The heuristics of addiction: Decision making in problem drinking and recovery
Robert B. Cutler
Abstract
Recent findings from judgment studies make it possible to re-conceptualize addiction in new terms: as a manifestation of normal judgment processes. This article focuses on the decision “Should I drink?” in non-intoxicated problem drinkers.
First, three characteristics of this decision are identified. They are: its unremitting reoccurrence in problem drinkers, the surprising complexity of that decision, and the high speed of its resolution. The result is that continued abstinence is very difficult because it requires making quick difficult decisions repeatedly over the relatively long time span that is required for recovery.
Next, five factors that cause normal human decision error are discussed in relation to problem drinking. They are: (1) Base Rate Neglect, which causes underestimation of the probability of a negative drinking outcome; (2) Duration Neglect, a cognitive illusion which causes overestimation of the positive effects of drinking; (3) the Affect Heuristic, which causes both overestimation of positives and underestimation of negatives (this effect abates with protracted abstinence); (4) Cognitive Load, which limits accessibility of corrective considerations at the decision moment, and (5) Narrow Framing, which causes both magnification of immediate positives and minimization of remote negatives. Each of these factors has implications for insightful decision making, possibly shedding light on why resolution of problem drinking has been characterized as resulting from “heightened reflective experience.”
3
Introduction
In 2002 the Nobel Prize was awarded to the psychologist Daniel Kahneman for his work in human decision making. Although this is a relatively new field it has become well accepted in recent years. Many of the practical applications of the findings have been in economics, and it was in the field of economics that Kahneman received the award. Judgment study concepts, such as heuristics and biases, offer a new way of viewing addictive behavior. Heuristics are the simple everyday “rules of thumb” that we use to make decisions and choose courses of action.
Judgment study findings may be applied to addiction because they deal with choice, and particularly with errors in choice. By identifying factors that cause errors it becomes possible to change error rates. Some decision theory concepts have begun to be applied in the addiction field, but here they have been primarily related to nicotine addiction and gambling, rather than to alcohol addiction 1 - 4. This may be because smoking and gambling are sometimes considered to be more under “rational” control whereas alcohol addiction is often regarded as an irrational obsession, due to its seemingly intractable and involuntary nature.
However, judgment heuristics distinguishes between intuitive "irrational" processes and the relatively slower “rational” processes (in the two-system model), but incorporates both processes in explaining how decisions are made. Indeed, the studies have shown that many of our decisions are not particularly rational. Modern decision theory is associated with an "attack on the rational agent model" of decision making. The concept of rationality has now been amended with the overarching concept of “bounded rationality” 5. The research has found that people do not usually think hard about the decisions that they make each day, but instead they choose a course of action that quickly comes to mind.
The focus here is restricted to non-intoxicated decision-making because alcohol causes an impairment of thought 6. Nearly all active problem drinkers spend parts of days and/or series of days in a row in a non-intoxicated state. Vaillant, in his 1995 book, "The Natural History of Alcoholism Revised" wrote that alcohol addiction is characterized by "unpredictable oscillations between use and abuse of alcohol … during any given month a majority of so-called alcoholics will be observed to be either abstinent or drinking asymptomatically… It is not an off-on phenomenon, and … remissions and relapses are common." 7 Sober periods of time are common, and most problem drinkers have tried unsuccessfully to quit 8. This makes it possible to consider decision making during those times.
Some decisions are not “free”; they are made under forced conditions, with serious consequences. For a problem drinker, the decision to drink (or not) is often, if not always, made under such coercive conditions, and the decision involves great risks and rewards. It is an emotion-laden decision made in the context of a tragedy. But it is a decision. The decision results from cognitive evaluations, and it is a conscious deliberate judgment about engaging in the varied and complicated sets of behaviors that are required in order to drink such as obtaining money, traveling to the beverage store or bar, conducting financial transactions, etc. This simple operational definition of drinking behavior as a decision or set of decisions allows us to apply the findings of judgment studies to alcohol use, misuse and recovery. The cognitive processes involved in the decision to drink (or not) are the subject of this article. The next section discusses three characteristics of the decision, and the final section describes five areas in which errors are made.
The Difficulty of the Decision
There are three decision characteristics that make the consistent choice to not drink a particularly difficult dilemma for a problem drinker. As described below, the decision is (1) complex, partly because outcomes are unpredictable, (2) rapid, because decisions are typically concluded very quickly, and (3) frequent, or continually reoccurring. In combination they put the non-intoxicated problem drinker in the unenviable position of repeatedly and hastily making formidable choices.
Complexity of the decision
A decision can be analyzed in terms of both the intuitive judgments and the deliberate thought processes associated with it. The intuitive elements are quick, and highly accessible. Kahneman likens the normal decision making process to a simple perceptual task (such as viewing a photograph) where the different elements of a situation are readily apparent 5. He points out that, when the decision task is more difficult, then a second process, a deliberative process, can modify the initial intuitive judgment. This is termed the two system theory. System one is the immediate intuitive judgment, and system two is a corrective “reasoning” process.
Just as decision theory gives a central role to the gains and losses that come to mind when a decision is to be made, the thought processes involved in the drinking decision have been divided into positive and negative factors 9 or expectancies 10. For the problem drinker, the rewards of drinking are the almost irresistible feelings of pleasure, well-being, and peace, the abatement of withdrawal symptoms, etc. The risks of drinking are a host of unpredictable consequences such as loss of social status, financial loss, and injury 11. Positives and negative aspects of the decision to drink will be discussed as shown in Table 1.
3
Table 1. The two sides of the decision
Should I Drink?Rewards or Gains / Certain Probability of feeling very good / (highly accessible)
AND consideration of
Risks or
Losses / Unknown Probability of feeling (unknown) bad / (relatively inaccessible)
3
The highly accessible content: Positive considerations
The rewards and risks of drinking differ in accessibility—some come to mind much more easily than others. In an active drinker the initial intuitive perception is of the reward. Heavy drinkers give precedence to the perceived positive benefits of drinking rather than to perceived negative effects 10 12 13. The active problem drinker certainly expects drinking to feel good, making these considerations highly accessible to the decision process.
The relatively inaccessible content: Negative considerations
Griffith Edwards referred to negative thoughts as "cognitions" that promote abstinence 14. Heavy drinkers are aware of the risks and negative consequences of their drinking, but these calculations are more uncertain and difficult 15 16. Orford 12 noted "the drawbacks of drinking are rarely clearly and separately delineated" in ongoing heavy drinkers. Drawbacks consist of sets of possibilities with varying probabilities and varying intensities. These are particularly difficult to predict because they would depend partly on that unknown quantity of alcohol that will be ultimately consumed. There are other associated possibilities that are also unknown, such as the actions of other people and of one's self when intoxicated. Heavy drinkers actually expect fewer negatives than do light drinkers 17 and they tend not to reflect on the negatives 18 19.
There have been a number of judgment studies involving uncertainty, and a primary finding is that unpredictability is poorly represented in quick intuitive judgments. Choices that involve uncertainty are more likely to be ignored. For these reasons negative drinking outcomes are relatively inaccessible. The decision literature also indicates that judgments of relatively inaccessible properties are much slower and more susceptible to interference by other thoughts (discussed below in the Cognitive Load section).
But overall, both elements of drinking, risks and rewards, are considered 12 during the decision moment. Norm theory shows that, to the extent that these are two competitive interpretations of reality, they suppress each other 20. The decision is made when a dominant role is assigned to one side. The general risk/reward concept has been termed the decision balance process in drinkers 21 22. A Decision Balance Scale has been developed for smokers 3. See Orford for a more general discussion on applying a theory of gain and loss to problem drinking 12.
Further complexity
As shown in Table 1, a number of investigators have further subdivided both reward and risk into two parts, a likelihood and a specific value (or "strength", “intensity”, or "valence") 3 23 24. Some have even proposed adding a third element, the multiplicative product of likelihood and value 25 - 27.
The concept of the problem as a decision balance is probably unrealistic 12 because it appears to be too simplistic. It might prove to be more useful to consider it more flexibly, for example, a process with two decision balances, one related to the risks and rewards of drinking, another to the risks and rewards of not drinking. In addition to considering the decision in terms of its “prototype attributes” as I have been doing here, the decision maker can consider it in “extensional attribute” terms, e.g., when an active drinker is contemplating extending the number of drinking days by one (with the expected minimal costs of that decision versus the relatively minimal or non existent benefits of a first day of abstinence), or when an abstinent individual is contemplating extending the number of non-drinking days by one (with the attendant potential costs and benefits). Kahneman argues that the elements of extensional attributes are low in accessibility. 5
Additionally, as noted earlier, there are many different types of interacting and competing positives and negatives in drinking. It may be less productive to think in terms of a dichotomous balance – with the focus on two opposing forces, and more productive to think in terms of the accessibility of corrective thoughts. Kahneman writes “people rely on a limited number of heuristic principles, which reduce the complex tasks of accessing probabilities and predicting values to simpler judgmental operations. In general, these heuristics are quite useful, but sometimes they lead to severe and systematic errors” (emphasis added) 28, and “people who are confronted with a difficult question sometimes answer an easier one instead” and “The essence … is that respondents offer a reasonable answer to a question that they have not been asked.” 5 For example, instead of answering the question “Should I drink?” the decision maker may instead answer “Will it be possible to obtain alcohol?” or “If I drink will it be possible to avoid (some specific negative consequence)?”
Inability to categorize factors leading to resolution of drinking
There have been a number of attempts to list all the different types of factors leading to resolution of drinking, organize them into meaningful categories, and find relationships between them and drinking outcome. Investigators have proven no better than problem drinkers (the actual decision-makers who are of interest here) in conceptualizing these factors in a coherent way. For example, one author listed 19 categories with “Intrapsychic factors” and “Specific frightening/humiliating experience” at the top of the list 9, while another listed 14 different categories with “Health-related” and “Negative personal effects” at the top 29. Another classification scheme began by organizing negative consequences by time, e.g., proximal (immediate), distal (longer term) with continued drinking (or relapse to drinking) as a third category 30. The range of categorizations in different studies has been reviewed by Blomqvist 31 32 . One author referred to the categorization process as "arbitrary" 29.
The difficulty is probably because there are so many different types of negatives and of positives and so many different individual situations. And just as drinking is different for the same person at different times (for example, morning versus evening, and Saturday versus Monday), the decision to drink or not requires a different judgment each time.
Change in decision over drinking career
Over the months and years of addiction, serious problems often grow more serious. Negative consequences of drinking may become increasingly obvious 33, and accessible to the decision making process. Conversely, one “puzzle” of recovery is that relapse is still likely after weeks and months of sobriety 34 in spite of the benefits of abstinence. Benefits include greatly improved mood 7 and greatly reduced stress 9. (Ironically, as Kahneman points out 5, being in a good mood makes judgment error more likely.) However, with an increase in length of time abstinence, the negative consequences of drinking may become less accessible to the decision making process. Many previous negative consequences may no longer appear to be relevant. In both phases of the drinking career the decision changes over time and therefore remains challenging.
The Difficulty of the Decision: Duration and Frequency
Two other general characteristics of the decision making process are particularly relevant. First, most decisions are resolved nearly instantaneously. Typically these are “decision moments” that last only one or two seconds. As noted earlier, Kahneman compares the “decision moment” to a perceptual task, where the different aspects of a situation are readily apparent and quickly resolvable. Important elements might not be considered when making quick choices.