Viewers often fail to see objects, such as pedestrians and other vehicles. Such failures-to-see have great importance in safety as well as in many other domains such marketing, medicine, and sports, and elsewhere. I have discussed seeing in general and why failure-to-see is so common. Here, I provide a broad overview of the specific reasons for failures-to-see. I present these in tables that are adapted from the book Roadway Human Factors: From Science To Application (Green, 2017), which discusses them in detail. However, such categorizations inevitably begin to fray at the edges. They are didactic and an attempt at cognitively simplifying the large number of visual factors affecting seeing.
The reasons can very broadly be divided into visibility and cognitive. The first table shows 5 visibility factors. External physical factors are partial or complete sightline obstructions. Physiological factors can be optical, neural, or motor. Psychophysical factors usually boil down to those affecting contrast detection. Lastly, these five entries can readily be decomposed further. There are more than 16 subfactors in contrast alone (Green, 2017).
|Physical||External Obstruction||Physical obstructions completely or partially block the sightline and prevent retinal image formation: e.g. buildings A-pillars, windshields.|
|Physiological||Optical||Optical imperfections. Blur caused by mis-accommodation and spherical aberration. Cataracts and opacity.|
|Neural||Retinal limitations, e.g., field size, photoreceptor spacing, and scatomata.|
|Motor||Eye muscle movements also reduce vision, e.g., saccadic suppression and blinking.|
|Psychophysical||Contrast||Factors that determine whether an object differs sufficiently from the background to be visible: e.g., size, adaptation level, retinal eccentricity, etc.|
In many cases, however, the cause of failure-to-see is cognitive – the viewer does not consciously see an object which is theoretically highly visible. The usual culprit is attention. The failure to see is especially likely in given circumstances. There is vast a literature on attention and cognition that provides many potential explanations for failure to see in a given set of circumstances. The table below boils this literature down to a single list of 28 cognitive failure-to-see “constructs” that are again grouped into rough categories purely for didactic purposes. Further, the chart constructs require further elaboration.
|Cognitive (Learned Adaptation)||Expectancy||Attention tuned to the location and objects of highest anticipated meaningfulness, so others not noticed.|
|Automaticity||Tasks performed with minimal attentional control, so little attentional supervision to notice difficulties.|
|Familiarity Blindness||Failure to notice scene information that has been irrelevant in the past.|
|Cognitive(Loading)||Tunnel vision||Attention concentrated in the center of the visual field, so information in peripheral vision is unnoticed.|
|Inexperience||Novices have not yet learned to reduce foveal loading by chunking or by automatic control.|
|Stress Hypervigilance||Stress focuses attention on a narrow set of information, so other information not noticed.|
|Mental Workload||Attention consumed by one task leaves less capacity to perform others.|
|Cognitive(Selection)||Inattentional blindness||Failure to notice even highly visible objects located in the center of the visual field.|
|Change blindness||Failure to notice a feature alternation in a scene that has changed over time.|
|Mind wandering (Internal distraction)||Attention is focused internally rather than to the external world.|
|Repetition blindness||Second in a sequence of two identical objects is not noticed.|
|Foreground bias||Viewers tend to fixate objects in the foreground and to ignore objects in the background.|
|Cognitive (Background)||Clutter masking||Background contours interfere with attention to foreground objects. It is sometimes a psychophysical effect.|
|Overshadowing||Attention attracted to the most salient scene object and away from other information.|
|Cue generalization||Viewers direct attention to the most easily discriminated cue, such as color.|
|Crowding||Visible objects in peripheral vision merge and cannot be identified.|
|Motion induced blindness (MIB)||Object projecting retinally stable images tend to disappear when seen against moving backgrounds.|
|Cognitive(Search)||Satisfaction of search||Termination of serial search before reaching the critical information.|
|Inhibition of return||Once a scene area is searched, the probability of making a return saccade to re-search is reduced.|
|Visual space asymmetry||Different parts of the 3-D visual space are specialized for different attentional tasks.|
|Switching Costs/Attentional blink||Switching attention from one focus to another takes time and effort.|
|Cognitive(Reduced Capacity)||Fatigue/Lack Of sleep||Difficult to define, but usually explained as lowered performance due to lack of sleep.|
|Vigilance decrement||Ability to notice information falls within the first half hour in routine tasks.|
|Low Workload/Boredom/Monotony||Low arousal level with longer time spent in a dull and unchanging environment.|
|Circadian Rhythm||Arousal lower in troughs of the daily 24-hour arousal cycle.|
|Age||Older viewers are both slower and have lower attentional capacity.|
|Cognitive (Decision)||Biases & Heuristics||Attention and decision are guided by mental short cuts designed to increase efficiency.|
|Satisficing||Attention becomes unnecessary once a reasonable solution achieved.|
The cognitive construct list is long, but the phenomena overlap considerably because they are different manifestations of our fundamental cognitive architecture and its consequences:
1) conscious awareness is limited by capacity,
2) attention is selective,
3) attention is efficient, trying to use the easiest and simplest selection criterion,
4) attention generally selects information that is most meaningful based on past experience and on expectation of the future. This overlap is why I call them “constructs” rather than variables or causes1. The criterion for inclusion in the list is only that the authors frequently use the “construct” to explain a cognitive failure-to- see. In many cases, the construct name specifies little more than an experimental paradigm, e.g. “change blindness”.
Lastly, the constructs also represent different granularities of explanation. “Inattentional blindness” (failure-to-see central objects) and “tunnel vision” (failure to see peripheral objects) are really just descriptive terms for phenomena that are ultimately caused by more specific constructs (expectation, hypervigilance, mental workload, etc.). They can be useful as umbrella terms when merely describing the functional loss, but they are not very specific. They are symptoms of deeper causes. They are not explanations.
1 The definition of “causation” is a knotty philosophical problem. In practice, however, it is often merely a “symptom” that doesn’t require an explanation. Causal reasoning normally works through a chain where symptoms have causes which then become symptoms for other causes. Here’s an example. A driver is blamed because he responded too slowly in seeing a pedestrian at night. Slow perception-response time is a symptom and low visibility is the cause. Low visibility is then the symptom, and a burned out streetlamp is the cause. The burned out streetlamp is now the symptom, and poor maintenance is the cause. Poor maintenance is the symptom… and so on. The chain stops when the symptom need not be explained. In this example, a human factors analysis doesn’t need to know what caused the streetlamp to be burned out, so the cause need not be determined. For an investigating utility company, the causal chain would continue until the poor maintenance were explained.