As I was thumbing through the initial chapter of Nudge, a great book on behavioural economics by Thaler & Sunstein, some of the arguments they proposed irked me. The first chapter explains the three original heuristics of behavioural economics- anchoring, availability and representativeness. I will try to point out a few logical inconsistencies that came to me while reading this. Of course, neither have I finished the book yet, nor reached even close to the authors’ understanding of behavioural economics, so take the arguments presented here with a grain of salt.
The first heuristic- the anchoring heuristic- is described as the irrational tendency of people to, when asked to guess a value, attach their guess to a number that holds weight in their mind, even though the two values are logically not connected to each other. Their guesses tend to be much closer to their “anchoring” values than the mean. My problem with this heuristic is that if people are asked to guess a value they don’t know, technically each value they could guess has an equal probability of being right, which is why they might as well make a guess close to the value they already know, even they’re not logically related. This is at least the thought process I went through.
The second heuristic- the availability heuristic- is explained as the tendency of people to make judgements based on the most recent experiences they can draw upon. The example the authors use to illustrate this is the that of flood insurance, stating that people tend to buy flood insurance right after a flooding, and that spending on flood insurance decreases as the most recent flood becomes older, regardless of their actual flood risk. The issue with this example, and the larger heuristic, is that (to borrow the example of flood insurance once again) people will obviously judge their flood risk based on the most recent flood, because as the time passes, the actual possibility of a flood decreases, whereas when a flood occurs, it serves as a testament to the flood risk in that area*.
The third heuristic- representativeness- is explained through the famous proposition of Linda- a philosophy major who is single, outspoken, and deeply concerned with issues of discrimination and social justice. The example asks you to pick- is Linda a bank teller or is Linda a bank teller who is active in the feminist movement? Most people pick the latter, even though obviously being a bank teller is more likely than being a bank teller AND being active in the feminist movement. The issue I have with this is that it, once again, is very vague in its description of the subject’s thought process. When people are presented with (what seems like) a binary choice- they assign the first option of a bank teller with NOT being in the feminist movement, because the other option is a bank teller who IS in the feminist movement. And now, they estimate the probability of Linda, an outspoken philosophy major who cares about social justice, being in the feminist movement versus not being in it, and naturally choose the former. In essence, the first option is not clearly defined to the subject, and they assume that the first option means not being in the feminist movement, because it seems like a binary choice between feminist and non-feminist. The “bias” present does not seem to actually exist, as if the options were clearly defined, people would pick the first**.
Ultimately, my main issue with these “heuristics” are that they are not actually contradictory to the idea of human behaviour being rational and they also seem very vague descriptions of biases in human behaviour.
Again, take the arguments presented in this article with a grain of salt. My views could change after I finish the book, or I might be enlightened by some of the further research I plan on doing (because I’m sure I’m not the first one to have raised these criticisms).
* I am aware that I haven’t explained my criticism very well here, because the argument I was making does not read well as I edit this, but this paper by Wanke, Schwarz and Bless explains it much clearer, but is also denser- https://www.sciencedirect.com/science/article/abs/pii/0001691893E0072A
** I haven’t conducted an actual to back up this claim, although it makes logical sense to me. The original “Tom W.” experiment conducted by Kahneman to prove the representativeness bias is also plagued with issues, because while the number of humanities students is greater than those that took computer science, the probability of a person like Tom W (neat, very intelligent) being in computer science is greater than being in humanities. Here’s the paper if you’re not familiar with the study I’m talking about (Tversky, Kahneman, 1973) — https://content.apa.org/doi/10.1037/h0034747