Brain games for seniors: what the research actually shows

If you've bought your parent a brain-training app, you've made a reasonable bet on weak evidence. Most commercial brain training shows *near-transfer* (you get better at the app) without the *far-transfer* that matters in life: driving better, remembering names, paying bills.

• Aerobic exercise. 30 min brisk walking; stronger evidence than any brain game. Hits the hippocampus directly.
• Daily social connection. A daily conversation does more than any puzzle.
• Sleep. Deep sleep clears beta-amyloid. Chronic poor sleep raises dementia risk substantially.
• Learning a new skill (instrument, language, woodworking) builds new neural pathways; crosswords use existing ones. Dr. Tanzi (Harvard/MassGeneral) calls this the L in SHIELD: building synaptic reserve before symptoms appear.
• Reminiscence Therapy. SMD 0.78–2.34 across 42 trials: best-evidenced non-pharma approach for older adults.

There's a category of game most articles lump in with Lumosity but shouldn't. Finish-the-lyric ("You ain't nothin' but a ..."), complete-the-movie-quote, era trivia from her decade, zodiac questions, dad jokes she groans at: these pull from autobiographical, musical, and emotional memory systems that degrade *last* in Alzheimer's. Different category from abstract puzzle apps.

The mechanism is anatomical. The medial prefrontal cortex, default mode network, and amygdala (the autobiographical-memory + emotional-tagging circuit) hold up far longer than the hippocampus does. Music processing regions in the caudal anterior cingulate and ventral pre-SMA are largely preserved into late-stage Alzheimer's (Jacobsen et al. 2015, Brain). A 75-year-old who can't remember breakfast can still finish a 1955 lyric. That's not a parlor trick; that's a brain region that's still working.

Why Lumosity-style abstract puzzles don't transfer: they train *near-transfer* skills (you get better at the puzzle) on circuits that are weakening anyway. Reminiscence-anchored games tap circuits that are *still strong*, in a frame that's emotionally rewarding, and they bridge into conversation, the actual evidence-backed intervention. The game is the doorway; the reminiscence is the room.

The 'brain games protect against dementia' claim rests on weak evidence. The FTC settled with Lumosity in 2016 for $2M over deceptive advertising; the complaint found the company lacked 'competent and reliable scientific evidence.'

A meta-analysis pooling hundreds of studies (Sala & Gobet, 2019) found far-transfer effects indistinguishable from zero. Even ACTIVE (Rebok 2014) showed memory effects fading by year 5.

Seventy neuroscientists signed an open letter in 2014 warning there is 'no compelling scientific evidence' that training generalizes. The stronger base is multi-domain engagement: reminiscence, exercise, social connection.

• Lumosity, BrainHQ, similar: better at the games, weak evidence of life transfer.
• Sudoku, crosswords, jigsaw puzzles: fine activities, low cognitive uplift in studies.
• Supplements (Prevagen, ginkgo): almost no evidence; some have faced FTC enforcement.
• 'Memory-boosting' courses with grandiose claims and no peer-reviewed data.

• Daily: 30-minute walk, conversation with family, 7-8 hours sleep.
• Weekly: 2-3 hours strength/balance, a social outing.
• Ongoing: a learning project (language, instrument, craft), 30 min × 3/week.
• Quarterly: hearing and vision check.
• Daily, optional: a Daily Call in Family Voices (AI based on Reminiscence Therapy), with family or a service like Familiar. Familiar also texts photos to the receiver's phone live during the call so reminiscence has visual anchors.