A general framework for personal decision making

Jonathan Kahan
16 min readFeb 3, 2020


I recently moved to a new city with an extremely hectic real estate market. I braced myself for weeks of desperate searches and apartment visits. And as it happens, I fell in love with the very first apartment I visited. What a stroke of luck!

But wait, wasn’t I rushing it a bit? After all, if the first apartment I saw was so nice, maybe I could find an even better deal somewhere else? Before signing, I had to check my gut feeling, filter it through the lens of rationality.

So I listed other apartments I knew were available but I hadn’t seen yet. I added some hypothetical other apartments with some characteristics I was looking for. Even though those weren’t places I had seen, I assumed that upon searching I could find such places.

I then tried to get my priorities straight. What am I looking for in an apartment? Price is one indicator. The choice of whether to stay in a more bourgeois or a more hip neighbourhood would affect the trade-off between comfort and fun, as would the one between living alone or with flatmates.

So I chose cost, comfort and fun as the three key prioritisation drivers. I weighted them against each other and found that I cared about each of them equally.

I then tried to give a score to each accommodation possibility for each of the three drivers. Some of these possibilities, of course, were merely hypothetical, and would have required further searches to be actually uncovered, so I discounted those by a risk factor.

In the end it turned out that, despite the fact that the apartment I had seen involved trading-off some of the fun for more comfort, it was indeed the best choice. Few hours later, I signed my contract with the landlord, and I haven’t looked back since.

I am particularly happy with this decision because the process that I applied to make it was such as to take into account both my gut feeling and logical reasoning, completely eliminating the feeling of regret that, as a child of the FOMO generation, I would undoubtedly have experienced if I had only taken reason or feeling into account.

This process prompted me to think about decision making in general. Decisions in personal life are all very different, and some of them don’t look anything like choosing an apartment. Is it somehow possible, however, to come up with one, encompassing framework for decision making?

Decision and regret

Whether we are picking stocks, choosing where to live or whether to get married, we almost always make decisions with incomplete information. We try to optimise the final outcome, but the number and entity of confound factors is such that despite our best efforts, the outcome is still most likely going to be different from what we expect. For example, after a thorough analysis, I may come to the conclusion that it is indeed worth it to rent apartment x. After sleeping there for the first night, however, I may realize that the neighbors throw a crazy party every Wednesday making sleeping impossible. Incomplete information right there.

On the other hand, many of the things that truly affect our life in a deep way, we somehow do not recognize as decisions. These include mainly things that we are NOT doing. For example, the thought of learning to dance salsa may have never crossed one’s mind, until one day it does and it changes his/her life. The abstract possibility of one starting a salsa course was always present. So in a sense, every single day up until that point he/she has been making the decision not to learn salsa.

This is why it’s so hard to say we can improve our decisions in absolute terms. Our spectrum of consideration is necessarily limited, and once the decision is taken, most likely there will be factors that weren’t accounted for.

What then can we expect from a decision making framework? Actually, mostly to minimise regret. Whatever the outcome of our decision, we want to be in a condition in which we can say “I took the best decision I could have under the circumstances I was in and with the information I had”. By being able to say this, we are not improving our present situation. But we definitely live better with the consequence of our decisions.

So if it’s hard to say that we can definitely improve our decisions by using a framework, we can definitely find a framework that can help us be happy, be at peace with our decisions.

Free Will

Before starting with the framework, let’s get the issue of free will out of the way.

Someone may ask whether we are making any decision at all, given the fact that any rational reasoning has been often shown to just be rationalising a gut decision. At an even deeper level, one may question free will in the first place, given that any decision we take will necessarily be the consequence of previous causes. This is a fundamental philosophical debate, but it is actually irrelevant when making a decision. Whether or not there really is something as free will, in our daily life we are faced with decisions, and we do live with the weight of constant decision making. Even if free will were an illusion, most people feel like they rejoice for what they perceive to be the consequence of a good decision, and suffer for those of a bad one. Our goal in improving decision making is maximising the joy and minimising the regret. Whether free will is real or not, this will make for a happier life.
Ultimately, deciding is closing doors. It’s eliminating possibilities that were potential, and are now ruled out. If we feel that we are faced with a decision, we will inevitably bear the weight of responsibility for the outcomes, even though these may have actually bear little correlation with our decision.

The Framework,

Deciding self, Assessing self

As we have seen, hindsight plays a crucial role in this model, since only in hindsight are we able to determine the quality of our decision. But who is the person who will be assessing whether the outcome of our decision? This is not so straightforward.

  1. Most day-to-day decisions don’t change us. If I need to choose a brand for my new car, the parameters that I use to take the decision should be fairly clear (eg. price, comfort, engine performance, etc.) Once I bought the car, I will be able to assess the quality of my decision right away, based on those same parameters I’ve used in the decision phase. The deciding self is therefore the same as the assessing self. We can call this type of situation a reflexive problem.
  2. Some of the most important decisions in our life change us. We may not think we will enjoy being parents now according to our existing parameters (eg. amount of free time, hours of sleep, etc.), but the experience of parenthood will inevitably change what we consider a good decision. When assessing our decision to become parents once kids are born, for example, new parameters will inevitably emerge, while some of the parameters we have used when making the decision will fade into irrelevance. This feedback loop makes taking a rational decision very tricky: the deciding self is different from the assessing self. Just like the double slit experiment in quantum physics, post-hoc observation of the outcome changes the outcome. Should we be optimizing for our current self or for our future self?
    We can call this a prospective problem.
  3. There are also decisions in which the assessing self is the same as the deciding self, but in which the outcome of my decision is strongly influenced by the decision of a third party. To further complicate things, this often happens in recursive situations: I take a decision, based on which a third party will take his decision, based on which I’ll have to take a second decision… and so on. An example would be a chess match, or a prisoner dilemma type of situation.
    We can call this a reciprocal problem, or a game*.

Having established a more or less exhaustive model of the type of problems that prompt personal decision-making, let’s see how each of these can be analysed.

Reflexive problems

Reflexive problems are the simplest kind of problem, involving only my present self. There are two key variables that define this type of problem:

  1. The first variable is what I care about. For example, when choosing a car I will have at least three or four things I care about: performance, price, safety, etc. We may call these drivers or parameters.
  2. The second variable is the number of options I have available: am I only choosing between two cars or do I have a whole range to choose from?

A good way to start to tackle these problem is therefore to list all drivers (sometimes done as a pros-cons list) and all options. This may not be as straightforward as it seems:

  • For what concerns drivers, it’s all too easy to get lazy and just consider what other people or “common sense” value as drivers. Drivers are highly subjective, and it’s essential to take that into account. For example, when deciding whether to take a job, money is one very important factor for many people, but that may not be the case for me.
  • As for options, we need to consider the uncertainty factor. Some of the options may be clear, certain and available, while others may carry more uncertainty. For example, I may know of a car I could buy that gives me a good price/performance combination, but I haven’t found a seller yet. It’s useful to list both certain and more unlikely options, but we need to remember that and keep it into account.
    Also, sometimes options are not discreet, in the sense that they conform as a spectrum rather than clear choices. In this case I must first find a way to discretize them. This is the case for example when picking dates for flights when having a whole month available for my trip. I should first choose a few combinations of outbound and inbound flights, and compare those to limit complexity.

When all drivers and options are listed, it’s time to give a weight to each driver. Drivers are compared to each other one by one in a matrix to evaluate to what extent we value on more than the other. If you are uncertain, it’s good to bear in mind that each parameter has only a limited influence, and that the entire decision will likely have less consequence than you think.

We then proceed listing all drivers with their weights on the x axis of a matrix, and the different options on the why. We give a score to each option for each of the parameters and we apply a discount factor, eg. -20% or -50% if the option has some measure of uncertainty. We sum all the scores for each of the options, and here we go! we have our rational decision.

As illustrated in the opening anecdote, I think this type of prioritisation works best in real life when used to test our gut feeling. It’s good to get a sense of what we would instinctively do. Whether that proves rational or irrational, we do want to be aware of that. Depending on the situation, it may end up ultimately guiding our decision anyways, despite the result obtained with the prioritisation (often the case in things like relationships, where we sometimes consciously go for the irrational option); or it may serve as a future reminder of our biases. Either way, we’ll make our decision with clarity and awareness, and whatever the outcome, we’ll minimise regret.

To summarise: Reflexive problems are best tackled by clearly identifying options and drivers, weighting the latter and then giving a score to each of the options while discounting for uncertainty.

Reciprocal problems

Reciprocal problems are perhaps the most theorised of the three types of problems we have seen: there is an entire discipline, game theory, covering this type of situation. Game theory is deep, complex and highly quantitative, so I won’t cover it here. As general advice, however, we can take two rules of thumb:

  1. Look for equilibrium.
    In game theory, non-cooperative games have their solution in Nash equilibria. A Nash equilibrium is a situation in which each player plays his/her optimal strategy taking into account the decision of the other player. In day to day life, looking at what third parties will do, and trying to predict their next move based on our own move is often overlooked common sense.
  2. Look for win-win, reframe zero-sum.
    Zero sum games are situations in which whatever one party takes, the other party looses. Winner-takes-all is rarely a good logic to follow in person-to-person interactions, particularly when we can expect that the interaction will be repeated over time. Therefore we must strive to transform zero sum games into positive sum by thinking out of the box bringing to light the hidden, unstated needs of both players.

Let’s examine one reciprocal problem.
I’m hosting a party with some friends and I’m serving cake. I love cake. Should I cut myself a bigger slice, knowing that my friends will get less?

At first sight, this looks like a zero-sum game: the more cake I take, the less there will be for others. It is easy however to transform the situation:

  1. Will I really be only playing once? If these are my friends, we’ll certainly get together again. If I cut myself a bigger slice today, when I’m gonna be someone else’s guest tomorrow, he will take the biggest slice, and so on. This is clearly not a pleasant situation to be in.
  2. Assuming that all I want is to maximise my slice of cake, are we actually sure that everyone in the group likes cake equally? Some of my friends may like pralines better. So I may propose to get the pralines out of my refrigerator and share them, so that those who like cake may have more cake and those who like pralines will have those.
  3. More importantly, is this game really about cake-maximization? For most people, it obviously won’t be. We’ll want to maximise our own happiness from eating cake, but also our friend’s happiness from eating cake. We might even want to maximise our own happiness from our friends having a good opinion of us, from considering us generous, and so on. Therefore, I may decide to sacrifice some of my cake and prioritize things I value more.

Beyond the triviality of this example, asking “what is the game really about, what do other players really want” is a great way to reframe apparent zero-sum games into positive-sum games. This kind of reasoning is routinely used in negotiation situations, but it arguably has an important place also in personal decision making.

But wait, there is a reason why we don’t think of most reciprocal decisions in personal life as games. Since the interactions we care about are mostly with friends, family or acquaintances, there are two additional concepts that come into play: fairness and identification.

  1. Reciprocal problems involving identification are tricky because, while the deciding self and the assessing self may be the same, the self goes beyond my mind and body, to include one or more other people, who’s outcome I may equally want to maximise.
  2. While in interactions with close friends and family we don’t tend to think in terms of game due to identification, most people have a sense in which personal outcome maximisation in non-hostile environments needs to be balanced by a fairness principle. After all, barring tough negotiation cases, living in civilized society we usually deal with people just like us; even when we don’t expect our interaction with the other person will be repeated, most of us intuitively expect that most interactions can and should be pleasant for both parties.
    Fairness can have two interpretations, one deontological and one utilitarian.
    - Deontological interpretation: without getting deep into the philosophical tradition, a good principle to navigate reciprocal problem with fairness is Kant’s categorical imperative, or the simpler golden rule dating back to the Bible and other religious traditions: do unto others as you would have them do unto you. The shortcomings of this rule are well known, but in most situation in life the principle holds.
    - Utilitarian interpretation: The utilitarian tradition is about maximising the best possible outcome for the larges possible number of people. It’s a very different conception of fairness that in some extreme cases can lead to very weird outcomes (should my save my daughter from the fire or a Picasso that, once sold, could save thousands of children’s lives?), but that in most cases can really do wonders. Let’s see another example.

My sister has asked me to do a chore as a favour, for example to go pick up a sensitive document for her at the post office. I am super busy with work and going to pick up the package may cause me some real hassle down the line. I could ask a friend who I know is less busy to do it for me, but I know this would slightly disappoint my sister because she has asked me to do that for her.

I am faced with a choice: I could either consider my sister as an extended part of me thanks to the principle of identification, and go get the package. Or I could use utilitarianism and come to the conclusion that asking my friend to pick up the package will maximise happiness for everyone.

If I don’t identify with the other person, but still want to be fair, I could get away by trying to blend a quasi-deontological principle with an utilitarian one, by asking the following question: If I used a utilitarian principle, would the other person understand it? Would I make it a general rule that in this situation one use a utilitarian principle?

To summarise: when facing a reciprocal problem, we first have to ask whether the identification or fairness principles apply. If they don’t, it’s a game. In a game, we should try to understand the other player’s point of view as clearly as possible in order to find the things that matter to them that we can concede at little cost to us, thus turning the game into positive sum.

Prospective problems

Prospective problems are the toughest case. How can I even take a decision if the person who will experience and evaluate its consequences is inscrutable?

To add to the challenge, while day-to-day decisions usually concern reflexive problems, prospective problems tend to show up in situations were decisions are truly life changing: should I go for a major career shift? Should I get married? Should I have children?

The best we can do in many cases is just to be aware of our gut feeling, collect all the data available and act based on the existing information, hoping that things will turn out for the best.

At a more attentive analysis, however, there’s more we can do. Let’s start by subdividing prospective problems into two kinds:

  • Some problems leave room for experimentation and iteration. At some point between the shift from present self to future self, you’ll have a chance to change your mind and go the other way. We call these reversible problems.
  • Other problems are less forgiving: once the decision is taken, the consequences will stay with us no matter what. We can call these non-reversible problems.

Reversible problems should be approached the way we approach complex situations whenever we can: by iterating our decision and testing out different solutions. For example, if I’m unhappy with my current career there’s a lot of ways I can test alternatives before fully committing to a shift: I could

The career example is useful also to explain another concept. If I’m considering quitting my job because I’m unsatisfied with my career, I could be tempted to consider it a non-reversible problem. After all, most companies won’t hire again an employee who quit. On the other hand, in the larger scheme of things, a career as a whole is indeed reversible. As long as I have clear what I consider my success factors to be, I can get back on track. This is why, especially early on, it’s worth to experiment with one’s career.
Obviously there are costs related to every iteration: starting from scratch is daunting and often involves a steep learning curve. This has to be kept in mind when thinking of how many iterations we’ll need to get the problem right.

Non-reversible problems can seem daunting. Here more than anywhere else it’s important to be forgiving with our past self when assessing our decision ex-post. Some things we just couldn’t know. But there are few techniques that can help us, if not improve our decision, at least make clarity.

  • First of all, we can try to reframe non-reversible problems as reversible. While decisions such as donating an organ are inevitably non-reversible, some other decisions are on the boundary. A good example is getting married. We may take it for granted that there’s no way back because that’s how marriage has been considered in most societies most of the time. On the other hand, we can also choose embrace the idea that marriage, like many other things, is related to a specific phase of our life. It works now, it may not work in the future, and that’s ok: we’ll find another arrangement. Some people seem to indeed take marriage as an iterative process and they are often just fine.
  • If we assume we only want to get married once, we can use a formula devised by a clever mathematician in the 1960s to solve the so-called secretary problem: sample 38.6% of the total possibilities you may want to try in a lifetime, then choose the first option that’s better than any of those you’ve seen. The math behind the specific 38.6 number is way too complicated for me, but it makes a great deal of logical sense to try about a third of the options and then choose the best you’ve seen so far.
    This can be applied to a range of situations in which you have a chance to see a few samples before you settle for one option.
  • If none of the above options apply, try to think in terms of values. This can be broken down into the following questions:
  1. Do I want to be the kind of person who makes this decision?
  2. To what extent will this decision change my values? How will they likely change?
  3. Do I hold any values which I don’t expect will ever change? In which directions do these values point me?

To conclude, here is a simplified decision tree for your decisions:

The decisions decision tree
  • Note 1: Reciprocal and prospective problems are not strictly speaking mutually exclusive: a decision in which the assessing self is different from the deciding self can, and often will, have its outcomes influenced by third party decisions. However, I argue that it’s worth it to consider it as its own category, as its prospectivity feature adds a layer of uncertainty such as to minimize the relative weight of the third party’s decision.