Cognitive Biases: How to Check a Conclusion

Suppose a colleague does not reply to a message until the evening. At first, it is simply a wait. Then an explanation appears: “He is avoiding me.” You remember yesterday’s brief greeting, an awkward joke last week and a meeting you were not invited to. Every detail seems to support the conclusion. His offer to help with a task that morning somehow drops out of the story.

This is one way a tendency to notice support for an existing explanation can appear. The individual facts remain real, but the selection is uneven. Understanding what is happening in the relationship means noticing both that selection and the information still missing. Silence during the afternoon does not, by itself, explain the colleague’s motives.

We continually judge people and events with incomplete information. Sometimes we use a simple guide: recalling a similar event, recognizing a familiar pattern or relying on an initial explanation. Such a judgment strategy is called a heuristic. It can help us act quickly, but under some conditions it leads to systematic errors, or cognitive biases.

Not every quick decision is wrong, and not every mistake is a distinct bias. A calculation might be wrong through inattention, or the information supplied might be inaccurate. The issue here is a recurring pattern in judgment: for example, the vividness of one incident repeatedly outweighing information about how often it actually occurs.

A classic 1974 paper by Amos Tversky and Daniel Kahneman describes several heuristics used in judgments under uncertainty and errors that can arise from them. It provides a useful basis for examining decisions, not a list of diagnoses that determines someone’s character.

The Ukrainian-language poster from the previous edition is retained below. Open it separately or download it to read the small labels. It can help you locate a topic; examining a particular decision still requires the circumstances of that decision.

Returning to the colleague, once “He is avoiding me” seems plausible, it is possible to collect suspicious episodes while barely looking for another explanation. Even a neutral reply can start to seem cold. The more effort invested in a story, the less welcome a detail that does not fit may become.

A useful question is what information would make me reconsider. Then ask whether there is a way to obtain it. “I sent you a question about tomorrow’s task. When might you be able to reply?” The answer may not be pleasant, but it supplies more information about the present situation than another reading of old messages.

An alternative explanation remains a hypothesis too. “He must just be busy” does not become a fact because it is reassuring. Both explanations can remain open while you decide what to do before clarification: continue the available part of the task, communicate a deadline or ask another project participant.

After hearing a friend’s account of lost luggage, travel may feel much riskier. The story is easy to recall, full of detail and recognizable frustration. That makes it available to memory, but does not reveal how often luggage is lost across all journeys. Estimating frequency requires comparable data over a defined period.

A small sample presents another difficulty. Two unsuccessful orders with one company justify investigating those orders, but do not establish a conclusion about an entire industry. How the cases were selected, how many exist in total and whether these differ from the rest all matter. An unpleasant experience need not be dismissed; it should not be made to show more than it does.

When a person resembles a familiar occupational image, we may also mistake resemblance for evidence of the qualities needed. A candidate’s confident speech recalls a successful manager, and checking their experience starts to seem less necessary. Resemblance alone does not show how this person has handled the tasks that matter to us.

The same claim appears in a feed, a friend’s account and a short video. Gradually, it feels known from several sources. Checking may reveal that all three accounts began with one unsupported publication. The number of repetitions is not the number of independent observations.

In experiments by Lisa Fazio and colleagues published in 2015, repetition could increase perceived truth, including when relevant knowledge was available. This does not mean that everyone will necessarily believe a repeated falsehood. It helps explain why recognition alone is insufficient verification.

Before sharing a claim, look for the original publication, its date and the distinction between observation and interpretation. If websites cite one another, trace the chain to the first report. When the original source cannot be found, “not yet verified” is more honest than adding confidence through your own retelling.

Ideas about age, occupation, gender or background can shape expectations before meeting someone. During an interview, this may show up as different questions for people with similar experience: one is invited to discuss achievements, while another is asked in advance whether they can handle the workload. The first impression begins to influence which evidence is collected at all.

In a work assessment, it helps to define the task requirements beforehand, ask candidates comparable questions and record examples of what they have done. This makes the grounds for a decision visible, although it does not guarantee complete impartiality. Saying “I treat everyone equally” does not replace comparing how the assessments actually proceeded.

Some assumptions can operate without clear awareness. “Unconscious biases” describes this feature, rather than naming another distinct error alongside availability or repetition. Knowing a bias’s name, whether applied to yourself or someone else, is not enough to establish its presence in a particular decision.

Choose a situation in which you are ready to settle on an answer. First, write down observable facts without explaining anyone else’s motives. Beside them, record your conclusion and what supports it: direct experience, someone else’s account, a repeated publication or comparison with a similar person.

Then name at least one alternative explanation and the information that could distinguish the possibilities. For the work message, that might be the colleague’s reply; for a news claim, the original source; for a candidate, a relevant work sample. Match the check to the importance of the decision. Buying bread does not require a research project, while a decision affecting someone else’s work or safety deserves closer examination.

After checking, you may change the conclusion, retain it with clearer grounds or acknowledge that there is not enough information yet. This is more useful than demanding that you never think automatically. What matters is keeping important decisions open to revision when something new appears.

CBT SHIFT separates facts, automatic thoughts, and alternative explanations.