📊 epidemiology

Psychosocial Factors Outweigh Short-Term Environmental Exposures in Subjective Cognitive Difficulties: A Causal AI Study

This causal AI study utilizing All of Us and BRFSS data reveals that while short-term environmental exposures show initial associations with subjective cognitive difficulties, these links largely attenuate when controlling for temporal variation, whereas psychosocial factors such as mental health burden, loneliness, and impaired social functioning demonstrate consistently stronger and more robust associations across multiple analytical frameworks.

Original authors: Cao, C., Ma, S.

Published 2026-06-25

📖 4 min read☕ Coffee break read

CC BY 4.0

Original authors: Cao, C., Ma, S.

Original paper licensed under CC BY 4.0 (https://creativecommons.org/licenses/by/4.0/). ⚕️ This is an AI-generated explanation of a preprint that has not been peer-reviewed. It is not medical advice. Do not make health decisions based on this content. Read full disclaimer

The Big Question: Is the Weather Making You Forgetful?

Imagine your brain is a garden. For a long time, scientists have wondered if sudden changes in the weather—like a heatwave, a blizzard, or a spike in smog—are the main reason people feel like their memory is slipping or they can't focus.

This study asked a very specific question: Do short-term environmental changes (like a sudden cold snap) cause subjective cognitive difficulties (feeling forgetful or distracted), or are other factors more important?

To find the answer, the researchers acted like high-tech detectives. They didn't just look at one snapshot in time; they tracked thousands of people over several years (2018–2024) using data from the "All of Us" research program and cross-checked it with a national health survey (BRFSS).

The Investigation: Separating the "Place" from the "Moment"

The researchers faced a tricky problem. Imagine you live in a very cold city. You might feel forgetful in the winter. But is it the cold air making you forgetful, or is it that your city has different schools, hospitals, and stress levels than a warm city?

To solve this, they used a special "time-travel" method (called fixed-effects analysis).

The Old Way (Pooled Analysis): They looked at everyone together. This is like saying, "People in cold places feel more forgetful than people in warm places."
The New Way (Within-Location Analysis): They looked at the same person in the same place over time. This is like asking, "When you experience a sudden cold shock compared to a normal day, do you feel more forgetful?"

The Result: When they switched to the "time-travel" method, the link between the weather and forgetfulness almost vanished.

The Analogy: It's like thinking a loud thunderstorm causes your car to stall. But when you check the same car on a rainy day vs. a sunny day, you realize the car stalls because of an old engine part, not the rain. The initial link was just a coincidence because the "old engine" cars happened to be in the rainy places.

The study found that while a quick "temperature shock" showed a tiny link to attention problems in the broad data, that link disappeared when they looked closely at the same people over time. The weather simply wasn't the main culprit.

The Real Culprit: The "Psychosocial" Storm

If the weather isn't the main problem, what is? The study found that psychosocial factors (how we feel inside and how we connect with others) were the heavy hitters.

Think of your brain as a house. The weather is like the wind outside the window. It might rattle the glass a little, but it doesn't knock the house down. However, psychosocial factors are like the foundation of the house. If the foundation is shaky, the whole house wobbles.

The study ranked the "shakiness" of the foundation:

Mental Health Burden: This was the strongest link. Feeling mentally overwhelmed was like a heavy storm shaking the house's foundation.
Social Functioning: How well you interact with others mattered a lot.
Loneliness: Feeling isolated was a major factor.
Literacy/Education: How comfortable you are with reading and understanding information also played a big role.

The Comparison:
The researchers compared the "weight" of these factors.

Environmental factors (like blizzards or smog) were as light as a feather.
Psychosocial factors were as heavy as a boulder.

In fact, the link between loneliness and feeling forgetful was nearly 10 times stronger than the link between any weather event and feeling forgetful.

The "AI" Check-Up

To make sure they weren't missing anything, the researchers used advanced Artificial Intelligence (AI) tools. They used these tools to look for hidden patterns, like a mechanic using a super-computer to scan a car for invisible cracks.

The AI confirmed the human findings: The "weather" signals were weak, and the "human connection/mental health" signals were loud and clear.

The Bottom Line

This study tells us that when people say, "I feel like my brain isn't working right," we shouldn't immediately blame the weather, the air quality, or a sudden cold snap.

Instead, the study suggests we should look at the human experience:

Are they lonely?
Are they stressed or depressed?
Do they feel socially isolated?

The Takeaway:
If you want to help someone who feels forgetful or distracted, fixing the thermostat or cleaning the air filters might help a tiny bit, but fixing their loneliness, stress, and mental health will make a massive difference. The "weather" outside matters much less than the "weather" inside our minds and our relationships.

Technical Summary: Psychosocial Factors Outweigh Environmental Exposures in Subjective Cognitive Difficulties

Problem Statement
Existing observational literature suggests links between short-term environmental exposures (e.g., temperature extremes, air pollution) and cognitive performance or attention. However, interpreting these associations is challenging because environmental exposures are heavily confounded by geography, season, infrastructure, and socioeconomic conditions. Cross-sectional analyses often fail to distinguish whether observed associations reflect acute environmental effects or persistent contextual differences between locations. Furthermore, subjective cognitive difficulties (SCD) are known to be strongly influenced by psychosocial factors such as mental health, loneliness, and social functioning, yet the relative importance of environmental versus psychosocial drivers remains unclear when analyzed through rigorous causal inference frameworks.

Methodology
The study utilized longitudinal data from the All of Us Research Program (2018–2024), linking daily environmental exposure records (temperature, snowfall, precipitation, blizzard exposure, PM2.5) to repeated participant-level cognitive and attention outcomes via ZIP3 geographic codes.

Identification Strategy: To disentangle within-location temporal variation from between-location contextual differences, the authors employed a series of increasingly restrictive causal inference models:
- Pooled Models: Standard regression across all data.
- Fixed-Effects Models: ZIP3-specific fixed effects ( $\alpha_z$ ) and calendar-time effects ( $\gamma_t$ ) to remove time-invariant geographic and demographic confounders.
- Lagged and Event-Study Analyses: To assess temporal dynamics and pre-trends surrounding environmental shocks.
- Doubly Demeaned Models: To further isolate variation.
Causal AI and Machine Learning: The study integrated advanced methods to assess heterogeneity and latent structures, including:
- Double Machine Learning (DML) and Causal Forests for high-dimensional confounding and non-linear relationships.
- Conditional Variational Autoencoders (cVAE) to model latent psychosocial representations.
Independent Replication: Findings were validated using the 2024 Behavioral Risk Factor Surveillance System (BRFSS), a cross-sectional national survey, focusing on the association between psychosocial factors (loneliness, mental health days) and SCD.
Mediation Analysis: Explored whether environmental effects operated through psychological, behavioral, or physiological pathways.

Key Contributions

Causal Framework Application: The paper applies a rigorous quasi-experimental framework (fixed effects, event studies) to environmental health research, demonstrating how pooled associations often attenuate when controlling for stable geographic and contextual factors.
Integration of Causal AI: It combines traditional longitudinal econometrics with modern causal AI (DML, causal forests, latent representation learning) to evaluate the robustness of findings against non-linearities and high-dimensional confounding.
Comparative Effect Size Analysis: The study systematically compares the magnitude of environmental exposures against psychosocial factors, providing empirical evidence that psychosocial variables dominate environmental predictors for SCD.
External Validation: It confirms the primacy of psychosocial correlates in an independent, nationally representative dataset (BRFSS), reinforcing the generalizability of the findings.

Results

Environmental Exposures: In pooled analyses, temperature shocks showed a modest association with attention impairment ( $\beta = 0.0315, p = 0.032$ ). However, this association substantially attenuated in fixed-effects models ( $\beta = 0.011, p = 0.617$ ) and other within-location specifications, becoming statistically non-significant. Similarly, associations between environmental exposures and subjective cognitive difficulty were consistently near zero across all model specifications (pooled, fixed-effects, event-study).
Psychosocial Factors: In stark contrast, psychosocial variables demonstrated substantially larger and statistically significant associations with SCD:
- Mental-health burden: $\beta = -0.114$ ( $p < 0.001$ )
- Social functioning: $\beta = -0.102$ ( $p < 0.001$ )
- Literacy-related measures: $\beta = -0.082$ ( $p < 0.001$ )
- Loneliness: $\beta = 0.053$ ( $p < 0.001$ )
Replication (BRFSS): The BRFSS analysis confirmed these patterns. Loneliness was strongly associated with subjective cognitive decline (OR = 1.81, 95% CI 1.76–1.86). Mediation analysis in BRFSS suggested that approximately 28.6% of the association between loneliness and SCD was mediated through poor mental-health days.
Mediation Pathways: Exploratory mediation analyses in the All of Us cohort found little evidence that short-term environmental exposures influenced cognitive outcomes through the examined psychological, behavioral, or physiological pathways. Indirect effects were uniformly small.
AI Findings: Causal AI analyses (DML, causal forests, cVAE) produced findings consistent with the primary longitudinal models, confirming that latent psychosocial burden is a stronger predictor of SCD than environmental shocks.

Significance and Claims
The paper claims that psychosocial factors are more strongly and consistently associated with subjective cognitive difficulties than short-term environmental exposures when analyzed through longitudinal causal-inference and causal AI frameworks.

The authors argue that the attenuation of environmental associations under within-location identification strategies suggests that many pooled associations in existing literature likely reflect broader geographic and contextual differences (e.g., regional socioeconomic status, infrastructure) rather than acute environmental effects occurring within the same location over time.

Consequently, the study concludes that interventions targeting loneliness, psychological distress, and impaired social functioning may have greater potential to reduce subjective cognitive burden than approaches focused solely on mitigating short-term environmental fluctuations. The findings highlight the necessity of distinguishing environmental signals from contextual and psychosocial influences in large observational datasets to avoid misattributing cognitive health outcomes to environmental causes.