PSYCH 644 Week 2 Attention and Perception Example

Why Inspectors Miss the Rare Crack: Attention, Perception and Visual Search on an Aerospace Parts Inspection Bench

[Student Name]

University of Phoenix

PSYCH/644: Psychology of Learning and Cognition

Week 2 Assignment

[Instructor Name]

[Date]

The aerospace shop, its inspectors and defect data are composites written for a model paper; research findings come from the sources listed.

Attention and perception determine what reaches awareness from the flood of information around us. They are efficient but limited, and their limits produce predictable errors. This paper analyzes one such error at my workplace using research on visual search, perceptual load and divided attention.

The Problem

Saguaro Precision in Tucson machines brackets and housings from titanium alloy for jet engines. Every finished part goes to final inspection, where one of six inspectors examines surfaces under a lighted magnifier for cracks, burrs, tool marks and dimensional problems before parts ship. Inspectors handle about 350 parts per shift. Cracks, the most serious defect, appear in about one part in eight hundred. Last quarter, an engine manufacturer's receiving inspection found a hairline crack in a bracket that had passed our inspection. No one was hurt, but the finding triggered a customer audit and a review of our inspection process.

The inspector who passed the part had eleven years of experience and a strong record. In practice tests with defects planted at higher rates, she found nearly every crack.

How Perception Works

Perception combines bottom-up processing, driven by the features of what we see, with top-down processing, shaped by knowledge, expectations and goals. Experienced inspectors use top-down knowledge to recognize defects quickly: a crack has a characteristic thin, dark line that differs from a tool mark. But expectations also shape what we look for and how long we look. An inspector who has examined eight hundred parts without finding a crack expects the next part to be clean.

Rare Targets Are Often Missed

Wolfe et al. (2005) asked observers to search images of baggage for a target object, varying how often the target appeared. When targets appeared in half the trials, observers missed about seven percent of them. When targets appeared in only one percent of trials, observers missed about thirty percent. Analysis of response times showed that observers ended searches more quickly when targets were rare, deciding "nothing here" before searching thoroughly. Telling observers that rare targets were important did not eliminate the effect.

The finding maps directly onto our inspection bench. Cracks are far rarer than one percent. Our inspectors' performance on practice tests, where defects were common, overstates their performance on real production, where defects are rare. The miss was not mainly a matter of skill or effort; it reflected a systematic feature of searching for rare targets.

Eleven years of experience could not overcome the fact that, eight hundred times in a row, the right answer had been "no crack."

Perceptual Load and Selective Attention

Lavie (2005) reviewed research on selective attention under load. Perceptual load theory holds that when a task demands most of our perceptual capacity, irrelevant information is excluded, but when load is low, spare capacity spills over to distractors. Load on the mind's executive functions works the other way round: keep working memory busy with paperwork and questions, and stray sights and sounds slip in more easily. That split tells a manager which distractions to remove first.

Our inspection bench sits beside the deburring area, with grinders running, a radio playing and a production board showing parts waiting. During busy periods, inspectors also track paperwork and answer questions from machinists. These demands load cognitive control, which Lavie's review suggests increases interference from distractors, and they encourage quick decisions.

Dividing Attention

Wickens (2008) reviewed multiple resource theory, which proposes that tasks interfere with each other more when they draw on the same resources, such as visual versus auditory input, spatial versus verbal processing and manual versus vocal responses. Two visual tasks interfere more than a visual and an auditory task. The theory has guided the design of displays and workplaces where people must do several things at once.

Visual inspection competes most with other visual demands, such as glancing at the production board or reading paperwork, and less with listening. Questions from machinists, which require both listening and looking up from the magnifier, interrupt the visual search entirely.

Fatigue and Vigilance

Inspection also requires vigilance, sustained attention over long periods for infrequent signals. Vigilance research has long shown that detection declines over time on watch, particularly with rare signals and monotonous conditions. Our inspectors work eight-hour shifts with a lunch break and two short breaks, often inspecting for two or more hours without interruption.

Training Inspectors to See

Perceptual expertise can be trained. Experienced inspectors learn to tell a crack from a scratch partly through years of exposure, and newer inspectors can build the same recognition faster with a library of examples. We have kept scrapped cracked parts for years without using them for training. A reference set of real defects and look-alike marks, reviewed monthly by all inspectors, would sharpen the top-down knowledge that guides search, and pairing it with feedback would show each inspector where their own judgments drift.

Lighting and Magnification

The physical setup also matters. Our magnifiers have fixed lighting from one side, which makes some crack orientations nearly invisible. Adjustable, multi-angle lighting, standard in many inspection settings, would make surface features appear more consistently and reduce the perceptual difficulty of each search. This change addresses bottom-up processing, making the signal stronger rather than asking the inspector to work harder to detect a weak one.

What Will Not Work

Telling inspectors to "be more careful" or adding a second inspection with the same conditions would likely help little, since both inspectors would face the same prevalence effect. Disciplining the inspector would ignore the evidence that the miss reflected predictable limits.

Recommendations

First, raise effective prevalence by inserting known defective parts, previously scrapped parts with real cracks, into the inspection stream at a rate of about one in fifty, with immediate feedback when an inspector finds or misses one. Research following Wolfe and colleagues' work suggests that increasing target frequency and providing feedback can reduce rare-target misses. Second, set a minimum inspection time per part based on surface area, so searches are not ended too quickly. Third, move the bench away from the deburring area, remove the production board from inspectors' line of sight and route machinists' questions through a lead. Fourth, limit continuous inspection to forty-five minutes, rotating inspectors to other tasks between blocks. Fifth, consider dye penetrant inspection for the most critical parts, which makes cracks visible under ultraviolet light and reduces reliance on unaided search.

Measuring Results

We will track detection of planted defects by inspector and time of shift, customer returns and inspection time per part.

Limits

Laboratory baggage-screening tasks differ from inspecting metal parts, and our crack rate is even lower than in the experiments. The recommendations should be tested and adjusted.

Conclusion

Skilled inspectors missed a rare crack for reasons research on attention and perception predicts: rare targets lead searchers to quit early, cognitive load and visual distractions narrow attention and vigilance declines over time. Changes to the task, the environment and the frequency of feedback are more likely to help than exhortation.

References

Lavie, N. (2005). Distracted and confused?: Selective attention under load. Trends in Cognitive Sciences, 9(2), 75-82. https://doi.org/10.1016/j.tics.2004.12.004

Wickens, C. D. (2008). Multiple resources and mental workload. Human Factors, 50(3), 449-455. https://doi.org/10.1518/001872008X288394

Wolfe, J. M., Horowitz, T. S., & Kenner, N. M. (2005). Rare items often missed in visual searches. Nature, 435(7041), 439-440. https://doi.org/10.1038/435439a