What does Doctor Watson think about when voting?

How Kahneman and Tversky's behavioral economics explains voting decisions, based on intuition, heuristics, and biases. Note from the Academic Bridge, March 2015.

It should be noted that this branch of economics was already known to specialists. However, it has gained wider dissemination and reached a broader audience. This is due to the publication of the book "Thinking, Fast and Slow" by Daniel Kahneman , winner of the 2002 Nobel Prize in Economics.

📄 Download the report (PDF)

 


Behavioral economics integrates insights from psychological research into economic analysis, especially human judgment and decision-making under uncertainty.

To be clear, it focuses on studying the specific decision-making process, and this work is carried out with certain particularities. One of these is that the theory is constructed and tested experimentally. That is, most of what is stated has a tested empirical basis. Debatable, with subtleties, nothing definitive; but tested.

Besides the novelty of the topic, there is another interesting reason, which is directly related to the Academic Bridge section of our website. It concerns the way in which economic knowledge and theory are constructed.

When we watch the news or read the newspaper, it is quite common to find phrases like the following: “an increase in the interest rate would favor an increase in savings,” or “if a depreciation of the exchange rate is expected, the public will turn to the purchase of dollars.”

The assumption that we act rationally

What underlies these assertions is the idea that agents—people—behave rationally. That is, faced with a given stimulus, which in our examples has been price movements, agents behave by giving the best possible response, based on the information they have. In other words, they maximize their benefit.

In this way, the rationality hypothesis solves a huge number of difficulties in theory-making. In very general terms, it assumes homogeneous behavior for all agents involved in the economy. The differences between the various decisions made lie in the particular constraints each agent faces, such as their budget or access to information. They can also simply lie in their preferences.

The problem with the theory, let's say “traditional”, is that according to the criteria of behavioral economics, we rarely behave rationally. Being rational implies having an enormous information processing capacity and behavior permanently oriented towards calculation.

This academic scenario is where the works of Kahneman and Tversky emerge.

Some would argue that the appropriate phrase should be “orthodox economics,” rather than “traditional economics.” However, many unorthodox schools of thought also implicitly or explicitly employ some variant of rationality that aligns with this approach.

And it's important to clarify: its use isn't necessarily good or bad. Assuming rational behavior allows us to draw relevant conclusions at low cost.

The problem would seem to lie in clarifying at what level of analysis the application of this assumption makes sense, regardless of whether it is considered orthodox or unorthodox, especially considering the practical consequences that could result from misuse.

Why talk about this in an election year?

Finally, another good reason to discuss some aspects of behavioral economics is that this is an election year . A huge amount of money and resources are being spent on predicting the winning candidate. The goal is to know who will be the favored one to win this round of the political game. Citizens vote, and in that ritual, they grant enormous power to someone distant and vague, someone to occupy a seat or a seat in Congress. And, above all, to make decisions.

In accordance with the above, throughout this article we will outline the general principles of the theory. We will use different interpretations of certain aspects of the electoral process as examples. Undoubtedly, the underlying question in our examples is: how do we decide our vote? Some might say: based on ideology, personal liking, self-interest…

All valid reasons. Notwithstanding the above, advances in behavioral economics show that the intuitive aspect—not the rational one—carries enormous weight. This occurs even in decisions such as exercising sovereignty through voting.

What's more, if the reader is patient and dares to finish reading this note, I would bet that a question will remain floating in the air: how do our representatives decide?

And then the psychologists appeared

The idea that man can be defined as a rational animal comes from afar. Although the concept has undergone multiple changes since Aristotle's definition to date, it should not be surprising that the variants of this approach have to fight for the podium.

The first problem is not having the intuition that perhaps we are not so rational, but proving it. And not only that, but also providing a better answer to the rationality hypothesis.

An answer that can also be integrated into the current body of knowledge. That convinces members of a scientific community to abandon some of the concepts with which they have worked with relative success for a long time.

That it passes the technical requirements of a moment, and so on and so forth. Undoubtedly, a lot of power relations are involved in this phenomenon, but so is evidence.

Without ignoring power relations, let's focus on the latter: there is evidence that shows that in certain circumstances a decision can be rational and evidence that a large part of our decisions are not.

Two systems of thought: intuition and reason

The answer provided by the behavioral economics approach to enable a systematic study of those behaviors perceived as anomalous by conventional economic theory is the splitting of the cognitive structure between two systems of thought.

The architecture proposed in the writings of D. Kahneman and A. Tversky recognizes the existence of two systems called system 1 and system 2. These can be associated with what is commonly called intuition and reason, respectively.

      1. To system 1 The characteristics of being are attributed to it: fast, parallel, automatic, effortless, associative, slow learning and emotional.

      1. To system 2 The characteristics of being are attributed to it: slow, serial, controlled, effortful, governed by rules, flexible and neutral.

    Deciding without calculating: the weight of intuition

    One of the theory's first findings is that behavior is not guided by calculations, but rather by the impressions and information available to the individual at the time of making a decision. A key property of intuitive thoughts is that they arise spontaneously.

    They occur without conscious searching or calculation, and without effort. For this reason, most assessments and decisions are made intuitively, and the rules that govern them are based primarily on visual analogies.

    Otherwise, the reasoning is deliberate and effortful. Therefore, we shouldn't be surprised by a description in which the agents are not accustomed to rigorous thinking. In such a description, they often settle for merely a plausible judgment.

    In other words, rational thought is possible but comes at a cost, given that the mental capacity required for it is limited. Under this same premise, it also makes sense that System 1 would operate by updating the state of the world with less wear and tear than would occur under the command of System 2.

    How difficult it is to get informed in order to vote

    And here a first example emerges regarding the electoral scenario . Using the tools provided by behavioral economics, it could be interpreted that the average voter would be more inclined to pay attention to looking at candidates' photos and more vulnerable to advertising than to learning about the platform or casting a vote according to their own preferences.

    Now, a word of caution: political economy theories that assume only a rational individual also cannot explain why a voter would be willing to make an informed decision. The explanation from this perspective is simple.

    Getting extensive information to make the best voting decision is an expensive activity, while the benefit of voting well is extremely marginal.

    Let's look at a concrete example of this. In the legislative elections held on October 27, 2013, approximately 30.635.464 people were eligible to vote, of whom 23.641.116 cast their ballots. That is, 77% of the electorate. The effective influence of a single vote would be 0,00000423%. A very small benefit from the perspective of our hypothetical rational voter.

    Sherlock Holmes and Doctor Watson

    Another popular science book that uses a very original and quite easy-to-remember analogy is Maria Konnikova's book, aptly titled: How to Think Like Sherlock Holmes?

    The framework proposed in the book is extremely effective. The author suggests we think of System 1 / intuition as Dr. Watson and System 2 / reason as Sherlock Holmes.

    Fantastic! Not only because both characters fit the description perfectly, but also because in the rooms of Baker Street, after a client presents a case and leaves, the characters compare notes and converse. Sherlock withdraws into long periods of silence to meditate on the various aspects of a case. Meanwhile, it is Watson who takes the lead in the story. The same seems to happen in our own minds: we are a Watson.

    How intuition works: heuristics and biases

    Let's see then what behavioral economics has to offer about the workings of intuition, which is a fairly generic way of understanding how we will vote on average.

    In order to avoid the cost of computing, system 1 reduces the processing of complex tasks to simpler operations, through an attribute substitution mechanism.

    It is said that a representation is mediated by a heuristic when the individual evaluates a specific attribute of an object in that representation, substituting it with some other property of that object that is easier for him to interpret.

    In this way, costs are avoided by providing a quick response to the situation or problem that arises. But these costs reappear in the form of serious and systematic errors, which in theory are called biases.

    Unlike rational action theory, these biases respond to a structure and are therefore predictable.

    Consider, for example, the widespread concept of "money illusion," which is identified when, in a context of inflation, agents perceive themselves as richer because they see their nominal income increase, without taking into account the increase in prices.

    The notion of heuristics could contribute to providing a foundation for the existence of the phenomenon, from an extremely novel and enriching level.

    The three mechanisms described by Kahneman and Tversky

    In the pioneering works of D. Kahneman and A. Tversky (1974), three mechanisms used in the formulation of judgments under uncertainty are described:

        1. La availability/accessibility heuristic, which is used when asking to estimate the frequency of a class or the plausibility of a particular development.

        1. La representativeness heuristic, which is usually used when asking to judge the probability that an object or event A belongs to class or process B.

        1. El adjustment from an anchor, which is usually used in numerical prediction, when a relevant reference value is available.

      The availability heuristic

      The first heuristic describes the ease with which some thoughts reach the mind before others, and consequently provides one of the keys to understanding how intuition works.

      Under this mechanism, system 1 reacts more easily when presented with attributes of an object that are familiar to it.

      In this way, when imputing probabilities, the heuristic operation replaces the feasibility of the occurrence of an event, weighting not only the expected frequencies, but:

          1. Due to the degree of association that the characteristics of that event have with the agent's prior knowledge.

          1. Because of its prominence.

          1. Or by the temporal distance between the observation and the occurrence of the event that the agent is evaluating.

        In other words, from this point of view, an average citizen will analyze the most obvious data and the data they find at hand, without delving into a more sophisticated search.

        So, if the availability heuristic works in this way, one cannot deny the wealth of new questions and intuitions that can be derived from this theory and which, of course, should be put to the test to be conclusive.

        Availability and electoral behavior

        Having made that warning, and following the same order of enumeration as before, let's see how some of these ideas could be translated into electoral behaviors:

        • If a candidate credibly displays certain personal characteristics, whether fictitious or not, that have the particularity of being shared by a part of the electorate, it is highly likely that the candidate will become a figure with greater presence as a feasible option within the that reference group.

        The same applies to candidates with a greater relative advertising impact. This is something that, given the amount of resources spent on campaigns, is widely known. Even so, a word of caution: a rational voter might be tempted to vote for the candidate with the greatest presence, not because they associate greater presence with any desirable characteristic, but because they may be calculating that this way they aren't wasting their vote on another candidate with no chance.

        This is a crucial characteristic, especially for undecided voters. The argument being made is that, when making decisions, the weight given to recent stimuli can have a decisive impact on the election being considered. In other words, a candidate's history, their successes and failures, carries much less weight than one might expect. That is to say, a large majority of candidates can effectively fade from memory simply by letting time pass.

        Alexander Todorov's candidate photos

        Alexander Todorov, from the Department of Psychology at Princeton University, is conducting research on how first impressions are formed.

        Generally speaking, some of their experiments use photos of candidates who actually competed in an election in one region of a country. These photos are shown to citizens living in another region. The goal is to ensure that the respondents are unfamiliar with the candidates and the election results from the first region.

        Todorov's work shows that the impressions made solely by photographs of unknown candidates on citizens of the second region significantly explain the election results in the first. In other words, opinions about photographs in one population are a good predictor of election results in another population.

        Representativeness and prototypes

        In recent work, the representativeness heuristic appears as part of a more general class called the prototype heuristic. The prototype of a set is characterized by the average values ​​of the main properties of its members.

        The high accessibility of the prototype information fulfills an important adaptive function, by allowing new stimuli to be efficiently categorized by comparing the particular characteristics of an element with those of the set; but at the cost of losing detailed information in that classification.

        To put it simply, the prototype heuristic provides theoretical support for the well-known saying: "Tell me who your friends are, and I'll tell you what you're like." But be careful, this only applies to first impressions.

        In the specific case of the representativeness heuristic , probabilities are evaluated by the degree to which A is representative of B. That is, by the degree to which A resembles B.

        Thus, when imputing probabilities, the representativeness heuristic replaces the likelihood of an event occurring. It no longer weighs expected frequencies, but rather an emotional response that allows the extension of attributes from one element to another.

        Representativeness in a list of candidates

        In terms of electoral examples, it's impossible not to consider the difference in name recognition the average citizen has between the top candidates on the list—or simply the first one—and the candidates further down the list. Let's consider this entirely hypothetical scenario. Imagine a list of candidates for parliament. The first candidate is qualified (whatever that means) for the position they are running for. However, the four candidates who follow are not.

        Let's say the first candidate manages to capture the electorate's attention as a viable option. Due to the representativeness heuristic, the characteristics of the good candidate are likely to be transferred to the image of the bad candidates. This results in a voting intention that, on average, can end up being contrary to the voter's own preferences. It would be, however, if the voter considered the overall characteristics of the group.

        The anchoring effect

        Finally, regarding the anchoring effect , it cannot be strictly speaking a heuristic. It is more accurately described as an adjustment procedure, moving from a starting point to a final estimate. The starting point can be a piece of information, data, or initial calculation. The final estimate is entirely dependent on the starting point or anchor.

        This means that different starting points generate different estimates that are biased towards the initial values.

        The intuitive example Kahneman uses in his book is the feeling of slowness a driver experiences when transitioning from highway driving to city driving. The anchoring effect means that when going from 130 km/h to 40 km/h, the perceived speed seems much lower than that 40 km/h.

        Contributions and limitations of behavioral economics

        In light of the postulation of this model of the cognitive system , the contribution made seems to be immense, while at the same time it is difficult to annex within the previous corpus.

        Behavioral economics differs from standard economics by using a more realistic, and therefore more complex, model to explain the behavior of agents in a given context. This entails sacrificing generality for greater realism.

        Indeed, one of the criticisms that Kahneman himself mentions in one of his works is the lack of systematization of his theories into models of formal and generalizable behavior.

        In this way, the field of research has been opened. On the one hand, explanations with different levels of rigor are appearing, by the most varied means, which are generating the reaction that “not everything can be explained by behavior.”

        It would seem logical that this should happen: partly the topic is new and has appeal for the reader, partly the alternative explanations based on statistical or econometric procedures are more demanding on the public's education.

        Is statistical thinking intuitive?

        In fact, one of the issues with which Kahneman and Tversky began their joint work was to resolve the question of whether or not statistical thinking is intuitive.

        Heuristics and biases (and the corresponding tests) seem to indicate that this is not the case. We need to awaken our System 2 to be able to reach conclusions consistent with those rules.

        However, the game is also open for academic production. There are recent works that show progress in the possibility of formalizing behaviors based on heuristics and biases, mixed with Bayesian analysis.

        In subsequent posts in this section, we will try to share some of these works. However, this year's focus will revolve around electoral issues.

        For now, the question of how we vote remains unclear. However, there's a sense that the voter is viewed more as someone susceptible to emotions than as someone who reads online platforms.

         

        To know a bit more…

        The experiments conducted with photographs of candidates, mentioned as examples, were based on two works by Alexander Todorov. These are detailed below and, of course, do not exhaust the topic:

            1. Ballew, Charles. & Todorov, Alexander (2007). «Predicting political elections from rapid and unreflective face judgments». PNAS November 13, 2007 vol. 104 no. 46 17948-17953.

            1. Todorov, Alexander; Mandisodza, Anesu; Goren, Amir & Hall, Crystal (2005). "Inferences of Competence from Faces Predict Election Outcomes." SCIENCE, June 2005, Vol 308 no. 10.

          Finally, the conceptual structure and order of this article are based on two works by Kahneman. Both are intended as a kind of essay summarizing the advances that have occurred in the field of behavioral economics.

              1. Kahneman, Daniel & Amos Tversky (1974) “Judgment Under Uncertainty: Heuristics and Biases.” Science, 185, 1124-1131.

              1. Kahneman, Daniel (2003). “Maps of bounded rationality: Psychology for behavioral economics.” AER, Dec 2003, 1449-1475.

            Open access, for use and citation

            The report is available freely and without charge, under a Creative Commons license (CC BY-NC-SA 4.0):

            ¿En qué piensa el Doctor Watson a la hora de votar? Informes del Observatorio UNR N.º 1 — Serie Puente Académico N.º 1, marzo de 2015. ISSN 2683-9067 / 2683-9091. Autor: Germán A. Tessmer.