This is today’s paper:
Vagal
nerve stimulation induces vascular oscillations and enhances long-term learning
I have a longstanding interest in what
you might call improving cognitive function or treating intellectual
disability.
Having a strong intellect does not
guarantee that you can function well in society, but a basic level of
intellectual ability is clearly a prerequisite for many aspects of independent
living.
Adaptive function is the term used to
describe all the basic skills that most of us take for granted.
My particular interest has been
whether it is possible to fast-track skill acquisition in previously slow
learners using biology and chemistry.
More broadly, many people with average
IQ struggle to retain skills that are being taught to them. Here, the problem
may lie at a later stage of learning: consolidation. Without successful
consolidation, you do not retain what you have been taught and therefore cannot
apply it later. It is not necessarily that you could not understand it in the
first place.
A learner needs to be able to:
acquire →
consolidate → retain → generalize
new skills.
In many respects, this may be more
important than IQ itself.
Acquisition — Can you learn it initially?
Consolidation — Does the new learning become biologically stabilised?
Retention — Can you still remember or perform it later?
Generalization — Can you use it in different situations?
Adaptive function — Does it help you function independently in real
life?
After all, what determines whether a
person can function independently is not simply how many points they score on
an intelligence test. It is whether they can learn to communicate, read,
calculate, cook, travel independently, use a computer, perform a job, manage
money, and acquire whatever other skills their particular life requires.
For those with classic autism, the
initial barrier of skill acquisition can often prove insurmountable.
For children who are above the modern
threshold for intellectual disability, the problem may occur at a later hurdle.
Historically, the threshold used to
define intellectual disability was much less strict. As a result, perhaps 18%
of children could fall below the threshold. Today, the definition is much
narrower, with intellectual disability affecting approximately 2.5% of the
population. That means that a large group of children who might once have been
recognised as having significant learning limitations are now expected to
succeed within mainstream education.
They may have sufficient intellectual
ability to understand and initially acquire what they are taught, but struggle
to consolidate, retain and apply it sufficiently well, ultimately falling
behind at school.
Today's post is primarily about these
people, although many of the same ideas may still be relevant to improving
adaptive function in classic autism.
The problem in schools
The scale of the educational problem
is striking.
In England, the 2026 high school
results (called GCSEs) showed that 28.2% of 16-year-olds did not achieve a pass
grade (grade 4 or above) in maths and 28.6% did not achieve grade 4 or above in
English.
If you fail you have to resit the
exam.
The same analysis reported that 84.7%
of maths resit candidates and 80.2% of English-language resit candidates again
failed to achieve a pass grade in 2026.
If someone fails an examination, there
are many possible explanations.
Perhaps they were not taught the
material effectively.
Perhaps attendance, poverty, anxiety,
motivation, language, sensory problems, attention or working memory are
involved.
But in at least some individuals,
there may also be a problem with acquiring, consolidating or retaining the
required skills.
If repeating the educational process
produces another failure in four out of five resit candidates, simply providing
more of the same may not be enough.
And the consequences extend far beyond
an examination.
Failure to acquire foundational
literacy and numeracy can restrict access to further education, apprenticeships
and employment. Young people without qualifications are at substantially
greater risk of becoming NEET — not in education, employment or training.
There are 1 million NEETs in the UK
aged 16 to 24 years old.
The ability to acquire foundational
skills is a major determinant of life opportunities.
Autism provides another way of looking at the problem
This is one reason I have become
increasingly interested in the concept of adaptive behaviour.
Someone with autism or intellectual
disability may not need to become more intelligent in some abstract sense.
They may need to become better able to
learn a particular set of useful skills.
That could mean:
- communicating wants and needs
- dressing and personal hygiene
- preparing food
- following a sequence of instructions
- travelling independently
- learning a vocational task
- using money
- learning academic skills
- coping with a new environment
- transferring a learned skill from one
situation to another.
These are adaptive skills.
And there is an important distinction
between diagnosis and learning capacity.
Monty has classic autism. Yet, with
his pharmacotherapy and sustained educational support, he was able to pass
GCSE-level maths and English, albeit at the age of 18 and on the first attempt.
That does not mean autism does not
affect learning. It very much does.
It means that a diagnostic label does
not tell us how much an individual can ultimately learn.
The useful question is:
What
is preventing this particular person from learning, and can we change it?
In Monty's case, years of teaching and
behavioural support had already been trying to give him these skills. What
changed, alongside the Polypill, was his ability to consolidate what that
support was providing. That is one individual's trajectory, not proof that the
Polypill was the active ingredient — but it is the observation that started me
asking the question this article is built around.
ABA: an application of "practice makes perfect"
One reason behavioural interventions
such as applied behaviour analysis are interesting is that they attack the
problem directly.
·
Break a
complicated behaviour into components.
·
Teach one
component.
·
Practice it
repeatedly.
·
Reinforce
successful performance.
·
Gradually
increase the difficulty.
·
Combine the
components.
·
Repeat until the
behaviour becomes reliable.
In that sense, ABA can be viewed as a
highly structured application of a very old idea:
Practice
makes perfect.
Of course, modern behavioural
intervention is considerably more sophisticated (when done correctly!!) than
simply making someone repeat something. It involves task analysis, prompting,
shaping, reinforcement, generalization and measurement.
The evidence is also more complicated
than either its strongest advocates or its strongest critics sometimes suggest.
Studies and meta-analyses report improvements in some adaptive and
developmental outcomes, but the evidence varies in quality and effects are not
uniform.
The important point here is not
whether ABA is the perfect treatment for autism.
It is that:
Intensive,
structured practice can produce measurable changes in what a person can do.
That raises a fascinating biological
question.
What if the effectiveness of intensive
practice partly depends on something happening inside the brain?
The therapy provides the enormous
amount of structured practice.
But the brain still has to change in
response to that practice.
The brain has to be plastic
Learning requires plasticity.
A new skill involves changes in neural
circuits. Some of those changes are temporary. Others become consolidated and
eventually become relatively automatic.
So perhaps there are two ways to
improve learning:
Increase
the quality and quantity of practice.
Or:
Make
the brain more receptive to the practice.
Behavioural intervention primarily
attacks the first problem.
Pharmacology attacks the second.
Neuromodulation might provide another
route.
There is already a large cognitive-enhancement industry
The idea of modifying the brain to
improve cognition is hardly new.
There is already a huge industry
devoted to it, ranging from prescription stimulants to old-fashioned
nootropics, metabolic interventions and nutritional supplements.
Prescription
stimulants
The clearest examples are
methylphenidate and the amphetamine-based drugs used to treat ADHD.
In people with ADHD, these drugs can
improve attention and several neurocognitive functions. A meta-analysis of
randomized trials found small improvements in working memory, vigilance and
response inhibition with methylphenidate.
Interestingly, the effects are not
necessarily confined to people with ADHD. In healthy adults, meta-analyses find
small effects on particular cognitive domains rather than a general increase in
intelligence.
And there is an especially relevant
finding for the argument of this article: a controlled experiment found that
methylphenidate enhanced declarative memory consolidation in healthy
volunteers.
That is much more interesting than
simply saying that stimulants make people more alert.
It suggests that at least some
pharmacological cognitive enhancement may operate at the level we are
interested in - not simply performing better during learning, but retaining
more of what was learned afterwards.
Piracetam,
vinpocetine and the nootropics
Then there are the classic nootropics
such as piracetam and vinpocetine.
Piracetam has been investigated for
decades, but the modern evidence is considerably less impressive than its
reputation might suggest. Recent systematic review evidence has not established
a clear memory benefit over placebo.
Vinpocetine is particularly
interesting because its proposed mechanisms include effects on cerebral
circulation and neuronal metabolism. But again, mechanistic plausibility is not
the same as clinical efficacy. A Cochrane review concluded that the evidence
was insufficient to establish vinpocetine as an effective treatment for
cognitive impairment.
Nevertheless, these drugs introduce
another important idea.
Blood
flow and brain metabolism
Some compounds marketed or
investigated as cognitive enhancers work partly by altering cerebral blood
flow, oxygen delivery or brain metabolism.
This does not mean that simply
increasing cerebral blood flow makes someone more intelligent. Cerebral
perfusion is tightly regulated, and more is not necessarily better.
But neurons require oxygen and
metabolic substrates, and vascular responses are closely coupled to neural
activity.
This makes the vascular finding in the
new Tohoku VNS study particularly intriguing.
The researchers found rhythmic changes
in cerebellar blood volume following VNS, and the magnitude of these
oscillations was associated with subsequent long-term learning.
Perhaps vascular and metabolic changes
are not merely passive consequences of neural activity.
Perhaps, under some circumstances, the
vascular/metabolic state of the brain helps determine how effectively learning
is consolidated.
Ketones
and omega-3s
The same logic has inspired very
different approaches.
Ketone esters are being investigated
as a way of changing the brain's fuel supply by rapidly increasing
beta-hydroxybutyrate (BHB). Omega-3 supplements, particularly EPA and DHA, are
marketed extensively for "brain health" and sometimes even for
increasing intelligence.
There are interesting biological
mechanisms and some encouraging findings in particular populations.
But neither should currently be
described as a proven way of raising IQ.
Again, the distinction matters:
changing
a biological variable → changing a cognitive measure → improving learning →
improving real-world function
are four different claims.
And that distinction is central to
what I am proposing here.
An older hypothesis: TRH and "retuning" the brain
There is another thread in my own
thinking that now looks unexpectedly relevant.
More than a decade ago, I was trying
to understand something particularly striking about Monty.
Why did certain forms of intense
sensory stimulation sometimes seem to produce such dramatic changes in his
behaviour?
Being on the upper deck of a ferry in
the wind, sitting on an open-top bus or experiencing similar forms of strong
sensory stimulation appeared to have effects that seemed disproportionate to
the stimulus itself.
I began looking for physiological
situations that might produce a similar state change.
This led me to thyrotropin-releasing
hormone (TRH).
TRH is best known for its role in the
hypothalamic-pituitary-thyroid axis, but it is also a neuropeptide with effects
on neural activity, arousal and neurotransmitter systems.
I eventually proposed what I called
the Peter Hypothesis of TRH-induced behavioural homeostasis in autism.
The idea was that abnormal TRH
signalling might contribute to some aspects of sensory and behavioural
dysregulation in some autistic individuals, and that increasing TRH activity
might potentially "retune" aspects of brain function.
This was — and remains — a hypothesis,
not an established explanation of autism.
But it led to a practical question.
Native TRH has a very short half-life
and is not particularly convenient as a therapeutic agent.
I therefore became interested in
taltirelin (Ceredist), an orally active TRH analogue developed in Japan for
spinocerebellar degeneration.
My own subsequent experience with
Ceredist was striking, but it is important to be clear about what that means
and what it does not mean.
The TRH literature provides another
reason for investigating the hypothesis: experimental studies of TRH and
TRH-like peptides have reported effects on memory and memory consolidation,
particularly in animal models.
That brings us surprisingly close to
the question we are now asking about VNS.
The gut–vagus–TRH connection
More recently, another piece of the
puzzle appeared.
I wrote about research involving
rifaximin, the gut microbiome, the vagus nerve and TRH.
Rifaximin is particularly interesting
because it acts largely within the gut rather than penetrating the brain in
significant amounts.
Yet experimental work has suggested
that rifaximin-induced changes in the gut can influence brain and peripheral
TRH/TRH-like peptide systems.
The proposed pathway was:
rifaximin → gut
microbiota → vagal signalling → brainstem/hypothalamic signalling → TRH
This was based on animal research and
remains a hypothesis rather than an established human mechanism.
But conceptually it is fascinating.
Perhaps we do not always need to
administer a brain-active molecule directly.
Perhaps we can sometimes influence the
body's own signalling system that causes the brain to change.
That gives us an interesting contrast:
Direct
pharmacological route
taltirelin/Ceredist → TRH receptors
versus
Indirect
brain-body route
gut → vagus → brain → endogenous TRH
signalling.
And now there is a third possibility.
The vagus nerve enters the learning story
This brings us to vagus nerve
stimulation (VNS).
I actually wrote about transcutaneous
VNS back in 2018, so the technology itself is not new to me.
What has changed is the question we
can now ask of it.
VNS has been used clinically for
years, particularly in epilepsy, and is also used in some forms of depression
and stroke rehabilitation.
The new question is whether VNS can be
used not merely to treat a neurological disorder, but to enhance learning
itself.
There is now preliminary human
evidence that it can influence neuroplasticity and learning.
Recent human studies of transcutaneous
VNS have reported effects on motor learning and neural plasticity, although
results are not uniformly positive.
Some studies have found improved
learning under particular stimulation conditions, while others have found no
benefit.
This is important.
It means that VNS should not be
thought of as a generic "brain booster."
The timing, location, intensity and
type of stimulation may matter.
Then came the Tohoku experiment:
Vagal
nerve stimulation induces vascular oscillations and enhances long-term learning
A recent study from Tohoku University
has made the idea considerably more interesting.
Researchers trained mice on a
cerebellar motor-learning task and then delivered VNS after the training
session.
The stimulation did not simply make
the animals perform better immediately.
Instead, the improvement emerged
later.
The researchers interpreted this as
evidence that VNS was affecting post-training processes involved in long-term
consolidation.
They also observed rhythmic changes in
blood volume around the cerebellar flocculus after VNS.
Animals with larger VNS-induced
vascular oscillations tended to show better learning several days later.
The important point is therefore not
simply:
"VNS improves learning."
It is:
VNS
delivered after learning may alter the biological environment in which learning
becomes a lasting memory.
This is still an animal experiment.
It does not establish that the same
thing happens in humans.
And the vascular oscillations are
associated with the learning effect; the experiment does not prove that they
are themselves the causal mechanism.
But it introduces a very interesting
concept:
Perhaps
the period immediately after learning is a therapeutic window.
The vagus may have several routes to memory consolidation
This is where the new finding connects
unexpectedly well with my earlier TRH work.
We now have two very different ideas
involving the vagus and memory consolidation.
The Tohoku study suggests:
electrical vagal
stimulation → vascular/metabolic changes → enhanced post-training consolidation
My earlier gut-brain hypothesis
suggests:
gut manipulation
→ vagal signalling → TRH → effects on neural state and memory
These are obviously not the same
mechanism, and the evidence supporting them is at very different stages.
But they share an important feature:
The vagus may
provide a route through which signals from outside the brain influence the
biological processes that determine whether learning becomes a lasting memory.
This makes the vagus more interesting
than simply another way of increasing alertness.
It raises the possibility that the
vagus forms part of a broader brain-body learning system, in which neural,
hormonal, metabolic and vascular signals influence what happens after a
learning experience.
And there may be several ways of
accessing that system.
We might stimulate it electrically.
We might influence the gut signals
reaching it.
We might alter downstream cholinergic
or noradrenergic systems.
Or we might influence hormones such as
TRH.
These approaches are not equivalent.
But they point towards the same broad
question:
Can we manipulate
the biological state of the learner so that practice produces a larger or more
durable change?
Which VNS are we actually talking about?
Once we start talking about vagus
nerve stimulation as a possible way of enhancing learning, another problem
immediately appears.
There are now many devices being sold
as "vagus nerve stimulators."
They are not all the same.
Some are well-characterized medical or
research devices that have been used in peer-reviewed human studies.
Others are consumer wellness products
for which the evidence is much thinner.
Some use completely different
stimulation parameters or stimulate different parts of the body.
Putting two electrodes on the ear and
calling the resulting device a "vagus nerve stimulator" does not
demonstrate that it is producing the biological effect seen in a research
study.
The question should therefore not
simply be:
Does this gadget
stimulate the vagus nerve?
It should be:
Has this
particular technology, at this particular anatomical site and with these
particular stimulation parameters, actually produced a reproducible biological
or clinical effect in controlled human research?
That is a much higher standard.
NEMOS
and research-grade auricular VNS (it fits to your ear)
![Home [B] – tVNS Technologies GmbH](https://blogger.googleusercontent.com/img/a/AVvXsEgukuv0wACgPtrifqyI2OWKWTSULNlFsbfBOMp_gJhcZU5C7SvX4DiKlmk7uLjk6ty63w7ND9Mhg1ecxEio1cXvTmhlYOezuPfIBU7gX615Qze-lU5phk3mX6DEbkpconxMUImCInzSb8yqe3u54V0DZTt3CQj10qcNYqjohAGoSjENthy-q1Et8ZZov7o)
The NEMOS/tVNS system is one of the
best-established research platforms for auricular VNS.
Studies have commonly used stimulation
at the cymba conchae, often around 25 Hz with pulse widths around 200
microseconds and intensity individually adjusted for tolerability.
The important point is not that every
NEMOS study has been positive.
It hasn't.
The important point is that the device
and stimulation approach have a substantial peer-reviewed research history.
That makes it very different from an
anonymous electrical stimulator bought online.
Parasym/Nurosym
Another interesting example is the
Parasym technology now marketed as Nurosym.
The technology has been used in
peer-reviewed human research, including randomized studies.
That doesn't mean Nurosym has been
proven to enhance learning.
It hasn't.
Nor should manufacturer-reported
numbers of clinical studies be confused with the number of independent positive
randomized trials.
But it does mean that the technology
has considerably more research provenance than a generic "vagus
nerve" gadget.
If I were interested in testing this
hypothesis experimentally, I would therefore prefer a device with a documented
research history rather than simply the cheapest device advertised as VNS.
Cervical
VNS is another category (it attaches to your neck)
Devices such as gammaCore provide
non-invasive cervical VNS rather than auricular stimulation. Round your neck
and not in your ear.
gammaCore has a substantial clinical
research base and is used medically for headache disorders.
But cervical VNS and auricular VNS
should not automatically be treated as equivalent.
A result obtained with one should not
simply be transferred to the other.
And
the cheap devices?
This is where I would be most
cautious.
A cheap electrical stimulator may
genuinely produce physiological effects.
It may even stimulate some vagal
afferents.
But unless we know:
- where the electrodes are positioned
- the current being delivered
- frequency
- pulse width
- waveform
- intensity
- duration
- and timing
we cannot assume that it reproduces
the intervention used in a research study.
Cheap does not necessarily mean
ineffective.
But:
cheap + no
parameter transparency + no independent human evidence = unknown.
I would not describe such a device as
a scientifically validated VNS intervention.
The timing problem may be even more important than the device
There is another reason not to think
of VNS as a simple wellness treatment.
The stimulation may need to be appropriately
timed.
Some human experiments have found
benefits when stimulation is paired with particular phases of learning or
movement.
Other studies have found no effect.
And some have found evidence that
poorly timed or nonspecific stimulation can actually interfere with learning.
The Tohoku experiment therefore makes
the question of timing particularly interesting.
Perhaps the optimal sequence will
eventually turn out to be:
high-quality
training → VNS → consolidation
rather than:
VNS → training
We simply don't know yet.
There may also be a pharmacological version of VNS
This brings us back to the earlier TRH
hypothesis.
Perhaps the vagal system does not
always have to be stimulated electrically.
The vagus is intimately connected with
cholinergic signalling and the so-called inflammatory reflex.
And the gut can influence the brain
partly through vagal signalling.
TRH provides another possible
downstream pathway.
This gives us several conceptually
different ways of modifying the brain-body system:
Electrical
neuromodulation
VNS →
brainstem/autonomic pathways → neuromodulation
Gut-mediated
signalling
microbiome →
vagus → brain → endogenous signalling
Pharmacological
modulation
taltirelin/Ceredist
or another drug → specific neural/hormonal pathway
These are not interchangeable.
But perhaps the future will involve
identifying which biological bottleneck is relevant in an individual and
choosing the intervention accordingly.
Exercise: another route to a plasticity-permissive brain?
There may be another route into this
broader brain-body learning system: exercise.
We have known for decades that
physical activity is good for the brain. Exercise is associated with improved
cognitive health and, in animal studies, changes in neurotrophic signalling and
hippocampal neurogenesis.
But a fascinating new line of research
suggests that we may be missing part of the story.
Perhaps contracting muscle does not
merely improve the brain indirectly through better cardiovascular health.
Perhaps muscle actively sends signals
that change the biological state of the brain.
Recent experiments in mice and cell
cultures have suggested a possible pathway involving astrocytes. Exercise was
associated with a mechanically activated, contractile state in hippocampal
astrocytes, while factors released from contracting skeletal muscle could
induce changes in astrocyte behaviour in laboratory experiments.
Exercise
engages a mechanically activated astrocyte state linking muscle activity to
hippocampal plasticity
Read the full paper on bioRxiv
This is still early science. The
mechanism has not been established in humans, and it would be premature to
claim that exercise improves learning through astrocyte contraction.
But the hypothesis is intriguing.
The possible sequence is:
muscle
contraction → peripheral signals → astrocyte state change → altered neural
environment → increased capacity for plasticity
That may be more important than the
simplistic idea that exercise merely "creates more neurons".
Exercise
after learning
There is already evidence that
exercise performed after learning can influence later memory.
This is particularly interesting
because the exercise occurs after the information has already been acquired.
In some experiments, timing appears to
matter. Exercise immediately after learning is not always optimal; exercise
performed later during the consolidation period may sometimes produce stronger
effects.
That raises a different question from
the usual one:
Does exercise make us more alert while
we are learning?
Perhaps the more interesting question
is:
Can exercise alter the biological
processes that determine whether recently acquired information becomes a
lasting memory?
One possible model is that learning
creates a temporary neural trace.
Exercise then generates a complex
physiological signal involving muscle-derived molecules, metabolism, autonomic
activity and perhaps astrocyte responses.
These signals may alter the biological
environment in which recently activated neural circuits are attempting to
consolidate.
In other words, learning provides
the information while exercise may help
provide the conditions for lasting plastic change.
A
possible parallel with VNS
This creates an interesting parallel
with vagus nerve stimulation.
The Tohoku experiment discussed
earlier suggested:
training → VNS →
altered post-training biological state → enhanced long-term learning
The exercise hypothesis might be:
training →
exercise → peripheral and brain signalling → altered plasticity/consolidation
state
These mechanisms are clearly not
identical.
VNS primarily accesses the brain
through vagal and brainstem pathways.
Exercise produces a much broader
physiological response involving muscle, metabolism, circulation, autonomic
activity and potentially muscle-to-brain signalling.
But they may share an important
conceptual feature.
Both could potentially influence the
period after learning when neural changes are being stabilized.
Practice plus biology
This brings us back to the central
question of this article.
Perhaps learning interventions have
two components.
The first is what does the
person need to practise?
The second is what biological
conditions allow the brain to respond most effectively to that practice?
Behavioural intervention, education
and rehabilitation primarily provide the first.
Pharmacology, sleep, exercise and
neuromodulation may influence the second.
The future may therefore involve
deliberately combining them.
For example:
This could
connect several apparently unrelated problems
A child struggling with literacy, an
autistic child learning a functional behaviour, a person with intellectual
disability learning independent-living skills, an adult struggling to learn a
new job and an older person with mild cognitive impairment are obviously not
experiencing the same neurological condition.
But they can all face a common
functional problem:
Can the brain
acquire, consolidate, retain and generalize the skills needed for that person's
life?
That may be a more useful target for
cognitive enhancement than simply trying to increase IQ.
The real endpoint
should be adaptive function
A treatment might improve a score on a
computerized memory test by 8%. Interesting.
But it would be far more meaningful if
it allowed someone to learn a new vocational procedure with fewer repetitions,
prepare a meal independently, retain mathematics they have repeatedly been
taught or learn strategies that help them remain independent.
We can measure this progression:
Acquisition — How much practice is required?
Consolidation — Does the new learning become stabilised?
Retention — What remains after a week or a month?
Generalization — Can the skill be used in different
situations?
Adaptive function — Does it actually make the person more
independent?
Conclusion
Perhaps the
future is not a single cognitive enhancer, but a personalized combination of:
- Pharmacotherapy — to address a
specific biological bottleneck.
- Structured
practice — to
provide the information and repetitions needed to learn.
- Exercise — potentially to
support plasticity and consolidation.
- Perhaps
neuromodulation, including VNS — to influence plasticity or post-learning
consolidation.
Different
people will have different bottlenecks, so not everyone will need the same
combination.
The goal is
not simply to increase IQ, but to help each person acquire, retain, generalize
and use the skills their life requires.
