Seasonality in human cognitive brain responses

The researchers tested whether human brain responses to two different cognitive tasks — sustained attention and working memory — vary systematically across the calendar year. They did not give participants any intervention or treatment. Instead, they brought healthy young adults into the lab at different times of year, kept them in an environment with no natural light, no clocks, no weather cues, and no seasonal information for 4.5 days, and then measured their brain activity using functional magnetic resonance imaging (fMRI) while they performed the tasks.

**Outcome measures:**

Brain activation (blood-oxygen-level-dependent signal, or BOLD) in specific regions during a sustained attention task (the Psychomotor Vigilance Task, PVT)

Brain activation in specific regions during a working memory task (an n-back task with three levels of difficulty: 0-back, 1-back, and 2-back)

Reaction times and accuracy on both tasks (behavioural performance)

**Comparators:** There was no control group. The comparison was across seasons — participants tested in winter (December–February), spring (March–May), summer (June–August), and autumn (September–November). The researchers then modelled brain activity as a sinusoidal function over the calendar year to identify the timing (phase) and magnitude (amplitude) of seasonal peaks and troughs.

**Sample size:** 28 healthy young adults (16 women, 12 men)

**Age:** Mean age 24.5 years (range not explicitly stated, but described as "young adults")

**Setting:** Université de Liège, Belgium. Participants were recruited from the local community.

**Inclusion criteria:** Right-handed, non-smokers, no history of sleep disorders, no psychiatric or neurological conditions, no shift work in the past 6 months, no transmeridian travel in the past 2 months, no medication affecting the central nervous system, no excessive caffeine consumption (<300 mg/day), no excessive alcohol consumption (<14 units/week), regular sleep-wake schedule (bedtime between 22:00–00:00, wake time between 07:00–09:00) confirmed by actigraphy for 7 days before the lab stay.

**Exclusion criteria:** Any of the above, plus MRI contraindications (metal implants, claustrophobia, pregnancy).

**Important note:** This is a cross-sectional study — each participant was tested only once, at a single time of year. The researchers did not follow the same people across all four seasons. This means individual seasonal trajectories could not be measured; instead, the study compared different groups of people tested at different times of year.

**Cognitive tasks:**

**Sustained attention:** Psychomotor Vigilance Task (PVT) — a 10-minute reaction-time task where participants press a button as soon as a millisecond counter appears on screen. Outcome: mean reaction time, number of lapses (reaction times >500 ms), and brain activation in the frontoparietal attention network.

**Working memory:** n-back task — participants saw a series of letters and had to indicate whether the current letter matched the one presented n positions earlier. Three conditions: 0-back (match a target letter), 1-back (match the previous letter), 2-back (match the letter two back). Outcome: accuracy, reaction time, and brain activation in the dorsolateral prefrontal cortex, parietal cortex, and other working memory regions.

**Brain imaging:**

Functional MRI (3 Tesla scanner) measuring BOLD signal — a proxy for neural activity based on blood oxygen levels.

Structural MRI scans were also collected to rule out anatomical differences.

**Environmental control:**

Participants lived in a windowless, sound-attenuated suite for 4.5 days.

No natural light, no clocks, no phones, no internet, no television, no seasonal cues.

Temperature and humidity were controlled (temperature ~21°C).

Light levels were fixed at ~15 lux during wake periods (dim indoor lighting) and <0.02 lux during sleep (complete darkness).

Meals were provided at fixed times, but participants were not told the time of day.

Sleep schedules were fixed: 8 hours in bed (23:00–07:00) for the first 3 nights, then 7 hours in bed (23:00–06:00) on the final night before scanning.

**Seasonal classification:**

Participants were assigned to a season based on the date of their fMRI scan: winter (December solstice to March equinox), spring (March equinox to June solstice), summer (June solstice to September equinox), autumn (September equinox to December solstice).

The exact dates of testing spanned 2.5 years (January 2011 to July 2013), with participants tested roughly evenly across the year.

**Study design:** Cross-sectional observational study. Participants were tested at a single time point, and the researchers compared brain activity across groups tested in different seasons. This is not a longitudinal study — the same people were not followed across seasons.

**Randomisation:** None. Participants were not randomly assigned to a season; they were recruited and tested at whatever time of year they volunteered. This is a major limitation because any pre-existing differences between people who happen to show up in summer versus winter could confound the results (e.g., personality traits, baseline cognitive ability, socioeconomic status, or even undiagnosed seasonal affective disorder).

**Blinding:** The participants were blind to the purpose of the study regarding seasonality — they were told it was a study of sleep and cognition. The researchers analysing the fMRI data were also blind to the season of testing during the initial processing. However, the study was not double-blind in the traditional sense because the experimenters knew what season it was when running the sessions.

**Duration of environmental control:** 4.5 days (108 hours) in a season-free environment before the fMRI scan. This is a critical design feature: if seasonal effects had disappeared after 4.5 days of constant conditions, the conclusion would be that the brain was simply reacting to immediate environmental cues (weather, light, temperature). Because the effects persisted, the researchers argue that the brain has an endogenous seasonal rhythm — an internal calendar that continues to tick even when external cues are removed.

**Statistical approach:**

The primary analysis modelled brain activation (BOLD signal) as a sinusoidal function of the day of the year (1–365). This is a standard approach for detecting seasonal rhythms.

The model estimated two parameters: amplitude (how much brain activity varies across the year) and phase (the time of year when activity peaks).

Statistical significance was assessed using a mixed-effects model with participants as a random factor, and correction for multiple comparisons across brain voxels (cluster-level correction at p < 0.05, with a cluster-forming threshold of p < 0.001).

Behavioural data (reaction times, accuracy) were analysed using linear mixed models with season as a categorical factor and task difficulty as a within-subject factor.

**What this design can prove:**

It can show that brain activity during cognitive tasks varies systematically across the calendar year, even after 4.5 days in a constant environment.

It can estimate the timing and magnitude of these seasonal peaks and troughs.

It can demonstrate that different cognitive processes have different seasonal patterns (attention peaks in summer, working memory peaks in autumn).

**What this design cannot prove:**

It cannot prove that these seasonal rhythms are caused by an internal biological clock rather than by lingering effects of the pre-lab environment (e.g., differences in vitamin D levels, prior light exposure, or prior sleep patterns that persist for more than 4.5 days).

It cannot prove that the same individuals would show the same seasonal pattern if followed across a full year — because each person was only tested once, the study compares different people at different times, not the same people across time.

It cannot prove causality — the study is purely descriptive. It cannot tell us what mechanism (e.g., melatonin, cortisol, vitamin D, photoperiod history) drives the seasonal variation.

It cannot rule out that the seasonal pattern is driven by differences in the participant pool across seasons (selection bias).

**Major methodological weaknesses:**

1. **Cross-sectional design:** The biggest weakness. With only 28 participants spread across the year, there are roughly 7 people per season. Individual differences in brain activity are enormous, so a few outliers could drive the apparent seasonal pattern.

2. **Small sample size:** 28 participants is very small for detecting seasonal effects, especially when each person is only measured once. The study is underpowered for reliable phase estimation.

3. **No replication sample:** The results were not validated in an independent group of participants.

4. **No control for pre-existing seasonal differences:** Participants were not matched across seasons for baseline cognitive ability, mood, or other factors that could affect brain activity.

5. **Only one measurement per participant:** Without repeated measures, it is impossible to distinguish true seasonal rhythms from random between-person variation.

**Primary outcome: Brain activation during sustained attention (PVT)**

Brain activity in the frontoparietal attention network (including the inferior frontal gyrus, middle frontal gyrus, and intraparietal sulcus) showed a significant seasonal rhythm (p < 0.05, cluster-corrected).

**Peak activation:** Around the summer solstice (late June).

**Trough activation:** Around the winter solstice (late December).

The amplitude of the seasonal variation was approximately 10–15% of the mean BOLD signal in these regions (exact percentage not reported, but estimated from figures).

This pattern was observed in both the left and right hemispheres, with similar timing.

**Primary outcome: Brain activation during working memory (n-back)**

Brain activity in the working memory network (dorsolateral prefrontal cortex, posterior parietal cortex, and anterior cingulate cortex) also showed a significant seasonal rhythm (p < 0.05, cluster-corrected).

**Peak activation:** Around the autumn equinox (late September).

**Trough activation:** Around the spring equinox (late March).

The amplitude was similar to the attention task (~10–15% of mean BOLD signal).

This pattern was consistent across all three difficulty levels (0-back, 1-back, 2-back), but was most pronounced for the most demanding condition (2-back).

**Secondary outcome: Behavioural performance**

**Sustained attention (PVT):** No significant seasonal variation in reaction times or lapse frequency (p > 0.05 for all comparisons). Despite large seasonal changes in brain activity, participants performed equally well across the year.

**Working memory (n-back):** No significant seasonal variation in accuracy or reaction times (p > 0.05 for all comparisons). Again, brain activity changed but behaviour did not.

**Key contrast between tasks:**

The seasonal timing was completely different for the two tasks — attention peaked in summer, working memory peaked in autumn. The phase difference was approximately 90 degrees (3 months apart on the calendar).

This dissociation is important because it rules out a simple "brain works better in summer" explanation. Instead, different cognitive processes appear to have their own seasonal rhythms.

**Additional analyses:**

The seasonal pattern was not explained by differences in sleep duration, sleep quality, or subjective alertness on the day of testing (all measured and found to be stable across seasons).

The pattern was also not explained by differences in mood (measured by the Profile of Mood States questionnaire), which did not vary significantly across seasons in this healthy sample.

Structural brain scans showed no seasonal differences in brain volume or grey matter density, suggesting the effects are functional rather than structural.

**For sustained attention:** Brain activity in key attention regions is approximately 10–15% higher in summer compared to winter. To put this in perspective, this is roughly the same magnitude of difference seen between well-rested and sleep-deprived individuals in some studies — but here it happens naturally across the year, even when people are well-rested and in constant conditions.

**For working memory:** Brain activity peaks in autumn at roughly 10–15% above the spring trough. This is comparable to the difference between performing an easy (0-back) and moderately difficult (1-back) version of the task — meaning the seasonal effect is large enough to shift brain activity by about one "difficulty level" worth of neural engagement.

**Behavioural null result:** Despite these large brain changes, performance on both tasks did not vary across seasons. This means the brain is apparently compensating — recruiting more neural resources at certain times of year to maintain stable performance. This is a common finding in cognitive neuroscience: brain activity can change without behaviour changing, because the brain adapts to maintain output.

**Practical implication for self-experimenters:** If you measure only behaviour (reaction time, accuracy), you might miss seasonal effects entirely. The effects are happening at the neural level, and they may only become behaviourally visible when the system is challenged (e.g., under sleep deprivation, stress, or in people with cognitive vulnerabilities).

**Acknowledged by authors:**

Cross-sectional design: "Because of the cross-sectional design, we cannot exclude the possibility that the observed seasonal variations reflect differences between participant groups rather than true seasonal rhythms."

Small sample size: "The relatively small sample size (n = 28) limits the statistical power to detect seasonal effects, particularly for behavioural measures."

Single laboratory environment: "The 4.5-day constant routine may not be sufficient to wash out all seasonal influences from the outside world."

No measurement of potential mediators: "We did not measure melatonin, cortisol, or other hormonal markers that could mediate seasonal effects on brain function."

Healthy young sample: "These findings may not generalise to older adults, children, or clinical populations."

**Additional critical limitations:**

**No replication:** The study has not been replicated by independent groups, and the specific seasonal phases (summer peak for attention, autumn peak for working memory) have not been confirmed.

**Multiple comparisons:** The researchers tested two tasks, multiple brain regions, and multiple difficulty levels, increasing the risk of false positives. While they used cluster correction, the small sample makes the correction less reliable.

**Seasonal definition:** The study used meteorological seasons (3-month blocks), but the sinusoidal model assumes a smooth annual cycle. If the true pattern is more complex (e.g., two peaks per year), the model would miss it.

**No control for vitamin D:** Vitamin D levels vary dramatically across seasons and affect brain function. The 4.5-day lab stay would not change vitamin D status, which reflects months of sun exposure.

**No control for prior light exposure:** The amount of daylight participants experienced before entering the lab could have lasting effects on circadian and seasonal biology that persist beyond 4.5 days.

**Geographic specificity:** The study was conducted in Belgium (latitude ~50°N), where seasonal changes in day length are extreme (8 hours in winter, 16 hours in summer). Results may not generalise to equatorial regions or to places with less seasonal variation.

**Publication bias:** The study reports only significant seasonal effects. It is possible that other brain regions or cognitive tasks were tested and showed no seasonal variation, but these null results were not reported.

For someone running their own n=1 experiment:

**What to test:**

Track your own cognitive performance across at least one full year (ideally 2–3 years to confirm patterns).

Focus on two distinct cognitive domains: sustained attention (e.g., reaction time on a simple vigilance task) and working memory (e.g., n-back task or digit span).

Also track subjective measures: mental effort required to concentrate, perceived cognitive fatigue, and mood.

**Minimum meaningful duration:**

At least 12 months, with measurements at least once per month (ideally weekly).

You need data from all four seasons to detect a sinusoidal pattern.

One measurement per season is not enough — you need multiple data points within each season to distinguish seasonal variation from random day-to-day fluctuation.

A minimum of 52 weekly measurements (one per week for a year) would give reasonable power to detect a seasonal rhythm.

**What to measure (specific metrics