Question 1 of 7: Compatibility, Display Modality, and Signal Detection Theory

Question 1: Compatibility, Display Modality, and Signal Detection Theory (20 marks: i–6, ii–7, iii–7)

Question text not reproduced: the examination questions are © Engineers and Geoscientists BC. Open the official past paper (linked at the top of this page) to read the question, then follow the worked solution below.

(i) Types of Compatibility in Human Information Processing

Compatibility describes how well a control, display, or task matches an operator's existing expectations and mental model of "how things should behave" — a well-matched design is learned faster, produces fewer errors, and is used with less conscious mental effort than an equally functional but poorly matched one. Four types are commonly distinguished. Spatial compatibility concerns the physical arrangement of controls and displays: a control's position should correspond to the position of the display or system element it affects (e.g., the front-right burner control on a stove should be the one physically nearest the front-right burner) — mismatched spatial layouts force the operator to consciously map location to function rather than acting on reflex. Movement (or direction-of-motion) compatibility concerns whether the direction a control moves matches the expected direction of the resulting display or system response, governed by population stereotypes (clockwise rotation or upward/rightward motion of a control is expected to increase a displayed value or turn a system "on"). Conceptual compatibility concerns whether codes, symbols, and abbreviations used in a design match the meanings and associations a population already holds (red universally means stop/danger; a stylized fuel-pump icon means "fuel") rather than requiring the operator to learn a new, arbitrary code. Modality compatibility (also called stimulus-response modality compatibility) concerns whether the sensory channel of an input signal is naturally suited to the response channel it demands — a verbal/auditory instruction pairs naturally with a verbal/vocal response, while a spatial visual cue pairs naturally with a manual/spatial response; cross-modal mismatches (e.g., requiring a vocal response to a spatial task) slow information processing even when both channels are individually available.

(ii) When to Use Auditory vs. Visual Presentation

The choice between an auditory and a visual display follows a well-established set of situational guidelines (after Deatherage), because the two senses have complementary strengths. Use an auditory display when: the message is short and simple (auditory memory for long or complex messages is poor); the message will not need to be referred back to later; the message calls for immediate action or is time-critical (a warning or alarm); the visual environment is overloaded, too dark, subject to glare, or the operator's eyes must be free for another task; the receiving location or operator moves about rather than remaining at a fixed viewing position; and continuous monitoring of a visual display is impractical. Use a visual display when: the message is long, complex, or highly detailed (visual scanning re-reading is easy; auditory replay is not); the message will be referred to later or must be stored for reference; the message does not require an immediate response; the auditory environment is overloaded, noisy, or itself requires the ears free (e.g., verbal communication); and the receiving operator remains at one location where a fixed display can be viewed. In practice, safety-critical or time-critical alerts are frequently presented redundantly in both modalities — an auditory alarm to capture attention immediately, backed by a visual display carrying the detailed follow-up information.

(iii) Signal Detection Theory (SDT)

Signal detection theory models an operator's decision on any detection task (is a target present on a radar screen? does this weld show a genuine flaw? is that noise the alarm or background hum?) as an act of statistical decision-making under uncertainty, not a simple threshold "can-see/cannot-see" phenomenon. Both "noise alone" and "signal + noise" produce a spread (not a single fixed value) of sensory evidence, because of random variability in the stimulus and in the observer's own perceptual/neural noise — the two conditions are modelled as two overlapping probability distributions along a single evidence axis.

The observer sets an internal decision criterion (β) somewhere along this axis and reports "signal present" whenever the sensory evidence on a given trial exceeds it. This produces four possible outcomes: a hit (signal was present, evidence exceeded β, correctly reported), a miss (signal was present but evidence fell below β, missed), a false alarm (no signal was present, but noise-alone evidence happened to exceed β, incorrectly reported), and a correct rejection (no signal present, evidence below β, correctly reported as absent). Two independent parameters fall out of this model: sensitivity (d′) is the separation between the means of the two distributions relative to their spread — it reflects the true discriminability of the signal from noise (better equipment, better training, or a less noisy environment raises d′) and cannot be changed just by choosing a different criterion. Response bias (β) is where the observer chooses to set the decision line, and it is a free choice driven by the perceived costs and benefits of each outcome — moving β left increases hits but also increases false alarms (a "liberal" criterion, appropriate when misses are very costly, e.g., a cancer screening or a smoke-detector alarm), while moving β right decreases false alarms but also decreases hits (a "conservative" criterion, appropriate when false alarms are very costly, e.g., an automated shutdown that halts an expensive production line). SDT's key insight for a design engineer is that a raw "percent correct" or "hit rate" figure is meaningless on its own, because it conflates the two parameters — a high hit rate could reflect either genuinely good sensitivity or simply a very liberal (over-alarming) criterion, and only by also examining the false-alarm rate can the two be disentangled.