How Dopamine & Serotonin Shape Decisions, Motivation & Learning | Dr. Read Montague

If any single win completely satisfied you, the drive would stop. Brains are built to keep tracking the next step, or you wouldn’t keep going.

Welcome to Huberman Lab, where we turn science into practical tools for daily life. Today I’m joined by Dr. Reed Montague to unpack how dopamine truly works across learning, motivation, and persistence, how it dances with serotonin, what SSRIs might be doing under the hood, and how ideas from AI help explain our behavior.

When someone asks you what dopamine does, how do you answer?

It’s not just pleasure. It’s a core learning signal that shapes motivation and can influence how you feel, though linking the mechanics to feeling is still tricky.

Can you give examples of dopamine’s role in learning?

Across animals, dopamine rises and falls as they learn, and different brain sites carry distinct signals. The key rule is temporal difference learning, where your brain updates from one prediction to the next long before the final outcome, a logic also used in modern reinforcement learning.

In real life you rarely get immediate feedback, like dating where expectations keep shifting. Does dopamine encode those rolling expectations even without a clear endpoint?

Exactly. That algorithm chains predictions across time better than old expectation-versus-outcome rules and matches dopamine data from animals to humans, while also powering AI systems that learned to beat world champions.

How do those ongoing updates translate into motivation?

Fast prediction errors track moment-to-moment expectations, and a slower envelope relates to how energized you feel about acting.

Are products like social media built to keep us foraging by constantly updating expectations?

Continuous goals keep living systems moving. The phrase dopamine hit is a blunt shorthand; most of the learning comes from rolling expectations punctuated by the occasional surprise.

If a shared experience is spectacular, does that higher dopamine state bias future expectations, and how does that contrast with low dopamine conditions?

A raised baseline shifts how new signals are interpreted, while a lowered baseline dampens learning from fluctuations. In Parkinson’s, massive dopamine loss flattens value signals and behavior can freeze because everything feels equally unimportant.

I think of motivation as urgency, including how thoughts move. Does dopamine steer thought as well as action?

Neuromodulators stabilize brain states and sequences, so they guide chains of thought and the value of options. Addiction can be seen as that learning machinery hijacked by an unanticipated chemical signal.

Do ADHD medications that raise dopamine narrow focus or just make more things sticky?

They likely stabilize and tighten thought sequences. Bees show a similar explore-versus-exploit split set by amine ratios, and humans carry both modes in different proportions.

Could short-form media overtrain the exploratory mode and erode long-horizon pursuit?

In artificial systems you must balance exploration and exploitation or learning skews. Some real contexts demand rapid switching, but people need training for that, and managing kids’ screens is a constant, imperfect negotiation.

I lock my phone away to protect focus; can resisting become rewarding, the way anorexia seems to flip reward onto restraint?

Yes, resistance can feel rewarding, and in pathology that drive becomes dangerous.

Given your athletic background, did embracing discomfort feel strengthening?

I loved grinding through hard training because it built calm under pressure and personal standards. My kids’ school now bans phones during the day while figuring out how to handle AI in the classroom.

Let’s bring in serotonin. How does it interact with dopamine?

Dopamine tracks positive learning and drive, while serotonin signals waiting and negative outcomes, and they often move in opposite directions in humans. SSRIs leave more serotonin around and some of it enters dopamine terminals, which can blunt reward learning in some contexts.

SSRIs help many, but can they also dampen reward by shifting how events are coded?

Blocking reuptake keeps serotonin in play longer, and some rides dopamine transporters into those terminals, so the effect depends on where it lands and how downstream circuits read the signal.

What does hunger do to these dynamics?

In rodents, deep hunger flips dopamine to encode aversive prediction errors, and stress biases judgments in people, so harsh feedback drives overgeneralization while gentle guidance teaches better.

When stress gets extreme, it can flip the dopamine system so it stops chasing good outcomes and starts working to stop things from getting worse, which is exactly why mild pain paired with a threat of worse pain can coerce people more effectively than outright brutality.

Total harm shuts learning down, but small doses of threat make relief feel rewarding, and you can see versions of that dynamic in harsh families and other everyday settings.

On the other side, when dopamine runs too hot, like with heavy cocaine use, judgment skews and everything feels like a green light, so what does chronic overindulgence do to reward?

It raises the bar so high that normal life rarely clears it, and almost nothing natural feels satisfying.

I once rescued a dog that had been abused so badly it struck first and trusted nothing, which is what happens when you push a nervous system past the edge and it reorganizes around pure defense.

We all know people who got bent that way, and it’s heartbreaking.

As we learn to intervene better, I hope we can help people caught in those loops, because drugs of abuse push many into the same trap even after they know where it leads.

So far we’re saying dopamine carries graded positive expectation while serotonin flags negatives, and under relentless hardship dopamine can reset to fight for bare safety instead of growth.

In survival states, anticipating bad outcomes gets reinforced so you learn about threats, and that shift keeps you alive even if it feels awful.

I’ve seen someone leave chaotic relationships and only after years of calm did their system stop bracing for damage and start flourishing, which makes me wonder what serotonin is doing during the worst periods.

In humans, serotonin signals generally move opposite to dopamine during these probes, while rodent results are scattered because the behaviors are hard to model.

I’m blown away by the idea that SSRIs raise serotonin and a lot of that spills into dopamine terminals, dulling the charm of good things.

A key study showed serotonin loads into dopamine circuitry when reuptake is blocked, so the positive channel carries a negative flavor, which could blunt learning about rewards or even make some negatives feel attractive.

Great work needs champions, and with that spirit, tell us about recording dopamine and serotonin through the nose because those experiments are wild.

In deep brain stimulation surgeries we piggyback microelectrodes and decode fast neurochemistry, and, thanks to Christina Zelano, we can also slide a tiny tube to the olfactory epithelium in healthy volunteers and track dopamine, serotonin, norepinephrine, pH, and peroxide during odors, meditation, breathing, and decisions.

Do nose signals mirror classic midbrain patterns, with dopamine rising for positive expectation and serotonin rising for negative expectation?

They do, and the nasal readouts look like what we expect from midbrain neurons to cues and outcomes.

It’s labeled minimally invasive, though I aged out of being a participant the week I turned sixty-six.

I sleep in two chunks, often dozing around eight or nine and waking near midnight, then I’m up at about three-thirty or four because I need deep quiet to think.

That first stint likely covers your deep sleep and hormone surge, and the second gives you enough REM to keep your mood steady.

Life taught me not to over-index on other people’s opinions, and science taught me to hold my ground because at the edge you get tested hard and often.

Funding panels force you to kill most proposals, the job is two jobs just to do the work, and still the grind trains a daily loop of small goals and long arcs.

Kids need reps at losing and recovering, which is why sports matter, and the same lesson plays out when students go from helpless to indispensable and then it’s time for them to fly.

Hard training leaves you too tired to get in trouble, wrestling teaches panic control, and our ancient wiring still needs tests modern life rarely provides.

Skateboarding taught me pain, failure, and the patience to keep going, which maps perfectly onto lab work and expectation management.

What happens to dopamine and serotonin when we shift inward during meditation and controlled breathing?

With deep electrodes and nasal probes, we see dopamine and norepinephrine ride the inhale–exhale rhythm during natural breathing, and when breathing is paced by instruction the signals wobble because cognitive control gets involved.

In an ultimatum game, breathing and peroxide track metabolic demand right when the model needs updating, and dopamine seems to gate the learning moment.

I think of dopamine as a universal currency for motivation across music, sport, work, and even online conflict, and part of resilience is not letting others drain your bankroll.

It’s really a coordinated system and dopamine even nudges mitochondria to make ATP, yet we still lack a clean bridge between the brain’s computations and its striking energy efficiency.

How do sleep and meditation restore motivation when two sleepless days can flatten it completely?

Sleep gives the system time to erase, consolidate, recycle, and reset homeostasis, and meditation likely mixes physiological shifts with housekeeping in the algorithms.

If we push dopamine up with drugs, does time feel faster or slower?

Timing is embedded in dopamine’s learning machinery and there are many clocks, and while drugs can warp time sense there isn’t one simple direction for all cases.

I’m lousy with time and spatial cues, stereo blind, and I relied on motion to compete, which shows how brains route around constraints.

I grew up with hours of unstructured wandering, but kids today swim in nonstop input, and while we worry, we don’t yet know the fix even as we lean on tools like Claude that can synthesize across fields.

Reinforcement learning breakthroughs like AlphaGo and AlphaFold show game-like training can solve enormous problems, and biology runs related algorithms under the hood.

If someone wants to understand their own learning and motivation loops without surgery, is there a practical way in?

A former postdoc is building Nebula Neuro, a consumer nasal probe that streams dopamine and serotonin to your phone so you can see how thoughts, tasks, or social scenarios move your signals.

Give me one use case that would make life easier for your kids.

Real-time feedback while reading could teach concentration by coupling live neurochemistry to gentle, closed-loop nudges that help them stay engaged.

If we can track real transmitters in real people and let neural networks learn from step-by-step behavior, we can finally map comprehension and decision patterns at scale, even if the models learn things we did not anticipate.

People worry about that, but it is exciting, and it mirrors how AI and brain machine interface went from fringe to front row, much like meditation, breathwork, psychedelics, and peptides cycling from mocked to mainstream.

Reinforcement learning was dismissed as weak, then it leapt to systems that can describe complex scenes and reason convincingly.

Let’s take questions from the audience, starting with this: how much of what we hear about so-called dopamine hits is real neuroscience versus hype?

Unexpected rewards do shift dopamine, but that is only a slice of the story.

Is it too simple to frame depression as serotonin and schizophrenia as dopamine, and what would you add?

Both chemicals fluctuate in both conditions, but dopamine’s role in schizophrenia is real because blocking its receptors reduces hallucinations, and giving L-dopa to someone without Parkinson’s can push them toward paranoia; still, the diagnosis is broad and will likely fragment as we record transmitters in patients.

Another asks about a serotonin to dopamine ratio for quitting decisions and the line between grit and sunk cost.

The missing piece is expectation formation and updating across states, which shapes release as much as the chemicals themselves, and AI should help us unpack that in the coming decades.

GROCK AI jumped in with a neat chemical answer about persistence and sunk cost; how did it do?

Decent if the brain were only chemicals, but it ignored the electrical dynamics that set expectations and gate release, which is half the mechanism.

What does the public most often get wrong about dopamine, and about serotonin?

Dopamine is not pleasure, and for serotonin, people miss that drugs which raise it are diverse, can be problematic over time, and that response varies widely; many psychotropic outcomes are heavily driven by expectation, which we still do not understand mechanistically.

How do dopamine responses change when external rewards are removed and it is all internal, like meditation or private creative work?

We do not know yet, and it would be fantastic to measure whether purely internal states, even in sensory deprivation, can generate chase-worthy signals.

Which has more influence, external input or internal feedback, and how do people maintain motivation when things are not going well?

They are intertwined because clear expectations and goal maintenance govern release and drive, and while I used to push myself to the edge with high-intensity protocols inspired by Arthur Jones, I would not recommend the puke-to-prove-it approach.

Many ask about serotonin syndrome and the side effects people attribute to SSRIs.

Those drugs hit many receptor subtypes, so side effects are a target-rich outcome, and even the act of being on a mood-altering drug can change how you feel.

Thank you for bringing real nuance on dopamine, serotonin, and where measurement and AI are taking the field; you have been fearless about solving hard problems in people, and we need more of that.

Thanks for having me; this was a blast.

To learn more, see the show notes; please subscribe on YouTube, follow on Spotify and Apple, leave a rating and comments, and support the sponsors.

My first book, Protocols: an operating manual for the human body, is available for pre-sale at ProtocolsBook.com, and you can follow Huberman Lab on all major platforms and join our zero-cost newsletter at HubermanLab.com; we do not share your email, and thanks again for your interest.