Startle and Surprise

Startle is the involuntary reflex to a sudden, intense stimulus: a loud bang, a sudden alarm, a violent jolt. Surprise is the cognitive response to an event that does not match what the pilot expected, whether it arrives loudly or in silence. They often come together, and their combined effect, the startle effect, is a brief but real loss of the ability to perceive, think and act well, at exactly the moment a correct response is most needed.

The subject matters because modern flight decks are reliable: failures are rare, so they are unexpected when they come. Investigators have pointed to startle and surprise in loss of control accidents such as Air France 447, guidance on upset prevention and recovery from ICAO, EASA and the FAA stresses both, and EASA's syllabus for crew resource management lists surprise and startle, and resilience, among its topics.

Startle and surprise defined

Human performance research separates the two reactions:

The two can occur separately. A loud event the crew is expecting, such as the fire bell during a test, can still produce the reflex without any surprise; a quiet but unexpected indication, such as a fuel figure that does not fit the plan, produces surprise without startle. The worst case combines them: an unexpected bang or warning in a phase of flight where nobody is prepared for it. The term startle effect is commonly used for that combined physiological and mental reaction to a sudden, unexpected and intense event, such as an autopilot disconnection, a loud bang, a stall or terrain warning, or an abrupt upset.

Physiological and cognitive effects

The startle reflex is fast and involuntary, a flinch and a surge of alertness that the pilot cannot suppress. It sets off the fight-or-flight response, in which the sympathetic nervous system and adrenaline raise the heart rate, blood pressure and breathing rate and prepare the muscles for action: a response built for running or fighting, not for reasoning in a seat (see stress). The FAA's upset recovery guidance describes the result as an adrenaline surge, tunnelled attention and a risk of freezing or of an incorrect reflex input.

The cognitive effects matter more than the physical ones:

• Attention narrows. Under high arousal the range of cues sampled contracts, peripheral information and warning flags go unseen, and auditory information is commonly the first to be discarded, which is how a clearly audible warning can go unheard (see attention, vigilance and distraction).
• Thinking slows. Recalling a procedure, interpreting instruments and weighing options are degraded, memory items may be delayed or forgotten, and time can seem distorted.
• Action becomes reflexive. High arousal produces faster but less accurate responses, and a prepared response can be fired by the wrong stimulus, an error of commission: a pilot primed for an engine shutdown drill may shut down an engine at the noise of a falling tray (see human information processing). At a stall warning the instinctive input is often to pull, as on Colgan Air 3407 and Air France 447, whereas the recovery needs the angle of attack reduced first: the 737's stall recovery begins with the pilot flying holding the control column firmly, then disconnecting the autopilot and autothrottle and applying nose-down elevator.
• Or nothing happens. Some pilots freeze and, for a time, do not act at all.

View full schematic

The reflex itself passes quickly. The damaging part is the time needed to rebuild a picture of the situation that has just been shown to be wrong, during which any instinctive input goes on acting on the aircraft. The length and severity of that disruption vary between individuals, and both fall with experience, with procedures learned beyond the minimum standard and with realistic training.

Why they matter on the flight deck

Modern aircraft fail rarely, which makes each failure a surprise. Human performance texts note that failures on automated flight decks are less frequent than before but less predictable, and much quicker and more intense when they come. Automation also leaves the pilot out of the loop: monitoring rather than controlling, then handed a confusing situation, with an alarm, at the moment the automation gives up (see flight deck automation).

Accident investigations show both the cost of startle and the cure:

The pattern is consistent. Where the first response was an instinctive control input or a hurried action, the outcome was bad; where the crew stabilised the aircraft, divided the work and took the time the situation allowed, even an extreme event was survivable (see human factors accident case studies).

Expectation and loss of situation awareness

Surprise is a failure of expectation, and expectation is built into perception. Situation awareness, in Endsley's model, has three levels: perceiving the elements of the situation, comprehending what they mean together, and projecting what will happen next (see situational awareness). A pilot's mental model of the flight drives all three, and it resists change: the brain completes incoming information from context and expectation, and confirmation bias favours evidence that fits the model over evidence that does not. At Tenerife, expectation turned an unclear message into the clearance the crew wanted to hear.

When the world departs from the model, the pilot must notice the mismatch, discard the model and build a new one, all while the aircraft keeps flying. That is why surprise costs time even when nothing is loud. Automation adds a common case, automation surprise: the aircraft does something the crew did not expect, usually because of a mode they did not select or did not understand, and someone asks what it is doing now. The recommended response is not to reprogramme but to reduce the level of automation and fly the aircraft.

In a two-pilot crew the other pilot is a resource. The pilot monitoring, less loaded at that instant, may keep a better picture and should say what he or she sees. A shared mental model, built by briefings and kept up by callouts, means that one pilot's surprise can be caught by the other, because it is unlikely that both hold the same wrong picture. The reverse also holds: two startled pilots who act without communicating, as when both made inputs at once on Air France 447, multiply the problem.

Managing startle in flight

Some events allow no pause. A TCAS resolution advisory, a terrain pull-up warning, a windshear escape or an approach to the stall each has a trained, memorised response that must be flown at once, and training aims to make that response the reflex.

For everything else, the priorities are the familiar aviate, navigate, communicate, and a short sequence helps the pilot get there:

1. Recognise the startle. Naming it, even silently, starts the recovery.
2. Fly the aircraft. Set or confirm a safe attitude and thrust and stabilise the flight path; with unreliable data, fly the memorised pitch and power.
3. Breathe and pause. A deliberate breath and a short pause damp the physiological response before any major or irreversible action.
4. Verbalise. Say what is happening and what you are doing; speaking re-engages reasoning and keeps the other pilot informed.
5. Apply the trained response. Memory items, then the checklist, then the decision.

Operators teach their own mnemonics for this, such as STAR (often stop, think, act, review, though the expansion varies), and the details vary by type. The principle is constant: except close to the ground, at take-off or on landing, time in flight rarely dictates a very quick decision, and crews can gain time by holding, slowing down, going around or delaying. Irreversible actions, such as shutting down an engine, are confirmed by both pilots first, the lesson of Kegworth, and any transfer of control is stated aloud.

Preparation reduces the startle before it happens. A briefing that decides in advance what the crew will do after an engine failure on take-off turns a startling event into the execution of a rehearsed plan. Rehearsal has its own trap, the error of commission, so the plan includes the check that confirms the cue before the action.

Training for the unexpected

Training cannot remove the startle reflex, but it can shorten the recovery and make the right response more likely:

• Upset prevention and recovery training (UPRT). Since ICAO's amendments of 2014, UPRT has been part of the training for multi-pilot type ratings and of airline training programmes. Under Part-FCL an advanced UPRT course, including dual flight in an aeroplane, is required before a first multi-pilot type rating, and on-aeroplane training exposes trainees to real g loads, disorientation and surprise that a simulator cannot reproduce. The FAA's UPRT guidance discusses startle and surprise, and its training includes unexpected scenarios so that pilots practise the first seconds of a response.
• Unpredictable simulator scenarios. A check in which every failure is known in advance trains the procedure but not the recovery from surprise. Evidence-based training deliberately builds unexpected events and distractions into line-oriented scenarios.
• CRM training. Surprise and startle are part of the EASA CRM syllabus, and after Air France 447 the BEA recommended training in the handling of surprise as well as in manual handling at high altitude.
• Overlearning and rehearsal. Responses learned beyond the minimum standard are more resistant to stress, which is why drills are practised long after the first correct attempt, and mental rehearsal and thorough briefings prepare decisions in advance (see memory and learning).

The aim of all of these is resilience: the capacity of a pilot and a crew to absorb an unexpected event, keep control of the aircraft and adapt the plan. No training promises that nothing will ever surprise a crew; it can make sure that a surprised crew still flies the aeroplane.

Library articles are written for study and exam preparation. They do not replace your aircraft's approved documentation, your operator's procedures or the regulations themselves.