Science-Based Mental Training & Visualization for Improved Learning

High-Level Overview

In this solo episode, Andrew Huberman explores mental training and visualization, a practice studied since Francis Galton's 1880 paper on the statistics of mental imagery and now supported by hundreds of studies showing it can speed up learning in music, mathematics, sports, dance and more. He starts with the neuroplasticity that makes it possible, including the roles of focused attention, agitation, sleep, long-term potentiation and long-term depression, then walks through classic experiments by Roger Shepard and Stephen Kosslyn showing that imagined experiences unfold at the same speed and spatial scale as real ones. From there he builds five core principles of effective mental training and a concrete protocol of short epochs, repetitions and weekly frequency.

The second half covers how much mental training to combine with real practice, what to do during injury or layoffs, timing relative to sleep, age and sex differences, first-person versus third-person visualization, the cerebellum and motor cortex, "go" and "no-go" learning in the stop signal task, and aphantasia, synesthesia and social cognition. The tone is that of a detailed, practical lecture with frequent recaps and a few lighthearted asides (including a badly hummed AC/DC riff and a detour into UFO sightings). It moves steadily from neuroscience to principles to step-by-step protocols.

The Discussion Explores:

• Developmental versus self-directed adaptive neuroplasticity, and the two-step process of focus followed by sleep
• Long-term potentiation and long-term depression, and why removing wrong movements matters as much as strengthening right ones
• Evidence that imagined scenes and movements unfold at the same speed and scale as real ones
• Five principles of effective mental training and visualization
• A practical protocol: 15-second epochs, 50 to 75 repetitions, three to five sessions a week
• Combining real and mental training, and using visualization during injury or time off
• First-person versus third-person imagery, eyes open versus closed, and watching video of yourself
• The cerebellum and motor cortex, go and no-go learning, and aphantasia, synesthesia and social cognition

Deep Dive into Main Ideas

Neuroplasticity: focus, agitation and sleep

Huberman calls neuroplasticity the holy grail of the nervous system: its ability to change in response to experience in adaptive ways. He distinguishes developmental plasticity, the largely passive change from birth to roughly age 25 (tapering gradually, not ending on a birthday), from adult or self-directed adaptive plasticity, which can begin in adolescence and continue into one's 80s, 90s or beyond, and which we can aim at whatever we want to learn. Not all plasticity is good; a concussion changes the brain in maladaptive ways.

Self-directed plasticity requires two things. First, focused, dedicated attention, which is often accompanied by agitation and frustration. That discomfort reflects the release of norepinephrine and epinephrine, and Huberman stresses it is a signal that things are going right, telling neurons that something needs to change. Second, deep rest, especially a good night's sleep on the first night after training, when the actual rewiring happens. If that first night goes badly, he says not to despair: there are second- and third-night effects too. Meditation and non-sleep deep rest also support the process.

Building up and sculpting down: LTP and LTD

The two main forms of plasticity he focuses on are long-term potentiation (LTP), the strengthening of connections summed up in the "fire together, wire together" mantra popularized by Stanford neurobiologist Carla Shatz, and long-term depression (LTD), which despite its name is often excellent for learning. LTD quiets or silences synapses, and much of motor learning is about eliminating incorrect movements. A toddler smearing food on their face with a spoon does not lack connections; they have too many neurons connected to too many others. Over years, excess connections are pruned until movements are precise.

The same is true for adults learning a tennis serve or golf swing: early attempts arc all over the place, while a mastered serve traces the same line again and again. Learning a language similarly requires suppressing sounds from your native language. Huberman acknowledges other forms of plasticity exist, such as spike-timing-dependent plasticity, but says LTP and LTD capture what matters here, because good mental training engages both the strengthening of the right connections and the weakening of inappropriate ones. He also gives away a key point early: mental training does not replace real-world practice, but it can greatly enhance the speed and stability of learning.

What happens when we imagine: principle one

Huberman invites listeners to imagine a yellow cube beside a red rose, then fly above and behind them. Brain imaging shows the visual cortex activates in similar but not identical ways to actually seeing such objects. The same holds for sound (he hums the opening of AC/DC's "Back in Black") and touch (imagining stroking a chinchilla). About 90 to 95% of people can do this, while roughly 5 to 15% find it harder, and a small subset have aphantasia, an inability to visualize.

What almost everyone struggles with is imagining long, elaborate scenes. That leads to the first principle: keep visualizations brief, on the order of 15 to 20 seconds or less, simple and sparse, and repeat them many times with high accuracy. For people with aphantasia, keeping the imagery short and simple is especially important.

Imagination runs at real-world speed: principles two and three

Roger Shepard at Stanford found that the time it takes to generate and mentally rotate an image scales with its complexity. Stephen Kosslyn, now at Harvard, had people memorize a map of an island with a loading dock, a restaurant, trees and other landmarks, then mentally travel between them. The time it took scaled linearly with the actual distance between landmarks on the map. Huberman stresses the importance: the processing speed and spatial relationships of imagined experiences match those of real experiences, so at the level of neurons, imagining behaves very much like doing.

Yet when it comes to learning, imagined and real are not equivalent, which is the second principle: mental training is an augment, not a replacement. Bistable images and impossible figures illustrate why. Looking at the classic faces-vase image, we can switch between seeing two faces and a vase but never both at once. When people try to imagine such a figure, they cannot make it flip in their mind, unless they first trace or draw it with a pen, after which they can. The combination of real motor and perceptual experience with imagination gives visualization its depth and flexibility. Hence the third principle: mental training is far more effective when paired with performing the same or very similar tasks in the real world.

Labels and eye movements: principles four and five

People manipulate the faces-vase image more easily than abstract impossible figures partly because labeling it recruits the brain's large face-processing machinery, such as the fusiform face area. Huberman touches on prosopagnosia and on so-called super recognizers, roughly half a percent of people, who can pick out faces from partial profiles in crowds and are valued by security agencies. He even applies the idea to UFO sightings: once someone labels an ambiguous object a UFO, that is the memory that gets stamped in. The fourth principle is to assign labels to what you visualize that match your real-world training. A golfer might call a specific swing "golf swing 1a in both mental and physical practice, while a putt becomes 1b." He cautions that you cannot simply lie to yourself and learn; the imagery has to accurately match reality.

Other experiments show that people move their eyes up behind closed lids when imagining a checkered ceiling, and down when imagining something on the ground, like the rattlesnake Huberman recently saw on a spring hike in California. It also takes longer to inspect a mouse's whiskers than an elephant's trunk in the mind's eye, because you have to zoom in on a small object. The fifth principle builds on this equivalence between imagery and perception: deliberately moving your eyes toward where the imagined action takes place, even with eyes closed, recruits more of the relevant neural circuitry.

The protocol: epochs, repetitions and frequency

Drawing on a systematic review titled Best practice for motor imagery: a systematic literature review on motor imagery training elements in five different disciplines, spanning education, medicine, music, psychology and sport, Huberman lays out the numbers. Use epochs of about 15 seconds. If a golf swing takes five seconds in your mind (and about the same in real life), do three swings in one 15-second epoch, rest about 15 seconds, and repeat, for a total of 50 to 75 repetitions per session. Do this three to five times per week, which has the strongest support. He advises not obsessing over exact timing.

Mental training works best for skills you can already perform correctly in the real world at least some of the time; it can raise, say, a 10% success rate substantially. For a language, choose a sentence you have said successfully but want to say more smoothly. Once a skill is consolidated and improved in the real world, you do not need to keep visualizing it to maintain it, and can move on to another skill.

How much, and what to do when you cannot train

Synthesizing several papers, Huberman concludes that real-world training beats mental training, and mental training beats no training. Hour for hour, 10 hours of physical practice beats 10 hours of mental practice, and nine hours physical plus one hour mental does not beat 10 hours physical. But adding even half an hour or an hour of mental training on top of the maximum real-world practice you can do produces significantly better speed, accuracy and consistency.

Mental training is especially valuable during injury, travel or layoffs, and studies of stroke patients, people with traumatic brain injury and those with limb or connective tissue injuries show it can maintain or even improve skills safely. He shares an anecdote from a Stanford psychology colleague about a recruited tennis player who was injured in their first year, reframed the layoff as a chance to do more mental training, and came back better than before. Total layoffs, he says, are a bad thing if you want to improve or retain skills.

Timing, sleep, age and sex

By Huberman's read of the literature, it does not seem to matter much whether mental training happens right after physical practice or on a separate day, but it should be placed so you can sleep well that night; doing it when you should be sleeping is a bad idea. He recommends aiming for sufficient quality sleep on at least 80% of nights, points listeners to his sleep episodes and toolkit and the Matthew Walker guest series, and cites Walker's 2003 paper with Robert Stickgold, Sleep and the time course of motor skill learning, on the importance of sleep after training. He invites anyone aware of peer-reviewed data on optimal timing to share it in the YouTube comments.

Early reports of sex differences in visualization were not borne out; the bulk of the literature shows no significant differences between males and females. For people 65 and older, the review suggests that combining physical and mental training may help them gain and consolidate skills better than physical training alone.

First person, third person, eyes open or closed

First-person visualization, seeing and feeling the action from the inside as if wearing a head camera, is significantly more effective than third-person visualization of yourself from across the room. For purely cognitive tasks, closing your eyes and performing them internally is naturally first person. Huberman was surprised to learn that many hallmark studies were done with eyes open, including having people watch videos of themselves. Brain imaging shows that watching yourself on video is experienced more like first person than watching someone else. So if you use a third-person approach, watch or listen to real recordings of yourself on a short loop rather than imagining yourself from outside.

For public speaking, where the goal is more about calm than exact words, he suggests one of the few exceptions: loop a short sequence such as walking to the podium and looking around the room, while deliberately keeping yourself calm, which he describes as a kind of autonomic training.

The cerebellum and motor cortex

A 2022 study titled Mental practice modulates functional connectivity between the cerebellum and the primary motor cortex had people perform a finger-tapping sequence (thumb to index, middle, ring and pinky: 1-2, 1-3, 1-4, 1-5), then either do an attention task or 50 imagined repetitions, and then test again. Huberman explains that primary motor cortex (M1) holds upper motor neurons that direct the spinal cord's lower motor neurons, while the cerebellum, a highly folded "mini brain" involved in balance, eye movements, timing and motor learning, largely inhibits motor cortex. Improving a skill tends to involve reducing that inhibition, a kind of inhibition of the brake.

Quoting the paper, "we found that mental practice enhanced both the speed and accuracy" of the real-world sequence after just 50 imagined trials. Using transcranial magnetic stimulation, researchers found mental training increased net excitation in cerebellum-to-motor cortex communication, rather than changing the direct motor pathways to the spinal cord. As always, participants could already perform the sequence in the real world.

Go, no-go and the stop signal task

Skilled performance involves not only doing the right actions but withholding wrong ones, and Huberman estimates that half to 75% of motor learning is about restricting inappropriate movements, utterances or thoughts. The basal ganglia handle this go versus no-go balance. He describes the stop signal task developed by Gordon Logan of Vanderbilt and William Cowan: press left or right to match an arrow within about 500 milliseconds, but occasionally a red stop signal appears 100 to 350 milliseconds after the arrow, and you must withhold any key press. Short delays are easy to stop; longer ones are very hard.

In a study titled Motor imagery combined with physical training improves response inhibition in the stop signal task, participants trained once a day for five days with about 30 practice trials, then were tested on about 144 go trials and 48 stop trials. Both physical and mental training alone improved performance significantly, but the combination outperformed either, in some cases nearly doubling the improvement. That is an exception to the rule that physical practice always wins hour for hour: if your problem is doing the wrong thing at the wrong time, combining mental and physical training is especially valuable, and practice should mix go and no-go trials. Much of this improvement likely reflects long-term depression of specific connections.

Aphantasia, synesthesia and social learning

People with aphantasia range from having no mental imagery to having weak imagery. Huberman describes two friends with synesthesia, one of whom links piano keys to specific colors, such as E flat with an amber hue, without being a better pianist for it. Research, including a paper titled What is the relationship between aphantasia, synesthesia and autism? and a review titled Aphantasia: the science of visual imagery extremes, shows that aphantasia is linked to weak visual imagery but can co-occur with synesthesia. People with aphantasia also tend to show more features associated with the autism spectrum, though cause and effect are unclear, not everyone on the spectrum has aphantasia, and the categories themselves are under revision. He promises a future episode with expert guests.

He closes the loop by noting that social cognition, learning what is appropriate in a given context, from smiling to staying silent to sitting still, is as much a learned pattern as finger tapping, and has direct parallels to visualization and go and no-go learning. He also notes meta-plasticity: the more you practice visualization, the better you get at it.

Conclusion

Huberman's message is that mental training and visualization is a reliable, science-backed way to learn faster, as long as it is done correctly. Keep visualizations brief and simple, repeat them 50 to 75 times per session in roughly 15-second epochs three to five times a week, label them the same way you label real practice, use first-person imagery and even your eye movements, and above all pair them with real-world practice of skills you can already perform. Mental practice will not replace real training, but it adds significantly to it, keeps skills alive during injury or time away, and is especially powerful for learning what not to do. Throughout, focus during practice and sleep afterward remain the foundation of lasting change, and he encourages people not to obsess over precision, since nearly every study finds consistent practice improves real-world performance.

This is a summary of "Science-Based Mental Training & Visualization for Improved Learning" from Huberman Lab, giving you the GIST so you can decide if you want the REST. Listen to the full episode.