This week: cognitive psychologist Alan Castel walked Huberman through the science of memory selectivity — why experts remember more by remembering less, and why older adults can outperform younger ones on memory tasks that reward relevance over volume. And the Essentials vault dropped the bipolar disorder episode with the lithium mechanism data.

GM.

The most counterintuitive episode this week is Alan Castel on Huberman. Castel studies memory at UCLA, and his core finding upends the narrative that memory simply declines with age. It does decline — in volume. But memory selectivity — the ability to remember what matters and forget what doesn't — actually improves with age and expertise. Older adults remember fewer items but a higher proportion of important ones.

⏱ ~5 min read

Two findings from the same week. One: memory isn't a recording system. It's a reconstruction system that actively selects what to encode based on perceived value — and this selection gets better, not worse, with age. Two: lithium — the oldest psychiatric medication still in use — protects neurons through a mechanism that connects to ageing, inflammation, and neuroplasticity in ways we're only beginning to understand.

The thread this week: the systems that seem to be declining may actually be specialising — doing less, but doing it better.

Value
The variable that determines what you remember — not repetition or effort, but perceived importance at the time of encoding.

Castel's key experiment: give people a list of words, each paired with a point value (1–10 points). Tell them they'll be tested and scored based on what they remember. Young adults remember more total words. But older adults remember a higher proportion of high-value words — they're better at allocating their limited memory resources to what matters most. Castel calls this 'value-directed remembering' and argues it's a cognitive strategy, not a deficit.

The mechanism: as total memory capacity declines with age, the prefrontal cortex compensates by becoming more selective in what it encodes. Younger brains encode everything with roughly equal priority. Older brains prioritise — encoding high-value information more deeply and allowing low-value information to fade. The result is a smaller but higher-quality memory store.

Castel's research on expertise shows the same pattern. Chess grandmasters remember meaningful board positions with extraordinary accuracy but are no better than novices at remembering randomly placed pieces. Medical experts remember diagnostic-relevant patient details but forget irrelevant ones faster than trainees. Expertise is, in part, a memory-selectivity skill — knowing what to encode and what to release.

The memory complaint most people have — 'I can't remember as much as I used to' — may be the wrong metric. The better question: are you remembering the right things?

Dr. Alan Castel on Huberman Lab · July 2026

Memory doesn't decline the way you think — it gets more selective
Alan Castel on value-directed remembering, memory selectivity with age, and why expertise is partly a memory-editing skill.
Dr. Alan Castel · Huberman Lab · July 2026

Castel is a memory researcher at UCLA who specialises in cognitive ageing. His work is a corrective to the 'memory inevitably declines' narrative — not by denying the decline, but by showing that the decline is accompanied by an improvement in selectivity that most ageing research ignores.

🎙️ Listen → Huberman Lab, Dr. Alan Castel, July 2026

Castel gives you the memory model. The bipolar episode gives you the lithium surprise. Together they reveal: the brain is constantly editing itself.

The bipolar disorder Essentials episode Huberman dropped the same week connects to Castel's memory model through lithium — a simple ion that has been used to treat bipolar disorder since 1949 and whose mechanism we're still discovering. Lithium is neuroprotective: it increases grey matter volume, promotes neurogenesis, and reduces neuroinflammation. These effects occur at therapeutic doses in bipolar patients and, intriguingly, at trace doses in population studies.

Epidemiological data from Texas, Japan, and Austria shows that communities with higher lithium levels in drinking water have lower rates of suicide, dementia, and all-cause mortality. The effect sizes are small but consistent across geographies. The mechanism: lithium inhibits GSK-3β, an enzyme involved in inflammation, tau phosphorylation (relevant to Alzheimer's), and apoptosis. At therapeutic doses, it also promotes BDNF expression — the same brain-derived neurotrophic factor that exercise, meditation, and novel movement stimulate.

The connection to Castel's memory work: lithium's neuroprotective effects preferentially support the hippocampus and prefrontal cortex — the same regions involved in memory encoding and the selectivity advantage Castel describes. Maintaining these regions may be one of the mechanisms through which the brain preserves memory quality (selectivity) even as quantity declines.

The broader pattern across this issue: the brain doesn't just degrade with age. It remodels — shedding capacity it doesn't need, strengthening functions that matter more. Memory selectivity improves. Expertise deepens. The brain with lithium support maintains the regions that drive these improvements. Ageing is not just loss. It's specialisation — if the supporting infrastructure holds.

The brain isn't just a recorder that degrades with age. It's an editor that gets better at its job — if you keep the editing suite in good repair.

P.S. Quick test: what are the three most important things you learned this week? Not what you read — what actually stuck. Hit reply with your three. We're curious how selective your memory already is.

What did you think of today's BrainBite?