mode network, a set of brain regions that stays active when a person is at rest rather than working on a task (PMID 30225715). Nobody in that study was asked to remember, calculate or concentrate on anything. A difference on a scan is not a difference in performance.

The second, from the Mental Health Research Center in Moscow, Russia in 2020, used the same approach in 52 healthy participants with a placebo arm, comparing Semax against Selank. It reported differences in how strongly the right amygdala connected to nearby temporal-lobe regions. The authors label that analysis “Post hoc” (PMID 32342318), meaning the comparison was chosen after the data were in hand. Again the endpoint is connectivity, not behaviour.

Only one human record touches memory directly. A 1999 report in a Russian-language Russian journal followed 73 patients with lasting brain injury after a period of low oxygen; the record lists no institution, so its country comes from its journal and language. The finding was that “The therapy was effective in patients with mnestic disorders” — mnestic meaning memory-related — with two other treatments given alongside and no control group described (PMID 10199046).

Why does the focus and attention claim keep appearing?

Two sources feed it, and neither is a study of attention. The first is the Russian literature’s own vocabulary. The group that developed the compound published a fifteen-year programme overview titled “[A nootropic adrenocorticotropin analog 4-10-semax (l5 years experience in its design and study)]” (PMID 9173745). That word is a category label the originating researchers applied to their own compound in a retrospective account of their own programme. It is not a measured result.

The second is the 2007 hypothesis paper from Taiwan. Medical Hypotheses publishes proposals rather than experiments, and this one says as much in its own sentence. It is the origin of most “Semax for ADHD” writing online, and in the two decades since, no clinical study has tested the proposal. When a claim’s only citation is the paper that proposed it, the honest description is that the idea exists, not that it was checked.

What have rodent studies reported about memory and behaviour?

Rodent work holds most of the behavioural record, bounded to particular laboratory models. A 2003 mouse study reported that a set of compounds including Semax “fully or completely prevent from the amnesia development” in its model (PMID 12924240). Semax was one of six agents grouped in that sentence, so the paper does not separate it from the others.

Nulls sit inside the positive papers too. A 2024 rat study at the National Research Center “Kurchatov Institute” in Moscow reported that treatment reversed several effects of chronic stress, then records in the same abstract that “In the forced swim test, no effects of the CUS procedure or peptides on the duration of rat immobility were detected” (PMID 39442746).

How solid is the proposed mechanism?

The explanation usually given runs through brain-derived neurotrophic factor, a protein involved in how neurons grow and connect. The rat data do not support a simple version of it. The 2009 study from the Institute of Molecular Genetics in Moscow found neurotrophin gene expression fell in the hippocampus and retina at twenty minutes and rose in the frontal cortex, and the authors chose the word “multidirectional” (PMID 19662538). Direction depends on region and time point.

The dopamine story, which is the neurochemistry people associate with attention, is weaker still. A 2005 rodent study at a Moscow pharmacology institute reported that “This peptide alone failed to alter the tissue and extracellular concentrations of dopamine and its metabolites”, while noting that given before amphetamine it “dramatically enhanced the effects of the latter” (PMID 16362768). On its own, no measured dopamine change.

One finding matters for anyone thinking about a healthy brain rather than an injured one. In 2024 the originating Russian group examined rat brain cells under normal conditions and found the peptides tested “predominantly caused decrease in expression of the genes associated with the immune system”. Their written conclusion: “it is necessary to study influence of the peptide on the brain cells under normal physiological conditions, including understanding the risks of their use” (PMID 39418522). The researchers closest to this compound say the healthy-brain case is the one not yet worked out. Background on the wider class sits in our neuropeptide research section, and specification data on the Semax product record.

What the research does not show

No Semax study in humans has ever been conducted outside Russia. Every human record we verified is from a Russian institution or a Russian-language Russian journal. The only non-Russian primary research is laboratory work with no people in it: two Italian in-vitro studies and one Indonesian rat study. Nobody outside Russia has repeated any of these results, in either direction.

PubMed indexes no double-blind Semax study at all. The two imaging studies had placebo arms, but neither record states that participants or assessors were blinded, and there are no meta-analyses and no systematic reviews. The single record PubMed tags as a randomised controlled trial calls itself open-label in its own abstract, and its primary result was negative: “Semax (1% solution) does not influence either the course of CPD or the dynamics of clinical estimates” (PMID 18379501). That trial studied motor neuron disease rather than cognition, but it is the strongest design in the record and it missed its primary endpoint.

No study has measured cognitive performance in healthy people. Not memory recall, not sustained attention, not reaction time, not learning rate. The healthy-volunteer work measured scans.

There is one human safety observation, and it sits in the memory report. The 1999 Russian paper noted: “In some cases EEG showed episodes of paroxysmal activity after semax injection” (PMID 10199046). That is the authors’ own caution, and a page about cognitive research should not omit the one human signal inside it.

Reviews from outside Russia say the same thing. A 2026 US review states that “there is a current lack of clinical trials” (PMID 41490200). A second 2026 review, which searched regulatory databases through January 2026, places Semax among non-approved peptides that “lack long-term safety data and systematic validation” (PMID 42021992). One rodent study also described Semax’s dose-effect relationships as “bell-shaped” — the relationship reverses past a point (PMID 20486550).

Frequently asked questions

Has Semax been tested for memory or attention in healthy people?

No. Both healthy-volunteer studies, from Moscow institutions, used resting-state brain imaging as their endpoint, and neither administered a cognitive test. For the fuller human record, see the Semax human trials overview.

Where does the ADHD connection come from?

From a 2007 paper in Medical Hypotheses, a journal that publishes proposals rather than experiments. The paper contains no data and describes itself as proposing an idea. No study has since tested it.

Is the BDNF explanation established?

Not as usually stated. The rat study most often cited for it reports change in both directions depending on brain region and time point, and the authors call the pattern multidirectional.

Why is all the human research Russian?

Semax was developed in Russia and studied there as a domestic pharmaceutical. Russian authors describe it as a registered domestic drug, though we could not confirm that against a registry record. What it has never had is a research programme anywhere else. Our overview of what Semax is studied for covers the other research areas.

References

1. PMID 30225715 — Effects of Semax on the Default Mode Network of the Brain. Bull Exp Biol Med, 2018. doi:10.1007/s10517-018-4234-3
2. PMID 32342318 — Functional Connectomic Approach to Studying Selank and Semax Effects. Dokl Biol Sci, 2020. doi:10.1134/S001249662001007X
3. PMID 10199046 — [Use of semax at a follow-up of patients with posthypoxic encephalopathy]. Anesteziol Reanimatol, 1999. No DOI in the PubMed record.
4. PMID 9173745 — [A nootropic adrenocorticotropin analog 4-10-semax (l5 years experience in its design and study)]. Zh Vyssh Nerv Deiat Im I P Pavlova, 1997. No DOI in the PubMed record.
5. PMID 16996699 — Med Hypotheses, 2007; Taipei Veterans General Hospital, Taiwan. A hypothesis paper containing no data. doi:10.1016/j.mehy.2006.07.017
6. PMID 12924240 — mouse amnesia model, 2003. Title not carried in our verification record; cited by PMID. No DOI in the PubMed record.
7. PMID 23025044 — mouse exploratory-behaviour study in BALB/c and C57BL/6 strains, 2012. Title not carried in our verification record; cited by PMID. No DOI in the PubMed record.
8. PMID 39442746 — Antidepressant-like and antistress effects of the ACTH(4-10) synthetic analogs Semax and Melanotan II on male rats in a model of chronic unpredictable stress. Eur J Pharmacol, 2024. doi:10.1016/j.ejphar.2024.177068
9. PMID 19662538 — Comparison of the temporary dynamics of NGF and BDNF gene expression in rat hippocampus, frontal cortex, and retina under Semax action. J Mol Neurosci, 2010. doi:10.1007/s12031-009-9270-z
10. PMID 16362768 — rodent dopamine study, Neurochem Res, 2005. Title not carried in our verification record; cited by PMID. doi:10.1007/s11064-005-8826-8
11. PMID 39418522 — Changes of Transcriptomic Activity in Rat Brain Cells under the Influence of Synthetic Adrenocorticotropic Hormone-Like Peptides. Biochemistry (Mosc), 2024. doi:10.1134/S0006297924090104
12. PMID 18379501 — [The study of chronic partial denervation and quality of life in patients with motor neuron disease treated with semax]. Zh Nevrol Psikhiatr Im S S Korsakova, 2007. No DOI in the PubMed record.
13. PMID 20486550 — rodent hypoxia-model comparison, 2010. Title not carried in our verification record; cited by PMID. No DOI in the PubMed record.
14. PMID 41490200 — 2026 review, US authors. doi:10.5435/JAAOSGlobal-D-25-00236
15. PMID 42021992 — 2026 review; searched FDA and WADA databases through January 2026. doi:10.3389/fragi.2026.1790247

Methodology: this page draws only on PubMed records retrieved and transcribed verbatim in our own primary-source verification pass, plus PubChem’s record for CID 9811102. Every quoted string above is reproduced exactly as the source wrote it. Last verified August 26, 2026.

All compounds sold by Artemis Labs are for laboratory research use only. Nothing on this page is medical advice, and no statement has been evaluated by the FDA.