Question about anchoring and adjustment (Tversky and Kahneman) and why anchoring often yields correct judgments

I am working on a problem set and I am having some difficulty wrapping my head around a concept introduced by Tversky and Kahneman. My problem set asks the question "Why does anchoring nevertheless often yield correct judgment?"

Based on their paper, I thought they were trying to say that anchoring often yields incorrect judgment. I've been thinking about it more and I'm not sure if I'm headed in the right direction.

Something they state in their paper is that people tend to overestimate conjunctive events and underestimate disjunctive events. Disjunctive events are associated with risk evaluation. As given by their example, a complex system, such as the human body, will malfunction if any of the essential components fail. Even if the liklihood of failure is slight, the probability of overall failure is high if many components are involved, however people tend to understimate the rate of failure in disjunctive events.

However, I believe that in this situation, anchoring does yield correct judgement because people do not think about the highly complex structures that the human body is made of. For the most part, we have control over our health and we can prevent from components of our bodies from failing. So by living healthy lives, we are more likely to live longer, therefore decreasing the chances of actual failure.

I don't know if I am on the right path or if my logic is off in my interpretation above.

• Why did you think they were trying to say that anchoring yields incorrect judgment? It is incorrect theoretically in controlled circumstances. But their big point is that in practical situations cognitive biases serve as useful heuristics, which explains why they are so pervasive. – Conifold Dec 16 '19 at 5:27

'Anchoring' merely means that people tend to base their actions and decisions on expectations projected from their immediate perceptions. If you know statistics, you can think of it as a Bayesian approach, where we calculate probabilities dynamically as we gain information. This tendency can be manipulated — by psychology researchers, salespeople, politicians, and whatnot — but in much of our life it is a perfectly adequate way of estimating outcomes.

If someone smiles at you, you will think they like you, and 98 times out of 100 you will be correct (unless it's a salesperson or politician). If a tiger turns and runs in your direction, you will assume it is inclined to eat you, because there aren't that many reasons a tiger would charge at you; there's not much point in gathering more data before you act.

Effectively, this means we have a salience bias. We tend to pay attention to things that are new, unusual, disruptive, shocking, high-valence, etc, and we tend to believe that 'normal' or 'routine' things will continue on exactly as they are, and discount potential changes. The conjunctive event bias occurs because once we see an event we expect it to continue appearing, and that anything that occurs after will be related to it. The disjunctive event bias occurs because we expect events to conform to a pattern, and discount the likelihood of non-conforming events.

With respect to your particular 'health' question, I believe you're overthinking it. People underestimate the risk of ill-health because their daily, moment-by-moment experience of life is health. When we get up in the morning and start to eat breakfast (or whatever), our anchor point is 'being alive'. It's normal, routine, regular, and we see no reason why it won't continue indefinitely like that. If we suddenly get a bad flu, of course, our anchor point changes; then (while we're sick) we start thinking about all the things that can go wrong with our body. This has nothing in particular do with thinking that we are in control of our health, except in the trivial sense that leading a 'healthy life' becomes part of our anchoring routine.

Why does anchoring nevertheless often yield correct judgment?

Probably because there's rarely a single correct judgement, and many choices are acceptable. A textbook example of anchoring is when people presented with an unrelated high number are then likely to agree to a higher price when haggling. Sure, this might be suboptimal, but in the end they still have a deal.

Same goes for health: if people overestimate the risk of cancer and go for checkups too often, that probably won't hurt anything except their wallet. If they underestimate the risk of flu and don't vaccinate, they will likely nevertheless survive.

"Living healthy lives" is certainly beneficial; the problem is that different people understand the concept differently. Lots of people adopt crazy and harmful diets based on incomplete and misleading information; anchoring might be a factor in some of the cases. Antivaccination movement is another example of catastrophically wrong risk assessment rooted in cognitive biases.