If you look above you can see the bias codex. If I remember correctly, this is the 2016 model. It looks beautiful. It is beautiful. It's also a bit much. A bit much doesn't even begin to cover the 2020 model. I did not use that image because it's ugly. No offense to neither it nor its creator(s), but with over 180+ biases in list form, no longer aesthetically surrounding the brain in a circular shape, it's just ugly. And overwhelming. I often ask myself: how many biases do we need? I'm not a keen believer in a Grand Unifying Theory of human behaviour, because I think context plays too much of a role to make that viable. It's not just context either. There is a large role to play for culture and its interaction with individual personality traits such as extraversion, but also risk attitude and temporal reasoning (e.g. present bias).
Maybe the question "how may biases do we need", is the wrong question. Maybe a better question is: how many biases are too many? And to be quite frank, I think we might have gone past that number already. Why are there so many biases to begin with? Well, given that behaviours can differ wildly depending on context, culture and personality (to mention just a few factors), there are many different behaviours, or ranges of behaviour, to explain. This doesn't just pertain to biases either. Most behavioural science enthusiasts are aware of temporal discounting: the present is worth more than the future. But how much more? We might all agree on this idea, but it can be measured in several different ways: exponential, hyperbolic and quasi-hyperbolic discounting. All three measures have received scientific testing and support, although the general agreement has come to favour quasi-hyperbolic discounting, where the present is vastly different from t+1 (one time period from the present), but there is a similar difference between moving from time 2 to time 3, or from time 100 to 101. The focus remains on the present. Why am I giving this example? Well, biases are quite similar. Unsurprisingly, given the amount of biases we currently have, some biases try to explain the same phenomenon, or share (massive) overlap. Good examples are any biases related to loss aversion. Examples of these are the endowment effect, which is linked to reference dependency and can be argued to be grounded in risk attitude (risk aversion) as well. A similar connection can be made between the self-serving bias, the optimism bias, the present bias and the planning fallacy, or the planning fallacy and the exponential bias; these work quite well in explaining why we are so inaccurate at estimating how long something will take. Also keep in mind that in many of these cases the word "bias" isn't explicitely mentioned in the title, they can also be addressed as effects, fallacies etc., but they remain to be biases. You can try this yourself as well: pick a single behaviour that you would like to explain, and see how many biases you can find that (somewhat) relate to this behaviour. You can even make this a competitive game, playing together with your fellow behavioural nerds. Make a card with a behaviour on it, and the person who can come up with the most biases related to the behaviour (and making their case convincingly) wins. Good luck!
So far I've been utterly inept at answering either the question: "how many biases do we need?", nor have I answered the question: "how many biases is too many?" Given that there is a wide range of behaviours to explain, I'm not sure how many biases is too many. And it might depend on the amount of behaviours we actually want to explain. But you have to admit that 180+ biases seems to get slightly out of hand. How did we get here? Why are there so many damn biases?! I think the latter question is one I am able to answer. Its roots lie within the god awful publishing system where novelty and shock value are king. What do you think is easier to publish: an academic piece of work confirming the existence of a bias, or an academic piece of work finding a whole new bias? Think about it...
It's obviously going to be the latter. A new bias is great. The thing is, what I'm starting to notice, and I have a feeling you will have noticed it too, is that these biases often aren't as new as they are made out to be.
So the biases are growing and growing, and it will not be much longer before the list gets even longer and, according to me, uglier. Moreover, there is no real incentive to stop this growth, because everyone wants a nice publication and a bias associated with them. I can't really blame anyone there.
But now that I'm on my soapbox anyway, I would like to make a remark with regards to this research: a lot of this research is great at identifying a (seemingly) new bias explaining a behaviour pattern we already recognise, but so what? A lot of the research on biases doesn't go much further than the recognition of a "deviant" pattern. Biases have been gaining recognition since the 1970s, I'm going to need a bit more than that...
Telling me you've just found out yet another bias isn't going to cut it. The more biases we name, the less value the "new" bias will have, but also the less value our entire collection of biases we will have. Behavioural science is not a naming game. What does it say about us if we can come up with 180+ biases, but cannot make all 180+ of them workable?
With workable I mean: recognising them, actively counteracting them, designing around them (incl. UX), moving them into AI and machine learning. If we cannot do that, we're really no better than the average 2 year old pointing at a tree and calling it by its name.
So long behavioural science.