The Three-Process Model of Alertness

Predicting when a person will be alert, drowsy, or dangerously fatigued is not merely an academic exercise. It is a matter of life and death. The Chernobyl disaster occurred during a night shift. The Exxon Valdez ran aground in the early morning hours. The majority of single-vehicle highway fatalities happen between midnight and 6 AM. In each case, human alertness — or the lack of it — was a decisive factor.

In 1987, Torbjorn Akerstedt and Simon Folkard published a model that would become the foundational framework for understanding human alertness across the 24-hour cycle. Their Three-Process Model of Alertness provided, for the first time, a mathematically precise description of how homeostatic sleep pressure, circadian rhythm, and sleep inertia combine to produce the characteristic daily pattern of wakefulness that every human experiences.

This article examines the model's architecture, its predictions, its 35+ years of validation, and why it matters for anyone interested in understanding how time of day shapes cognitive performance — including how nutrition tracking apps like KCALM build upon this foundation.

Paper Overview

Title: A Three-Process Model of the Regulation of Alertness-Sleepiness

Authors: Torbjorn Akerstedt and Simon Folkard

Published: 1987, in Sleep and Biological Rhythms (later expanded in subsequent publications through the 1990s and 2000s)

Context: Built upon the foundational two-process model of sleep regulation proposed by Alexander Borbely in 1982

The Problem: Why Predicting Alertness Matters

Before the Three-Process Model, work scheduling in safety-critical industries was based on rules of thumb and collective bargaining agreements rather than biological science. A 12-hour night shift was considered equivalent to a 12-hour day shift. Rotating schedules were designed for administrative convenience, not human physiology.

The consequences were measurable. Studies in the 1970s and 1980s documented that:

• Aviation: Pilot error rates increase 1.5-2x during the circadian nadir (02:00-06:00)
• Medicine: Medical errors in ICUs peak during night shifts and early morning hours
• Transportation: Truck driver accident rates correlate strongly with time-of-day and hours of continuous wakefulness
• Military: Sustained operations beyond 24 hours produce cognitive impairment equivalent to a blood alcohol concentration of 0.10% — above the legal driving limit

The Three Processes

The model describes alertness as the output of three interacting processes, each with distinct biological underpinnings and mathematical formulations.

Process S: Homeostatic Sleep Drive

Process S represents the accumulating pressure to sleep that builds during wakefulness and dissipates during sleep. The longer you have been awake, the greater the drive to sleep. The longer you have slept, the more that drive is reduced.

Mathematically, Process S rises exponentially during wakefulness and falls exponentially during sleep:

• During wakefulness: S(t) increases toward an upper asymptote with a time constant of approximately 18.2 hours
• During sleep: S(t) decreases toward a lower asymptote with a time constant of approximately 4.2 hours

Process C: Circadian Rhythm

Process C is the endogenous circadian oscillation driven by the suprachiasmatic nucleus (SCN) in the hypothalamus. Independent of how long you have been awake or asleep, your body maintains an approximately 24-hour rhythm of alertness and sleepiness.

Akerstedt and Folkard modeled Process C using a dual-harmonic sinusoidal function — the sum of a 24-hour fundamental and a 12-hour harmonic:

Process W: Sleep Inertia (Wake-Up Process)

Process W accounts for the transient period of reduced alertness immediately after waking — the grogginess, impaired cognition, and disorientation that can last from minutes to over an hour after sleep ends. This process was Akerstedt and Folkard's addition beyond Borbely's original two-process model.

Sleep inertia is not trivial. Studies show that cognitive performance in the first 15-30 minutes after awakening can be worse than performance after 24 hours of sustained wakefulness. For on-call workers who must make critical decisions immediately upon waking — physicians, firefighters, military personnel — Process W has direct operational significance.

Process W dissipates exponentially after waking, with a time constant of approximately 1 hour, though intensity varies based on the depth of prior sleep and the circadian phase at which awakening occurs.

How the Processes Interact

The Three-Process Model produces a predicted alertness value at any given time by combining all three processes:

Alertness(t) = S(t) + C(t) + W(t)

This simple additive structure produces remarkably realistic alertness curves. For a person on a typical schedule (sleeping 23:00-07:00):

The Post-Lunch Dip: Not What You Think

One of the most practically important findings embedded in the Three-Process Model is the nature of the post-lunch dip in alertness that most people experience between roughly 14:00 and 16:00.

The common assumption is that this dip is caused by eating lunch — that digestion diverts blood flow from the brain, producing drowsiness. The Three-Process Model demonstrates that this is primarily incorrect. The post-lunch dip is predominantly a circadian phenomenon, driven by the 12-hour harmonic of Process C. It occurs whether or not a person eats lunch.

This has been confirmed in multiple controlled studies:

• Subjects who skip lunch still experience the dip
• The dip occurs at approximately the same clock time regardless of when (or whether) the subject ate
• The timing of the dip aligns with the 12-hour harmonic prediction, not with time elapsed since eating

This interaction between circadian alertness and nutritional state is precisely the kind of relationship that KCALM's Mental Bandwidth Score is designed to capture.

Validation: 35 Years of Evidence

The Three-Process Model has been validated across an extraordinary range of settings and populations since its initial publication:

• Military operations: Used by NATO nations for planning sustained operations; validated against performance data from field exercises lasting 48-72 hours without sleep
• Aviation: Adapted by civil aviation authorities for crew scheduling; validated against in-flight performance measures and cockpit actigraphy data
• Shift work: Applied to predict alertness patterns in rotating shift schedules across manufacturing, healthcare, and emergency services
• Space operations: NASA has used model-based fatigue predictions for International Space Station crew scheduling
• Road safety: Incorporated into fatigue risk management systems used by transportation agencies worldwide

Relevance to KCALM: The Circadian Baseline Layer

KCALM's Mental Bandwidth Score — a composite estimate of a user's cognitive readiness at any given moment — uses the Three-Process Model as its circadian baseline layer. The logic is straightforward:

The Three-Process Model provides the "when" — the temporal structure of alertness. KCALM adds the "what" — the nutritional inputs that modulate that structure.

Limitations

The Three-Process Model, despite its success, has well-documented limitations:

• Individual differences: The model uses population-average parameters. Real humans vary substantially in chronotype (morning larks vs. night owls), sleep need (6-9+ hours), and sensitivity to sleep deprivation. Some individuals show minimal impairment after 24 hours awake; others are severely degraded
• No nutrition inputs: The original model has no mechanism for food, caffeine, or hydration effects on alertness. This is the gap that KCALM and similar systems aim to fill
• Simplified circadian component: The dual-harmonic sinusoidal is an approximation. Real circadian rhythms are influenced by light exposure, social cues, meal timing (Zeitgebers), and can shift with jet lag or irregular schedules
• Assumed sleep quality: Process S treats all sleep as equivalent, but sleep architecture (time in slow-wave sleep, REM, light sleep) significantly affects restoration
• Static parameters: The model does not adapt its parameters based on individual data over time — a limitation that modern machine learning approaches can address

Conclusion

The Three-Process Model of Alertness remains, nearly four decades after its publication, the conceptual backbone of fatigue science. Its elegant decomposition of alertness into homeostatic, circadian, and inertial components has proven robust across thousands of studies and dozens of applied domains. For the field of nutrition and cognitive performance, it provides the essential temporal framework — the daily canvas upon which nutritional effects are painted.

Understanding that the post-lunch dip is circadian rather than alimentary, that sleep inertia can be worse than sleep deprivation, and that alertness follows predictable mathematical trajectories — these insights transform how we think about optimizing daily performance through both sleep and nutrition.

Citations:

Akerstedt, T., & Folkard, S. (1987). A three-process model of the regulation of alertness-sleepiness. In R. Ogilvie & S. Broughton (Eds.), Sleep and Biological Rhythms. New York: Oxford University Press.

Borbely, A. A. (1982). A two process model of sleep regulation. Human Neurobiology, 1(3), 195-204.

Akerstedt, T., & Folkard, S. (1997). The three-process model of alertness and its extension to performance, sleep latency, and sleep length. Chronobiology International, 14(2), 115-123.

Folkard, S., & Akerstedt, T. (2004). Trends in the risk of accidents and injuries and their implications for models of fatigue and performance. Aviation, Space, and Environmental Medicine, 75(3), A161-A167.

Dawson, D., & Reid, K. (1997). Fatigue, alcohol and performance impairment. Nature, 388, 235.

Van Dongen, H. P. A., Maislin, G., Mullington, J. M., & Dinges, D. F. (2003). The cumulative cost of additional wakefulness: dose-response effects on neurobehavioral functions and sleep physiology from chronic sleep restriction and total sleep deprivation. Sleep, 26(2), 117-126.