Tunnel Vision and Blame Attribution (Post 3)

Confirmation bias. Tunnel vision. Whatever you call it, the phenomenon of collecting information to support or investigate a particular hypothesis regarding a to-be-investigated event can compromise objectivity and lead to erroneous conclusions. Although Popper solved the problem of confirmation bias for science, the problem is far from solved in the minds of the general population. One domain in which this effect is especially apparent is industrial accidents - one must shift from a worker role into an investigator role, and any number of cognitive biases enter into play, such as the Fundamental Attribution Error.

How, then, can we remedy tunnel vision and prevent undue bias in investigations? Education is one idea - if people are encouraged to consider alternatives and adopt a pseudo-Popperian mindset, by which they attempt to falsify their preconceptions rather than confirm them, a reduction in bias may result.

In this experiment, participants read about an industrial accident after being assigned to one of two bias conditions - they were given an initial description that blamed an accident either on an incompetent worker or on a piece of faulty equipment - and were educated or not educated on the phenomenon of tunnel vision, hypothesis testing, and the consideration of alternatives. They were then asked to allocate blame. Unsurprisingly, people tended to blame whoever they were intially biased against, but education played a very small role - it failed to interact with bias.


Additional evidence was then provided supporting either the worker or the equipment as the responsible party, and participants were asked how supportive the evidence was of their initial blame allocation. Education had no effect on blame allocation, interestingly, and people tended to consider additional evidence as slightly supportive of their original hypothesis - no matter what the original hypothesis was!

Finally, more blame allocation was asked for. This revealed a large effect, whereby people initially biased against the worker who were educated in the FAE and tunnel vision revised their initial attribution of responsibility such that the worker and the equipment were equally at fault. Conversely, people with an initial anti-equipment bias tended to shift blame onto the worker if they were educated.

Initial biases are clearly resilient - though education has some effect, people who are aware of the fundamental attribution error are still all too likely to commit it.