I want to dedicate a few words to the anchoring or the harvesting bias, also known as the sunk cost bias. I will use these terms interchangeably. The bias tends to drive us to consider past decisions and investments that need not have any bearing on present choices we face.
I will first introduce the bias at an unnecessarily high level of formal language. I will then proceed with some examples, the sheer triviality of which might severely counterpose the texture presented in the formal introduction. I will finally diverge along a tangent not entirely inline (read: in total discordance) with the preceding content, at the whim of my imagination. (Is this not the idea behind interdisciplinary research?)
The slippery past
The past does contain valuable information. But how we handle that information makes the difference between learning and fallacy-prone decision making.
Observing past behavior with the respective past outcomes having already occurred is necessary (albeit not necessarily conclusive) in learning. We can use the particular trajectory leading from cause to effect to reinforce the behavioral choices we make in the present or the future. Monte Carlo simulations and Machine Learning, for example, construct models that rely on past information to approximate the future.
Past behavior in the context of decisions we have yet to make is an entirely different story. It may seem that a decision initiated in the past should have bearing on the present choices. But all the information necessary for making a rational decision is already encoded in the present. We face only present choices and the consequences they carry.
This leads to a key point. In rationalizing decision making, we place emotional emphasis on how much we experienced or invested into something in the past. However, that is pseudo-information, having no bearing on future consequences. Rather, it is a matter of reformulating the problem by incorporating the relevant consequences of past choices into the present, and then eliminating the past entirely from the problem, mitigating the cognitive trap.
For the probability-theory inclined, I will mention, in parentheses, that a Markov chain provides a precise mathematical framework maintaining this past-invariance decision making property.
How to be pedantic with decision making
I will list a number of examples having occurred to me or to people around me. I will indicate the cognitively misguided way of observing the problem, followed by the reformulation that considers only the present and the future. Most of these examples involve comparatively trivial situations that nevertheless drain unnecessary energy if permitted.
You’re in the middle of a public guided tour, but suddenly lose interest for whatever reason.
Fallacy: You paid for the tour, travelled to reach the group, and/or already invested some time into the experience, therefore you should bear it to the finish.
Reality: You can either lose time and energy on the tour you no longer care for, or abandon the group, save your energy, and gain more time to pursue more interesting endeavors.
Remarks: How much, if anything, you paid for the tour, or the time you already spent on it no longer factors into your decision. Those are sunk costs. The reality involves your present physical and emotional state, and how you wish to proceed. Bonus: free tours usually involve a tipping element towards the end. By leaving early, you save additional energy from the moral consideration of tipping for something you didn’t value.
You don’t really enjoy beer consumption, but win one at a raffle. Alternatively, you purchase one at a bar due to a momentary lapse of reason, or because it’s notably cheaper than other alcoholic beverages (which you might also not care for).
Fallacy: You already have the beverage and should drink it. Also, having a bevarage smoothes the social interaction and alleviates anxiety.
Reality: How you acquired the beverage is of no relevance. If you paid for it, there’s unlikely a refund opportunity. You face the choice of either finishing the beverage you find unsavory at best, compromising your health in the process, or abandoning it and cutting your (irredeamable) losses.
Remarks: How you feel and are perceived in a social setting is a product of your inner projection rather than beverage consumption.
You order a meal at a restaurant. (This is beginning to resemble a Voight-Kampff test.) You find the meal either slightly unsavory or of a serving size nothing short of fatal. After some moderate progress, you realize that continuing this meal will not contribute to your physical well being in any foreseeable way.
Fallacy: You paid for the meal. If you don’t finish it, it will be wasted. You don’t know the next time you might eat. You don’t want to bother with a take-out box (clearly out of concern for natural resources.) You are tormented by chastising childhood memories of unfinished meals.
Reality: As far as I know, the food will biodegrade either way, so waste shouldn’t be of concern. The only concern you should have relates to inner-energy and physical and mental performance - and likewise, not how much or where you paid for the meal.
Remarks: Here’s an anecdote. At 25-26 years of age, I consumed 2 pounds of pasta at a Chicago Italian restaurant… at night. This was a plate likely designed for a take-out box to span 3-4 nights. But I accepted the challenge imposed by no-one and consumed the portion in it’s entirety. Upon leaving the restaurant, I first experienced severe nausea. Then I suffered insomnia for three nights.
You’re in the process of reading a book you no longer enjoy.
Fallacy: You’re a respectable portion into the book, having invested time into the reading, and therefore should finish. You’re not a person who abandons initiated tasks.
Reality: The only tangible factor is whether you actually need to finish this book, for whatever personal or professional reason you may have. If so, in abandoning the book, you would possibly lose additional time in rereading the already covered portion later, if you return to it after significant time, or you would sacfifice whatever benefit in reading the particular book. That is a realistic concern. If, however, the book carries no tangible reward other than self-gratification of simply incrementing your finished book tally by sitting through a book you don’t understand nor enjoy, then you should abandon it, irrespective of whether you read 50 pages or 550. Any continuous pursuit is a waste of time.
Remarks: I abandoned many books as deep as 80 percent into the experience. Some of those I found linguistically too challenging to enjoy even at a surface level. Many I didn’t connect with. Others I simply couldn’t focus on at the time due to subject matter entirely incohesive with the then state of my mind. At the time of this writing, I abandoned the last two books I started.
You arrive at a gathering or some public event. Maybe you’re with friends or maybe solo. You don’t enjoy the setting, perhaps due to overwhelming noise, crowd, nature of conversations, lack of positive vibe, lack of inspiration, or an overall sensation of rapid energy dissipation.
Fallacy: You should bear the situation until the end or some reasonable time investment. If you’re with friends, you would lose team player points in suddenly abandoning. In any case, you don’t grow in social endurance by abandoning environments at a sign of discomfort. You already showed up, which is half the battle as they tend to say.
Reality: This situation is naturally geared more towards introverts. If you’re in a place that doesn’t severely drain you, regardless of your energy cycle, then feel free to stay and see what happens. But if you find yourself in a precarious situation, you can leave after three minutes even.
Remarks: I have done this more times than I can recall, as quickly as immediately upon arrival. If your friends fail to empathise, you should reconsider who you surround yourself with. With respect to personal growth in sustaining challenging environments, I’ll say the following. You usually have some intuition on how you intend to proceed. If you intend to be proactive, interact, learn something and take risks, then yes, consider staying, as this is how you grow. If, otherwise, you intend to activate your force field, conserve energy, and passively float around, then you might find better means to spend your time.
You’re enrolled in an individual course or an entire accredited program. You already completed a significant amount of coursework. Suddenly, you find yourself either losing interest or no longer clearly see the marginal returns in continuing.
Fallacy: You should continue with your learning, having already invested so much time, energy, and possibly money. Finishing this content is the responsible act, respected by society and family. You’re not a person who abandons major projects.
Reality: How much money or time you invested so far is irrelevant, unless you can expect some refund or recompensation. The relevant factors include the following:
What additional tangible value to you and you alone you can gain by finishing this content, and at what time, financial, and emotional cost?
What you can gain by abandoning the program immediately, and using the time, energy, and expenses saved towards other purposes, or perhaps by pursuing similar material independently, more effectively, and much less expensively?
Corrolary: As an alternate formulation of the problem, consider the following. Suppose you already possess the knowlege acquired from the incomplete program, in which you’re otherwise not enrolled. Would you sign up for a new program containing strictly the three aforementioned missing courses for the same time and cost per each, presuming it also offers a diploma of the same stature?
Remarks: Even if you’re one semester short of finishing a highly ranked Masters program, but don’t need the accrediation or the diploma (what marginal benefit does one provide?), what do you gain in paying for the extra semester, when you already possess most of the valuable knowledge? Education is not a binary process, but a product of non-linear and often diminishing returns over time, especially at a point when you’ve learned sufficiently to manage your time more effectively by independent means from then on.
Unless you’re in a heavily regulated career such as Law, Medicine, Architecture, Structural Engineering, etc, you have much flexibility in knowledge pursuit. Online material, books, and independent projects can arm you with more practical or theoretical exposure even than many expensive accrediated programs. I often meet people enrolled in such programs strictly for Visa purposes (not strictly, but effectively, since otherwise much cheaper and more effective means exist of acquiring the equivalent knowledge). If that is you, review your options. I cannot comment, having myself never been in this situation. If you’re not otherwise constrained by any tangible factors, analyze your marginal returns in continuing with the status quo. Previously invested time and social repercussions are factors of no relevance.
Purchases and possessions.
I can’t resist in building a connection to the possession minimalism I eagerly preach. I had eliminated an impressive amount of possessions throughout my life, and continue with the trend. It has become a hobby and a dogma. It clears the playing field of unnecessary distractions and facilitates the needed focus on the more important decisions and projects.
These possessions fall into several following categories, but I’ve largely applied the same present-future Markovian analysis in constructing elimination strategies.
Small, yet perpetually accumulating items: media, books, papers, artwork, school materials, notebooks, gifts, electronics, clothes, childhood toys, etc. These items had been acquired by a number of different means. Some had been purchased, others gifted, and some developed by own effort.
- Is the item adding significant marginal value to my life? (I emphasize marginal, since it is not a question of whether the item has independent value, but rather if it contributes on top of what you already possess?)
- If not, can it be sold AND would the effort justify the income?
- If yes, sell.
- If not, donate or eliminate.
In analyzing value, by removing the notion of sentimentality (which I respect in moderate quantities but find distracting on a large scale), the workflow becomes simple and effective. Notice that nowhere have I incorporated past variables such as how I acquired it or how much I paid for it.
Larger items: furniture, cars, real-estate. I mention these in the same bullet point, since at their core, sentimentality and cultural entanglement aside, I fail to see the distinction. Their imposing dimensions and symbolic implications easily lead into the sunk cost bias. I also don’t distinguish between owned, rented, or even business generating possessions as far as the analysis is concerned. The elimination workflow is mostly identical, although slightly differently formulated in light of the more complex value-analysis function.
- Does the possession add value that justifies the cost and energy in maintaining it? (It might not, for example, if an existing or alternatively acquired possession could satisfy your parameters at a much lower financial and/or emotional cost.)
- If not, can it be sold AND would the effort justify the income (or the elimination of further loss)?
- If yes, sell.
- If not, donate or eliminate by other means.
I emphasize that despite the complex analysis of owned or rented real estate or vehicles, there is nothing inherently sacred or mythical in their possession. You are free to analyze and eliminate objects present in your life to make space for more important projects or ambitions. Notice, again, that nowhere have I raised the question of how you arrived to own these possessions, since these are no longer relevant factors. What’s relevant is their present emotional and financial value in your life, and the forthcoming choices at your disposal.
Digital possessions. Digitizing everything is not entirely energy efficient, in any sense of the concept. Keeping track of endless accounts, correspondences, notifications, updates and information incurs mental (as well as the frequently neglected natural) energy, although often difficult to quantify. Regretfully I tend to postpone much digital reorganization in favor of the physical, for the immediately apparent spacial and temporal freedom the latter provides.
Digital content may not occupy physical space, but it demands energy to mentally compartmentalize and maintain. For this reason, I see no reason not to apply the same decision process to digital content. You can eliminate email accounts, social media accounts, obsolete web sites and multimedia, no longer accessed notes, and any other digital remnants no longer conductive to your aspirations. Despite cultural entanglement, there is no need to keep a digital trail of your entire life.
In spite of the aforementioned relatively simple cases of the sunk cost bias, it knows no limits. The bias is as applicable to your career choices and relationships, although wrapped in layers of stigma, scrutiny and culturally imposed expectations. I will not explore further domains, but highly recommend questioning and requestioning what is of real importance to your life as opposed to belonging to the realm of pseudo-importance or mysticism.
Interestingly, if like myself, you take on an overabundance of tasks and projects, you become faced with inevitably having to scale down to make room for the lesser few of greater importance. Some of us simply have to experiment much longer (decades even) with a wide array of activities to discover the few that lend themselves to sustained passion. See the rather old, but applicable Tim Ferris post The Top 5 Reasons to Be a Jack of All Trades for a similar perspective.
I’m not an optimal strategist in choosing projects/directions that I sustain over the long term. That is, the majority become extinct, but a minority survive natural selection. Genetic algorithms model this behavior quiet fittingly if we map the problem of identifying your passions to that of iteratively experimenting with a larger pool of activities, crossing over some to produce a new mixture, and eliminating the subset of the less valuable specimens in favor of the more promising, with a small element of randomness. As a consequence, however, there’s a conception among some of my peers that I’m always changing or not finishing what I start (possibly in accordance with my own perception.)
To bridge this with the main topic, each activity you pursue also lends itself to the sunk cost analysis. Irrespective of how much time and energy you dedicated to one over the past, it would behoove you to eliminate the activity when it notably distracts you from your current mission. And you might have to deal with the reality of being seen as that random, constantly changing person, but who cares, really?
The only remaining point I wish to add to this rather long post concerns the very analysis of the decisions at your disposal. Sometimes the parameters are many, and even after having eliminated superfluous past considerations and encoded the necessary information into the present, the picture seems somewhat fuzzy. I occasionally resort to a mathematical analysis, not only to maximize the possibility of being objective with the data, but because I prefer to solve problems on paper. Here are a few options of how I might approach such a problem.
Classical narrative. Sometimes the simple nature of writing automatically maps out your thoughts and even unravels the necessary threads in the subconscious, resulting in all you need to make an informed decision. I try to write daily, as short as a paragraph in a journal to clear the mind. This doesn’t always happen.
Decision matrix analysis. My method may deviate from the classical, but for the more subtle decisions, I often resort to assigning numerical utilities to the different choices and outcomes they may yield. If the decisions involve multiple parameters, I will assign a specific weight to each and possibly apply normalization if the parameter scales vary. The normalization formula to reduce any parameter to the [0,1] scale is
xnorm = (x - xmin) / (xmax - xmin).
In the very simplified problem domain of decisions (sell, keep) each involving parameters (financial, emotional), I may assign weights (0.4, 0.6) to the two parameters. The utility of the each decision becomes the linear combination of the two vectors:
Utility(sell) = 0.4(sellfinancial) + 0.6(sellemotional)
Of course, it’s up to you to assign the appropriate values and insure scale invariance.
If the decision can lead to more than one outcome, I map each decision to a probability distribution of outcomes, and the decision utility becomes the expected utility across all possible outcomes. The matrix can quickly grow in complexity, but here’s a simplified case with two outcomes:
Outcome1, 0.5 Outcome2, 0.5 Emo, 0.6 Fin, 0.4 Emo, 0.6 Fin, 0.4 Sell 9 4 5 2 Keep 2 8 2 4
We assign a probability to each of the two outcome (in this case uniform), and calculate the state utility as the expected utility across the outcomes:
Utilityexp(sell) = 0.5 * (0.6 * 9 + 0.4 * 4) + 0.5 * (0.6 * 5 + 0.4 * 2) = 5.4
Utilityexp(keep) = 0.5 * (0.6 * 2 + 0.4 * 8) + 0.5 * (0.6 * 2 + 0.4 * 4) = 3.6
So provided that the arbitrary values I chose for the different parameters conform to the same scale (1 - 10 in this case), the expected utility of selling outweighs the expected utility of keeping the hypothetical item.
Markov Decision Process. For a decision process of greater complexity involving combinations of multiple actions to reach certain outcomes, and with an element of randomness, you might prefer an Markov Decision Process (MDP). I will not elaborate further in respect for space considerations.
- Monte Carlo Method
- Markov chain
- The Top 5 Reasons to Be a Jack of All Trades
- Genetic algorithms
- Markov Decision Process
Questions, comments? Connect.