Sleep, insomnia and cognition: can cognitive training help?

Why does chronic insomnia matter for memory?

Chronic insomnia is one of the most common complaints of later life: across surveys its prevalence in older adults ranges from about 20% to nearly 50%, higher in women than in men¹. After age 60, the time to fall asleep, night-time awakenings and lighter sleep all increase, while total sleep time and sleep efficiency fall³.

That decline does not stay confined to the night. In a home-based study of 99 community-dwelling older adults (mean age 72.3 years), 35 with insomnia and 64 good sleepers were compared by actigraphy and cognitive testing³. The insomnia group averaged 74.5% sleep efficiency versus 91.6%, and showed impaired performance across five domains: memory span, allocating attention to a target, time estimation, executive functioning, and integrating visual and semantic information³. The authors concluded that "insomnia may have detrimental effects on some cognitive functions in healthy older adults", the memory signature that distinguishes chronic insomnia from an ordinary bad night.

First-line drug therapy is a poor long-term fit: sedative-hypnotics carry risks of adverse effects and dependence, and their effectiveness in insomnia is reported to wane rapidly after about 30 days of use¹, one reason researchers have tested non-pharmacological, learning-based approaches.

Can cognitive training actually improve sleep?

The strongest test to date is a randomized controlled trial of 51 older adults aged 65–85, all diagnosed with chronic insomnia and confirmed by a week of baseline actigraphy¹. Participants were randomly allocated to a personalized, home-based computerized cognitive training program (n = 34) or to an active control group (n = 17) that completed Microsoft Word and Paint assignments not engaging high-level cognitive functioning. Both arms trained 20 to 30 minutes per session, three times a week, for 8 weeks (24 sessions); sleep was measured for seven nights before and after with a wrist actigraph plus a sleep diary.

After eight weeks, the training group improved on four of five actigraphic measures. The time to fall asleep fell from 38.4 to 24.8 minutes, a 13.7-minute reduction, below the 31-minute clinical criterion for insomnia (p < 0.001). Sleep efficiency rose from 73.5% to 80.3% (+6.7 points, p < 0.001), wake time after sleep onset dropped from 72.1 to 58.9 minutes (p < 0.05), and nightly awakenings fell from 10.7 to 9.0 (p < 0.05)¹. The active control group showed none of these gains, all five sleep variables were statistically unchanged, and between-group effect sizes favored training for both sleep onset latency (Cohen's d = −0.70) and sleep efficiency (d = 0.70)¹.

Cognition moved alongside sleep, where the memory thread of chronic insomnia becomes concrete. After controlling for age, the training group improved on five cognitive abilities relative to controls: avoiding distractions, naming, general memory, visual memory and working memory, while the control group's working memory significantly declined¹. Specific skills tracked specific sleep outcomes: improved visual scanning predicted earlier sleep onset, improved naming predicted fewer awakenings and less wake after sleep onset, and improved "avoiding distractions" predicted longer total sleep time¹. The changes moved together, though, as the authors stressed, a correlation between two improvements cannot prove one caused the other.

Does it work in working-age adults too?

A 2023 Phase I/II clinical trial extended the question below age 60, enrolling 12 patients with diagnosed insomnia disorder (mean age 44, range 25–55, 75% male) from a hospital sleep unit in Spain². A dose-escalation phase first set 45 minutes as the maximum tolerated session length; participants then completed seven home-based sessions of nine 5-minute gamified tasks on alternate days across 15 days, about 5 hours 15 minutes of training.

From day 1 to day 15, global sleep quality on the Pittsburgh Sleep Quality Index improved significantly (mean change −7.4, p = 0.014, a large effect, d = 0.84)². Mood and worry moved with it: depressive symptoms on the BDI-II fell (−6.2, p = 0.015), worry on the Penn State Worry Questionnaire fell (−4.2, p = 0.005), and trait anxiety on the STAI dropped sharply (−18.3, p < 0.001)². Not every instrument shifted: the Insomnia Severity Index and the BRIEF-A executive-function inventory showed no significant change, attributed partly to the short two-week window². No session produced extreme fatigue, and the authors reported improvements "in the full absence of safety issues." The gap this addresses is large: in Spain alone, about 4.7 million people are diagnosed with insomnia within a population of roughly 47 million, while CBT-I reaches an estimated 1% of them².

How does this fit with first-line insomnia care?

None of this displaces the standard of care. The European guideline for the diagnosis and treatment of insomnia positions cognitive behavioral therapy for insomnia (CBT-I) as first-line for chronic insomnia in adults of all ages, with pharmacology reserved as a shorter-term adjunct⁴; a meta-analysis confirmed CBT-I produces clinically meaningful sleep improvements while documenting that most patients still rely on medication instead⁵. The cognitive-training studies sit upstream of treatment claims: early-stage tests of whether a learning-based activity, distinct from CBT-I, was associated with better sleep and cognition in people with insomnia. They are a complement to, not a replacement for, first-line therapy, and distinct from consumer "sleep apps" that track or coach sleep without targeting cognition.

What the evidence does NOT show

These are early, encouraging results reported by the study authors, not a settled case. Three limits matter most. First, the samples were small: the randomized trial analyzed 51 completers (61% of 84 eligible) and the Spanish trial just 12 patients, with a 50% drop-out rate². Second, the 2023 trial had no control group and ran only 15 days, so its sleep gains cannot be cleanly separated from placebo effects, daily clinician contact or natural fluctuation, and its long-term durability was, in the authors' words, "still unknown"². Third, the 2008 study was observational, it shows insomnia and weaker cognition co-occur, not that one causes the other³.

Two further cautions. In the randomized trial the arms were not perfectly matched on age (the training group was older, p = 0.02), one author was an employee of the company whose program was tested¹, and the cognition–sleep link was correlational, causal direction, the authors noted, "cannot be inferred from correlation analysis"¹. All three trials tested specific adaptive programs, whose results do not transfer to every "brain game" on the market. No study here establishes that cognitive training treats, cures or prevents insomnia, or that it reduces dementia risk.

What this means in practice

For adults with chronic insomnia, the reviewed studies suggest structured, adaptive cognitive training was well tolerated and was followed by improved objective sleep in one randomized trial and improved self-reported sleep quality in another, alongside gains in memory and attention, while CBT-I remains the guideline-recommended first-line therapy⁴⁵. Larger, longer, independently controlled trials are still needed before any firmer conclusion.

For the broader picture, see the related guides on sleep, cognition and the aging brain, cognitive training in healthy older adults, stress, worry and memory, cognition after stroke and brain injury, and the pillar overview of the cognitive-decline evidence base. The two anchor trials are broken down in full on the study pages for the 2023 digital-therapeutics insomnia trial and the 2013 sleep-quality randomized trial.