Evaluating Mental Models
This post will help you determine if you are applying the right mental model to your problem, and what to do if you are not. What follows is a mental model I created to help evaluate the application of mental models. Consider the various kinds of flashlights as they relate to mental models:
- Inoperable Flashlight - A poorly chosen mental model. It sheds no new light on the problem.
- Weak Flashlight - A shallow mental model. It reveals an insufficient understanding of the total environment, exposing only the most obvious details.
- Penlight - A misleading mental model. A penlight gives you a bright but narrow beam of light which reveals the environment in slices. The clarity of the slice can mislead you to believing you’re seeing everything. This model is right in only a small number of cases, or right for the wrong reasons.
- Flood Flashlight - A well-chosen mental model. The flood light has a wide beam that projects deeply into the darkness and at the periphery of your vision. This light gives you a holistic understanding of the environment. An alternative to the flood flashlight could be a flashlight in each hand e.g. using more than one mental model to get a more complete understanding of the terrain.
I created the flashlight model in reaction to the trend of people claiming that mental models, on their own, will help you make better decisions. If someone says that to you, be wary. Mental models are like hammers. If I told you hammers will help you make better houses you know instinctively that it also depends on how skilled the craftsman is, and what they’re using the hammer for. In my experience, mental models are most effective when used together (in aggregate or in juxtaposition) as a way of framing the affordances, limitations and attributes of a problem. As Maslow’s saying goes, “if all you have is a hammer, everything looks like a nail.” Even if you could build a house with just a hammer (break the wood instead of sawing) no one would want to live there. A key to using mental models effectively is to fill your bag with a variety of tools (other mental models) that are fit for purpose, and then develop the personal experience to navigate the nuances of when and how to apply them.
Mental models are also necessarily reductive. For them to be efficient, they must helpfully narrow the unknowns of a problem by framing them against a system you already understand. When I say “companies work like machines,” that gives you an entire suite of metaphors and prior understandings about how machines work. You can then overlay those understandings onto companies to give you manufactured clarity on how they operate. Of course, because models are abstractions, this activity can be horribly reductive and lead to all sorts of negative outcomes. For example, developing processes that treat employees like interchangeable and disposable machine parts. When mental models are used incorrectly, they share a lot in common with biases which steer you (often unconsciously) towards a decision you sometimes forget is a choice.
There are a couple of reasons the flashlight model is compelling:
- The flashlight will never illuminate everything. There will still be ambiguity, and unknowns in the same way that a mental model will always be a simplification of the actual system.
- The model implies you will only find what you are looking for if know where to point the light in the right direction in the first place. You must actively scan with the model to understand the problem.
- Flashlights are tools used to explore the unknown or unfamiliar, and so are mental models. Since time immemorial light itself has been associated with knowledge and learning (e.g. age of enlightenment.)
I would love your thoughts on this post. Does the description and use of mental models resonate with you? Does the flashlight model make sense, and where are the limitations?
Note: After creating this post I found that a writer named Brian Lui also landed upon a similar metaphor to describe Mental Models. The nuances in our coclusions are different, but I felt I should link to his post since it predates mine: https://brianlui.dog/2018/02/15/wrong-models-are-good/